Research ArticleComputer Vision-Based Portable System forNitroaromatics Discrimination

Nuria Loacutepez-Ruiz1 Miguel M Erenas2 Ignacio de Orbe-Payaacute2 Luis F Capitaacuten-Vallvey2

Alberto J Palma1 and Antonio Martiacutenez-Olmos1

1CITIC-UGR Department of Electronic and Computer Technology ECsens University of Granada 18071 Granada Spain2Analytical Chemical Department ECsens University of Granada 18071 Granada Spain

Correspondence should be addressed to Alberto J Palma ajpalmaugres

Received 12 February 2016 Revised 21 April 2016 Accepted 22 May 2016

Academic Editor Giorgio Pennazza

Copyright copy 2016 Nuria Lopez-Ruiz et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

A computer vision-based portable measurement system is presented in this report The system is based on a compact reader unitcomposed of a microcamera and a Raspberry Pi board as control unitThis reader can acquire and process images of a sensor arrayformed by four nonselective sensing chemistries Processing these array images it is possible to identify and quantify eight differentnitroaromatic compounds (both explosives and related compounds) by using chromatic coordinates of a color spaceThe system isalso capable of sending the obtained information after the processing by aWiFi link to a smartphone in order to present the analysisresult to the final userThe identification andquantification algorithmprogrammed in theRaspberry board is easy andquick enoughto allow real time analysis Nitroaromatic compounds analyzed in the range of mgL were picric acid 24-dinitrotoluene (24-DNT) 13-dinitrobenzene (13-DNB) 35-dinitrobenzonitrile (35-DNBN) 2-chloro-35-dinitrobenzotrifluoride (2-C-35-DNBF)135-trinitrobenzene (TNB) 246-trinitrotoluene (TNT) and tetryl (TT)

1 Introduction

The term explosive encompasses a wide variety of chemicalsand mixtures of which nitroaromatics are one of the mostcommon groups Some of the most known are TNT(246-trinitrotoluene) RDX (135-trinitro-135-triazacy-clohexane) PETN (pentaerythritol tetranitrate) HMX(1357-tetranitro-1357-tetraazacyclooctane) EGDN (ethy-lene glycol dinitrate) and TATP (triacetone triperoxide)Furthermore molecules which are not themselves explosivescompounds but are used for their synthesis and present withsome of the properly named explosives should also be targetanalytes (24-dinitrotoluene DNT or several phthalateseg) [1 2] Regarding explosives in solution the wastedischarges from manufacturing of explosives and fabricationof finished munitions as well as the postwar destruction ofout-of-date bombs rockets and ammunition could producecontamination of the soil groundwater and surroundinglakes Moreover dissolved explosive material can also be

detected beyond the boundaries of these installations intonearby farmland causing increased public concern [3ndash7]

Therefore the reliable detection of explosives and relatedcompounds is a sensing challenge for different fields such asmilitary operations public security criminal investigationand pollution control [8ndash10] In these fields different cir-cumstances must be faced such as low levels of explosivesin most practical scenarios multiple interferences fromcommon household and personal care products variableenvironmental conditions and the need for low-cost on-sitereal time and highly portable detection methods

Various high resolution analytical techniques and strate-gies including ionmobility spectrometry [11 12] chromatog-raphy [13 14] mass spectrometry [15 16] nuclear quadrupoleresonance [17] energy dispersive X-ray diffraction [18]and Surface Enhanced Raman Spectroscopy [19 20] havebeen reported for the sensitive and selective detection ofthese explosives Nevertheless these methods suffer fromdrawbacks such as cumbersome pretreatment of samples

Hindawi Publishing CorporationJournal of SensorsVolume 2016 Article ID 7087013 10 pageshttpdxdoiorg10115520167087013

2 Journal of Sensors

interference from other compounds trained staff or bulkyinstrumentation [14] An alternative approach overcomingsome of these drawbacks relies on the biomolecular recog-nition capability of the biological element (eg antibody)for a target molecule (eg antigen) as fiber optic probes[21] However analyzing complex mixtures of compoundswith this high selective approach requires the design andsynthesis of receptors for each component in the mixturewhich is often an extremely time-consuming and expensivetask

On the other hand resembling mammalian senses suchas olfaction and taste instead of identifying an analyte byits strong affinity for one particular receptor recognition canbe achieved by the composite response of the entire array ofsemiselective receptors The result is a characteristic patternor fingerprint for each analyte An important advantage ofthis approach for molecule identification is that the systemis limited by the number of different patterns possible foran array of receptors rather than the number of differentreceptors Thus the different receptors need not be highlyspecific for any particular analyte Moreover this arrayapproach allows the discrimination of analytes or analytemixtures that have not been exhaustively characterized Col-orimetric sensor arrays have been extensively developed forthis purpose because of the potential for high sensitivity goodselectivity rapidity of analysis portability of instrumentationand overall cost-effectiveness [2]

However sensor array approaches produce a largeamount of data which is usually not interpretable by visualinspection of the dataset or by using basic calibration proce-dures like a simple linear regressionTherefore chemometricmethods are usually applied to reduce the dimensionalityof the data for simplifying visual interpretation Amongthem principal component analysis (PCA) hierarchicalcluster analysis (HCA) linear discriminant analysis andartificial neural networks (ANN) are the most commontechniques [22] They can be somewhat time-consumingcomputational methods which could prevent a true real timemeasurement

Image processing is a common technique for measuringcolorimetric response of optical sensors [1] including sensorsfor explosive determination [2 6 23] In this work thisstrategy has been adopted Low-cost on-site and real timemeasurements are some of the objectives of the developedprototype in line with current portable analytical instru-mentation [1] Thus we have developed a fully portableand wireless measurement system for the identification anddetermination of explosives and related compounds whenthey are mixed in a solution It is based on a colorimetricsensor array an electronic reader unit with a microcamerafor the acquisition of an image of the sensor array and asoftware application running in a smartphone capable ofshowing the results in an easy and direct way A set of eightnitroaromatic compounds have been analyzed with a sensorarray using four different sensing chemistries This has beenpossible by programming a simple algorithm which allowsthe simultaneous identification and quantification of thesecompounds in a mixture of dissolved analytes The mainnovelty of our design compared to others [2 4ndash6 23] relies on

being a low-cost compact and wireless system which allowsgetting real time and in situ information about a large numberof nitroaromatic compounds even mixed in a solution

2 Experimental

21 Reagents All reagents were of analytical-reagent-gradeunless stated otherwise Reagents used for the preparation ofthe sensing membranes were creatinine potassiumhydroxide aniline and NN-diethylaniline and ethanol allpurchased from Sigma (Sigma-Aldrich Quımica SA Spain)Nitroaromatic analytes used in this study were picric acid24-dinitrotoluene (24-DNT) 13-dinitrobenzene (13-DNB)35-dinitrobenzonitrile (35-DNBN) and 2-chloro-35-dini-trobenzotrifluoride (2-C-35-DNBF) purchased from Sigma-Aldrich and standards 1000mgL of 135-trinitrobenzene(TNB) 246-trinitrotoluene (TNT) and tetryl (TT) boughtfrom Chem Service (West Chester Pennsylvania USA)Solvents used to dilute the analytes in order to preparestandard solutions were methanol and acetonitrile purchasedfrom Sigma-Aldrich Aqueous solutions were made usingreverse-osmosis type quality water (Milli-RO 12 plus Milli-Qstation from Millipore conductivity 182MΩsdotcm) Filterpaper (ref 1250 basis weight 87 gm2 thickness 0180 120583mretention 10ndash15 120583m) from Filter-Lab Barcelona Spaindouble-sided adhesive tape from Miarco (Spain) and whitesheets from Schwan-Stabilo (Germany) were used to preparesensing membranes

22 Instruments and Software Theelectrical characterizationof the system was carried out using the following labora-tory instrumentation a mixed signal oscilloscope MSO4101(Tektronix Oregon USA) 812-bit Digital Multimeter 3158A(Agilent Technologies California USA) 15MHz waveformgenerator 33120A (Agilent Technologies) a DC power sup-ply E3630A (Agilent Technologies) and balance DV215CD(Ohaus Co New Jersey USA) For testing and calibra-tion purposes the software Matlab (MathWorks Inc Mas-sachusetts USA) andAdobe Photoshop (Adobe Systems IncCalifornia USA) were used as well

23 System Description

231 Sensing Array Preparation The disposable sensor arraywas prepared by laser-cutting technique because it is a cost-efficient simple and reproducible process using standardlaboratory filter paper (Filter-Lab ref 1250) The pattern wasdesigned using Illustrator software (Adobe Systems) and thenexported as an FS file to the controller software of a desktoplaser engraver Rayjet with a CO

2laser source (Trotec Laser

GmbH Wels Austria) For the fabrication a piece of paperwas affixed using double-sided adhesive tape to awhite plasticsheet The removal of weeding was performed manuallyThe design consists of 4 circles of 5mm diameter arrangedon 2 columns and 2 rows with dimensions indicated inFigure 2The size of the array and spots and their distributionwere designed taking into account the field of view of themicrocamera and the number of spots in each picture The

Journal of Sensors 3

Microcamera

LEDSensor array

LEDRaspberry Pi

Battery

WiFimodule

Android device

Portable reading unit

Figure 1 Block diagram of the instrument showing both the portable reader unit and the Android mobile device with the programmedapplication to present the results

ROI

Sensor array

ROR

5mm

25mm

3mm

25mm

3mm

25mm

18mm

18

mm

Figure 2 Dimensions of the sensor array and selection of the ROIsand ROR

device was prepared by drop-casting in each sensing area theneeded reagents under ambient atmospheric conditions

232 Description of the Measurement System The developedmeasurement electronic system is aimed at registering andquantifying the color variation of the sensing membraneswhen a volume of a solution containing some of the nitroaro-matic targets described above is added over them Theprototype is composed of a reader and processing hardwareunitwhich sends the information to amobile device (such as asmartphone or tablet) with a customizedAndroid applicationto show the result of the analysis carried out to the final userThe information gathering is based on the color variationsof a colorimetric chemical sensor array when exposed tothe analytes The capture and processing of the sensor arrayimage are respectively carried out by a microcamera and amicroprocessor included in the reader unit and results aresent to the mobile device of a remote user via a WiFi linkAlthough other systems have been already reported followingthis approach [24] our system presents the characteristic ofbeing fully portable and remotely operable

Sensing Module As it has been already explained the instru-ment is based on the acquisition and processing of the imageof the sensor array in order to quantify the color variationsand to relate them to the presence of some nitroaromaticcompound Figure 1 shows a diagram of the prototype Theprocessing electronics is enclosed in a dark box of dimensions13 times 13 times 5 cm3 with a slot for the insertion of the sensor array

As it can be seen the core of the instrument is acommercial Raspberry Pi platform the Raspberry Pi 2modelB (Raspberry Pi Foundation United Kingdom) This is alow-cost device that includes a 900MHz quad-core ARMCortex-A7 CPU and 1GB of RAMThese specifications makethis platform a powerful tool for fast image processingIn addition the Raspberry Pi 2 model B counts with 4-port USB for connection of external modules a camerainterface (CSI) and a micro-SD card slot which allowsexpanding the memory capacity in order to store the takenimages Besides the Raspberry Pi board provides 40 generalpurpose inputoutput (GPIO) ports allowing us to includesome external elements such as the light source used forthe system The microcamera used is the model RaspberryPi camera (Raspberry Pi Foundation United Kingdom)which is compatible with the Raspberry Pi platform It is ahigh resolution camera up to 5 megapixels based on theimage sensor OmniVision 5647 (OmniVision TechnologiesCalifornia USA) This camera allows acquiring both videoand still photographs It is connected to the Raspberry corethrough the available CSI

To obtain uniform and stable illumination for the acqui-sition of the image two diffuse white light emitting diodes(LEDs) model ASMT-MWH (Avago Technologies Singa-pore) are disposed in the inside of the box The operationof the LEDs is controlled by the Raspberry Pi core throughtwo digital outputs that switch them on and offThe intensityof the emitted light is manually adjusted by means of apotentiometer so that overexposition and subexposition areavoided during the acquisition procedure

An external module for setting up a WiFi network hasbeen added to the Raspberry Pi board This is the modelMiniature WiFi (Adafruit Industries New York USA) thatis fully compatible with the Raspberry Pi platform to which

4 Journal of Sensors

is connected through a USB port It implements wirelessstandards IEEE 80211n (draft) IEEE 80211g and IEEE80211b The whole system is battery-powered A large-sizedrechargeable battery pack is included in the design (AdafruitIndustries New York USA) Inside is a massive 10000mAhlithium ion battery that provides 5 VDC and 2 A via a USB-Aport The total weight of the reading unit is 150 g

Android Application An application based on Android oper-ative system has been developed in order to allow theremote control of the instrument through smart devicessuch as smartphones or tablets In order to implement theapplication Eclipse v2230 was chosen as the integrateddevelopment environment (IDE) since the plugin needed tointegrate Android is more developed than for other IDEsand also a phone emulator can be used for debuggingpurposes The application establishes a WiFi-based wirelesscommunication with the Raspberry Pi board through whichthe user requires a measurement and receives the resultingdata presenting them to the user through a simple andintuitive interface

233 Measurement Procedure The reagents used for thecolorimetric determination of the nitroaromatic compoundsare creatinine potassium hydroxide aniline and NN-diethylaniline Creatinine is dissolved in NaOH 01M with acreatinine concentration of 100mmolL KOH is dissolved inethanol and its concentration is 100 gL finally aniline andNN-diethylaniline are used without dilution The standardscontaining the analytes picric acid 24-dinitrotoluene 13-dinitrobenzene 35-dinitrobenzonitrile and 2-chloro-35-dinitrobenzotrifluoride are prepared by dissolving them inacetonitrile methanol (1 1) mixture In the case of 135-trinitrobenzene 246-trinitrotoluene and tetryl a 1000mgLstandard solution in acetonitrile methanol (1 1) is usedThe dilutions used in this study are obtained from thesedissolutions

The measurement procedure consists in the addition of5 120583L of sample to each of the paper spots present in theplastic cardboard and later 5 120583L of one of the reagent solutionsin each spot Each reagent solution must be added in apredetermined position in the card After the addition 15minutes is waited in order to let the reactions occur andto capture a picture of the complete colorimetric signalAfterwards the card is introduced in the reader unit and animage is taken from the cardboard

Safety Note Some of the nitroaromatic compounds analyzedin this study show explosive character For this reason thehandling of the samplesmust be carried out with precautionspaying extra attention to the amount of the manipulatedcompounds temperature variations and possible impacts

3 Results and Discussion

31 Nitroaromatics Recognition The recognition of eightnitroaromatics under study was performed using a whiteplastic board containing an array of paper circles fabricatedby craft-cutting technique Three different reactions were

used for detection and determination of these compoundsThe reaction of aromatic polynitrocompounds with car-banions results in the formation of Janovsky 120590-complexes[25] The red-orange color and the intensity depend on thenitroaromatic substance that is present and its concentrationThe reagent used is creatinine in basic medium (Jaffe reac-tion) [26 27] A second reaction takes place using potas-sium hydroxide in ethanolic solution that results in coloredMeisenheimer type complexes in some cases and coloredanions in others [28 29] Another reaction responsible for theformation of colored products with nitroaromatics involvesthe use of basic reagents such as aromatic amines based on acharge transfer mechanism in which the aromatic amine actsas a 120587 donor whereas the nitroaromatics serve as 120587 acceptor[28] We used this last reaction with two different reagentsaniline and NN-diethylaniline increasing in this way thediscrimination power of the array

32 Image Processing The core of the reader unit that is theRaspberry Pi board is programmed to carry out processingof the acquired images in order to obtain quantification of thecolor of eachmembraneThis image processing consists of analgorithm that provides only one parameter related directlyto the color change registered for each reaction The firststep of this algorithm consists in the selection of the regionsof interest (ROIs) in the sensor array These are the centralzone of each membrane discarding the edges where somenonuniformities of the membrane can be found In additiona region in the blank space of the board is also selected asa region of reference (ROR) as it is depicted in Figure 2This fifth region is used to normalize the color quantificationof the previous four regions corresponding to the sensitivemembranes in order to correct color measurement driftscaused by intensity changes in the biasing of the whiteLEDs or small nonuniformities in the illumination due tomisalignment of the membrane board and the white LEDs

After these ROIs are selected the algorithm extractsthe value of the red green and blue (R G and B) colorcoordinates of each region as the most frequent values (thestatistical mode) of R G and B of the pixels inside eachROI These values define the three chromaticities that allowunequivocally specifying a color in the RGB color spaceFrom these values normalized coordinates R1015840 G1015840 and B1015840for the ROIs corresponding to the sensitive membranes areobtained as the ratio between the RGB coordinates of thisregion and the RGB values of the fifth region correspondingto the blank space betweenmembranes in the board which isconsidered as thewhite reference Although these normalizedvalues of the RGB coordinates allow quantifying the color ofthemembranes these data are reduced to only one to simplifythe information on the color variations With this aim theEuclidean distance of the color of each membrane before andafter the reaction with the explosive sample is consideredThis distance is defined as the difference between the initialcolor coordinate and final color coordinate of the analyzedsamples [30]

119863 = radic(R1015840 minus R10158400)2

+ (G1015840 minus G10158400)2

+ (B1015840 minus B10158400)2

(1)

Journal of Sensors 5

20mgL10mgL5mgL1mgL

(a)

20mgL10mgL5mgL1mgL

(b)

Figure 3 Color variations of potassium hydroxide with 2-chloro-35-dinitrobenzotrifluoride (a) and creatinine with 35-dinitrobenzonitrile(b) for different analyte concentrations from 1 to 20mgL

where 11988310158400is the normalized color coordinate (red green

or blue) before the reaction with the sample and 1198831015840 is thecorresponding normalized coordinate after the reaction

By means of the colorimetric distance 119863 only oneparameter is used to measure directly the color variation thatthe membrane suffers when it is reacted in the presence of anitroaromatic compound

33 Calibration Curves The calibration of the measurementsystem was performed using separated solutions of theanalytes picric acid 24-dinitrotoluene 13-dinitrobenzene35-dinitrobenzonitrile 2-chloro-35-dinitrobenzotrifluor-ide ranged from 1 to 20mgL and 135-trinitrobenzene246-trinitrotoluene and tetryl ranged from 1 to 100mgLwith four replicas per solution and for each one of thefour mentioned reagents above This procedure results in38 calibration curves with four replicas for each of themErrors have been calculated as the standard deviationof experimental data The calibration was carried out asdescribed in the measurement procedure section obtaininga color change on the paper when the reagents are added InFigure 3 some examples of color variation of two reagentscan be observed when they are reacted with differentconcentrations of analytes

The Euclidean color distances have been extracted for allthe analytes in concentration ranging from 1 to 100mgLThe results show that this parameter presents a tendencythat can be fitted to a simple calibration curve (linearexponential or logarithmic) for two reagents at least Figure 4

shows some examples of the fitting curves obtained for somereactions between some of the reactants and nitroaromaticcompounds

In Table 1 the different fitting equations found in the cali-bration procedure are presented where119910 is the concentrationof the nitroaromatic compound and 119863 is the calculatedEuclidean colorimetric distance

34 Analyte Determination Algorithm Principal componentanalysis (PCA) is a standard technique for samples separationandor classification in a solution with a mixture of analytes[31 32] In this work we propose developing an even simpleralgorithm able to calculate the presence and concentration ofcertain nitroaromatic compounds some of them explosivesthus making the use of computationally costly multivariabletechniques unnecessary As it can be seen in Table 1 simpletwo-parameter linear exponential or logarithmic fittingequations can be used to predict the concentration of theanalytes Nevertheless the reactants are nonselective andcolor variations are produced when they react with differentanalytes Therefore it is necessary to select the proper fittingfunction when a color variation is produced in some of thesensing membranes This is accomplished in two steps asshown in Figure 5

Firstly by comparing the measured Euclidean colorimet-ric distances 119863

119894for each reaction (four values for the four

membranes in the sensor array) to the possible variationrange of this parameter obtained in the calibration curve ofthe corresponding reaction that is if a value of119863 falls in the

6 Journal of Sensors

Table 1 Fitting equations for nitroaromatic compounds

Nitroaromatic compound Aniline NN-Diethylaniline Creatinine Potassium hydroxide

Picric acid mdash mdash 119910 = 32629119863 minus 30636

1198772= 0998

119910 = 38665119863 minus 21012

1198772= 0999

24-DNT 119910 = 30098119863 + 053931

1198772= 0997

119910 = 80684 ln119863+ 351551198772= 0998

mdash 119910 = 01521411989038807119863

1198772= 0981

13-DNB 119910 = 08704811989037975119863

1198772= 0999

mdash mdash 119910 = 02654311989011940119863

1198772= 0974

35-DNBN 119910 = 91612119863 + 046375

1198772= 0993

119910 = 37418119863 minus 12879

1198772= 0985

119910 = 07228111989032627119863

1198772= 0999

119910 = 06602511989019234119863

1198772= 0969

2-Cl-35-DNT 119910 = 97915119863 + 071300

1198772= 0997

mdash mdash 119910 = 34155119863 minus 33118

1198772= 0983

TNB mdash mdash 119910 = 43885119863 minus 19694

1198772= 0997

119910 = 34040119863 minus 16816

1198772= 0983

TNT mdash mdash 119910 = 84385 ln119863+ 259951198772= 0988

119910 = 53987119863 minus 99810

1198772= 0983

TT mdash mdash 119910 = 52070119863 minus 30026

1198772= 0986

119910 = 3147211989041415119863

1198772= 0966

000

050

100

150

0 5 10 15 2024-DNT (mgL)

Eucli

dean

dist

ance

D

(a)

000

005

010

015

020

0 20 40 60 80 100TNT (mgL)

Eucli

dean

dist

ance

D

(b)

Figure 4 Relationship between concentration of two analytes and measured Euclidean colorimetric distance for potassium hydroxide (a)and creatinine (b)

variation range obtained for a given reaction this reactionis considered as a possible solution On the contrary thisreaction is discarded as well as the presence of this analyteAfter this process several reactions may still be solutionsof the problem case For the final selection of the properreaction it is taken into account that according to Table 1each analyte produces at least two color variations in thesensor array that is there are at least two reactants thatchange their color when they are immersed in the analyte

sample This fact allows using at least two fitting equationsfor the determination of each analyte If the result of thesepredictions for the given values of 119863 is the same theobtained value of 119910 is taken as the correct prediction of thecorresponding analyte We have checked that this procedureresults in univocal solution in the analyzed framework

This separation algorithm has been implemented in thecore of the Raspberry Pi board and it has been tested topredict the presence and concentration of the analytes as

Journal of Sensors 7

Analyte selection

Analyte discarded

No

Evaluation of the fitting equations

Yes

Same solution

Analyte discarded

No

Analyte identified

Yes

Analyte determined

Measurement of Di (4 values)

with variation ranges

Comparison of Di

corresponding range

Every Di inside

Figure 5 Flux diagram of the nitroaromatic identification and determination algorithm

it is detailed below Good results are achieved proving thatthe portable instrument presented here is useful to detect thepresence of a variety of nitroaromatic compounds

35 Validation and Specifications The resolution and sensi-tivity of the system have been theoretically calculated fromthe expressions of Table 1 The sensitivity 119878 is evaluated asthe derivative of the fitting function for the nitroaromaticcompound 119910

119878119910=120597119891 (119863)

120597119863 (2)

where 119891(119863) represents any of the fitting functions in Table 1The resolution can be evaluated from the sensitivity as the

product of this parameter by the error in the determinationof the variable119863 [33]

Δ119910 = 119878119910sdot Δ119863 (3)

From (1) the resolution of the Euclidean distance119863 is

Δ119863 =120597119863

120597R1015840ΔR1015840 + 120597119863120597R10158400

ΔR10158400+120597119863

120597G1015840ΔG1015840 + 120597119863120597G10158400

ΔG10158400

+120597119863

120597B1015840ΔB1015840 + 120597119863120597B10158400

ΔB10158400

(4)

And the resolution of the normalized color coordinates iscalculated as

1198831015840=119883

119883ref997904rArr Δ119883

1015840=1

119883ref(1 +119883

119883ref)Δ119883 (5)

where 119883 = R G B and Δ119883 = Δ119883ref is the resolution inthe measurement of the color coordinate (R G or B) This isgiven by the resolution of the microcamera which generatesimages with 8 bits per channel Therefore Δ119883 is taken as 128= 1256

Taking into account that for the calculus of the analytesconcentration at least two functions can be used the resolu-tion in its determination is considered as the worst case ofthese two possibilities

The limit of detection (LOD) is calculated using the stan-dard criteria LOD = 119910

119887+ 3119904119887 where 119910

119887is the average blank

signal and 119904119887is the standard deviation of the blank which is

determined using ten replicas [34] The values of sensitivityresolution and LOD achieved for the determination of thenitroaromatic compounds are listed in Table 2 The units ofboth the resolution and sensitivity are the same since thedistance119863 used in (1)ndash(3) is a dimensionless parameter

As in every electronic system some noise is presentthat can affect the measurement In this case two main

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

2 Journal of Sensors

interference from other compounds trained staff or bulkyinstrumentation [14] An alternative approach overcomingsome of these drawbacks relies on the biomolecular recog-nition capability of the biological element (eg antibody)for a target molecule (eg antigen) as fiber optic probes[21] However analyzing complex mixtures of compoundswith this high selective approach requires the design andsynthesis of receptors for each component in the mixturewhich is often an extremely time-consuming and expensivetask

On the other hand resembling mammalian senses suchas olfaction and taste instead of identifying an analyte byits strong affinity for one particular receptor recognition canbe achieved by the composite response of the entire array ofsemiselective receptors The result is a characteristic patternor fingerprint for each analyte An important advantage ofthis approach for molecule identification is that the systemis limited by the number of different patterns possible foran array of receptors rather than the number of differentreceptors Thus the different receptors need not be highlyspecific for any particular analyte Moreover this arrayapproach allows the discrimination of analytes or analytemixtures that have not been exhaustively characterized Col-orimetric sensor arrays have been extensively developed forthis purpose because of the potential for high sensitivity goodselectivity rapidity of analysis portability of instrumentationand overall cost-effectiveness [2]

However sensor array approaches produce a largeamount of data which is usually not interpretable by visualinspection of the dataset or by using basic calibration proce-dures like a simple linear regressionTherefore chemometricmethods are usually applied to reduce the dimensionalityof the data for simplifying visual interpretation Amongthem principal component analysis (PCA) hierarchicalcluster analysis (HCA) linear discriminant analysis andartificial neural networks (ANN) are the most commontechniques [22] They can be somewhat time-consumingcomputational methods which could prevent a true real timemeasurement

Image processing is a common technique for measuringcolorimetric response of optical sensors [1] including sensorsfor explosive determination [2 6 23] In this work thisstrategy has been adopted Low-cost on-site and real timemeasurements are some of the objectives of the developedprototype in line with current portable analytical instru-mentation [1] Thus we have developed a fully portableand wireless measurement system for the identification anddetermination of explosives and related compounds whenthey are mixed in a solution It is based on a colorimetricsensor array an electronic reader unit with a microcamerafor the acquisition of an image of the sensor array and asoftware application running in a smartphone capable ofshowing the results in an easy and direct way A set of eightnitroaromatic compounds have been analyzed with a sensorarray using four different sensing chemistries This has beenpossible by programming a simple algorithm which allowsthe simultaneous identification and quantification of thesecompounds in a mixture of dissolved analytes The mainnovelty of our design compared to others [2 4ndash6 23] relies on

being a low-cost compact and wireless system which allowsgetting real time and in situ information about a large numberof nitroaromatic compounds even mixed in a solution

2 Experimental

21 Reagents All reagents were of analytical-reagent-gradeunless stated otherwise Reagents used for the preparation ofthe sensing membranes were creatinine potassiumhydroxide aniline and NN-diethylaniline and ethanol allpurchased from Sigma (Sigma-Aldrich Quımica SA Spain)Nitroaromatic analytes used in this study were picric acid24-dinitrotoluene (24-DNT) 13-dinitrobenzene (13-DNB)35-dinitrobenzonitrile (35-DNBN) and 2-chloro-35-dini-trobenzotrifluoride (2-C-35-DNBF) purchased from Sigma-Aldrich and standards 1000mgL of 135-trinitrobenzene(TNB) 246-trinitrotoluene (TNT) and tetryl (TT) boughtfrom Chem Service (West Chester Pennsylvania USA)Solvents used to dilute the analytes in order to preparestandard solutions were methanol and acetonitrile purchasedfrom Sigma-Aldrich Aqueous solutions were made usingreverse-osmosis type quality water (Milli-RO 12 plus Milli-Qstation from Millipore conductivity 182MΩsdotcm) Filterpaper (ref 1250 basis weight 87 gm2 thickness 0180 120583mretention 10ndash15 120583m) from Filter-Lab Barcelona Spaindouble-sided adhesive tape from Miarco (Spain) and whitesheets from Schwan-Stabilo (Germany) were used to preparesensing membranes

22 Instruments and Software Theelectrical characterizationof the system was carried out using the following labora-tory instrumentation a mixed signal oscilloscope MSO4101(Tektronix Oregon USA) 812-bit Digital Multimeter 3158A(Agilent Technologies California USA) 15MHz waveformgenerator 33120A (Agilent Technologies) a DC power sup-ply E3630A (Agilent Technologies) and balance DV215CD(Ohaus Co New Jersey USA) For testing and calibra-tion purposes the software Matlab (MathWorks Inc Mas-sachusetts USA) andAdobe Photoshop (Adobe Systems IncCalifornia USA) were used as well

23 System Description

231 Sensing Array Preparation The disposable sensor arraywas prepared by laser-cutting technique because it is a cost-efficient simple and reproducible process using standardlaboratory filter paper (Filter-Lab ref 1250) The pattern wasdesigned using Illustrator software (Adobe Systems) and thenexported as an FS file to the controller software of a desktoplaser engraver Rayjet with a CO

2laser source (Trotec Laser

GmbH Wels Austria) For the fabrication a piece of paperwas affixed using double-sided adhesive tape to awhite plasticsheet The removal of weeding was performed manuallyThe design consists of 4 circles of 5mm diameter arrangedon 2 columns and 2 rows with dimensions indicated inFigure 2The size of the array and spots and their distributionwere designed taking into account the field of view of themicrocamera and the number of spots in each picture The

Journal of Sensors 3

Microcamera

LEDSensor array

LEDRaspberry Pi

Battery

WiFimodule

Android device

Portable reading unit

Figure 1 Block diagram of the instrument showing both the portable reader unit and the Android mobile device with the programmedapplication to present the results

ROI

Sensor array

ROR

5mm

25mm

3mm

25mm

3mm

25mm

18mm

18

mm

Figure 2 Dimensions of the sensor array and selection of the ROIsand ROR

device was prepared by drop-casting in each sensing area theneeded reagents under ambient atmospheric conditions

232 Description of the Measurement System The developedmeasurement electronic system is aimed at registering andquantifying the color variation of the sensing membraneswhen a volume of a solution containing some of the nitroaro-matic targets described above is added over them Theprototype is composed of a reader and processing hardwareunitwhich sends the information to amobile device (such as asmartphone or tablet) with a customizedAndroid applicationto show the result of the analysis carried out to the final userThe information gathering is based on the color variationsof a colorimetric chemical sensor array when exposed tothe analytes The capture and processing of the sensor arrayimage are respectively carried out by a microcamera and amicroprocessor included in the reader unit and results aresent to the mobile device of a remote user via a WiFi linkAlthough other systems have been already reported followingthis approach [24] our system presents the characteristic ofbeing fully portable and remotely operable

Sensing Module As it has been already explained the instru-ment is based on the acquisition and processing of the imageof the sensor array in order to quantify the color variationsand to relate them to the presence of some nitroaromaticcompound Figure 1 shows a diagram of the prototype Theprocessing electronics is enclosed in a dark box of dimensions13 times 13 times 5 cm3 with a slot for the insertion of the sensor array

As it can be seen the core of the instrument is acommercial Raspberry Pi platform the Raspberry Pi 2modelB (Raspberry Pi Foundation United Kingdom) This is alow-cost device that includes a 900MHz quad-core ARMCortex-A7 CPU and 1GB of RAMThese specifications makethis platform a powerful tool for fast image processingIn addition the Raspberry Pi 2 model B counts with 4-port USB for connection of external modules a camerainterface (CSI) and a micro-SD card slot which allowsexpanding the memory capacity in order to store the takenimages Besides the Raspberry Pi board provides 40 generalpurpose inputoutput (GPIO) ports allowing us to includesome external elements such as the light source used forthe system The microcamera used is the model RaspberryPi camera (Raspberry Pi Foundation United Kingdom)which is compatible with the Raspberry Pi platform It is ahigh resolution camera up to 5 megapixels based on theimage sensor OmniVision 5647 (OmniVision TechnologiesCalifornia USA) This camera allows acquiring both videoand still photographs It is connected to the Raspberry corethrough the available CSI

To obtain uniform and stable illumination for the acqui-sition of the image two diffuse white light emitting diodes(LEDs) model ASMT-MWH (Avago Technologies Singa-pore) are disposed in the inside of the box The operationof the LEDs is controlled by the Raspberry Pi core throughtwo digital outputs that switch them on and offThe intensityof the emitted light is manually adjusted by means of apotentiometer so that overexposition and subexposition areavoided during the acquisition procedure

An external module for setting up a WiFi network hasbeen added to the Raspberry Pi board This is the modelMiniature WiFi (Adafruit Industries New York USA) thatis fully compatible with the Raspberry Pi platform to which

4 Journal of Sensors

is connected through a USB port It implements wirelessstandards IEEE 80211n (draft) IEEE 80211g and IEEE80211b The whole system is battery-powered A large-sizedrechargeable battery pack is included in the design (AdafruitIndustries New York USA) Inside is a massive 10000mAhlithium ion battery that provides 5 VDC and 2 A via a USB-Aport The total weight of the reading unit is 150 g

Android Application An application based on Android oper-ative system has been developed in order to allow theremote control of the instrument through smart devicessuch as smartphones or tablets In order to implement theapplication Eclipse v2230 was chosen as the integrateddevelopment environment (IDE) since the plugin needed tointegrate Android is more developed than for other IDEsand also a phone emulator can be used for debuggingpurposes The application establishes a WiFi-based wirelesscommunication with the Raspberry Pi board through whichthe user requires a measurement and receives the resultingdata presenting them to the user through a simple andintuitive interface

233 Measurement Procedure The reagents used for thecolorimetric determination of the nitroaromatic compoundsare creatinine potassium hydroxide aniline and NN-diethylaniline Creatinine is dissolved in NaOH 01M with acreatinine concentration of 100mmolL KOH is dissolved inethanol and its concentration is 100 gL finally aniline andNN-diethylaniline are used without dilution The standardscontaining the analytes picric acid 24-dinitrotoluene 13-dinitrobenzene 35-dinitrobenzonitrile and 2-chloro-35-dinitrobenzotrifluoride are prepared by dissolving them inacetonitrile methanol (1 1) mixture In the case of 135-trinitrobenzene 246-trinitrotoluene and tetryl a 1000mgLstandard solution in acetonitrile methanol (1 1) is usedThe dilutions used in this study are obtained from thesedissolutions

The measurement procedure consists in the addition of5 120583L of sample to each of the paper spots present in theplastic cardboard and later 5 120583L of one of the reagent solutionsin each spot Each reagent solution must be added in apredetermined position in the card After the addition 15minutes is waited in order to let the reactions occur andto capture a picture of the complete colorimetric signalAfterwards the card is introduced in the reader unit and animage is taken from the cardboard

Safety Note Some of the nitroaromatic compounds analyzedin this study show explosive character For this reason thehandling of the samplesmust be carried out with precautionspaying extra attention to the amount of the manipulatedcompounds temperature variations and possible impacts

3 Results and Discussion

31 Nitroaromatics Recognition The recognition of eightnitroaromatics under study was performed using a whiteplastic board containing an array of paper circles fabricatedby craft-cutting technique Three different reactions were

used for detection and determination of these compoundsThe reaction of aromatic polynitrocompounds with car-banions results in the formation of Janovsky 120590-complexes[25] The red-orange color and the intensity depend on thenitroaromatic substance that is present and its concentrationThe reagent used is creatinine in basic medium (Jaffe reac-tion) [26 27] A second reaction takes place using potas-sium hydroxide in ethanolic solution that results in coloredMeisenheimer type complexes in some cases and coloredanions in others [28 29] Another reaction responsible for theformation of colored products with nitroaromatics involvesthe use of basic reagents such as aromatic amines based on acharge transfer mechanism in which the aromatic amine actsas a 120587 donor whereas the nitroaromatics serve as 120587 acceptor[28] We used this last reaction with two different reagentsaniline and NN-diethylaniline increasing in this way thediscrimination power of the array

32 Image Processing The core of the reader unit that is theRaspberry Pi board is programmed to carry out processingof the acquired images in order to obtain quantification of thecolor of eachmembraneThis image processing consists of analgorithm that provides only one parameter related directlyto the color change registered for each reaction The firststep of this algorithm consists in the selection of the regionsof interest (ROIs) in the sensor array These are the centralzone of each membrane discarding the edges where somenonuniformities of the membrane can be found In additiona region in the blank space of the board is also selected asa region of reference (ROR) as it is depicted in Figure 2This fifth region is used to normalize the color quantificationof the previous four regions corresponding to the sensitivemembranes in order to correct color measurement driftscaused by intensity changes in the biasing of the whiteLEDs or small nonuniformities in the illumination due tomisalignment of the membrane board and the white LEDs

After these ROIs are selected the algorithm extractsthe value of the red green and blue (R G and B) colorcoordinates of each region as the most frequent values (thestatistical mode) of R G and B of the pixels inside eachROI These values define the three chromaticities that allowunequivocally specifying a color in the RGB color spaceFrom these values normalized coordinates R1015840 G1015840 and B1015840for the ROIs corresponding to the sensitive membranes areobtained as the ratio between the RGB coordinates of thisregion and the RGB values of the fifth region correspondingto the blank space betweenmembranes in the board which isconsidered as thewhite reference Although these normalizedvalues of the RGB coordinates allow quantifying the color ofthemembranes these data are reduced to only one to simplifythe information on the color variations With this aim theEuclidean distance of the color of each membrane before andafter the reaction with the explosive sample is consideredThis distance is defined as the difference between the initialcolor coordinate and final color coordinate of the analyzedsamples [30]

119863 = radic(R1015840 minus R10158400)2

+ (G1015840 minus G10158400)2

+ (B1015840 minus B10158400)2

(1)

Journal of Sensors 5

20mgL10mgL5mgL1mgL

(a)

20mgL10mgL5mgL1mgL

(b)

Figure 3 Color variations of potassium hydroxide with 2-chloro-35-dinitrobenzotrifluoride (a) and creatinine with 35-dinitrobenzonitrile(b) for different analyte concentrations from 1 to 20mgL

where 11988310158400is the normalized color coordinate (red green

or blue) before the reaction with the sample and 1198831015840 is thecorresponding normalized coordinate after the reaction

By means of the colorimetric distance 119863 only oneparameter is used to measure directly the color variation thatthe membrane suffers when it is reacted in the presence of anitroaromatic compound

33 Calibration Curves The calibration of the measurementsystem was performed using separated solutions of theanalytes picric acid 24-dinitrotoluene 13-dinitrobenzene35-dinitrobenzonitrile 2-chloro-35-dinitrobenzotrifluor-ide ranged from 1 to 20mgL and 135-trinitrobenzene246-trinitrotoluene and tetryl ranged from 1 to 100mgLwith four replicas per solution and for each one of thefour mentioned reagents above This procedure results in38 calibration curves with four replicas for each of themErrors have been calculated as the standard deviationof experimental data The calibration was carried out asdescribed in the measurement procedure section obtaininga color change on the paper when the reagents are added InFigure 3 some examples of color variation of two reagentscan be observed when they are reacted with differentconcentrations of analytes

The Euclidean color distances have been extracted for allthe analytes in concentration ranging from 1 to 100mgLThe results show that this parameter presents a tendencythat can be fitted to a simple calibration curve (linearexponential or logarithmic) for two reagents at least Figure 4

shows some examples of the fitting curves obtained for somereactions between some of the reactants and nitroaromaticcompounds

In Table 1 the different fitting equations found in the cali-bration procedure are presented where119910 is the concentrationof the nitroaromatic compound and 119863 is the calculatedEuclidean colorimetric distance

34 Analyte Determination Algorithm Principal componentanalysis (PCA) is a standard technique for samples separationandor classification in a solution with a mixture of analytes[31 32] In this work we propose developing an even simpleralgorithm able to calculate the presence and concentration ofcertain nitroaromatic compounds some of them explosivesthus making the use of computationally costly multivariabletechniques unnecessary As it can be seen in Table 1 simpletwo-parameter linear exponential or logarithmic fittingequations can be used to predict the concentration of theanalytes Nevertheless the reactants are nonselective andcolor variations are produced when they react with differentanalytes Therefore it is necessary to select the proper fittingfunction when a color variation is produced in some of thesensing membranes This is accomplished in two steps asshown in Figure 5

Firstly by comparing the measured Euclidean colorimet-ric distances 119863

119894for each reaction (four values for the four

membranes in the sensor array) to the possible variationrange of this parameter obtained in the calibration curve ofthe corresponding reaction that is if a value of119863 falls in the

6 Journal of Sensors

Table 1 Fitting equations for nitroaromatic compounds

Nitroaromatic compound Aniline NN-Diethylaniline Creatinine Potassium hydroxide

Picric acid mdash mdash 119910 = 32629119863 minus 30636

1198772= 0998

119910 = 38665119863 minus 21012

1198772= 0999

24-DNT 119910 = 30098119863 + 053931

1198772= 0997

119910 = 80684 ln119863+ 351551198772= 0998

mdash 119910 = 01521411989038807119863

1198772= 0981

13-DNB 119910 = 08704811989037975119863

1198772= 0999

mdash mdash 119910 = 02654311989011940119863

1198772= 0974

35-DNBN 119910 = 91612119863 + 046375

1198772= 0993

119910 = 37418119863 minus 12879

1198772= 0985

119910 = 07228111989032627119863

1198772= 0999

119910 = 06602511989019234119863

1198772= 0969

2-Cl-35-DNT 119910 = 97915119863 + 071300

1198772= 0997

mdash mdash 119910 = 34155119863 minus 33118

1198772= 0983

TNB mdash mdash 119910 = 43885119863 minus 19694

1198772= 0997

119910 = 34040119863 minus 16816

1198772= 0983

TNT mdash mdash 119910 = 84385 ln119863+ 259951198772= 0988

119910 = 53987119863 minus 99810

1198772= 0983

TT mdash mdash 119910 = 52070119863 minus 30026

1198772= 0986

119910 = 3147211989041415119863

1198772= 0966

000

050

100

150

0 5 10 15 2024-DNT (mgL)

Eucli

dean

dist

ance

D

(a)

000

005

010

015

020

0 20 40 60 80 100TNT (mgL)

Eucli

dean

dist

ance

D

(b)

Figure 4 Relationship between concentration of two analytes and measured Euclidean colorimetric distance for potassium hydroxide (a)and creatinine (b)

variation range obtained for a given reaction this reactionis considered as a possible solution On the contrary thisreaction is discarded as well as the presence of this analyteAfter this process several reactions may still be solutionsof the problem case For the final selection of the properreaction it is taken into account that according to Table 1each analyte produces at least two color variations in thesensor array that is there are at least two reactants thatchange their color when they are immersed in the analyte

sample This fact allows using at least two fitting equationsfor the determination of each analyte If the result of thesepredictions for the given values of 119863 is the same theobtained value of 119910 is taken as the correct prediction of thecorresponding analyte We have checked that this procedureresults in univocal solution in the analyzed framework

This separation algorithm has been implemented in thecore of the Raspberry Pi board and it has been tested topredict the presence and concentration of the analytes as

Journal of Sensors 7

Analyte selection

Analyte discarded

No

Evaluation of the fitting equations

Yes

Same solution

Analyte discarded

No

Analyte identified

Yes

Analyte determined

Measurement of Di (4 values)

with variation ranges

Comparison of Di

corresponding range

Every Di inside

Figure 5 Flux diagram of the nitroaromatic identification and determination algorithm

it is detailed below Good results are achieved proving thatthe portable instrument presented here is useful to detect thepresence of a variety of nitroaromatic compounds

35 Validation and Specifications The resolution and sensi-tivity of the system have been theoretically calculated fromthe expressions of Table 1 The sensitivity 119878 is evaluated asthe derivative of the fitting function for the nitroaromaticcompound 119910

119878119910=120597119891 (119863)

120597119863 (2)

where 119891(119863) represents any of the fitting functions in Table 1The resolution can be evaluated from the sensitivity as the

product of this parameter by the error in the determinationof the variable119863 [33]

Δ119910 = 119878119910sdot Δ119863 (3)

From (1) the resolution of the Euclidean distance119863 is

Δ119863 =120597119863

120597R1015840ΔR1015840 + 120597119863120597R10158400

ΔR10158400+120597119863

120597G1015840ΔG1015840 + 120597119863120597G10158400

ΔG10158400

+120597119863

120597B1015840ΔB1015840 + 120597119863120597B10158400

ΔB10158400

(4)

And the resolution of the normalized color coordinates iscalculated as

1198831015840=119883

119883ref997904rArr Δ119883

1015840=1

119883ref(1 +119883

119883ref)Δ119883 (5)

where 119883 = R G B and Δ119883 = Δ119883ref is the resolution inthe measurement of the color coordinate (R G or B) This isgiven by the resolution of the microcamera which generatesimages with 8 bits per channel Therefore Δ119883 is taken as 128= 1256

Taking into account that for the calculus of the analytesconcentration at least two functions can be used the resolu-tion in its determination is considered as the worst case ofthese two possibilities

The limit of detection (LOD) is calculated using the stan-dard criteria LOD = 119910

119887+ 3119904119887 where 119910

119887is the average blank

signal and 119904119887is the standard deviation of the blank which is

determined using ten replicas [34] The values of sensitivityresolution and LOD achieved for the determination of thenitroaromatic compounds are listed in Table 2 The units ofboth the resolution and sensitivity are the same since thedistance119863 used in (1)ndash(3) is a dimensionless parameter

As in every electronic system some noise is presentthat can affect the measurement In this case two main

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

Journal of Sensors 3

Microcamera

LEDSensor array

LEDRaspberry Pi

Battery

WiFimodule

Android device

Portable reading unit

Figure 1 Block diagram of the instrument showing both the portable reader unit and the Android mobile device with the programmedapplication to present the results

ROI

Sensor array

ROR

5mm

25mm

3mm

25mm

3mm

25mm

18mm

18

mm

Figure 2 Dimensions of the sensor array and selection of the ROIsand ROR

device was prepared by drop-casting in each sensing area theneeded reagents under ambient atmospheric conditions

232 Description of the Measurement System The developedmeasurement electronic system is aimed at registering andquantifying the color variation of the sensing membraneswhen a volume of a solution containing some of the nitroaro-matic targets described above is added over them Theprototype is composed of a reader and processing hardwareunitwhich sends the information to amobile device (such as asmartphone or tablet) with a customizedAndroid applicationto show the result of the analysis carried out to the final userThe information gathering is based on the color variationsof a colorimetric chemical sensor array when exposed tothe analytes The capture and processing of the sensor arrayimage are respectively carried out by a microcamera and amicroprocessor included in the reader unit and results aresent to the mobile device of a remote user via a WiFi linkAlthough other systems have been already reported followingthis approach [24] our system presents the characteristic ofbeing fully portable and remotely operable

Sensing Module As it has been already explained the instru-ment is based on the acquisition and processing of the imageof the sensor array in order to quantify the color variationsand to relate them to the presence of some nitroaromaticcompound Figure 1 shows a diagram of the prototype Theprocessing electronics is enclosed in a dark box of dimensions13 times 13 times 5 cm3 with a slot for the insertion of the sensor array

As it can be seen the core of the instrument is acommercial Raspberry Pi platform the Raspberry Pi 2modelB (Raspberry Pi Foundation United Kingdom) This is alow-cost device that includes a 900MHz quad-core ARMCortex-A7 CPU and 1GB of RAMThese specifications makethis platform a powerful tool for fast image processingIn addition the Raspberry Pi 2 model B counts with 4-port USB for connection of external modules a camerainterface (CSI) and a micro-SD card slot which allowsexpanding the memory capacity in order to store the takenimages Besides the Raspberry Pi board provides 40 generalpurpose inputoutput (GPIO) ports allowing us to includesome external elements such as the light source used forthe system The microcamera used is the model RaspberryPi camera (Raspberry Pi Foundation United Kingdom)which is compatible with the Raspberry Pi platform It is ahigh resolution camera up to 5 megapixels based on theimage sensor OmniVision 5647 (OmniVision TechnologiesCalifornia USA) This camera allows acquiring both videoand still photographs It is connected to the Raspberry corethrough the available CSI

To obtain uniform and stable illumination for the acqui-sition of the image two diffuse white light emitting diodes(LEDs) model ASMT-MWH (Avago Technologies Singa-pore) are disposed in the inside of the box The operationof the LEDs is controlled by the Raspberry Pi core throughtwo digital outputs that switch them on and offThe intensityof the emitted light is manually adjusted by means of apotentiometer so that overexposition and subexposition areavoided during the acquisition procedure

An external module for setting up a WiFi network hasbeen added to the Raspberry Pi board This is the modelMiniature WiFi (Adafruit Industries New York USA) thatis fully compatible with the Raspberry Pi platform to which

4 Journal of Sensors

is connected through a USB port It implements wirelessstandards IEEE 80211n (draft) IEEE 80211g and IEEE80211b The whole system is battery-powered A large-sizedrechargeable battery pack is included in the design (AdafruitIndustries New York USA) Inside is a massive 10000mAhlithium ion battery that provides 5 VDC and 2 A via a USB-Aport The total weight of the reading unit is 150 g

Android Application An application based on Android oper-ative system has been developed in order to allow theremote control of the instrument through smart devicessuch as smartphones or tablets In order to implement theapplication Eclipse v2230 was chosen as the integrateddevelopment environment (IDE) since the plugin needed tointegrate Android is more developed than for other IDEsand also a phone emulator can be used for debuggingpurposes The application establishes a WiFi-based wirelesscommunication with the Raspberry Pi board through whichthe user requires a measurement and receives the resultingdata presenting them to the user through a simple andintuitive interface

233 Measurement Procedure The reagents used for thecolorimetric determination of the nitroaromatic compoundsare creatinine potassium hydroxide aniline and NN-diethylaniline Creatinine is dissolved in NaOH 01M with acreatinine concentration of 100mmolL KOH is dissolved inethanol and its concentration is 100 gL finally aniline andNN-diethylaniline are used without dilution The standardscontaining the analytes picric acid 24-dinitrotoluene 13-dinitrobenzene 35-dinitrobenzonitrile and 2-chloro-35-dinitrobenzotrifluoride are prepared by dissolving them inacetonitrile methanol (1 1) mixture In the case of 135-trinitrobenzene 246-trinitrotoluene and tetryl a 1000mgLstandard solution in acetonitrile methanol (1 1) is usedThe dilutions used in this study are obtained from thesedissolutions

The measurement procedure consists in the addition of5 120583L of sample to each of the paper spots present in theplastic cardboard and later 5 120583L of one of the reagent solutionsin each spot Each reagent solution must be added in apredetermined position in the card After the addition 15minutes is waited in order to let the reactions occur andto capture a picture of the complete colorimetric signalAfterwards the card is introduced in the reader unit and animage is taken from the cardboard

Safety Note Some of the nitroaromatic compounds analyzedin this study show explosive character For this reason thehandling of the samplesmust be carried out with precautionspaying extra attention to the amount of the manipulatedcompounds temperature variations and possible impacts

3 Results and Discussion

31 Nitroaromatics Recognition The recognition of eightnitroaromatics under study was performed using a whiteplastic board containing an array of paper circles fabricatedby craft-cutting technique Three different reactions were

used for detection and determination of these compoundsThe reaction of aromatic polynitrocompounds with car-banions results in the formation of Janovsky 120590-complexes[25] The red-orange color and the intensity depend on thenitroaromatic substance that is present and its concentrationThe reagent used is creatinine in basic medium (Jaffe reac-tion) [26 27] A second reaction takes place using potas-sium hydroxide in ethanolic solution that results in coloredMeisenheimer type complexes in some cases and coloredanions in others [28 29] Another reaction responsible for theformation of colored products with nitroaromatics involvesthe use of basic reagents such as aromatic amines based on acharge transfer mechanism in which the aromatic amine actsas a 120587 donor whereas the nitroaromatics serve as 120587 acceptor[28] We used this last reaction with two different reagentsaniline and NN-diethylaniline increasing in this way thediscrimination power of the array

32 Image Processing The core of the reader unit that is theRaspberry Pi board is programmed to carry out processingof the acquired images in order to obtain quantification of thecolor of eachmembraneThis image processing consists of analgorithm that provides only one parameter related directlyto the color change registered for each reaction The firststep of this algorithm consists in the selection of the regionsof interest (ROIs) in the sensor array These are the centralzone of each membrane discarding the edges where somenonuniformities of the membrane can be found In additiona region in the blank space of the board is also selected asa region of reference (ROR) as it is depicted in Figure 2This fifth region is used to normalize the color quantificationof the previous four regions corresponding to the sensitivemembranes in order to correct color measurement driftscaused by intensity changes in the biasing of the whiteLEDs or small nonuniformities in the illumination due tomisalignment of the membrane board and the white LEDs

After these ROIs are selected the algorithm extractsthe value of the red green and blue (R G and B) colorcoordinates of each region as the most frequent values (thestatistical mode) of R G and B of the pixels inside eachROI These values define the three chromaticities that allowunequivocally specifying a color in the RGB color spaceFrom these values normalized coordinates R1015840 G1015840 and B1015840for the ROIs corresponding to the sensitive membranes areobtained as the ratio between the RGB coordinates of thisregion and the RGB values of the fifth region correspondingto the blank space betweenmembranes in the board which isconsidered as thewhite reference Although these normalizedvalues of the RGB coordinates allow quantifying the color ofthemembranes these data are reduced to only one to simplifythe information on the color variations With this aim theEuclidean distance of the color of each membrane before andafter the reaction with the explosive sample is consideredThis distance is defined as the difference between the initialcolor coordinate and final color coordinate of the analyzedsamples [30]

119863 = radic(R1015840 minus R10158400)2

+ (G1015840 minus G10158400)2

+ (B1015840 minus B10158400)2

(1)

Journal of Sensors 5

20mgL10mgL5mgL1mgL

(a)

20mgL10mgL5mgL1mgL

(b)

Figure 3 Color variations of potassium hydroxide with 2-chloro-35-dinitrobenzotrifluoride (a) and creatinine with 35-dinitrobenzonitrile(b) for different analyte concentrations from 1 to 20mgL

where 11988310158400is the normalized color coordinate (red green

or blue) before the reaction with the sample and 1198831015840 is thecorresponding normalized coordinate after the reaction

By means of the colorimetric distance 119863 only oneparameter is used to measure directly the color variation thatthe membrane suffers when it is reacted in the presence of anitroaromatic compound

33 Calibration Curves The calibration of the measurementsystem was performed using separated solutions of theanalytes picric acid 24-dinitrotoluene 13-dinitrobenzene35-dinitrobenzonitrile 2-chloro-35-dinitrobenzotrifluor-ide ranged from 1 to 20mgL and 135-trinitrobenzene246-trinitrotoluene and tetryl ranged from 1 to 100mgLwith four replicas per solution and for each one of thefour mentioned reagents above This procedure results in38 calibration curves with four replicas for each of themErrors have been calculated as the standard deviationof experimental data The calibration was carried out asdescribed in the measurement procedure section obtaininga color change on the paper when the reagents are added InFigure 3 some examples of color variation of two reagentscan be observed when they are reacted with differentconcentrations of analytes

The Euclidean color distances have been extracted for allthe analytes in concentration ranging from 1 to 100mgLThe results show that this parameter presents a tendencythat can be fitted to a simple calibration curve (linearexponential or logarithmic) for two reagents at least Figure 4

shows some examples of the fitting curves obtained for somereactions between some of the reactants and nitroaromaticcompounds

In Table 1 the different fitting equations found in the cali-bration procedure are presented where119910 is the concentrationof the nitroaromatic compound and 119863 is the calculatedEuclidean colorimetric distance

34 Analyte Determination Algorithm Principal componentanalysis (PCA) is a standard technique for samples separationandor classification in a solution with a mixture of analytes[31 32] In this work we propose developing an even simpleralgorithm able to calculate the presence and concentration ofcertain nitroaromatic compounds some of them explosivesthus making the use of computationally costly multivariabletechniques unnecessary As it can be seen in Table 1 simpletwo-parameter linear exponential or logarithmic fittingequations can be used to predict the concentration of theanalytes Nevertheless the reactants are nonselective andcolor variations are produced when they react with differentanalytes Therefore it is necessary to select the proper fittingfunction when a color variation is produced in some of thesensing membranes This is accomplished in two steps asshown in Figure 5

Firstly by comparing the measured Euclidean colorimet-ric distances 119863

119894for each reaction (four values for the four

membranes in the sensor array) to the possible variationrange of this parameter obtained in the calibration curve ofthe corresponding reaction that is if a value of119863 falls in the

6 Journal of Sensors

Table 1 Fitting equations for nitroaromatic compounds

Nitroaromatic compound Aniline NN-Diethylaniline Creatinine Potassium hydroxide

Picric acid mdash mdash 119910 = 32629119863 minus 30636

1198772= 0998

119910 = 38665119863 minus 21012

1198772= 0999

24-DNT 119910 = 30098119863 + 053931

1198772= 0997

119910 = 80684 ln119863+ 351551198772= 0998

mdash 119910 = 01521411989038807119863

1198772= 0981

13-DNB 119910 = 08704811989037975119863

1198772= 0999

mdash mdash 119910 = 02654311989011940119863

1198772= 0974

35-DNBN 119910 = 91612119863 + 046375

1198772= 0993

119910 = 37418119863 minus 12879

1198772= 0985

119910 = 07228111989032627119863

1198772= 0999

119910 = 06602511989019234119863

1198772= 0969

2-Cl-35-DNT 119910 = 97915119863 + 071300

1198772= 0997

mdash mdash 119910 = 34155119863 minus 33118

1198772= 0983

TNB mdash mdash 119910 = 43885119863 minus 19694

1198772= 0997

119910 = 34040119863 minus 16816

1198772= 0983

TNT mdash mdash 119910 = 84385 ln119863+ 259951198772= 0988

119910 = 53987119863 minus 99810

1198772= 0983

TT mdash mdash 119910 = 52070119863 minus 30026

1198772= 0986

119910 = 3147211989041415119863

1198772= 0966

000

050

100

150

0 5 10 15 2024-DNT (mgL)

Eucli

dean

dist

ance

D

(a)

000

005

010

015

020

0 20 40 60 80 100TNT (mgL)

Eucli

dean

dist

ance

D

(b)

Figure 4 Relationship between concentration of two analytes and measured Euclidean colorimetric distance for potassium hydroxide (a)and creatinine (b)

variation range obtained for a given reaction this reactionis considered as a possible solution On the contrary thisreaction is discarded as well as the presence of this analyteAfter this process several reactions may still be solutionsof the problem case For the final selection of the properreaction it is taken into account that according to Table 1each analyte produces at least two color variations in thesensor array that is there are at least two reactants thatchange their color when they are immersed in the analyte

sample This fact allows using at least two fitting equationsfor the determination of each analyte If the result of thesepredictions for the given values of 119863 is the same theobtained value of 119910 is taken as the correct prediction of thecorresponding analyte We have checked that this procedureresults in univocal solution in the analyzed framework

This separation algorithm has been implemented in thecore of the Raspberry Pi board and it has been tested topredict the presence and concentration of the analytes as

Journal of Sensors 7

Analyte selection

Analyte discarded

No

Evaluation of the fitting equations

Yes

Same solution

Analyte discarded

No

Analyte identified

Yes

Analyte determined

Measurement of Di (4 values)

with variation ranges

Comparison of Di

corresponding range

Every Di inside

Figure 5 Flux diagram of the nitroaromatic identification and determination algorithm

it is detailed below Good results are achieved proving thatthe portable instrument presented here is useful to detect thepresence of a variety of nitroaromatic compounds

35 Validation and Specifications The resolution and sensi-tivity of the system have been theoretically calculated fromthe expressions of Table 1 The sensitivity 119878 is evaluated asthe derivative of the fitting function for the nitroaromaticcompound 119910

119878119910=120597119891 (119863)

120597119863 (2)

where 119891(119863) represents any of the fitting functions in Table 1The resolution can be evaluated from the sensitivity as the

product of this parameter by the error in the determinationof the variable119863 [33]

Δ119910 = 119878119910sdot Δ119863 (3)

From (1) the resolution of the Euclidean distance119863 is

Δ119863 =120597119863

120597R1015840ΔR1015840 + 120597119863120597R10158400

ΔR10158400+120597119863

120597G1015840ΔG1015840 + 120597119863120597G10158400

ΔG10158400

+120597119863

120597B1015840ΔB1015840 + 120597119863120597B10158400

ΔB10158400

(4)

And the resolution of the normalized color coordinates iscalculated as

1198831015840=119883

119883ref997904rArr Δ119883

1015840=1

119883ref(1 +119883

119883ref)Δ119883 (5)

where 119883 = R G B and Δ119883 = Δ119883ref is the resolution inthe measurement of the color coordinate (R G or B) This isgiven by the resolution of the microcamera which generatesimages with 8 bits per channel Therefore Δ119883 is taken as 128= 1256

Taking into account that for the calculus of the analytesconcentration at least two functions can be used the resolu-tion in its determination is considered as the worst case ofthese two possibilities

The limit of detection (LOD) is calculated using the stan-dard criteria LOD = 119910

119887+ 3119904119887 where 119910

119887is the average blank

signal and 119904119887is the standard deviation of the blank which is

determined using ten replicas [34] The values of sensitivityresolution and LOD achieved for the determination of thenitroaromatic compounds are listed in Table 2 The units ofboth the resolution and sensitivity are the same since thedistance119863 used in (1)ndash(3) is a dimensionless parameter

As in every electronic system some noise is presentthat can affect the measurement In this case two main

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

4 Journal of Sensors

is connected through a USB port It implements wirelessstandards IEEE 80211n (draft) IEEE 80211g and IEEE80211b The whole system is battery-powered A large-sizedrechargeable battery pack is included in the design (AdafruitIndustries New York USA) Inside is a massive 10000mAhlithium ion battery that provides 5 VDC and 2 A via a USB-Aport The total weight of the reading unit is 150 g

Android Application An application based on Android oper-ative system has been developed in order to allow theremote control of the instrument through smart devicessuch as smartphones or tablets In order to implement theapplication Eclipse v2230 was chosen as the integrateddevelopment environment (IDE) since the plugin needed tointegrate Android is more developed than for other IDEsand also a phone emulator can be used for debuggingpurposes The application establishes a WiFi-based wirelesscommunication with the Raspberry Pi board through whichthe user requires a measurement and receives the resultingdata presenting them to the user through a simple andintuitive interface

233 Measurement Procedure The reagents used for thecolorimetric determination of the nitroaromatic compoundsare creatinine potassium hydroxide aniline and NN-diethylaniline Creatinine is dissolved in NaOH 01M with acreatinine concentration of 100mmolL KOH is dissolved inethanol and its concentration is 100 gL finally aniline andNN-diethylaniline are used without dilution The standardscontaining the analytes picric acid 24-dinitrotoluene 13-dinitrobenzene 35-dinitrobenzonitrile and 2-chloro-35-dinitrobenzotrifluoride are prepared by dissolving them inacetonitrile methanol (1 1) mixture In the case of 135-trinitrobenzene 246-trinitrotoluene and tetryl a 1000mgLstandard solution in acetonitrile methanol (1 1) is usedThe dilutions used in this study are obtained from thesedissolutions

The measurement procedure consists in the addition of5 120583L of sample to each of the paper spots present in theplastic cardboard and later 5 120583L of one of the reagent solutionsin each spot Each reagent solution must be added in apredetermined position in the card After the addition 15minutes is waited in order to let the reactions occur andto capture a picture of the complete colorimetric signalAfterwards the card is introduced in the reader unit and animage is taken from the cardboard

Safety Note Some of the nitroaromatic compounds analyzedin this study show explosive character For this reason thehandling of the samplesmust be carried out with precautionspaying extra attention to the amount of the manipulatedcompounds temperature variations and possible impacts

3 Results and Discussion

31 Nitroaromatics Recognition The recognition of eightnitroaromatics under study was performed using a whiteplastic board containing an array of paper circles fabricatedby craft-cutting technique Three different reactions were

used for detection and determination of these compoundsThe reaction of aromatic polynitrocompounds with car-banions results in the formation of Janovsky 120590-complexes[25] The red-orange color and the intensity depend on thenitroaromatic substance that is present and its concentrationThe reagent used is creatinine in basic medium (Jaffe reac-tion) [26 27] A second reaction takes place using potas-sium hydroxide in ethanolic solution that results in coloredMeisenheimer type complexes in some cases and coloredanions in others [28 29] Another reaction responsible for theformation of colored products with nitroaromatics involvesthe use of basic reagents such as aromatic amines based on acharge transfer mechanism in which the aromatic amine actsas a 120587 donor whereas the nitroaromatics serve as 120587 acceptor[28] We used this last reaction with two different reagentsaniline and NN-diethylaniline increasing in this way thediscrimination power of the array

32 Image Processing The core of the reader unit that is theRaspberry Pi board is programmed to carry out processingof the acquired images in order to obtain quantification of thecolor of eachmembraneThis image processing consists of analgorithm that provides only one parameter related directlyto the color change registered for each reaction The firststep of this algorithm consists in the selection of the regionsof interest (ROIs) in the sensor array These are the centralzone of each membrane discarding the edges where somenonuniformities of the membrane can be found In additiona region in the blank space of the board is also selected asa region of reference (ROR) as it is depicted in Figure 2This fifth region is used to normalize the color quantificationof the previous four regions corresponding to the sensitivemembranes in order to correct color measurement driftscaused by intensity changes in the biasing of the whiteLEDs or small nonuniformities in the illumination due tomisalignment of the membrane board and the white LEDs

After these ROIs are selected the algorithm extractsthe value of the red green and blue (R G and B) colorcoordinates of each region as the most frequent values (thestatistical mode) of R G and B of the pixels inside eachROI These values define the three chromaticities that allowunequivocally specifying a color in the RGB color spaceFrom these values normalized coordinates R1015840 G1015840 and B1015840for the ROIs corresponding to the sensitive membranes areobtained as the ratio between the RGB coordinates of thisregion and the RGB values of the fifth region correspondingto the blank space betweenmembranes in the board which isconsidered as thewhite reference Although these normalizedvalues of the RGB coordinates allow quantifying the color ofthemembranes these data are reduced to only one to simplifythe information on the color variations With this aim theEuclidean distance of the color of each membrane before andafter the reaction with the explosive sample is consideredThis distance is defined as the difference between the initialcolor coordinate and final color coordinate of the analyzedsamples [30]

119863 = radic(R1015840 minus R10158400)2

+ (G1015840 minus G10158400)2

+ (B1015840 minus B10158400)2

(1)

Journal of Sensors 5

20mgL10mgL5mgL1mgL

(a)

20mgL10mgL5mgL1mgL

(b)

Figure 3 Color variations of potassium hydroxide with 2-chloro-35-dinitrobenzotrifluoride (a) and creatinine with 35-dinitrobenzonitrile(b) for different analyte concentrations from 1 to 20mgL

where 11988310158400is the normalized color coordinate (red green

or blue) before the reaction with the sample and 1198831015840 is thecorresponding normalized coordinate after the reaction

By means of the colorimetric distance 119863 only oneparameter is used to measure directly the color variation thatthe membrane suffers when it is reacted in the presence of anitroaromatic compound

33 Calibration Curves The calibration of the measurementsystem was performed using separated solutions of theanalytes picric acid 24-dinitrotoluene 13-dinitrobenzene35-dinitrobenzonitrile 2-chloro-35-dinitrobenzotrifluor-ide ranged from 1 to 20mgL and 135-trinitrobenzene246-trinitrotoluene and tetryl ranged from 1 to 100mgLwith four replicas per solution and for each one of thefour mentioned reagents above This procedure results in38 calibration curves with four replicas for each of themErrors have been calculated as the standard deviationof experimental data The calibration was carried out asdescribed in the measurement procedure section obtaininga color change on the paper when the reagents are added InFigure 3 some examples of color variation of two reagentscan be observed when they are reacted with differentconcentrations of analytes

The Euclidean color distances have been extracted for allthe analytes in concentration ranging from 1 to 100mgLThe results show that this parameter presents a tendencythat can be fitted to a simple calibration curve (linearexponential or logarithmic) for two reagents at least Figure 4

shows some examples of the fitting curves obtained for somereactions between some of the reactants and nitroaromaticcompounds

In Table 1 the different fitting equations found in the cali-bration procedure are presented where119910 is the concentrationof the nitroaromatic compound and 119863 is the calculatedEuclidean colorimetric distance

34 Analyte Determination Algorithm Principal componentanalysis (PCA) is a standard technique for samples separationandor classification in a solution with a mixture of analytes[31 32] In this work we propose developing an even simpleralgorithm able to calculate the presence and concentration ofcertain nitroaromatic compounds some of them explosivesthus making the use of computationally costly multivariabletechniques unnecessary As it can be seen in Table 1 simpletwo-parameter linear exponential or logarithmic fittingequations can be used to predict the concentration of theanalytes Nevertheless the reactants are nonselective andcolor variations are produced when they react with differentanalytes Therefore it is necessary to select the proper fittingfunction when a color variation is produced in some of thesensing membranes This is accomplished in two steps asshown in Figure 5

Firstly by comparing the measured Euclidean colorimet-ric distances 119863

119894for each reaction (four values for the four

membranes in the sensor array) to the possible variationrange of this parameter obtained in the calibration curve ofthe corresponding reaction that is if a value of119863 falls in the

6 Journal of Sensors

Table 1 Fitting equations for nitroaromatic compounds

Nitroaromatic compound Aniline NN-Diethylaniline Creatinine Potassium hydroxide

Picric acid mdash mdash 119910 = 32629119863 minus 30636

1198772= 0998

119910 = 38665119863 minus 21012

1198772= 0999

24-DNT 119910 = 30098119863 + 053931

1198772= 0997

119910 = 80684 ln119863+ 351551198772= 0998

mdash 119910 = 01521411989038807119863

1198772= 0981

13-DNB 119910 = 08704811989037975119863

1198772= 0999

mdash mdash 119910 = 02654311989011940119863

1198772= 0974

35-DNBN 119910 = 91612119863 + 046375

1198772= 0993

119910 = 37418119863 minus 12879

1198772= 0985

119910 = 07228111989032627119863

1198772= 0999

119910 = 06602511989019234119863

1198772= 0969

2-Cl-35-DNT 119910 = 97915119863 + 071300

1198772= 0997

mdash mdash 119910 = 34155119863 minus 33118

1198772= 0983

TNB mdash mdash 119910 = 43885119863 minus 19694

1198772= 0997

119910 = 34040119863 minus 16816

1198772= 0983

TNT mdash mdash 119910 = 84385 ln119863+ 259951198772= 0988

119910 = 53987119863 minus 99810

1198772= 0983

TT mdash mdash 119910 = 52070119863 minus 30026

1198772= 0986

119910 = 3147211989041415119863

1198772= 0966

000

050

100

150

0 5 10 15 2024-DNT (mgL)

Eucli

dean

dist

ance

D

(a)

000

005

010

015

020

0 20 40 60 80 100TNT (mgL)

Eucli

dean

dist

ance

D

(b)

Figure 4 Relationship between concentration of two analytes and measured Euclidean colorimetric distance for potassium hydroxide (a)and creatinine (b)

variation range obtained for a given reaction this reactionis considered as a possible solution On the contrary thisreaction is discarded as well as the presence of this analyteAfter this process several reactions may still be solutionsof the problem case For the final selection of the properreaction it is taken into account that according to Table 1each analyte produces at least two color variations in thesensor array that is there are at least two reactants thatchange their color when they are immersed in the analyte

sample This fact allows using at least two fitting equationsfor the determination of each analyte If the result of thesepredictions for the given values of 119863 is the same theobtained value of 119910 is taken as the correct prediction of thecorresponding analyte We have checked that this procedureresults in univocal solution in the analyzed framework

This separation algorithm has been implemented in thecore of the Raspberry Pi board and it has been tested topredict the presence and concentration of the analytes as

Journal of Sensors 7

Analyte selection

Analyte discarded

No

Evaluation of the fitting equations

Yes

Same solution

Analyte discarded

No

Analyte identified

Yes

Analyte determined

Measurement of Di (4 values)

with variation ranges

Comparison of Di

corresponding range

Every Di inside

Figure 5 Flux diagram of the nitroaromatic identification and determination algorithm

it is detailed below Good results are achieved proving thatthe portable instrument presented here is useful to detect thepresence of a variety of nitroaromatic compounds

35 Validation and Specifications The resolution and sensi-tivity of the system have been theoretically calculated fromthe expressions of Table 1 The sensitivity 119878 is evaluated asthe derivative of the fitting function for the nitroaromaticcompound 119910

119878119910=120597119891 (119863)

120597119863 (2)

where 119891(119863) represents any of the fitting functions in Table 1The resolution can be evaluated from the sensitivity as the

product of this parameter by the error in the determinationof the variable119863 [33]

Δ119910 = 119878119910sdot Δ119863 (3)

From (1) the resolution of the Euclidean distance119863 is

Δ119863 =120597119863

120597R1015840ΔR1015840 + 120597119863120597R10158400

ΔR10158400+120597119863

120597G1015840ΔG1015840 + 120597119863120597G10158400

ΔG10158400

+120597119863

120597B1015840ΔB1015840 + 120597119863120597B10158400

ΔB10158400

(4)

And the resolution of the normalized color coordinates iscalculated as

1198831015840=119883

119883ref997904rArr Δ119883

1015840=1

119883ref(1 +119883

119883ref)Δ119883 (5)

where 119883 = R G B and Δ119883 = Δ119883ref is the resolution inthe measurement of the color coordinate (R G or B) This isgiven by the resolution of the microcamera which generatesimages with 8 bits per channel Therefore Δ119883 is taken as 128= 1256

Taking into account that for the calculus of the analytesconcentration at least two functions can be used the resolu-tion in its determination is considered as the worst case ofthese two possibilities

The limit of detection (LOD) is calculated using the stan-dard criteria LOD = 119910

119887+ 3119904119887 where 119910

119887is the average blank

signal and 119904119887is the standard deviation of the blank which is

determined using ten replicas [34] The values of sensitivityresolution and LOD achieved for the determination of thenitroaromatic compounds are listed in Table 2 The units ofboth the resolution and sensitivity are the same since thedistance119863 used in (1)ndash(3) is a dimensionless parameter

As in every electronic system some noise is presentthat can affect the measurement In this case two main

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

Journal of Sensors 5

20mgL10mgL5mgL1mgL

(a)

20mgL10mgL5mgL1mgL

(b)

Figure 3 Color variations of potassium hydroxide with 2-chloro-35-dinitrobenzotrifluoride (a) and creatinine with 35-dinitrobenzonitrile(b) for different analyte concentrations from 1 to 20mgL

where 11988310158400is the normalized color coordinate (red green

or blue) before the reaction with the sample and 1198831015840 is thecorresponding normalized coordinate after the reaction

By means of the colorimetric distance 119863 only oneparameter is used to measure directly the color variation thatthe membrane suffers when it is reacted in the presence of anitroaromatic compound

33 Calibration Curves The calibration of the measurementsystem was performed using separated solutions of theanalytes picric acid 24-dinitrotoluene 13-dinitrobenzene35-dinitrobenzonitrile 2-chloro-35-dinitrobenzotrifluor-ide ranged from 1 to 20mgL and 135-trinitrobenzene246-trinitrotoluene and tetryl ranged from 1 to 100mgLwith four replicas per solution and for each one of thefour mentioned reagents above This procedure results in38 calibration curves with four replicas for each of themErrors have been calculated as the standard deviationof experimental data The calibration was carried out asdescribed in the measurement procedure section obtaininga color change on the paper when the reagents are added InFigure 3 some examples of color variation of two reagentscan be observed when they are reacted with differentconcentrations of analytes

The Euclidean color distances have been extracted for allthe analytes in concentration ranging from 1 to 100mgLThe results show that this parameter presents a tendencythat can be fitted to a simple calibration curve (linearexponential or logarithmic) for two reagents at least Figure 4

shows some examples of the fitting curves obtained for somereactions between some of the reactants and nitroaromaticcompounds

In Table 1 the different fitting equations found in the cali-bration procedure are presented where119910 is the concentrationof the nitroaromatic compound and 119863 is the calculatedEuclidean colorimetric distance

34 Analyte Determination Algorithm Principal componentanalysis (PCA) is a standard technique for samples separationandor classification in a solution with a mixture of analytes[31 32] In this work we propose developing an even simpleralgorithm able to calculate the presence and concentration ofcertain nitroaromatic compounds some of them explosivesthus making the use of computationally costly multivariabletechniques unnecessary As it can be seen in Table 1 simpletwo-parameter linear exponential or logarithmic fittingequations can be used to predict the concentration of theanalytes Nevertheless the reactants are nonselective andcolor variations are produced when they react with differentanalytes Therefore it is necessary to select the proper fittingfunction when a color variation is produced in some of thesensing membranes This is accomplished in two steps asshown in Figure 5

Firstly by comparing the measured Euclidean colorimet-ric distances 119863

119894for each reaction (four values for the four

membranes in the sensor array) to the possible variationrange of this parameter obtained in the calibration curve ofthe corresponding reaction that is if a value of119863 falls in the

6 Journal of Sensors

Table 1 Fitting equations for nitroaromatic compounds

Nitroaromatic compound Aniline NN-Diethylaniline Creatinine Potassium hydroxide

Picric acid mdash mdash 119910 = 32629119863 minus 30636

1198772= 0998

119910 = 38665119863 minus 21012

1198772= 0999

24-DNT 119910 = 30098119863 + 053931

1198772= 0997

119910 = 80684 ln119863+ 351551198772= 0998

mdash 119910 = 01521411989038807119863

1198772= 0981

13-DNB 119910 = 08704811989037975119863

1198772= 0999

mdash mdash 119910 = 02654311989011940119863

1198772= 0974

35-DNBN 119910 = 91612119863 + 046375

1198772= 0993

119910 = 37418119863 minus 12879

1198772= 0985

119910 = 07228111989032627119863

1198772= 0999

119910 = 06602511989019234119863

1198772= 0969

2-Cl-35-DNT 119910 = 97915119863 + 071300

1198772= 0997

mdash mdash 119910 = 34155119863 minus 33118

1198772= 0983

TNB mdash mdash 119910 = 43885119863 minus 19694

1198772= 0997

119910 = 34040119863 minus 16816

1198772= 0983

TNT mdash mdash 119910 = 84385 ln119863+ 259951198772= 0988

119910 = 53987119863 minus 99810

1198772= 0983

TT mdash mdash 119910 = 52070119863 minus 30026

1198772= 0986

119910 = 3147211989041415119863

1198772= 0966

000

050

100

150

0 5 10 15 2024-DNT (mgL)

Eucli

dean

dist

ance

D

(a)

000

005

010

015

020

0 20 40 60 80 100TNT (mgL)

Eucli

dean

dist

ance

D

(b)

Figure 4 Relationship between concentration of two analytes and measured Euclidean colorimetric distance for potassium hydroxide (a)and creatinine (b)

variation range obtained for a given reaction this reactionis considered as a possible solution On the contrary thisreaction is discarded as well as the presence of this analyteAfter this process several reactions may still be solutionsof the problem case For the final selection of the properreaction it is taken into account that according to Table 1each analyte produces at least two color variations in thesensor array that is there are at least two reactants thatchange their color when they are immersed in the analyte

sample This fact allows using at least two fitting equationsfor the determination of each analyte If the result of thesepredictions for the given values of 119863 is the same theobtained value of 119910 is taken as the correct prediction of thecorresponding analyte We have checked that this procedureresults in univocal solution in the analyzed framework

This separation algorithm has been implemented in thecore of the Raspberry Pi board and it has been tested topredict the presence and concentration of the analytes as

Journal of Sensors 7

Analyte selection

Analyte discarded

No

Evaluation of the fitting equations

Yes

Same solution

Analyte discarded

No

Analyte identified

Yes

Analyte determined

Measurement of Di (4 values)

with variation ranges

Comparison of Di

corresponding range

Every Di inside

Figure 5 Flux diagram of the nitroaromatic identification and determination algorithm

it is detailed below Good results are achieved proving thatthe portable instrument presented here is useful to detect thepresence of a variety of nitroaromatic compounds

35 Validation and Specifications The resolution and sensi-tivity of the system have been theoretically calculated fromthe expressions of Table 1 The sensitivity 119878 is evaluated asthe derivative of the fitting function for the nitroaromaticcompound 119910

119878119910=120597119891 (119863)

120597119863 (2)

where 119891(119863) represents any of the fitting functions in Table 1The resolution can be evaluated from the sensitivity as the

product of this parameter by the error in the determinationof the variable119863 [33]

Δ119910 = 119878119910sdot Δ119863 (3)

From (1) the resolution of the Euclidean distance119863 is

Δ119863 =120597119863

120597R1015840ΔR1015840 + 120597119863120597R10158400

ΔR10158400+120597119863

120597G1015840ΔG1015840 + 120597119863120597G10158400

ΔG10158400

+120597119863

120597B1015840ΔB1015840 + 120597119863120597B10158400

ΔB10158400

(4)

And the resolution of the normalized color coordinates iscalculated as

1198831015840=119883

119883ref997904rArr Δ119883

1015840=1

119883ref(1 +119883

119883ref)Δ119883 (5)

where 119883 = R G B and Δ119883 = Δ119883ref is the resolution inthe measurement of the color coordinate (R G or B) This isgiven by the resolution of the microcamera which generatesimages with 8 bits per channel Therefore Δ119883 is taken as 128= 1256

Taking into account that for the calculus of the analytesconcentration at least two functions can be used the resolu-tion in its determination is considered as the worst case ofthese two possibilities

The limit of detection (LOD) is calculated using the stan-dard criteria LOD = 119910

119887+ 3119904119887 where 119910

119887is the average blank

signal and 119904119887is the standard deviation of the blank which is

determined using ten replicas [34] The values of sensitivityresolution and LOD achieved for the determination of thenitroaromatic compounds are listed in Table 2 The units ofboth the resolution and sensitivity are the same since thedistance119863 used in (1)ndash(3) is a dimensionless parameter

As in every electronic system some noise is presentthat can affect the measurement In this case two main

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

6 Journal of Sensors

Table 1 Fitting equations for nitroaromatic compounds

Nitroaromatic compound Aniline NN-Diethylaniline Creatinine Potassium hydroxide

Picric acid mdash mdash 119910 = 32629119863 minus 30636

1198772= 0998

119910 = 38665119863 minus 21012

1198772= 0999

24-DNT 119910 = 30098119863 + 053931

1198772= 0997

119910 = 80684 ln119863+ 351551198772= 0998

mdash 119910 = 01521411989038807119863

1198772= 0981

13-DNB 119910 = 08704811989037975119863

1198772= 0999

mdash mdash 119910 = 02654311989011940119863

1198772= 0974

35-DNBN 119910 = 91612119863 + 046375

1198772= 0993

119910 = 37418119863 minus 12879

1198772= 0985

119910 = 07228111989032627119863

1198772= 0999

119910 = 06602511989019234119863

1198772= 0969

2-Cl-35-DNT 119910 = 97915119863 + 071300

1198772= 0997

mdash mdash 119910 = 34155119863 minus 33118

1198772= 0983

TNB mdash mdash 119910 = 43885119863 minus 19694

1198772= 0997

119910 = 34040119863 minus 16816

1198772= 0983

TNT mdash mdash 119910 = 84385 ln119863+ 259951198772= 0988

119910 = 53987119863 minus 99810

1198772= 0983

TT mdash mdash 119910 = 52070119863 minus 30026

1198772= 0986

119910 = 3147211989041415119863

1198772= 0966

000

050

100

150

0 5 10 15 2024-DNT (mgL)

Eucli

dean

dist

ance

D

(a)

000

005

010

015

020

0 20 40 60 80 100TNT (mgL)

Eucli

dean

dist

ance

D

(b)

Figure 4 Relationship between concentration of two analytes and measured Euclidean colorimetric distance for potassium hydroxide (a)and creatinine (b)

variation range obtained for a given reaction this reactionis considered as a possible solution On the contrary thisreaction is discarded as well as the presence of this analyteAfter this process several reactions may still be solutionsof the problem case For the final selection of the properreaction it is taken into account that according to Table 1each analyte produces at least two color variations in thesensor array that is there are at least two reactants thatchange their color when they are immersed in the analyte

sample This fact allows using at least two fitting equationsfor the determination of each analyte If the result of thesepredictions for the given values of 119863 is the same theobtained value of 119910 is taken as the correct prediction of thecorresponding analyte We have checked that this procedureresults in univocal solution in the analyzed framework

This separation algorithm has been implemented in thecore of the Raspberry Pi board and it has been tested topredict the presence and concentration of the analytes as

Journal of Sensors 7

Analyte selection

Analyte discarded

No

Evaluation of the fitting equations

Yes

Same solution

Analyte discarded

No

Analyte identified

Yes

Analyte determined

Measurement of Di (4 values)

with variation ranges

Comparison of Di

corresponding range

Every Di inside

Figure 5 Flux diagram of the nitroaromatic identification and determination algorithm

it is detailed below Good results are achieved proving thatthe portable instrument presented here is useful to detect thepresence of a variety of nitroaromatic compounds

35 Validation and Specifications The resolution and sensi-tivity of the system have been theoretically calculated fromthe expressions of Table 1 The sensitivity 119878 is evaluated asthe derivative of the fitting function for the nitroaromaticcompound 119910

119878119910=120597119891 (119863)

120597119863 (2)

where 119891(119863) represents any of the fitting functions in Table 1The resolution can be evaluated from the sensitivity as the

product of this parameter by the error in the determinationof the variable119863 [33]

Δ119910 = 119878119910sdot Δ119863 (3)

From (1) the resolution of the Euclidean distance119863 is

Δ119863 =120597119863

120597R1015840ΔR1015840 + 120597119863120597R10158400

ΔR10158400+120597119863

120597G1015840ΔG1015840 + 120597119863120597G10158400

ΔG10158400

+120597119863

120597B1015840ΔB1015840 + 120597119863120597B10158400

ΔB10158400

(4)

And the resolution of the normalized color coordinates iscalculated as

1198831015840=119883

119883ref997904rArr Δ119883

1015840=1

119883ref(1 +119883

119883ref)Δ119883 (5)

where 119883 = R G B and Δ119883 = Δ119883ref is the resolution inthe measurement of the color coordinate (R G or B) This isgiven by the resolution of the microcamera which generatesimages with 8 bits per channel Therefore Δ119883 is taken as 128= 1256

Taking into account that for the calculus of the analytesconcentration at least two functions can be used the resolu-tion in its determination is considered as the worst case ofthese two possibilities

The limit of detection (LOD) is calculated using the stan-dard criteria LOD = 119910

119887+ 3119904119887 where 119910

119887is the average blank

signal and 119904119887is the standard deviation of the blank which is

determined using ten replicas [34] The values of sensitivityresolution and LOD achieved for the determination of thenitroaromatic compounds are listed in Table 2 The units ofboth the resolution and sensitivity are the same since thedistance119863 used in (1)ndash(3) is a dimensionless parameter

As in every electronic system some noise is presentthat can affect the measurement In this case two main

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

Journal of Sensors 7

Analyte selection

Analyte discarded

No

Evaluation of the fitting equations

Yes

Same solution

Analyte discarded

No

Analyte identified

Yes

Analyte determined

Measurement of Di (4 values)

with variation ranges

Comparison of Di

corresponding range

Every Di inside

Figure 5 Flux diagram of the nitroaromatic identification and determination algorithm

it is detailed below Good results are achieved proving thatthe portable instrument presented here is useful to detect thepresence of a variety of nitroaromatic compounds

35 Validation and Specifications The resolution and sensi-tivity of the system have been theoretically calculated fromthe expressions of Table 1 The sensitivity 119878 is evaluated asthe derivative of the fitting function for the nitroaromaticcompound 119910

119878119910=120597119891 (119863)

120597119863 (2)

where 119891(119863) represents any of the fitting functions in Table 1The resolution can be evaluated from the sensitivity as the

product of this parameter by the error in the determinationof the variable119863 [33]

Δ119910 = 119878119910sdot Δ119863 (3)

From (1) the resolution of the Euclidean distance119863 is

Δ119863 =120597119863

120597R1015840ΔR1015840 + 120597119863120597R10158400

ΔR10158400+120597119863

120597G1015840ΔG1015840 + 120597119863120597G10158400

ΔG10158400

+120597119863

120597B1015840ΔB1015840 + 120597119863120597B10158400

ΔB10158400

(4)

And the resolution of the normalized color coordinates iscalculated as

1198831015840=119883

119883ref997904rArr Δ119883

1015840=1

119883ref(1 +119883

119883ref)Δ119883 (5)

where 119883 = R G B and Δ119883 = Δ119883ref is the resolution inthe measurement of the color coordinate (R G or B) This isgiven by the resolution of the microcamera which generatesimages with 8 bits per channel Therefore Δ119883 is taken as 128= 1256

Taking into account that for the calculus of the analytesconcentration at least two functions can be used the resolu-tion in its determination is considered as the worst case ofthese two possibilities

The limit of detection (LOD) is calculated using the stan-dard criteria LOD = 119910

119887+ 3119904119887 where 119910

119887is the average blank

signal and 119904119887is the standard deviation of the blank which is

determined using ten replicas [34] The values of sensitivityresolution and LOD achieved for the determination of thenitroaromatic compounds are listed in Table 2 The units ofboth the resolution and sensitivity are the same since thedistance119863 used in (1)ndash(3) is a dimensionless parameter

As in every electronic system some noise is presentthat can affect the measurement In this case two main

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

8 Journal of Sensors

Table 2 Theoretical resolution sensitivity and limit of detection

Nitroaromaticcompound Resolution (120583gL) Sensitivity (mgL) LOD

(mgL)

Picric acid 32 (at 1mgL)07 (at 10mgL)

326 (at 1mgL)387 (at 10mgL) 023

24-DNT 65 (at 1mgL)67 (at 20mgL)

301 (at 1mgL)576 (at 20mgL) 013

13-DNB 07 (at 1mgL)14 (at 20mgL)

384 (at 1mgL)784 (at 20mgL) 023

35-DNBN 48 (at 1mgL)65 (at 10mgL)

374 (at 1mgL)374 (at 10mgL) 014

2-Cl-35-DNT 120 (at 1mgL)12 (at 20mgL)

3416 (at 1mgL)3416 (at 20mgL) 114

TNB 719 (at 1mgL)27 (at 100mgL)

3404 (at 1mgL)3404 (at 100mgL) 051

TNT 442 (at 1mgL)37 (at 100mgL)

17063 (at 1mgL)5555 (at 100mgL) 18

TT 10 (at 1mgL)206 (at 100mgL)

5207 (at 1mgL)34754 (at 100mgL) 011

sources of electronic noise should be considered On onehand small oscillations in the biasing of the white LEDsare used for the illumination of the sensors On the otherhand the noise is induced by the CMOSmicrocamera eitherelectric or quantification noise These sources of noise havebeen evaluated resulting that the high stabilization in thebiasing of the LEDs allows neglecting any fluctuation in theillumination in addition the noise induced by the digitalmicrocamera including random noise ldquofixed patternrdquo noiseand banding noise limits the specifications of the camera witha signal to noise ratio of 36 dB

In addition to the electronic noise some environmentalnoise in form of light interference can also affect the mea-surements That is the reason to enclose the whole system ina dark box

The presented specifications show a good performanceof the developed instrument for the detection and deter-mination of a variety of nitroaromatic compounds A verygood resolution in the order of 120583gL is achieved in everycase as well as very low LOD The detection limits obtainedare in the expected range for optical sensors includingcolorimetric sensors [24 35] or fluorescence quenching basedsensors [36ndash38] Although other techniques for nitroaro-matics sensing provide better LODs below 10 ppb such assurface plasmon resonance [39] or mass spectrometry [40]they require very complex instrumentation that makes thedevelopment of simple and portable systems based on themnonviable

In order to demonstrate the proper functioning of thedeveloped system several nitroaromatic standard solutionsof known concentration were tested and their concentrationwas predicted using the presented system The results arepresented in Table 3 where the prediction value is themean of four replicas and the last row shows the standarddeviation of these replicas Although the results presentedin Table 3 show accuracy lower than the expected one at

the view of the resolution data of Table 2 this is due to thenonidealities in the sensor array preparation This processis handmade and misalignments heterogeneities shadowsand other effects can occur therefore the response of theinstrument is degraded

As it can be observed the recovery rate defined as theratio between the prediction and the real concentrationobtained from the samples used for testing purpose variesfrom 963 for 2-Cl-35-DNT at 50mgL to 1060 forpicric acid at 5mgL Therefore our measurement systemis capable of determining accurately different nitroaromaticcompoundswith a low-cost and portable hardware displayingthe results in a smartphone

4 Conclusions

In this work a portable measurement system based onimage processing for the detection and determination ofnitroaromatic compounds explosives or precursors of explo-sives is presented The prototype includes a Raspberry Piboard as the core of the instrument This device controlsa microcamera used for the acquisition of an image of thearray of chemical nonselective sensors sensitive to severalnitroaromatic compounds that show a color variation in thepresence of these analytes In addition the image processingrequired to quantify the color variations in the sensingmembranes is carried out in the Raspberry Pi board Onlyfour different sensors are used in the array for the detectionof a large number of nitroaromatic compounds from whicheight are evaluated here A simple and reliable algorithm forthe detection and determination of single nitroaromatic com-pounds in mixtures is developed therefore avoiding complexmultivariate analysis techniques The results obtained fromthe measurements are sent via WiFi to a remote user bymeans of a custom developed application for Android-baseddevices such as tablets and smartphones A high resolutionin the order of 120583gL can be achieved with this instrumentassuming a good repeatability in the fabrication of the sensorsarray Even in the case of handmade fabrication and depo-sition of the sensors array where the repeatability is poora good prediction of nitroaromatic compounds has beenachieved

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by Projects PIN-17-2014 fromthe CEMIX-UGR CTQ2013-44545-R from the Ministeriode Economıa y Competitividad (Spain) and P10-FQM-5974from the Junta de Andalucıa (Proyecto de Excelencia) Theseprojects were partially supported by European RegionalDevelopment Funds (ERDF)

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

Journal of Sensors 9

Table 3 Validation values

Nitroaromatic compound Concentration (mgL) Prediction (mgL) Recovery rate () Standard deviation (mgL)

Picric acid 50 49 1060 06100 107 996 11

24-DNT 50 52 978 06200 201 992 04

13-DNB 100 99 986 06

35-DNBN 10 10 1003 01100 100 1004 03

2-Cl-35-DNT 50 48 963 03

TNB 250 262 996 19125 122 1001 09

TNT 1000 1011 1005 14TT 500 505 1010 12

References

[1] V George T F Jenkins D C Leggett et al ldquoProgress ondetermining the vapor signature of a buried landminerdquo inDetection and Remediation Technologies for Mines and MinelikeTargets IV vol 3710 of Proceedings of the SPIE pp 258ndash269April 1999

[2] J Yinon Ed Advances in Analysis and Detection of ExplosivesProceedings of the 4th International Symposium on Analysis andDetection of Explosives September 7ndash10 1992 Jerusalem IsraelSpringer Berlin Germany 1992

[3] E G Kayser and N E Burlinson ldquoMigration of explosives insoil analysis of RDX TNT and tetryl from a 14c lysimeterstudyrdquo Journal of Energetic Materials vol 6 no 1-2 pp 45ndash711988

[4] R F Spalding and J W Fulton ldquoGroundwater munitionresidues and nitrate near Grand Island Nebraska USArdquoJournal of Contaminant Hydrology vol 2 no 2 pp 139ndash1531988

[5] A J Palazzo and D C Leggett ldquoEffect and disposition of TNTin a terrestrial plantrdquo Journal of Environmental Quality vol 15no 1 pp 49ndash52 1986

[6] D L Pugh Milan Army Ammunition Plant ContaminationSurvey Aberdeen Proving Ground Maryland USA Toxic andHazardous Materials Agency 1982

[7] P K Sekhar J Zhou H Wang and E R Hamblin ldquoTracedetection of pentaerythritol tetranitrate using electrochemicalgas sensorsrdquo Journal of Sensors vol 2014 Article ID 234607 6pages 2014

[8] X Fu R F Benson J Wang and D Fries ldquoRemote underwaterelectrochemical sensing system for detecting explosive residuesin the fieldrdquo Sensors and Actuators B Chemical vol 106 no 1pp 296ndash301 2005

[9] US Department of Homeland Securit Basic Research FocusAreasMay 2009 US Government PrintingOfficeWashingtonDC USA 2009

[10] S-A Barshick ldquoAnalysis of accelerants and fire debris usingaroma detection technologyrdquo Journal of Forensic Sciences vol43 no 2 pp 284ndash293 1998

[11] R G Ewing D A Atkinson G A Eiceman and G J Ewing ldquoAcritical review of ion mobility spectrometry for the detection ofexplosives and explosive related compoundsrdquo Talanta vol 54no 3 pp 515ndash529 2001

[12] D D Fetterolf and T D Clark ldquoDetection of trace explosiveevidence by ion mobility spectrometryrdquo Journal of ForensicSciences vol 38 no 1 pp 28ndash39 1993

[13] J A Caulfield T J Bruno andK EMiller ldquoEnthalpy of solutionand kovats retention indices for nitroaromatic compoundson stationary phases using gas chromatographyrdquo Journal ofChemicalampEngineeringData vol 54 no 6 pp 1814ndash1822 2009

[14] G W Cook P T LaPuma G L Hook and B A EckenrodeldquoUsing gas chromatography with ion mobility spectrometry toresolve explosive compounds in the presence of interferentsrdquoJournal of Forensic Sciences vol 55 no 6 pp 1582ndash1591 2010

[15] R G Cooks Z Ouyang Z Takats and J M WisemanldquoAmbient mass spectrometryrdquo Science vol 311 no 5767 pp1566ndash1570 2006

[16] Y Zhang X Ma S Zhang C Yang Z Ouyang and XZhang ldquoDirect detection of explosives on solid surfaces by lowtemperature plasma desorption mass spectrometryrdquo Analystvol 134 no 1 pp 176ndash181 2009

[17] A N Garroway M L Buess J B Miller et al ldquoRemotesensing by nuclear quadrupole resonancerdquo IEEE Transactionson Geoscience and Remote Sensing vol 39 no 6 pp 1108ndash11182001

[18] R D Luggar M J Farquharson J A Horrocks and R J LaceyldquoMultivariate analysis of statistically poor EDXRD spectra forthe detection of concealed explosivesrdquo X-Ray Spectrometry vol27 no 2 pp 87ndash94 1998

[19] J M Sylvia J A Janni J D Klein and K M Spencer ldquoSurface-enhanced Raman detection of 24-dinitrotoluene impurityvapor as a marker to locate landminesrdquo Analytical Chemistryvol 72 no 23 pp 5834ndash5840 2000

[20] A Tao F Kim C Hess et al ldquoLangmuir-Blodgett silvernanowire monolayers for molecular sensing using surface-enhanced Raman spectroscopyrdquo Nano Letters vol 3 no 9 pp1229ndash1233 2003

[21] J S Caygill F Davis and S P J Higson ldquoCurrent trends inexplosive detection techniquesrdquoTalanta vol 88 pp 14ndash29 2012

[22] P C Jurs G A Bakken and H E McClelland ldquoComputationalmethods for the analysis of chemical sensor array data fromvolatile analytesrdquo Chemical Reviews vol 100 no 7 pp 2649ndash2678 2000

[23] E G Kayser andN E BurlinsonMigration of Explosives in SoilNaval Surface Weapons Center 1982

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

10 Journal of Sensors

[24] M O Salles G N Meloni W R de Araujo and T R L CPaixao ldquoExplosive colorimetric discrimination using a smart-phone paper device and chemometrical approachrdquo AnalyticalMethods vol 6 no 7 pp 2047ndash2052 2014

[25] S S Gitis andA Y Kaminskii ldquoJanovsky 120590-complexesrdquoUspekhiKhimii vol 47 pp 1970ndash2013 1978

[26] A R Butler ldquoThe Jaffe reaction II Kinetic study of theJanovsky complexes formed from creatinine (2-imino-1-methylimazolidin-4-one) and acetonerdquo Journal of the ChemicalSociety Perkin Transactions 2 no 8 pp 853ndash857 1975

[27] T F Jenkins and M E Walsh ldquoDevelopment of field screeningmethods for TNT 24-DNT and RDX in soilrdquo Talanta vol 39no 4 pp 419ndash428 1992

[28] R Jenkins and H J Yallop ldquoThe identification of explosives intrace quantities on objets near an explosionrdquo Explosivstoffe vol6 pp 139ndash141 1970

[29] M Marshall and J C Oxley Aspects of Explosives DetectionElsevier Oxford UK 2009

[30] L F Capitan-Vallvey N Lopez-Ruiz A Martınez-Olmos MM Erenas and A J Palma ldquoRecent developments in computervision-based analytical chemistry a tutorial reviewrdquo AnalyticaChimica Acta vol 899 pp 23ndash56 2015

[31] K L Diehl and E V Anslyn ldquoArray sensing using opticalmethods for detection of chemical and biological hazardsrdquoChemical Society Reviews vol 42 no 22 pp 8596ndash8611 2013

[32] Y Xin Q Wang T Liu L Wang J Li and Y Fang ldquoA portableand autonomousmultichannel fluorescence detector for on-lineand in situ explosive detection in aqueous phaserdquo Lab on a Chipvol 12 no 22 pp 4821ndash4828 2012

[33] N Lopez-Ruiz D Hernandez-Belanger M A Carvajal L FCapitan-Vallvey A J Palma and AMartınez-Olmos ldquoFast life-time and amplitude determination in luminescence exponentialdecaysrdquo Sensors and Actuators B Chemical vol 216 pp 595ndash602 2015

[34] N Lopez-Ruiz A Martınez-Olmos I M Perez De Vargas-Sansalvador et al ldquoDetermination of O

2using colour sensing

from image processing with mobile devicesrdquo Sensors andActuators B Chemical vol 171-172 pp 938ndash945 2012

[35] W Zhu J S Park J L Sessler and A Gaitas ldquoA colorimetricreceptor combined with a microcantilever sensor for explosivevapor detectionrdquo Applied Physics Letters vol 98 no 12 ArticleID 123501 2011

[36] P Baran Pati and S S Zade ldquoHighly emissive triphenylaminebased fluorophores for detection of picric acidrdquo TetrahedronLetters vol 55 no 38 pp 5290ndash5293 2014

[37] I S Kovalev O S Taniya N V Slovesnova et al ldquoFluorescentdetection of 24-DNTand 246-TNT in aqueousmedia by usingsimple water-soluble pyrene derivativesrdquo Chemistry vol 11 no5 pp 775ndash781 2016

[38] F Chu G Tsiminis N A Spooner and T M Monro ldquoExplo-sives detection by fluorescence quenching of conjugated poly-mers in suspended core optical fibersrdquo Sensors and Actuators BChemical vol 199 pp 22ndash26 2014

[39] K Nagatomo T Kawaguchi N Miura K Toko and KMatsumoto ldquoDevelopment of a sensitive surface plasmon reso-nance immunosensor for detection of 24-dinitrotoluene with anovel oligo (ethylene glycol)-based sensor surfacerdquoTalanta vol79 no 4 pp 1142ndash1148 2009

[40] S Saha M Kanti Mandal L Chuin Chen S Ninomiya YShida and K Hiraoka ldquoTrace level detection of explosivesin solution using leidenfrost phenomenon assisted thermal

desorption ambient mass spectrometryrdquo Mass Spectrometryvol 2 Article ID S0008 5 pages 2013

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of