research article application of temperature modulation-sdp...
TRANSCRIPT
Research ArticleApplication of Temperature Modulation-SDP on MOS GasSensors Capturing Soil Gaseous Profile for Discrimination ofSoil under Different Nutrient Addition
Arief Sudarmaji12 and Akio Kitagawa1
1Electrical Engineering and Computer Science Kanazawa University Kakuma-machi Kanazawa Ishikawa 920-1192 Japan2Agricultural Engineering University of Jenderal Soedirman Jalan Dr Soeparno Karangwangkal PurwokertoCentral Java 53123 Indonesia
Correspondence should be addressed to Arief Sudarmaji ariefunsoedgmailcom
Received 24 January 2016 Accepted 28 February 2016
Academic Editor Kourosh Kalantar-Zadeh
Copyright copy 2016 A Sudarmaji and A Kitagawa This is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited
A technique of temperature modulation-SDP (specified detection point) on MOS gas sensors was designed and tested on theirsensing performance to such complex mixture soil gaseous compound And a self-made e-nose was built to capture and analyzethe gaseous profile from sampling headspace of two soils (sandy loam and sand) with the addition of nutrient at different dose(without normal and high addition) It comprises (a) 6 MOS gas sensors which were driven wirelessly on a certain modulationthrough (b) a PSoC CY8C28445-24PVXI-based interface and (c) the Principal Component Analysis (PCA) and neural network(NN) as pattern recognition tools The gaseous compounds are accumulated in a static headspace with thermostatting and stirringunder controlled condition to optimize equilibration and gases concentration as well The patterns are trained by backpropagationalgorithm which employs a log-sigmoid function and updates the weights using search-then-converge schedule PCA resultsindicate that the sensor array used is able to differentiate the soil type clearly and may provide a discrimination as a response topresencelevel of the nutrients addition in soil Additionally the PCA enhances the classification performance ofNN to discriminateamong the predescribed nutrient additions
1 Introduction
One promising technique that may overcome the cross-selectivity problem on MOS gas sensor and also increase itssensitivity is temperature modulation [1] It alters the kineticsof the sensor due to a cyclic variation of operational tem-perature resulting from the changes of the working voltageperiodically The sensor temperature is determined by theamplitude of the voltage applied across its heater The cyclickinetics of MOS gas sensor will lead to different rates of reac-tion which seem to follow the givenmodulation so as to pro-vide a unique response for each gas dependent on either shapeor amplitude of modulation [1ndash3] The operating modulationvoltage may change periodically by either squarerectangularor sine waveform or triangularsaw-tooth [3]
Temperature modulation has successfully enhanced theselectivity of (thickthin) SnO
2in distinguishing a single or
multicomponent by applying certain modulation or analysisfor optimum result as reported in [4ndash6] even on catalytic gassensor [7] Moreover it also succeeded on the TGSs [8ndash11]In the early investigation of this technique pulsethermal-cycling technique was used which mostly was applied onTGSs as reported by Lee and Reedy [1] Most of investigationsutilized the modulation generator and digital acquisitionsignal recorder separately It is difficult to take the advantagesof temperature modulation for field in situ measurement
We introduced a technique namely the temperaturemodulation with specified detection point (temperaturemodulation-SDP) (see Section 2) which is able be appliedto drive a singlearray of MOS gas sensors [12] By applying
Hindawi Publishing CorporationJournal of SensorsVolume 2016 Article ID 1035902 11 pageshttpdxdoiorg10115520161035902
2 Journal of Sensors
Static
Modulated
MOS sensor
VC
VH
Vo
R1
RH RS
(a)
MOS sensor
SDP
Modulated
VC
VH
Modulated
R1
RH RS
of Vo
(b)
Detection point
75 middle of
VC
VH
SVH duty cycle
(c)
FIS
TGS
Detection point
(d)
Figure 1 Principle and basic schematic of (a) general temperature modulation and (b) temperature modulation-SDP with (c) the desirerectangular modulation and (d) the generated waveform modulation
rectangular temperature modulation-SDP we tested 6 MOSgas sensors (3 TGSs and 3 FISs) in a PSoC-based e-nosesystem where the result shows that there is a significantincrement of selectivity in discriminating 3 single volatilecompounds (toluene ethanol and ammonia) compared withstatic temperature operation [12]
Since those 6 MOS gas sensors (TGS2444 TGS2602TGS825 FISAQ1 FISSB30 and FIS12A) are designed by theirmanufacturers to sense the volatile compounds to furthertest for the sensitivity and selectivity we aim to assess thetemperature modulation-SDP in distinguishing a such com-plex compound in variable conditions We therefore testedthe performance of the rectangular temperature modulation-SDP for evaluating the influence of soil type and nutrientaddition on their responses Soils a complex mixture arecomposed mostly of minerals and organic materials waterair and countless organisms [13 14] Many gases mostlyvolatile organic compounds are found at soil atmosphere duetomicrobial activity in which the type and the concentrationsof VOCs produced may differ because of differences incommunity composition or nutrient availability [15ndash17] Soil
also is known to have a unique smell that can be sensedwith human olfaction system resulting from two specialmolecules (geosmin and methylisoborneol) due to the activ-ity of bacteria mostly belonging to the genus Streptomyces[18 19] We tested two soils (sandy loam and sand) with thefollowing addition of commercial compost in different dose(without normal and high)
2 The Temperature Modulation with SpecifiedDetection Point (SDP)
In principle the temperature modulation-SDP is similar togeneral temperature modulation (Figure 1(a)) yet besidesa modulation on heater unit (119881
119867) it also modulates the
sensing unit (119881119862) concurrently and in the same phase with
119881119867(Figure 1(b)) The 119881
119862is positioned on midpoint 75 of
ldquoonhighrdquo state of 119881119867(Figure 1(c)) The SDP means that the
time of output detection of MOS gas sensor is put at specifiedpoint The SDP ensures the same measurement point at eachoutput shape Moreover the 119881
119862which is associated with 119881
119867
may lead to prevention of the sensor from possible migration
Journal of Sensors 3
Table 1 MOS gas sensors used and typical gas target claimed
Number Sensor Gas target Working range1 TGS2444 Ammonia 1ndash100 ppm [20]2 TGS2602 Air contaminant 1ndash30 ppm of EtOH [44]3 TGS825 Hydrogen sulfide 5ndash100 ppm [36]4 FIS12A Methane 300ndash7000 ppm [45]5 FIS30SB Alcohol 1ndash100 ppm [46]6 FISAQ1 VOC (air quality) 10ndash10000 ppm [47]
To arraysensor
To arraysensor
VC
VH
VOH
VOC
Vo
SVH
SVC
R1
RH RS
(a)
To arraysensor
To arraysensor
VC
VH
VOH
VOC
Vo
SVH
SVC
R1
RH RS
(b)
Figure 2 The schematic of temperature modulation-SDP for array TGSs (a) and FISs (b) 119881119867and 119881
119862are static voltage 119881
119862is sensing circuit
voltage 119878119881119867
is modulation signal for 119881119867 and 119878
119881119862is modulation signal for 119881
119862
of heater materials into the sensing material which couldcause long term drift of sensingmaterialrsquos resistance to highervalues [20]
This technique allows a single chip (such as controllerprocessor or hybrid) to get the advantages of temperaturemodulation by concomitantly generating modulation signaland acquiring the output at a constant point as well in the chipitself even when using many MOS gas sensors Generally ahandheld device employs a single chip processor or controlleras the heart of system and its time consumption dependson the clock used and complexity of tasks and featuresinvolved (sequential multiplexing digitalanalog conversionIO handling timer interruption communicating with outerdevice etc) Lower-end chip will spendmore time Howeverthe flexibility to be set as a custom-developed system couldactually be an advantage in an application like a sensorshandling [21 22]
In this study we put on a rectangular modulation and asingle detection point atmiddle ofmodulation of sensing unitas shown in Figure 1(c) Figure 1(d) shows that the waveformsmodulation (captured byOscilloscope Tektronix TDS 2024B)at heater unit (yellow) and sensing unit (green) and the zoneof detection point of overall MOS gas sensors used (purple)meet the desired modulation in Figure 1(c)
Wedesigned the schematic of single temperaturemodula-tion-SDP for each array of TGSs and FISs respectively as
shown in Figure 2 since there is slight difference in config-uration on them It employs common modulation circuitsemploying FET (Field Effect Transistor) In particular onTGS244 we constructed an individual modulation circuitbecause it requires a recommended modulation as notedin its datasheet [20] Both TGSs and FISs are configuredin voltage divider as standard technique for measuringresistance changes [23]
3 The Self-Made E-Nose
We built a PSoC-based e-nose that consists of 3 main units(1) sensing unit 6 MOS gas sensors (Table 1) which areexpected to sense the soil volatile compounds since they arespecified to detect a particular volatile compounds in lowconcentration range and 2 environment sensors (LM35 andHSM30G) tomonitor temperature and humidity in chamber(2) a PSoC CY8C28445-24PVXI-based interface system and(3) PCA and NN as preprocessing and pattern recognitionrespectively
As shown in Figure 3 a single CY8C28445-24PVXI usedacts as a core of system which mainly functioned to generatedesired modulation signals to acquire all sensors outputand to communicate with computer wirelessly It connectsthrough radiofrequency using XBee (IEEE 802154) serialcommunication interfaced by a developed program under
4 Journal of Sensors
AMux
TGS2444
CPU
Out
TGS 825FISAQ1
TGS 2602
FISSB30FIS12A
HSM30GLM53
TX
RXXBee
XBee
PC
CY8C28445-24PVXI
MO
S se
nsor
out
put
FISscircuit
TGSscircuit
Envi
ronm
ent
sens
orou
tput
Modulationgenerator
ModulationgeneratorTimer8_2
Timer8_1
ADC_1
ADC_2
VH driverVC driver
PSOC1 24MHz
SVH_FIS
SVC_FIS
SVH_TGS
SVC_TGS
Figure 3 Block diagram of PSoC-based e-nose system for capturing soil gaseous profile
Visual BasicNet 2012 We configured some analog blocks(PGAs Multiplexer ADCs and Switched Capacitor) anddigital (Timer8 Counter and PWM for ADC and UART)blocks inside the PSoC to comply with the functions Formore detailed diagram and configuration of the PSoC referto our previous works [12 24] The PSoC firmware was builtusing PSoC Designer 54
We developed the software of PCA and NN using VisualStudio 2012 to analyze the profiles of the array sensorresponses corresponding to the soil samples The PCAsoftware is constructed by utilizing PCA routine in open-source AccordNET Framework 210 The NN was developedbased on backpropagation (BP) learning method in Multi-layer Perceptron Neural Network (MLPNN) architecture byemploying a log-sigmoid activation functionTheweights areupdated using global adapted learning parameter 120578 updatedby search-then-converge schedule It is a simple and non-adaptive annealing schedule Typically it starts with large 120578and gradually decreases as the learning proceeds in which theprocess of adapting 120578 is similar to that in simulated annealing[25] Basically the BP algorithm is a generalization of thedelta rule (Least-Mean Squares algorithm) also called thegeneralized delta rule which uses a gradient search techniqueto minimize a cost function equivalent to the Mean SquareError (MSE) between actual network outputs and the desired(target) output [25] The BP propagates the MSE backwardthrough the network and the weights (and biases) are thenadjusted by a gradient descent based algorithm Thus aclosed-loop control system is established in network
4 Material and Method
41 Soil Preparation and Sample Handling The sandy loamand sand soil were derived from the top 15 cmand landwithoutprior soil management Sandy clay loam soil was taken fromland around Kanazawa University (36∘32101584046338010158401015840N136∘42101584011545210158401015840E) while sand soil was taken fromaround coastal area of Uchinada Beach (36∘381015840391910158401015840N
136∘371015840378810158401015840E) a sand hill on Sea of Japan which is locatedabout 17 km from Kanazawa University The collected soilsamples were crushed and sieved manually at lt2mm afterplant derbies turfs and gravels were carefully removedAs soil treatments we added an amount of fermentationcompost The compost is given at averagenormal and highdoses as recommended in practical application that is 20and 30 tons haminus1 DM (Dry Matter) respectively [26] Takinginto account a general assumption that in 1 ha soil area15 cm deep that contains 2Mkg despite bulk density of soilvarying considerably [27] we therefore added the compostat 0 15 and 225mgg soil sample which approached dosesof 0 20 and 30 tons haminus1 DM respectively
The soil and compost samples were put into LLDPE(linear low-density polyethylene) plastic bag and sealed withparaffin Then we stored them in refrigerator at 5 plusmn 05∘Cto inactivate microbial activity in soil This temperatureis known as biologic zero temperature which recognizedthat most microbes in soil become relatively inactive attemperature below 5∘C [28 29] Prior to being used thesamples were air-dried up to room temperature
We prepared the samples into solution since soil containsmany soluble substances in water and liquid has biggerdiffusion coefficient than solid and thus leads to shorterdiffusion times We calculate the mass of soil sample using(1) to obtain the mass of pure water and compost additionwhere119898
119904expressesmass of soil (119892)119881V is volume of headspace
vial (mL) 120588119904is bulk density of soil (sandy loam = 144 gmL
and sand = 152 gmL [30 31]) 120588119908is density of pure water
= 0998 gmL 120573 (119881119866119881119878) is phase ratio in SH and 119908
119888is
water content (in fractional number) Table 2 resumes theproperties of parameters used and calculation results
119898119904=
119881V times 120588119904 times 120588119908
(120573 + 1) times (120588119908+ 119908119888times 120588119904) (1)
42 Measurement Procedures The soil gaseous compoundsare accumulated in a static headspace (SH) and the headspace
Journal of Sensors 5
Table 2 Properties of samples of soil fertilizer water and staticheadspace condition
Properties of SH ValueVolume of SH vial 90mLBulk density of sandy loam soil 144 gmLBulk density of sand soil 152 gmLPhase ratio 15Water content 100Density of pure water 0998 gmLCalculation resultsMass of sandy loam soil 2122 g
(i) Mass of compost adding at 20 tonha 0318 g(ii) Mass of compost adding at 30 tonha 0477 g
Mass of sand soil 2163 g(i) Mass of compost adding at 20 tonha 0324 g(ii) Mass of compost adding at 30 tonha 0287 g
(rpm)Stir
PowerHeat
Hot Top
Alcohol thermometer
Magnetic bar Water
Offminus +
Offminus +CorningPC-420D
(∘C)
Figure 4 Headspace conditioning with heating and stirring usingCorning PC-420D in SH sampling the layout of Corning modifiedfrom [32]
equilibration is optimized by both agitating (ie stirring)and thermostatting concurrently for all samples on thesame phase ratio We set 30 minutes 60∘C and 200 rpmof equilibration time temperature and stirring frequencyrespectively We utilized Corning PC-4200D to heat and stirthe sample in the SH vialWe used 90mL glass container withsealed cap as headspace vial which is put inside the 500mLopen beaker filled with 100mL water (Figure 4) It aims to
maintain the equilibrium relative humidity the same as thesoil sample And the headspacing was conducted inside aroom under controlled temperature By those ways all soilsamples were under the same treatments and environmentalconditions
The temperature modulation is set on 025Hz 75 dutycycle to drive all MOS gas sensors except for TGS2444 [20]which is on its recommended duty cycle The initial actionof the MOS gas sensors after a long inactive state is carriedout for one hour ofmeasuring the reference gas to allow themto reach a stable condition The gas sensors are expressed inresistance and the profile is defined by its sensitivity (119878) [4]where119877
0is sensor resistance of air and119877
119892is sensor resistance
of soil gaseous compound (see (2))
119878 =1198770
119877119892
(2)
The measurement of soil gaseous profiles is performedusing close measurement method by switching between thereference gas (filtered air with silica gel) as baseline andanalyte gas (soil gaseous compounds) The flow directionand rate of gas are controlled by 3-way valve and the Koflocmass flow controller (MFC) respectively The MFC are set at03 lpmAs shown in Figure 5 the reference gas flows throughpoint (a) (valve-1) point (c) (valve-2) and point (e) (valve-3)while the analyte gas flows through point (b) (valve-1) point(d) (valve-2) and point (e) (valve-3) The purging of sensorchamber was in open measurement mode by disconnectingthe hose of inlet pump fromvalve-2 directing valve-3 to point(f) and turning on the purge pump
At preresearch we observed 119877119892for 5 minutes after 119877
0
measurement to determine the response of each sensorand obtain the best starting measurement time for 119877
119892
measurement Significantly we found that overall sensorsreached a stable state after plusmn150 s (plusmn25min) which stronglyindicate that they are sensing stably the flow of gas thathave been spread evenly in the close measurement systemWe therefore took this time to be the starting point of 119877
119892
measurement Thus we set the total measurement time persample as 37 minutes covering the phases of the headspace(30 minutes) 119877
0measurement (1 minute) stabling time
(25 minutes) 119877119892measurement (1 minute) and purging (5
minutes) sequentially The sampling period of both 1198770and
119877119892measurement was 2 seconds and their averages were used
to represent the baseline and soil gaseous compound
5 Results and Discussion
51 Individual Sensitivity-Based Response ofMOSGas SensorsIndividual sensitivity-based soil gaseous profiles of MOS gassensors used on each soil type with the different dose of nutri-ent addition are shown in Figure 6 It reveals that the arrayof gas sensors was able to sense the soil gases andor volatilecompounds resulting from different samples and as well indi-cates that the method of the optimized SH seems suitable forprovidingaccumulating the concentration sufficientlyThoseindividual responses indicate that the technique of temper-ature modulation-SDP led the sensors to sense differently
6 Journal of Sensors
MOS array sensor
Gas Gasinlet outlet
Heaterand stirrer Purge pump
Valve-3
Valve-1
Watercontainer
Sensor chamberSoilcontainer
Inlet pumpMFC-1
MFC mass flow controller Interface
Silicagel
PSOC1(CY8C28445)
MFC-2
Valve-2(a)
(b)
(c)
(d)
(e)
(f) Waste
Figure 5 Experimental setup to capture the soil gaseous compounds using static headspace extraction in sample flow system (close)measurement
Sand SandSandSand Sand Sand Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
TGS2444 TGS2602 TGS825 FISAQ1 FISSB30 FIS12A
08
1
12
14
16
18
2
Sens
itivi
ty
20 tha30 tha
0 tha
Figure 6 Individual sensitivity of sensor the average and standard deviation of 5 replicates to 3 levels of compost addition in different soil
the amounts and types of soil gaseous compounds producedand released inside the SH atmosphere which correspondedto the soil type and doses of nutrient addition MoreoverFigure 6 also presents the standard deviation of the MOSsensors to five replicates of each measurement It relativelyshows the low variance among responses which indicates thesufficient consistency of sensors reproducibility in producingthe soil gaseous profiles on the same environment treatmentindependently throughout this study
As shown in Figure 6 formost of theMOS gas sensors butTGS2602 the sensitivity to the nutrient addition (20 tha and30 tha) was higher than without nutrient addition whetherfor the same soil type or between sandy loam and sand Sandyloam soil usually has more holding capacity of water andnutrient alongwith lower bulk density compared to sand soil
thus leading to having more organic matter content [31 33]andmicroorganism [34] In addition the use of a flow system(usually employing a pump) in sample detection causescooling of the sensor surface reducing the high increment oftemperature and humidity inside such sensor chamber (heatdissipation) [35] thus also influencing its response
Interestingly on TGS825 which is technically designedto respond to the hydrogen sulfide (H
2S) [36] it had the
highest sensitivity among the others for each soil type Itreveals that the H
2S concentration during the headspace
process was high and it is seen that the presence of nutrientaddition contributed significantly to H
2S accumulation in
the headspace (Figure 7) The response indicates that thereis much acid sulfate material in soil samples This gas canbe produced from the oxidation process of organic material
Journal of Sensors 7
SandSandy loam
12
14
16
18
2
22
Sens
itivi
ty
20 tha 30 tha0 tha 20 tha 30 tha0 tha
Figure 7 Response variances of TGS825 for five replicates betweensandy loam and sand soil in different dose of nutrient addition
containing sulfate acid due to bacterial activities in lowoxygen environment (like flooded soil) which depends onambient conditions such as temperature humidity and theconcentration of certainmetal ions [37]The result also showsthat the additional nutrient in sandy loam soil providedrelatively higher concentration than in sand soil and therewas a little cross-response in differentiating level of compostaddition between doses 20 tha and 30 tha
The operation of temperature modulation-SDP throughoscillating the heater voltage by square modulation does notonly cause altering the kinetics of both adsorption and reac-tion process at the surface of sensor (effect of the frequency)but also consequently lead the MOS gas sensor to run atlower effective temperature (effect of the duty cycle) as onthe TGS2444 which is driven by low duty cycle modulation[20] and shown to have high selectivity to ammonia gas [12]For particularmaterial the specificworking temperature pro-vides optimum sensitivity for sensing a certain gas evidently[38 39] Ou et al [39] found that under the low workingtemperature (ie 120∘C) a 2D metal disulfide-based gassensor has very high selectivity to NO
2in which the sensing
mechanism is dominated by charge transfer adsorptionbetween the surface-adsorbed NO
2gas molecules and metal
disulfide strongly due to paramagnetic behavior of NO2
Thus the combination of frequency and effective workingvoltage by duty cycle selection of temperature modulation-SDP had potential to sense sensitively the complex gas andorvolatile compounds of soil which then provide the uniquegaseous profiles
However like typical characteristic of the use of sensorarray in e-nose which does not allow individual sensor toidentify a specific or complex volatile compounds we foundthat there was no single sensor used which individuallyshowed a relation for characterization of the difference of soilconditions clearly and linearly with regard to soil type andnutrient additionThere was a cross-response on each sensorin differentiating the dose level of nutrient addition espe-cially between normal dose (20 tha) and high dose (30 tha)The complexity of soil gaseous compounds in potentiallyvarious kinds of gases especially volatile compounds [16 17]causes an inevitable cross-response onMOSgas sensor as also
PCA of sandy loam versus sand soil
Sandy loamSand
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus005minus02 minus01minus015 005 01 015 020Component 1
Figure 8 PCA plot showing discrimination between 2 soils withoutnutrient addition
founded by Rincon et al [40] who simulated a monitoringof VOC as soil contaminants through measuring 8 kindsof gases The cross-response of individual sensor may bereduced by projecting collectively into new dimension usingPCA as commonly used in e-nose
52 Performance of Discrimination of Soil under DifferentNutrient Addition Thepotential of nonparametric biologicalsystem for discriminating soil type as well as for differ-entiating between different nutrient additions treatmentsbased on its gaseous profile was tested Firstly the PCA asa nonsupervised technique was employed to find generalrelationships between samples while preserving most of thevariance within data PCA allow projecting variables ontofewer dimensions reflecting the most relevant analyticalinformation [41] This offers an advantage that the classifi-cation of unknowns is processed much faster thus reducingdetection time
Figure 8 shows the PCA plot of discrimination of twosoils both without addition of compost It shows a distinctzone of patterns volatile production between sandy loamsoil and sand soil where the principal component- (PC-) 1accounts for higher differentiation of cluster than PC-2 PC-1 and PC-2 cumulatively account for 7832 of the variancewithin the data set
Meanwhile Figure 9 shows the PCA plot for replicatesof each soil sample in distinguishing three doses of compostaddition It seem that PCA allow discriminating distinctlybetween soil conditions whether with or without compost(nutrient) addition indicated by separated blue zone evenwhen differentiating regardless of soil type (Figure 9(c))
It was only for sandy loam soil (Figure 9(a)) the levelof compost addition could be clustered clearly into threegroups as predefined previously while there was misiden-tification between soils with dose 20 tha and dose 30 tha
8 Journal of Sensors
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
Static
Modulated
MOS sensor
VC
VH
Vo
R1
RH RS
(a)
MOS sensor
SDP
Modulated
VC
VH
Modulated
R1
RH RS
of Vo
(b)
Detection point
75 middle of
VC
VH
SVH duty cycle
(c)
FIS
TGS
Detection point
(d)
Figure 1 Principle and basic schematic of (a) general temperature modulation and (b) temperature modulation-SDP with (c) the desirerectangular modulation and (d) the generated waveform modulation
rectangular temperature modulation-SDP we tested 6 MOSgas sensors (3 TGSs and 3 FISs) in a PSoC-based e-nosesystem where the result shows that there is a significantincrement of selectivity in discriminating 3 single volatilecompounds (toluene ethanol and ammonia) compared withstatic temperature operation [12]
Since those 6 MOS gas sensors (TGS2444 TGS2602TGS825 FISAQ1 FISSB30 and FIS12A) are designed by theirmanufacturers to sense the volatile compounds to furthertest for the sensitivity and selectivity we aim to assess thetemperature modulation-SDP in distinguishing a such com-plex compound in variable conditions We therefore testedthe performance of the rectangular temperature modulation-SDP for evaluating the influence of soil type and nutrientaddition on their responses Soils a complex mixture arecomposed mostly of minerals and organic materials waterair and countless organisms [13 14] Many gases mostlyvolatile organic compounds are found at soil atmosphere duetomicrobial activity in which the type and the concentrationsof VOCs produced may differ because of differences incommunity composition or nutrient availability [15ndash17] Soil
also is known to have a unique smell that can be sensedwith human olfaction system resulting from two specialmolecules (geosmin and methylisoborneol) due to the activ-ity of bacteria mostly belonging to the genus Streptomyces[18 19] We tested two soils (sandy loam and sand) with thefollowing addition of commercial compost in different dose(without normal and high)
2 The Temperature Modulation with SpecifiedDetection Point (SDP)
In principle the temperature modulation-SDP is similar togeneral temperature modulation (Figure 1(a)) yet besidesa modulation on heater unit (119881
119867) it also modulates the
sensing unit (119881119862) concurrently and in the same phase with
119881119867(Figure 1(b)) The 119881
119862is positioned on midpoint 75 of
ldquoonhighrdquo state of 119881119867(Figure 1(c)) The SDP means that the
time of output detection of MOS gas sensor is put at specifiedpoint The SDP ensures the same measurement point at eachoutput shape Moreover the 119881
119862which is associated with 119881
119867
may lead to prevention of the sensor from possible migration
Journal of Sensors 3
Table 1 MOS gas sensors used and typical gas target claimed
Number Sensor Gas target Working range1 TGS2444 Ammonia 1ndash100 ppm [20]2 TGS2602 Air contaminant 1ndash30 ppm of EtOH [44]3 TGS825 Hydrogen sulfide 5ndash100 ppm [36]4 FIS12A Methane 300ndash7000 ppm [45]5 FIS30SB Alcohol 1ndash100 ppm [46]6 FISAQ1 VOC (air quality) 10ndash10000 ppm [47]
To arraysensor
To arraysensor
VC
VH
VOH
VOC
Vo
SVH
SVC
R1
RH RS
(a)
To arraysensor
To arraysensor
VC
VH
VOH
VOC
Vo
SVH
SVC
R1
RH RS
(b)
Figure 2 The schematic of temperature modulation-SDP for array TGSs (a) and FISs (b) 119881119867and 119881
119862are static voltage 119881
119862is sensing circuit
voltage 119878119881119867
is modulation signal for 119881119867 and 119878
119881119862is modulation signal for 119881
119862
of heater materials into the sensing material which couldcause long term drift of sensingmaterialrsquos resistance to highervalues [20]
This technique allows a single chip (such as controllerprocessor or hybrid) to get the advantages of temperaturemodulation by concomitantly generating modulation signaland acquiring the output at a constant point as well in the chipitself even when using many MOS gas sensors Generally ahandheld device employs a single chip processor or controlleras the heart of system and its time consumption dependson the clock used and complexity of tasks and featuresinvolved (sequential multiplexing digitalanalog conversionIO handling timer interruption communicating with outerdevice etc) Lower-end chip will spendmore time Howeverthe flexibility to be set as a custom-developed system couldactually be an advantage in an application like a sensorshandling [21 22]
In this study we put on a rectangular modulation and asingle detection point atmiddle ofmodulation of sensing unitas shown in Figure 1(c) Figure 1(d) shows that the waveformsmodulation (captured byOscilloscope Tektronix TDS 2024B)at heater unit (yellow) and sensing unit (green) and the zoneof detection point of overall MOS gas sensors used (purple)meet the desired modulation in Figure 1(c)
Wedesigned the schematic of single temperaturemodula-tion-SDP for each array of TGSs and FISs respectively as
shown in Figure 2 since there is slight difference in config-uration on them It employs common modulation circuitsemploying FET (Field Effect Transistor) In particular onTGS244 we constructed an individual modulation circuitbecause it requires a recommended modulation as notedin its datasheet [20] Both TGSs and FISs are configuredin voltage divider as standard technique for measuringresistance changes [23]
3 The Self-Made E-Nose
We built a PSoC-based e-nose that consists of 3 main units(1) sensing unit 6 MOS gas sensors (Table 1) which areexpected to sense the soil volatile compounds since they arespecified to detect a particular volatile compounds in lowconcentration range and 2 environment sensors (LM35 andHSM30G) tomonitor temperature and humidity in chamber(2) a PSoC CY8C28445-24PVXI-based interface system and(3) PCA and NN as preprocessing and pattern recognitionrespectively
As shown in Figure 3 a single CY8C28445-24PVXI usedacts as a core of system which mainly functioned to generatedesired modulation signals to acquire all sensors outputand to communicate with computer wirelessly It connectsthrough radiofrequency using XBee (IEEE 802154) serialcommunication interfaced by a developed program under
4 Journal of Sensors
AMux
TGS2444
CPU
Out
TGS 825FISAQ1
TGS 2602
FISSB30FIS12A
HSM30GLM53
TX
RXXBee
XBee
PC
CY8C28445-24PVXI
MO
S se
nsor
out
put
FISscircuit
TGSscircuit
Envi
ronm
ent
sens
orou
tput
Modulationgenerator
ModulationgeneratorTimer8_2
Timer8_1
ADC_1
ADC_2
VH driverVC driver
PSOC1 24MHz
SVH_FIS
SVC_FIS
SVH_TGS
SVC_TGS
Figure 3 Block diagram of PSoC-based e-nose system for capturing soil gaseous profile
Visual BasicNet 2012 We configured some analog blocks(PGAs Multiplexer ADCs and Switched Capacitor) anddigital (Timer8 Counter and PWM for ADC and UART)blocks inside the PSoC to comply with the functions Formore detailed diagram and configuration of the PSoC referto our previous works [12 24] The PSoC firmware was builtusing PSoC Designer 54
We developed the software of PCA and NN using VisualStudio 2012 to analyze the profiles of the array sensorresponses corresponding to the soil samples The PCAsoftware is constructed by utilizing PCA routine in open-source AccordNET Framework 210 The NN was developedbased on backpropagation (BP) learning method in Multi-layer Perceptron Neural Network (MLPNN) architecture byemploying a log-sigmoid activation functionTheweights areupdated using global adapted learning parameter 120578 updatedby search-then-converge schedule It is a simple and non-adaptive annealing schedule Typically it starts with large 120578and gradually decreases as the learning proceeds in which theprocess of adapting 120578 is similar to that in simulated annealing[25] Basically the BP algorithm is a generalization of thedelta rule (Least-Mean Squares algorithm) also called thegeneralized delta rule which uses a gradient search techniqueto minimize a cost function equivalent to the Mean SquareError (MSE) between actual network outputs and the desired(target) output [25] The BP propagates the MSE backwardthrough the network and the weights (and biases) are thenadjusted by a gradient descent based algorithm Thus aclosed-loop control system is established in network
4 Material and Method
41 Soil Preparation and Sample Handling The sandy loamand sand soil were derived from the top 15 cmand landwithoutprior soil management Sandy clay loam soil was taken fromland around Kanazawa University (36∘32101584046338010158401015840N136∘42101584011545210158401015840E) while sand soil was taken fromaround coastal area of Uchinada Beach (36∘381015840391910158401015840N
136∘371015840378810158401015840E) a sand hill on Sea of Japan which is locatedabout 17 km from Kanazawa University The collected soilsamples were crushed and sieved manually at lt2mm afterplant derbies turfs and gravels were carefully removedAs soil treatments we added an amount of fermentationcompost The compost is given at averagenormal and highdoses as recommended in practical application that is 20and 30 tons haminus1 DM (Dry Matter) respectively [26] Takinginto account a general assumption that in 1 ha soil area15 cm deep that contains 2Mkg despite bulk density of soilvarying considerably [27] we therefore added the compostat 0 15 and 225mgg soil sample which approached dosesof 0 20 and 30 tons haminus1 DM respectively
The soil and compost samples were put into LLDPE(linear low-density polyethylene) plastic bag and sealed withparaffin Then we stored them in refrigerator at 5 plusmn 05∘Cto inactivate microbial activity in soil This temperatureis known as biologic zero temperature which recognizedthat most microbes in soil become relatively inactive attemperature below 5∘C [28 29] Prior to being used thesamples were air-dried up to room temperature
We prepared the samples into solution since soil containsmany soluble substances in water and liquid has biggerdiffusion coefficient than solid and thus leads to shorterdiffusion times We calculate the mass of soil sample using(1) to obtain the mass of pure water and compost additionwhere119898
119904expressesmass of soil (119892)119881V is volume of headspace
vial (mL) 120588119904is bulk density of soil (sandy loam = 144 gmL
and sand = 152 gmL [30 31]) 120588119908is density of pure water
= 0998 gmL 120573 (119881119866119881119878) is phase ratio in SH and 119908
119888is
water content (in fractional number) Table 2 resumes theproperties of parameters used and calculation results
119898119904=
119881V times 120588119904 times 120588119908
(120573 + 1) times (120588119908+ 119908119888times 120588119904) (1)
42 Measurement Procedures The soil gaseous compoundsare accumulated in a static headspace (SH) and the headspace
Journal of Sensors 5
Table 2 Properties of samples of soil fertilizer water and staticheadspace condition
Properties of SH ValueVolume of SH vial 90mLBulk density of sandy loam soil 144 gmLBulk density of sand soil 152 gmLPhase ratio 15Water content 100Density of pure water 0998 gmLCalculation resultsMass of sandy loam soil 2122 g
(i) Mass of compost adding at 20 tonha 0318 g(ii) Mass of compost adding at 30 tonha 0477 g
Mass of sand soil 2163 g(i) Mass of compost adding at 20 tonha 0324 g(ii) Mass of compost adding at 30 tonha 0287 g
(rpm)Stir
PowerHeat
Hot Top
Alcohol thermometer
Magnetic bar Water
Offminus +
Offminus +CorningPC-420D
(∘C)
Figure 4 Headspace conditioning with heating and stirring usingCorning PC-420D in SH sampling the layout of Corning modifiedfrom [32]
equilibration is optimized by both agitating (ie stirring)and thermostatting concurrently for all samples on thesame phase ratio We set 30 minutes 60∘C and 200 rpmof equilibration time temperature and stirring frequencyrespectively We utilized Corning PC-4200D to heat and stirthe sample in the SH vialWe used 90mL glass container withsealed cap as headspace vial which is put inside the 500mLopen beaker filled with 100mL water (Figure 4) It aims to
maintain the equilibrium relative humidity the same as thesoil sample And the headspacing was conducted inside aroom under controlled temperature By those ways all soilsamples were under the same treatments and environmentalconditions
The temperature modulation is set on 025Hz 75 dutycycle to drive all MOS gas sensors except for TGS2444 [20]which is on its recommended duty cycle The initial actionof the MOS gas sensors after a long inactive state is carriedout for one hour ofmeasuring the reference gas to allow themto reach a stable condition The gas sensors are expressed inresistance and the profile is defined by its sensitivity (119878) [4]where119877
0is sensor resistance of air and119877
119892is sensor resistance
of soil gaseous compound (see (2))
119878 =1198770
119877119892
(2)
The measurement of soil gaseous profiles is performedusing close measurement method by switching between thereference gas (filtered air with silica gel) as baseline andanalyte gas (soil gaseous compounds) The flow directionand rate of gas are controlled by 3-way valve and the Koflocmass flow controller (MFC) respectively The MFC are set at03 lpmAs shown in Figure 5 the reference gas flows throughpoint (a) (valve-1) point (c) (valve-2) and point (e) (valve-3)while the analyte gas flows through point (b) (valve-1) point(d) (valve-2) and point (e) (valve-3) The purging of sensorchamber was in open measurement mode by disconnectingthe hose of inlet pump fromvalve-2 directing valve-3 to point(f) and turning on the purge pump
At preresearch we observed 119877119892for 5 minutes after 119877
0
measurement to determine the response of each sensorand obtain the best starting measurement time for 119877
119892
measurement Significantly we found that overall sensorsreached a stable state after plusmn150 s (plusmn25min) which stronglyindicate that they are sensing stably the flow of gas thathave been spread evenly in the close measurement systemWe therefore took this time to be the starting point of 119877
119892
measurement Thus we set the total measurement time persample as 37 minutes covering the phases of the headspace(30 minutes) 119877
0measurement (1 minute) stabling time
(25 minutes) 119877119892measurement (1 minute) and purging (5
minutes) sequentially The sampling period of both 1198770and
119877119892measurement was 2 seconds and their averages were used
to represent the baseline and soil gaseous compound
5 Results and Discussion
51 Individual Sensitivity-Based Response ofMOSGas SensorsIndividual sensitivity-based soil gaseous profiles of MOS gassensors used on each soil type with the different dose of nutri-ent addition are shown in Figure 6 It reveals that the arrayof gas sensors was able to sense the soil gases andor volatilecompounds resulting from different samples and as well indi-cates that the method of the optimized SH seems suitable forprovidingaccumulating the concentration sufficientlyThoseindividual responses indicate that the technique of temper-ature modulation-SDP led the sensors to sense differently
6 Journal of Sensors
MOS array sensor
Gas Gasinlet outlet
Heaterand stirrer Purge pump
Valve-3
Valve-1
Watercontainer
Sensor chamberSoilcontainer
Inlet pumpMFC-1
MFC mass flow controller Interface
Silicagel
PSOC1(CY8C28445)
MFC-2
Valve-2(a)
(b)
(c)
(d)
(e)
(f) Waste
Figure 5 Experimental setup to capture the soil gaseous compounds using static headspace extraction in sample flow system (close)measurement
Sand SandSandSand Sand Sand Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
TGS2444 TGS2602 TGS825 FISAQ1 FISSB30 FIS12A
08
1
12
14
16
18
2
Sens
itivi
ty
20 tha30 tha
0 tha
Figure 6 Individual sensitivity of sensor the average and standard deviation of 5 replicates to 3 levels of compost addition in different soil
the amounts and types of soil gaseous compounds producedand released inside the SH atmosphere which correspondedto the soil type and doses of nutrient addition MoreoverFigure 6 also presents the standard deviation of the MOSsensors to five replicates of each measurement It relativelyshows the low variance among responses which indicates thesufficient consistency of sensors reproducibility in producingthe soil gaseous profiles on the same environment treatmentindependently throughout this study
As shown in Figure 6 formost of theMOS gas sensors butTGS2602 the sensitivity to the nutrient addition (20 tha and30 tha) was higher than without nutrient addition whetherfor the same soil type or between sandy loam and sand Sandyloam soil usually has more holding capacity of water andnutrient alongwith lower bulk density compared to sand soil
thus leading to having more organic matter content [31 33]andmicroorganism [34] In addition the use of a flow system(usually employing a pump) in sample detection causescooling of the sensor surface reducing the high increment oftemperature and humidity inside such sensor chamber (heatdissipation) [35] thus also influencing its response
Interestingly on TGS825 which is technically designedto respond to the hydrogen sulfide (H
2S) [36] it had the
highest sensitivity among the others for each soil type Itreveals that the H
2S concentration during the headspace
process was high and it is seen that the presence of nutrientaddition contributed significantly to H
2S accumulation in
the headspace (Figure 7) The response indicates that thereis much acid sulfate material in soil samples This gas canbe produced from the oxidation process of organic material
Journal of Sensors 7
SandSandy loam
12
14
16
18
2
22
Sens
itivi
ty
20 tha 30 tha0 tha 20 tha 30 tha0 tha
Figure 7 Response variances of TGS825 for five replicates betweensandy loam and sand soil in different dose of nutrient addition
containing sulfate acid due to bacterial activities in lowoxygen environment (like flooded soil) which depends onambient conditions such as temperature humidity and theconcentration of certainmetal ions [37]The result also showsthat the additional nutrient in sandy loam soil providedrelatively higher concentration than in sand soil and therewas a little cross-response in differentiating level of compostaddition between doses 20 tha and 30 tha
The operation of temperature modulation-SDP throughoscillating the heater voltage by square modulation does notonly cause altering the kinetics of both adsorption and reac-tion process at the surface of sensor (effect of the frequency)but also consequently lead the MOS gas sensor to run atlower effective temperature (effect of the duty cycle) as onthe TGS2444 which is driven by low duty cycle modulation[20] and shown to have high selectivity to ammonia gas [12]For particularmaterial the specificworking temperature pro-vides optimum sensitivity for sensing a certain gas evidently[38 39] Ou et al [39] found that under the low workingtemperature (ie 120∘C) a 2D metal disulfide-based gassensor has very high selectivity to NO
2in which the sensing
mechanism is dominated by charge transfer adsorptionbetween the surface-adsorbed NO
2gas molecules and metal
disulfide strongly due to paramagnetic behavior of NO2
Thus the combination of frequency and effective workingvoltage by duty cycle selection of temperature modulation-SDP had potential to sense sensitively the complex gas andorvolatile compounds of soil which then provide the uniquegaseous profiles
However like typical characteristic of the use of sensorarray in e-nose which does not allow individual sensor toidentify a specific or complex volatile compounds we foundthat there was no single sensor used which individuallyshowed a relation for characterization of the difference of soilconditions clearly and linearly with regard to soil type andnutrient additionThere was a cross-response on each sensorin differentiating the dose level of nutrient addition espe-cially between normal dose (20 tha) and high dose (30 tha)The complexity of soil gaseous compounds in potentiallyvarious kinds of gases especially volatile compounds [16 17]causes an inevitable cross-response onMOSgas sensor as also
PCA of sandy loam versus sand soil
Sandy loamSand
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus005minus02 minus01minus015 005 01 015 020Component 1
Figure 8 PCA plot showing discrimination between 2 soils withoutnutrient addition
founded by Rincon et al [40] who simulated a monitoringof VOC as soil contaminants through measuring 8 kindsof gases The cross-response of individual sensor may bereduced by projecting collectively into new dimension usingPCA as commonly used in e-nose
52 Performance of Discrimination of Soil under DifferentNutrient Addition Thepotential of nonparametric biologicalsystem for discriminating soil type as well as for differ-entiating between different nutrient additions treatmentsbased on its gaseous profile was tested Firstly the PCA asa nonsupervised technique was employed to find generalrelationships between samples while preserving most of thevariance within data PCA allow projecting variables ontofewer dimensions reflecting the most relevant analyticalinformation [41] This offers an advantage that the classifi-cation of unknowns is processed much faster thus reducingdetection time
Figure 8 shows the PCA plot of discrimination of twosoils both without addition of compost It shows a distinctzone of patterns volatile production between sandy loamsoil and sand soil where the principal component- (PC-) 1accounts for higher differentiation of cluster than PC-2 PC-1 and PC-2 cumulatively account for 7832 of the variancewithin the data set
Meanwhile Figure 9 shows the PCA plot for replicatesof each soil sample in distinguishing three doses of compostaddition It seem that PCA allow discriminating distinctlybetween soil conditions whether with or without compost(nutrient) addition indicated by separated blue zone evenwhen differentiating regardless of soil type (Figure 9(c))
It was only for sandy loam soil (Figure 9(a)) the levelof compost addition could be clustered clearly into threegroups as predefined previously while there was misiden-tification between soils with dose 20 tha and dose 30 tha
8 Journal of Sensors
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
Table 1 MOS gas sensors used and typical gas target claimed
Number Sensor Gas target Working range1 TGS2444 Ammonia 1ndash100 ppm [20]2 TGS2602 Air contaminant 1ndash30 ppm of EtOH [44]3 TGS825 Hydrogen sulfide 5ndash100 ppm [36]4 FIS12A Methane 300ndash7000 ppm [45]5 FIS30SB Alcohol 1ndash100 ppm [46]6 FISAQ1 VOC (air quality) 10ndash10000 ppm [47]
To arraysensor
To arraysensor
VC
VH
VOH
VOC
Vo
SVH
SVC
R1
RH RS
(a)
To arraysensor
To arraysensor
VC
VH
VOH
VOC
Vo
SVH
SVC
R1
RH RS
(b)
Figure 2 The schematic of temperature modulation-SDP for array TGSs (a) and FISs (b) 119881119867and 119881
119862are static voltage 119881
119862is sensing circuit
voltage 119878119881119867
is modulation signal for 119881119867 and 119878
119881119862is modulation signal for 119881
119862
of heater materials into the sensing material which couldcause long term drift of sensingmaterialrsquos resistance to highervalues [20]
This technique allows a single chip (such as controllerprocessor or hybrid) to get the advantages of temperaturemodulation by concomitantly generating modulation signaland acquiring the output at a constant point as well in the chipitself even when using many MOS gas sensors Generally ahandheld device employs a single chip processor or controlleras the heart of system and its time consumption dependson the clock used and complexity of tasks and featuresinvolved (sequential multiplexing digitalanalog conversionIO handling timer interruption communicating with outerdevice etc) Lower-end chip will spendmore time Howeverthe flexibility to be set as a custom-developed system couldactually be an advantage in an application like a sensorshandling [21 22]
In this study we put on a rectangular modulation and asingle detection point atmiddle ofmodulation of sensing unitas shown in Figure 1(c) Figure 1(d) shows that the waveformsmodulation (captured byOscilloscope Tektronix TDS 2024B)at heater unit (yellow) and sensing unit (green) and the zoneof detection point of overall MOS gas sensors used (purple)meet the desired modulation in Figure 1(c)
Wedesigned the schematic of single temperaturemodula-tion-SDP for each array of TGSs and FISs respectively as
shown in Figure 2 since there is slight difference in config-uration on them It employs common modulation circuitsemploying FET (Field Effect Transistor) In particular onTGS244 we constructed an individual modulation circuitbecause it requires a recommended modulation as notedin its datasheet [20] Both TGSs and FISs are configuredin voltage divider as standard technique for measuringresistance changes [23]
3 The Self-Made E-Nose
We built a PSoC-based e-nose that consists of 3 main units(1) sensing unit 6 MOS gas sensors (Table 1) which areexpected to sense the soil volatile compounds since they arespecified to detect a particular volatile compounds in lowconcentration range and 2 environment sensors (LM35 andHSM30G) tomonitor temperature and humidity in chamber(2) a PSoC CY8C28445-24PVXI-based interface system and(3) PCA and NN as preprocessing and pattern recognitionrespectively
As shown in Figure 3 a single CY8C28445-24PVXI usedacts as a core of system which mainly functioned to generatedesired modulation signals to acquire all sensors outputand to communicate with computer wirelessly It connectsthrough radiofrequency using XBee (IEEE 802154) serialcommunication interfaced by a developed program under
4 Journal of Sensors
AMux
TGS2444
CPU
Out
TGS 825FISAQ1
TGS 2602
FISSB30FIS12A
HSM30GLM53
TX
RXXBee
XBee
PC
CY8C28445-24PVXI
MO
S se
nsor
out
put
FISscircuit
TGSscircuit
Envi
ronm
ent
sens
orou
tput
Modulationgenerator
ModulationgeneratorTimer8_2
Timer8_1
ADC_1
ADC_2
VH driverVC driver
PSOC1 24MHz
SVH_FIS
SVC_FIS
SVH_TGS
SVC_TGS
Figure 3 Block diagram of PSoC-based e-nose system for capturing soil gaseous profile
Visual BasicNet 2012 We configured some analog blocks(PGAs Multiplexer ADCs and Switched Capacitor) anddigital (Timer8 Counter and PWM for ADC and UART)blocks inside the PSoC to comply with the functions Formore detailed diagram and configuration of the PSoC referto our previous works [12 24] The PSoC firmware was builtusing PSoC Designer 54
We developed the software of PCA and NN using VisualStudio 2012 to analyze the profiles of the array sensorresponses corresponding to the soil samples The PCAsoftware is constructed by utilizing PCA routine in open-source AccordNET Framework 210 The NN was developedbased on backpropagation (BP) learning method in Multi-layer Perceptron Neural Network (MLPNN) architecture byemploying a log-sigmoid activation functionTheweights areupdated using global adapted learning parameter 120578 updatedby search-then-converge schedule It is a simple and non-adaptive annealing schedule Typically it starts with large 120578and gradually decreases as the learning proceeds in which theprocess of adapting 120578 is similar to that in simulated annealing[25] Basically the BP algorithm is a generalization of thedelta rule (Least-Mean Squares algorithm) also called thegeneralized delta rule which uses a gradient search techniqueto minimize a cost function equivalent to the Mean SquareError (MSE) between actual network outputs and the desired(target) output [25] The BP propagates the MSE backwardthrough the network and the weights (and biases) are thenadjusted by a gradient descent based algorithm Thus aclosed-loop control system is established in network
4 Material and Method
41 Soil Preparation and Sample Handling The sandy loamand sand soil were derived from the top 15 cmand landwithoutprior soil management Sandy clay loam soil was taken fromland around Kanazawa University (36∘32101584046338010158401015840N136∘42101584011545210158401015840E) while sand soil was taken fromaround coastal area of Uchinada Beach (36∘381015840391910158401015840N
136∘371015840378810158401015840E) a sand hill on Sea of Japan which is locatedabout 17 km from Kanazawa University The collected soilsamples were crushed and sieved manually at lt2mm afterplant derbies turfs and gravels were carefully removedAs soil treatments we added an amount of fermentationcompost The compost is given at averagenormal and highdoses as recommended in practical application that is 20and 30 tons haminus1 DM (Dry Matter) respectively [26] Takinginto account a general assumption that in 1 ha soil area15 cm deep that contains 2Mkg despite bulk density of soilvarying considerably [27] we therefore added the compostat 0 15 and 225mgg soil sample which approached dosesof 0 20 and 30 tons haminus1 DM respectively
The soil and compost samples were put into LLDPE(linear low-density polyethylene) plastic bag and sealed withparaffin Then we stored them in refrigerator at 5 plusmn 05∘Cto inactivate microbial activity in soil This temperatureis known as biologic zero temperature which recognizedthat most microbes in soil become relatively inactive attemperature below 5∘C [28 29] Prior to being used thesamples were air-dried up to room temperature
We prepared the samples into solution since soil containsmany soluble substances in water and liquid has biggerdiffusion coefficient than solid and thus leads to shorterdiffusion times We calculate the mass of soil sample using(1) to obtain the mass of pure water and compost additionwhere119898
119904expressesmass of soil (119892)119881V is volume of headspace
vial (mL) 120588119904is bulk density of soil (sandy loam = 144 gmL
and sand = 152 gmL [30 31]) 120588119908is density of pure water
= 0998 gmL 120573 (119881119866119881119878) is phase ratio in SH and 119908
119888is
water content (in fractional number) Table 2 resumes theproperties of parameters used and calculation results
119898119904=
119881V times 120588119904 times 120588119908
(120573 + 1) times (120588119908+ 119908119888times 120588119904) (1)
42 Measurement Procedures The soil gaseous compoundsare accumulated in a static headspace (SH) and the headspace
Journal of Sensors 5
Table 2 Properties of samples of soil fertilizer water and staticheadspace condition
Properties of SH ValueVolume of SH vial 90mLBulk density of sandy loam soil 144 gmLBulk density of sand soil 152 gmLPhase ratio 15Water content 100Density of pure water 0998 gmLCalculation resultsMass of sandy loam soil 2122 g
(i) Mass of compost adding at 20 tonha 0318 g(ii) Mass of compost adding at 30 tonha 0477 g
Mass of sand soil 2163 g(i) Mass of compost adding at 20 tonha 0324 g(ii) Mass of compost adding at 30 tonha 0287 g
(rpm)Stir
PowerHeat
Hot Top
Alcohol thermometer
Magnetic bar Water
Offminus +
Offminus +CorningPC-420D
(∘C)
Figure 4 Headspace conditioning with heating and stirring usingCorning PC-420D in SH sampling the layout of Corning modifiedfrom [32]
equilibration is optimized by both agitating (ie stirring)and thermostatting concurrently for all samples on thesame phase ratio We set 30 minutes 60∘C and 200 rpmof equilibration time temperature and stirring frequencyrespectively We utilized Corning PC-4200D to heat and stirthe sample in the SH vialWe used 90mL glass container withsealed cap as headspace vial which is put inside the 500mLopen beaker filled with 100mL water (Figure 4) It aims to
maintain the equilibrium relative humidity the same as thesoil sample And the headspacing was conducted inside aroom under controlled temperature By those ways all soilsamples were under the same treatments and environmentalconditions
The temperature modulation is set on 025Hz 75 dutycycle to drive all MOS gas sensors except for TGS2444 [20]which is on its recommended duty cycle The initial actionof the MOS gas sensors after a long inactive state is carriedout for one hour ofmeasuring the reference gas to allow themto reach a stable condition The gas sensors are expressed inresistance and the profile is defined by its sensitivity (119878) [4]where119877
0is sensor resistance of air and119877
119892is sensor resistance
of soil gaseous compound (see (2))
119878 =1198770
119877119892
(2)
The measurement of soil gaseous profiles is performedusing close measurement method by switching between thereference gas (filtered air with silica gel) as baseline andanalyte gas (soil gaseous compounds) The flow directionand rate of gas are controlled by 3-way valve and the Koflocmass flow controller (MFC) respectively The MFC are set at03 lpmAs shown in Figure 5 the reference gas flows throughpoint (a) (valve-1) point (c) (valve-2) and point (e) (valve-3)while the analyte gas flows through point (b) (valve-1) point(d) (valve-2) and point (e) (valve-3) The purging of sensorchamber was in open measurement mode by disconnectingthe hose of inlet pump fromvalve-2 directing valve-3 to point(f) and turning on the purge pump
At preresearch we observed 119877119892for 5 minutes after 119877
0
measurement to determine the response of each sensorand obtain the best starting measurement time for 119877
119892
measurement Significantly we found that overall sensorsreached a stable state after plusmn150 s (plusmn25min) which stronglyindicate that they are sensing stably the flow of gas thathave been spread evenly in the close measurement systemWe therefore took this time to be the starting point of 119877
119892
measurement Thus we set the total measurement time persample as 37 minutes covering the phases of the headspace(30 minutes) 119877
0measurement (1 minute) stabling time
(25 minutes) 119877119892measurement (1 minute) and purging (5
minutes) sequentially The sampling period of both 1198770and
119877119892measurement was 2 seconds and their averages were used
to represent the baseline and soil gaseous compound
5 Results and Discussion
51 Individual Sensitivity-Based Response ofMOSGas SensorsIndividual sensitivity-based soil gaseous profiles of MOS gassensors used on each soil type with the different dose of nutri-ent addition are shown in Figure 6 It reveals that the arrayof gas sensors was able to sense the soil gases andor volatilecompounds resulting from different samples and as well indi-cates that the method of the optimized SH seems suitable forprovidingaccumulating the concentration sufficientlyThoseindividual responses indicate that the technique of temper-ature modulation-SDP led the sensors to sense differently
6 Journal of Sensors
MOS array sensor
Gas Gasinlet outlet
Heaterand stirrer Purge pump
Valve-3
Valve-1
Watercontainer
Sensor chamberSoilcontainer
Inlet pumpMFC-1
MFC mass flow controller Interface
Silicagel
PSOC1(CY8C28445)
MFC-2
Valve-2(a)
(b)
(c)
(d)
(e)
(f) Waste
Figure 5 Experimental setup to capture the soil gaseous compounds using static headspace extraction in sample flow system (close)measurement
Sand SandSandSand Sand Sand Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
TGS2444 TGS2602 TGS825 FISAQ1 FISSB30 FIS12A
08
1
12
14
16
18
2
Sens
itivi
ty
20 tha30 tha
0 tha
Figure 6 Individual sensitivity of sensor the average and standard deviation of 5 replicates to 3 levels of compost addition in different soil
the amounts and types of soil gaseous compounds producedand released inside the SH atmosphere which correspondedto the soil type and doses of nutrient addition MoreoverFigure 6 also presents the standard deviation of the MOSsensors to five replicates of each measurement It relativelyshows the low variance among responses which indicates thesufficient consistency of sensors reproducibility in producingthe soil gaseous profiles on the same environment treatmentindependently throughout this study
As shown in Figure 6 formost of theMOS gas sensors butTGS2602 the sensitivity to the nutrient addition (20 tha and30 tha) was higher than without nutrient addition whetherfor the same soil type or between sandy loam and sand Sandyloam soil usually has more holding capacity of water andnutrient alongwith lower bulk density compared to sand soil
thus leading to having more organic matter content [31 33]andmicroorganism [34] In addition the use of a flow system(usually employing a pump) in sample detection causescooling of the sensor surface reducing the high increment oftemperature and humidity inside such sensor chamber (heatdissipation) [35] thus also influencing its response
Interestingly on TGS825 which is technically designedto respond to the hydrogen sulfide (H
2S) [36] it had the
highest sensitivity among the others for each soil type Itreveals that the H
2S concentration during the headspace
process was high and it is seen that the presence of nutrientaddition contributed significantly to H
2S accumulation in
the headspace (Figure 7) The response indicates that thereis much acid sulfate material in soil samples This gas canbe produced from the oxidation process of organic material
Journal of Sensors 7
SandSandy loam
12
14
16
18
2
22
Sens
itivi
ty
20 tha 30 tha0 tha 20 tha 30 tha0 tha
Figure 7 Response variances of TGS825 for five replicates betweensandy loam and sand soil in different dose of nutrient addition
containing sulfate acid due to bacterial activities in lowoxygen environment (like flooded soil) which depends onambient conditions such as temperature humidity and theconcentration of certainmetal ions [37]The result also showsthat the additional nutrient in sandy loam soil providedrelatively higher concentration than in sand soil and therewas a little cross-response in differentiating level of compostaddition between doses 20 tha and 30 tha
The operation of temperature modulation-SDP throughoscillating the heater voltage by square modulation does notonly cause altering the kinetics of both adsorption and reac-tion process at the surface of sensor (effect of the frequency)but also consequently lead the MOS gas sensor to run atlower effective temperature (effect of the duty cycle) as onthe TGS2444 which is driven by low duty cycle modulation[20] and shown to have high selectivity to ammonia gas [12]For particularmaterial the specificworking temperature pro-vides optimum sensitivity for sensing a certain gas evidently[38 39] Ou et al [39] found that under the low workingtemperature (ie 120∘C) a 2D metal disulfide-based gassensor has very high selectivity to NO
2in which the sensing
mechanism is dominated by charge transfer adsorptionbetween the surface-adsorbed NO
2gas molecules and metal
disulfide strongly due to paramagnetic behavior of NO2
Thus the combination of frequency and effective workingvoltage by duty cycle selection of temperature modulation-SDP had potential to sense sensitively the complex gas andorvolatile compounds of soil which then provide the uniquegaseous profiles
However like typical characteristic of the use of sensorarray in e-nose which does not allow individual sensor toidentify a specific or complex volatile compounds we foundthat there was no single sensor used which individuallyshowed a relation for characterization of the difference of soilconditions clearly and linearly with regard to soil type andnutrient additionThere was a cross-response on each sensorin differentiating the dose level of nutrient addition espe-cially between normal dose (20 tha) and high dose (30 tha)The complexity of soil gaseous compounds in potentiallyvarious kinds of gases especially volatile compounds [16 17]causes an inevitable cross-response onMOSgas sensor as also
PCA of sandy loam versus sand soil
Sandy loamSand
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus005minus02 minus01minus015 005 01 015 020Component 1
Figure 8 PCA plot showing discrimination between 2 soils withoutnutrient addition
founded by Rincon et al [40] who simulated a monitoringof VOC as soil contaminants through measuring 8 kindsof gases The cross-response of individual sensor may bereduced by projecting collectively into new dimension usingPCA as commonly used in e-nose
52 Performance of Discrimination of Soil under DifferentNutrient Addition Thepotential of nonparametric biologicalsystem for discriminating soil type as well as for differ-entiating between different nutrient additions treatmentsbased on its gaseous profile was tested Firstly the PCA asa nonsupervised technique was employed to find generalrelationships between samples while preserving most of thevariance within data PCA allow projecting variables ontofewer dimensions reflecting the most relevant analyticalinformation [41] This offers an advantage that the classifi-cation of unknowns is processed much faster thus reducingdetection time
Figure 8 shows the PCA plot of discrimination of twosoils both without addition of compost It shows a distinctzone of patterns volatile production between sandy loamsoil and sand soil where the principal component- (PC-) 1accounts for higher differentiation of cluster than PC-2 PC-1 and PC-2 cumulatively account for 7832 of the variancewithin the data set
Meanwhile Figure 9 shows the PCA plot for replicatesof each soil sample in distinguishing three doses of compostaddition It seem that PCA allow discriminating distinctlybetween soil conditions whether with or without compost(nutrient) addition indicated by separated blue zone evenwhen differentiating regardless of soil type (Figure 9(c))
It was only for sandy loam soil (Figure 9(a)) the levelof compost addition could be clustered clearly into threegroups as predefined previously while there was misiden-tification between soils with dose 20 tha and dose 30 tha
8 Journal of Sensors
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
AMux
TGS2444
CPU
Out
TGS 825FISAQ1
TGS 2602
FISSB30FIS12A
HSM30GLM53
TX
RXXBee
XBee
PC
CY8C28445-24PVXI
MO
S se
nsor
out
put
FISscircuit
TGSscircuit
Envi
ronm
ent
sens
orou
tput
Modulationgenerator
ModulationgeneratorTimer8_2
Timer8_1
ADC_1
ADC_2
VH driverVC driver
PSOC1 24MHz
SVH_FIS
SVC_FIS
SVH_TGS
SVC_TGS
Figure 3 Block diagram of PSoC-based e-nose system for capturing soil gaseous profile
Visual BasicNet 2012 We configured some analog blocks(PGAs Multiplexer ADCs and Switched Capacitor) anddigital (Timer8 Counter and PWM for ADC and UART)blocks inside the PSoC to comply with the functions Formore detailed diagram and configuration of the PSoC referto our previous works [12 24] The PSoC firmware was builtusing PSoC Designer 54
We developed the software of PCA and NN using VisualStudio 2012 to analyze the profiles of the array sensorresponses corresponding to the soil samples The PCAsoftware is constructed by utilizing PCA routine in open-source AccordNET Framework 210 The NN was developedbased on backpropagation (BP) learning method in Multi-layer Perceptron Neural Network (MLPNN) architecture byemploying a log-sigmoid activation functionTheweights areupdated using global adapted learning parameter 120578 updatedby search-then-converge schedule It is a simple and non-adaptive annealing schedule Typically it starts with large 120578and gradually decreases as the learning proceeds in which theprocess of adapting 120578 is similar to that in simulated annealing[25] Basically the BP algorithm is a generalization of thedelta rule (Least-Mean Squares algorithm) also called thegeneralized delta rule which uses a gradient search techniqueto minimize a cost function equivalent to the Mean SquareError (MSE) between actual network outputs and the desired(target) output [25] The BP propagates the MSE backwardthrough the network and the weights (and biases) are thenadjusted by a gradient descent based algorithm Thus aclosed-loop control system is established in network
4 Material and Method
41 Soil Preparation and Sample Handling The sandy loamand sand soil were derived from the top 15 cmand landwithoutprior soil management Sandy clay loam soil was taken fromland around Kanazawa University (36∘32101584046338010158401015840N136∘42101584011545210158401015840E) while sand soil was taken fromaround coastal area of Uchinada Beach (36∘381015840391910158401015840N
136∘371015840378810158401015840E) a sand hill on Sea of Japan which is locatedabout 17 km from Kanazawa University The collected soilsamples were crushed and sieved manually at lt2mm afterplant derbies turfs and gravels were carefully removedAs soil treatments we added an amount of fermentationcompost The compost is given at averagenormal and highdoses as recommended in practical application that is 20and 30 tons haminus1 DM (Dry Matter) respectively [26] Takinginto account a general assumption that in 1 ha soil area15 cm deep that contains 2Mkg despite bulk density of soilvarying considerably [27] we therefore added the compostat 0 15 and 225mgg soil sample which approached dosesof 0 20 and 30 tons haminus1 DM respectively
The soil and compost samples were put into LLDPE(linear low-density polyethylene) plastic bag and sealed withparaffin Then we stored them in refrigerator at 5 plusmn 05∘Cto inactivate microbial activity in soil This temperatureis known as biologic zero temperature which recognizedthat most microbes in soil become relatively inactive attemperature below 5∘C [28 29] Prior to being used thesamples were air-dried up to room temperature
We prepared the samples into solution since soil containsmany soluble substances in water and liquid has biggerdiffusion coefficient than solid and thus leads to shorterdiffusion times We calculate the mass of soil sample using(1) to obtain the mass of pure water and compost additionwhere119898
119904expressesmass of soil (119892)119881V is volume of headspace
vial (mL) 120588119904is bulk density of soil (sandy loam = 144 gmL
and sand = 152 gmL [30 31]) 120588119908is density of pure water
= 0998 gmL 120573 (119881119866119881119878) is phase ratio in SH and 119908
119888is
water content (in fractional number) Table 2 resumes theproperties of parameters used and calculation results
119898119904=
119881V times 120588119904 times 120588119908
(120573 + 1) times (120588119908+ 119908119888times 120588119904) (1)
42 Measurement Procedures The soil gaseous compoundsare accumulated in a static headspace (SH) and the headspace
Journal of Sensors 5
Table 2 Properties of samples of soil fertilizer water and staticheadspace condition
Properties of SH ValueVolume of SH vial 90mLBulk density of sandy loam soil 144 gmLBulk density of sand soil 152 gmLPhase ratio 15Water content 100Density of pure water 0998 gmLCalculation resultsMass of sandy loam soil 2122 g
(i) Mass of compost adding at 20 tonha 0318 g(ii) Mass of compost adding at 30 tonha 0477 g
Mass of sand soil 2163 g(i) Mass of compost adding at 20 tonha 0324 g(ii) Mass of compost adding at 30 tonha 0287 g
(rpm)Stir
PowerHeat
Hot Top
Alcohol thermometer
Magnetic bar Water
Offminus +
Offminus +CorningPC-420D
(∘C)
Figure 4 Headspace conditioning with heating and stirring usingCorning PC-420D in SH sampling the layout of Corning modifiedfrom [32]
equilibration is optimized by both agitating (ie stirring)and thermostatting concurrently for all samples on thesame phase ratio We set 30 minutes 60∘C and 200 rpmof equilibration time temperature and stirring frequencyrespectively We utilized Corning PC-4200D to heat and stirthe sample in the SH vialWe used 90mL glass container withsealed cap as headspace vial which is put inside the 500mLopen beaker filled with 100mL water (Figure 4) It aims to
maintain the equilibrium relative humidity the same as thesoil sample And the headspacing was conducted inside aroom under controlled temperature By those ways all soilsamples were under the same treatments and environmentalconditions
The temperature modulation is set on 025Hz 75 dutycycle to drive all MOS gas sensors except for TGS2444 [20]which is on its recommended duty cycle The initial actionof the MOS gas sensors after a long inactive state is carriedout for one hour ofmeasuring the reference gas to allow themto reach a stable condition The gas sensors are expressed inresistance and the profile is defined by its sensitivity (119878) [4]where119877
0is sensor resistance of air and119877
119892is sensor resistance
of soil gaseous compound (see (2))
119878 =1198770
119877119892
(2)
The measurement of soil gaseous profiles is performedusing close measurement method by switching between thereference gas (filtered air with silica gel) as baseline andanalyte gas (soil gaseous compounds) The flow directionand rate of gas are controlled by 3-way valve and the Koflocmass flow controller (MFC) respectively The MFC are set at03 lpmAs shown in Figure 5 the reference gas flows throughpoint (a) (valve-1) point (c) (valve-2) and point (e) (valve-3)while the analyte gas flows through point (b) (valve-1) point(d) (valve-2) and point (e) (valve-3) The purging of sensorchamber was in open measurement mode by disconnectingthe hose of inlet pump fromvalve-2 directing valve-3 to point(f) and turning on the purge pump
At preresearch we observed 119877119892for 5 minutes after 119877
0
measurement to determine the response of each sensorand obtain the best starting measurement time for 119877
119892
measurement Significantly we found that overall sensorsreached a stable state after plusmn150 s (plusmn25min) which stronglyindicate that they are sensing stably the flow of gas thathave been spread evenly in the close measurement systemWe therefore took this time to be the starting point of 119877
119892
measurement Thus we set the total measurement time persample as 37 minutes covering the phases of the headspace(30 minutes) 119877
0measurement (1 minute) stabling time
(25 minutes) 119877119892measurement (1 minute) and purging (5
minutes) sequentially The sampling period of both 1198770and
119877119892measurement was 2 seconds and their averages were used
to represent the baseline and soil gaseous compound
5 Results and Discussion
51 Individual Sensitivity-Based Response ofMOSGas SensorsIndividual sensitivity-based soil gaseous profiles of MOS gassensors used on each soil type with the different dose of nutri-ent addition are shown in Figure 6 It reveals that the arrayof gas sensors was able to sense the soil gases andor volatilecompounds resulting from different samples and as well indi-cates that the method of the optimized SH seems suitable forprovidingaccumulating the concentration sufficientlyThoseindividual responses indicate that the technique of temper-ature modulation-SDP led the sensors to sense differently
6 Journal of Sensors
MOS array sensor
Gas Gasinlet outlet
Heaterand stirrer Purge pump
Valve-3
Valve-1
Watercontainer
Sensor chamberSoilcontainer
Inlet pumpMFC-1
MFC mass flow controller Interface
Silicagel
PSOC1(CY8C28445)
MFC-2
Valve-2(a)
(b)
(c)
(d)
(e)
(f) Waste
Figure 5 Experimental setup to capture the soil gaseous compounds using static headspace extraction in sample flow system (close)measurement
Sand SandSandSand Sand Sand Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
TGS2444 TGS2602 TGS825 FISAQ1 FISSB30 FIS12A
08
1
12
14
16
18
2
Sens
itivi
ty
20 tha30 tha
0 tha
Figure 6 Individual sensitivity of sensor the average and standard deviation of 5 replicates to 3 levels of compost addition in different soil
the amounts and types of soil gaseous compounds producedand released inside the SH atmosphere which correspondedto the soil type and doses of nutrient addition MoreoverFigure 6 also presents the standard deviation of the MOSsensors to five replicates of each measurement It relativelyshows the low variance among responses which indicates thesufficient consistency of sensors reproducibility in producingthe soil gaseous profiles on the same environment treatmentindependently throughout this study
As shown in Figure 6 formost of theMOS gas sensors butTGS2602 the sensitivity to the nutrient addition (20 tha and30 tha) was higher than without nutrient addition whetherfor the same soil type or between sandy loam and sand Sandyloam soil usually has more holding capacity of water andnutrient alongwith lower bulk density compared to sand soil
thus leading to having more organic matter content [31 33]andmicroorganism [34] In addition the use of a flow system(usually employing a pump) in sample detection causescooling of the sensor surface reducing the high increment oftemperature and humidity inside such sensor chamber (heatdissipation) [35] thus also influencing its response
Interestingly on TGS825 which is technically designedto respond to the hydrogen sulfide (H
2S) [36] it had the
highest sensitivity among the others for each soil type Itreveals that the H
2S concentration during the headspace
process was high and it is seen that the presence of nutrientaddition contributed significantly to H
2S accumulation in
the headspace (Figure 7) The response indicates that thereis much acid sulfate material in soil samples This gas canbe produced from the oxidation process of organic material
Journal of Sensors 7
SandSandy loam
12
14
16
18
2
22
Sens
itivi
ty
20 tha 30 tha0 tha 20 tha 30 tha0 tha
Figure 7 Response variances of TGS825 for five replicates betweensandy loam and sand soil in different dose of nutrient addition
containing sulfate acid due to bacterial activities in lowoxygen environment (like flooded soil) which depends onambient conditions such as temperature humidity and theconcentration of certainmetal ions [37]The result also showsthat the additional nutrient in sandy loam soil providedrelatively higher concentration than in sand soil and therewas a little cross-response in differentiating level of compostaddition between doses 20 tha and 30 tha
The operation of temperature modulation-SDP throughoscillating the heater voltage by square modulation does notonly cause altering the kinetics of both adsorption and reac-tion process at the surface of sensor (effect of the frequency)but also consequently lead the MOS gas sensor to run atlower effective temperature (effect of the duty cycle) as onthe TGS2444 which is driven by low duty cycle modulation[20] and shown to have high selectivity to ammonia gas [12]For particularmaterial the specificworking temperature pro-vides optimum sensitivity for sensing a certain gas evidently[38 39] Ou et al [39] found that under the low workingtemperature (ie 120∘C) a 2D metal disulfide-based gassensor has very high selectivity to NO
2in which the sensing
mechanism is dominated by charge transfer adsorptionbetween the surface-adsorbed NO
2gas molecules and metal
disulfide strongly due to paramagnetic behavior of NO2
Thus the combination of frequency and effective workingvoltage by duty cycle selection of temperature modulation-SDP had potential to sense sensitively the complex gas andorvolatile compounds of soil which then provide the uniquegaseous profiles
However like typical characteristic of the use of sensorarray in e-nose which does not allow individual sensor toidentify a specific or complex volatile compounds we foundthat there was no single sensor used which individuallyshowed a relation for characterization of the difference of soilconditions clearly and linearly with regard to soil type andnutrient additionThere was a cross-response on each sensorin differentiating the dose level of nutrient addition espe-cially between normal dose (20 tha) and high dose (30 tha)The complexity of soil gaseous compounds in potentiallyvarious kinds of gases especially volatile compounds [16 17]causes an inevitable cross-response onMOSgas sensor as also
PCA of sandy loam versus sand soil
Sandy loamSand
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus005minus02 minus01minus015 005 01 015 020Component 1
Figure 8 PCA plot showing discrimination between 2 soils withoutnutrient addition
founded by Rincon et al [40] who simulated a monitoringof VOC as soil contaminants through measuring 8 kindsof gases The cross-response of individual sensor may bereduced by projecting collectively into new dimension usingPCA as commonly used in e-nose
52 Performance of Discrimination of Soil under DifferentNutrient Addition Thepotential of nonparametric biologicalsystem for discriminating soil type as well as for differ-entiating between different nutrient additions treatmentsbased on its gaseous profile was tested Firstly the PCA asa nonsupervised technique was employed to find generalrelationships between samples while preserving most of thevariance within data PCA allow projecting variables ontofewer dimensions reflecting the most relevant analyticalinformation [41] This offers an advantage that the classifi-cation of unknowns is processed much faster thus reducingdetection time
Figure 8 shows the PCA plot of discrimination of twosoils both without addition of compost It shows a distinctzone of patterns volatile production between sandy loamsoil and sand soil where the principal component- (PC-) 1accounts for higher differentiation of cluster than PC-2 PC-1 and PC-2 cumulatively account for 7832 of the variancewithin the data set
Meanwhile Figure 9 shows the PCA plot for replicatesof each soil sample in distinguishing three doses of compostaddition It seem that PCA allow discriminating distinctlybetween soil conditions whether with or without compost(nutrient) addition indicated by separated blue zone evenwhen differentiating regardless of soil type (Figure 9(c))
It was only for sandy loam soil (Figure 9(a)) the levelof compost addition could be clustered clearly into threegroups as predefined previously while there was misiden-tification between soils with dose 20 tha and dose 30 tha
8 Journal of Sensors
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
Table 2 Properties of samples of soil fertilizer water and staticheadspace condition
Properties of SH ValueVolume of SH vial 90mLBulk density of sandy loam soil 144 gmLBulk density of sand soil 152 gmLPhase ratio 15Water content 100Density of pure water 0998 gmLCalculation resultsMass of sandy loam soil 2122 g
(i) Mass of compost adding at 20 tonha 0318 g(ii) Mass of compost adding at 30 tonha 0477 g
Mass of sand soil 2163 g(i) Mass of compost adding at 20 tonha 0324 g(ii) Mass of compost adding at 30 tonha 0287 g
(rpm)Stir
PowerHeat
Hot Top
Alcohol thermometer
Magnetic bar Water
Offminus +
Offminus +CorningPC-420D
(∘C)
Figure 4 Headspace conditioning with heating and stirring usingCorning PC-420D in SH sampling the layout of Corning modifiedfrom [32]
equilibration is optimized by both agitating (ie stirring)and thermostatting concurrently for all samples on thesame phase ratio We set 30 minutes 60∘C and 200 rpmof equilibration time temperature and stirring frequencyrespectively We utilized Corning PC-4200D to heat and stirthe sample in the SH vialWe used 90mL glass container withsealed cap as headspace vial which is put inside the 500mLopen beaker filled with 100mL water (Figure 4) It aims to
maintain the equilibrium relative humidity the same as thesoil sample And the headspacing was conducted inside aroom under controlled temperature By those ways all soilsamples were under the same treatments and environmentalconditions
The temperature modulation is set on 025Hz 75 dutycycle to drive all MOS gas sensors except for TGS2444 [20]which is on its recommended duty cycle The initial actionof the MOS gas sensors after a long inactive state is carriedout for one hour ofmeasuring the reference gas to allow themto reach a stable condition The gas sensors are expressed inresistance and the profile is defined by its sensitivity (119878) [4]where119877
0is sensor resistance of air and119877
119892is sensor resistance
of soil gaseous compound (see (2))
119878 =1198770
119877119892
(2)
The measurement of soil gaseous profiles is performedusing close measurement method by switching between thereference gas (filtered air with silica gel) as baseline andanalyte gas (soil gaseous compounds) The flow directionand rate of gas are controlled by 3-way valve and the Koflocmass flow controller (MFC) respectively The MFC are set at03 lpmAs shown in Figure 5 the reference gas flows throughpoint (a) (valve-1) point (c) (valve-2) and point (e) (valve-3)while the analyte gas flows through point (b) (valve-1) point(d) (valve-2) and point (e) (valve-3) The purging of sensorchamber was in open measurement mode by disconnectingthe hose of inlet pump fromvalve-2 directing valve-3 to point(f) and turning on the purge pump
At preresearch we observed 119877119892for 5 minutes after 119877
0
measurement to determine the response of each sensorand obtain the best starting measurement time for 119877
119892
measurement Significantly we found that overall sensorsreached a stable state after plusmn150 s (plusmn25min) which stronglyindicate that they are sensing stably the flow of gas thathave been spread evenly in the close measurement systemWe therefore took this time to be the starting point of 119877
119892
measurement Thus we set the total measurement time persample as 37 minutes covering the phases of the headspace(30 minutes) 119877
0measurement (1 minute) stabling time
(25 minutes) 119877119892measurement (1 minute) and purging (5
minutes) sequentially The sampling period of both 1198770and
119877119892measurement was 2 seconds and their averages were used
to represent the baseline and soil gaseous compound
5 Results and Discussion
51 Individual Sensitivity-Based Response ofMOSGas SensorsIndividual sensitivity-based soil gaseous profiles of MOS gassensors used on each soil type with the different dose of nutri-ent addition are shown in Figure 6 It reveals that the arrayof gas sensors was able to sense the soil gases andor volatilecompounds resulting from different samples and as well indi-cates that the method of the optimized SH seems suitable forprovidingaccumulating the concentration sufficientlyThoseindividual responses indicate that the technique of temper-ature modulation-SDP led the sensors to sense differently
6 Journal of Sensors
MOS array sensor
Gas Gasinlet outlet
Heaterand stirrer Purge pump
Valve-3
Valve-1
Watercontainer
Sensor chamberSoilcontainer
Inlet pumpMFC-1
MFC mass flow controller Interface
Silicagel
PSOC1(CY8C28445)
MFC-2
Valve-2(a)
(b)
(c)
(d)
(e)
(f) Waste
Figure 5 Experimental setup to capture the soil gaseous compounds using static headspace extraction in sample flow system (close)measurement
Sand SandSandSand Sand Sand Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
TGS2444 TGS2602 TGS825 FISAQ1 FISSB30 FIS12A
08
1
12
14
16
18
2
Sens
itivi
ty
20 tha30 tha
0 tha
Figure 6 Individual sensitivity of sensor the average and standard deviation of 5 replicates to 3 levels of compost addition in different soil
the amounts and types of soil gaseous compounds producedand released inside the SH atmosphere which correspondedto the soil type and doses of nutrient addition MoreoverFigure 6 also presents the standard deviation of the MOSsensors to five replicates of each measurement It relativelyshows the low variance among responses which indicates thesufficient consistency of sensors reproducibility in producingthe soil gaseous profiles on the same environment treatmentindependently throughout this study
As shown in Figure 6 formost of theMOS gas sensors butTGS2602 the sensitivity to the nutrient addition (20 tha and30 tha) was higher than without nutrient addition whetherfor the same soil type or between sandy loam and sand Sandyloam soil usually has more holding capacity of water andnutrient alongwith lower bulk density compared to sand soil
thus leading to having more organic matter content [31 33]andmicroorganism [34] In addition the use of a flow system(usually employing a pump) in sample detection causescooling of the sensor surface reducing the high increment oftemperature and humidity inside such sensor chamber (heatdissipation) [35] thus also influencing its response
Interestingly on TGS825 which is technically designedto respond to the hydrogen sulfide (H
2S) [36] it had the
highest sensitivity among the others for each soil type Itreveals that the H
2S concentration during the headspace
process was high and it is seen that the presence of nutrientaddition contributed significantly to H
2S accumulation in
the headspace (Figure 7) The response indicates that thereis much acid sulfate material in soil samples This gas canbe produced from the oxidation process of organic material
Journal of Sensors 7
SandSandy loam
12
14
16
18
2
22
Sens
itivi
ty
20 tha 30 tha0 tha 20 tha 30 tha0 tha
Figure 7 Response variances of TGS825 for five replicates betweensandy loam and sand soil in different dose of nutrient addition
containing sulfate acid due to bacterial activities in lowoxygen environment (like flooded soil) which depends onambient conditions such as temperature humidity and theconcentration of certainmetal ions [37]The result also showsthat the additional nutrient in sandy loam soil providedrelatively higher concentration than in sand soil and therewas a little cross-response in differentiating level of compostaddition between doses 20 tha and 30 tha
The operation of temperature modulation-SDP throughoscillating the heater voltage by square modulation does notonly cause altering the kinetics of both adsorption and reac-tion process at the surface of sensor (effect of the frequency)but also consequently lead the MOS gas sensor to run atlower effective temperature (effect of the duty cycle) as onthe TGS2444 which is driven by low duty cycle modulation[20] and shown to have high selectivity to ammonia gas [12]For particularmaterial the specificworking temperature pro-vides optimum sensitivity for sensing a certain gas evidently[38 39] Ou et al [39] found that under the low workingtemperature (ie 120∘C) a 2D metal disulfide-based gassensor has very high selectivity to NO
2in which the sensing
mechanism is dominated by charge transfer adsorptionbetween the surface-adsorbed NO
2gas molecules and metal
disulfide strongly due to paramagnetic behavior of NO2
Thus the combination of frequency and effective workingvoltage by duty cycle selection of temperature modulation-SDP had potential to sense sensitively the complex gas andorvolatile compounds of soil which then provide the uniquegaseous profiles
However like typical characteristic of the use of sensorarray in e-nose which does not allow individual sensor toidentify a specific or complex volatile compounds we foundthat there was no single sensor used which individuallyshowed a relation for characterization of the difference of soilconditions clearly and linearly with regard to soil type andnutrient additionThere was a cross-response on each sensorin differentiating the dose level of nutrient addition espe-cially between normal dose (20 tha) and high dose (30 tha)The complexity of soil gaseous compounds in potentiallyvarious kinds of gases especially volatile compounds [16 17]causes an inevitable cross-response onMOSgas sensor as also
PCA of sandy loam versus sand soil
Sandy loamSand
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus005minus02 minus01minus015 005 01 015 020Component 1
Figure 8 PCA plot showing discrimination between 2 soils withoutnutrient addition
founded by Rincon et al [40] who simulated a monitoringof VOC as soil contaminants through measuring 8 kindsof gases The cross-response of individual sensor may bereduced by projecting collectively into new dimension usingPCA as commonly used in e-nose
52 Performance of Discrimination of Soil under DifferentNutrient Addition Thepotential of nonparametric biologicalsystem for discriminating soil type as well as for differ-entiating between different nutrient additions treatmentsbased on its gaseous profile was tested Firstly the PCA asa nonsupervised technique was employed to find generalrelationships between samples while preserving most of thevariance within data PCA allow projecting variables ontofewer dimensions reflecting the most relevant analyticalinformation [41] This offers an advantage that the classifi-cation of unknowns is processed much faster thus reducingdetection time
Figure 8 shows the PCA plot of discrimination of twosoils both without addition of compost It shows a distinctzone of patterns volatile production between sandy loamsoil and sand soil where the principal component- (PC-) 1accounts for higher differentiation of cluster than PC-2 PC-1 and PC-2 cumulatively account for 7832 of the variancewithin the data set
Meanwhile Figure 9 shows the PCA plot for replicatesof each soil sample in distinguishing three doses of compostaddition It seem that PCA allow discriminating distinctlybetween soil conditions whether with or without compost(nutrient) addition indicated by separated blue zone evenwhen differentiating regardless of soil type (Figure 9(c))
It was only for sandy loam soil (Figure 9(a)) the levelof compost addition could be clustered clearly into threegroups as predefined previously while there was misiden-tification between soils with dose 20 tha and dose 30 tha
8 Journal of Sensors
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
MOS array sensor
Gas Gasinlet outlet
Heaterand stirrer Purge pump
Valve-3
Valve-1
Watercontainer
Sensor chamberSoilcontainer
Inlet pumpMFC-1
MFC mass flow controller Interface
Silicagel
PSOC1(CY8C28445)
MFC-2
Valve-2(a)
(b)
(c)
(d)
(e)
(f) Waste
Figure 5 Experimental setup to capture the soil gaseous compounds using static headspace extraction in sample flow system (close)measurement
Sand SandSandSand Sand Sand Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
Sandyloam
TGS2444 TGS2602 TGS825 FISAQ1 FISSB30 FIS12A
08
1
12
14
16
18
2
Sens
itivi
ty
20 tha30 tha
0 tha
Figure 6 Individual sensitivity of sensor the average and standard deviation of 5 replicates to 3 levels of compost addition in different soil
the amounts and types of soil gaseous compounds producedand released inside the SH atmosphere which correspondedto the soil type and doses of nutrient addition MoreoverFigure 6 also presents the standard deviation of the MOSsensors to five replicates of each measurement It relativelyshows the low variance among responses which indicates thesufficient consistency of sensors reproducibility in producingthe soil gaseous profiles on the same environment treatmentindependently throughout this study
As shown in Figure 6 formost of theMOS gas sensors butTGS2602 the sensitivity to the nutrient addition (20 tha and30 tha) was higher than without nutrient addition whetherfor the same soil type or between sandy loam and sand Sandyloam soil usually has more holding capacity of water andnutrient alongwith lower bulk density compared to sand soil
thus leading to having more organic matter content [31 33]andmicroorganism [34] In addition the use of a flow system(usually employing a pump) in sample detection causescooling of the sensor surface reducing the high increment oftemperature and humidity inside such sensor chamber (heatdissipation) [35] thus also influencing its response
Interestingly on TGS825 which is technically designedto respond to the hydrogen sulfide (H
2S) [36] it had the
highest sensitivity among the others for each soil type Itreveals that the H
2S concentration during the headspace
process was high and it is seen that the presence of nutrientaddition contributed significantly to H
2S accumulation in
the headspace (Figure 7) The response indicates that thereis much acid sulfate material in soil samples This gas canbe produced from the oxidation process of organic material
Journal of Sensors 7
SandSandy loam
12
14
16
18
2
22
Sens
itivi
ty
20 tha 30 tha0 tha 20 tha 30 tha0 tha
Figure 7 Response variances of TGS825 for five replicates betweensandy loam and sand soil in different dose of nutrient addition
containing sulfate acid due to bacterial activities in lowoxygen environment (like flooded soil) which depends onambient conditions such as temperature humidity and theconcentration of certainmetal ions [37]The result also showsthat the additional nutrient in sandy loam soil providedrelatively higher concentration than in sand soil and therewas a little cross-response in differentiating level of compostaddition between doses 20 tha and 30 tha
The operation of temperature modulation-SDP throughoscillating the heater voltage by square modulation does notonly cause altering the kinetics of both adsorption and reac-tion process at the surface of sensor (effect of the frequency)but also consequently lead the MOS gas sensor to run atlower effective temperature (effect of the duty cycle) as onthe TGS2444 which is driven by low duty cycle modulation[20] and shown to have high selectivity to ammonia gas [12]For particularmaterial the specificworking temperature pro-vides optimum sensitivity for sensing a certain gas evidently[38 39] Ou et al [39] found that under the low workingtemperature (ie 120∘C) a 2D metal disulfide-based gassensor has very high selectivity to NO
2in which the sensing
mechanism is dominated by charge transfer adsorptionbetween the surface-adsorbed NO
2gas molecules and metal
disulfide strongly due to paramagnetic behavior of NO2
Thus the combination of frequency and effective workingvoltage by duty cycle selection of temperature modulation-SDP had potential to sense sensitively the complex gas andorvolatile compounds of soil which then provide the uniquegaseous profiles
However like typical characteristic of the use of sensorarray in e-nose which does not allow individual sensor toidentify a specific or complex volatile compounds we foundthat there was no single sensor used which individuallyshowed a relation for characterization of the difference of soilconditions clearly and linearly with regard to soil type andnutrient additionThere was a cross-response on each sensorin differentiating the dose level of nutrient addition espe-cially between normal dose (20 tha) and high dose (30 tha)The complexity of soil gaseous compounds in potentiallyvarious kinds of gases especially volatile compounds [16 17]causes an inevitable cross-response onMOSgas sensor as also
PCA of sandy loam versus sand soil
Sandy loamSand
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus005minus02 minus01minus015 005 01 015 020Component 1
Figure 8 PCA plot showing discrimination between 2 soils withoutnutrient addition
founded by Rincon et al [40] who simulated a monitoringof VOC as soil contaminants through measuring 8 kindsof gases The cross-response of individual sensor may bereduced by projecting collectively into new dimension usingPCA as commonly used in e-nose
52 Performance of Discrimination of Soil under DifferentNutrient Addition Thepotential of nonparametric biologicalsystem for discriminating soil type as well as for differ-entiating between different nutrient additions treatmentsbased on its gaseous profile was tested Firstly the PCA asa nonsupervised technique was employed to find generalrelationships between samples while preserving most of thevariance within data PCA allow projecting variables ontofewer dimensions reflecting the most relevant analyticalinformation [41] This offers an advantage that the classifi-cation of unknowns is processed much faster thus reducingdetection time
Figure 8 shows the PCA plot of discrimination of twosoils both without addition of compost It shows a distinctzone of patterns volatile production between sandy loamsoil and sand soil where the principal component- (PC-) 1accounts for higher differentiation of cluster than PC-2 PC-1 and PC-2 cumulatively account for 7832 of the variancewithin the data set
Meanwhile Figure 9 shows the PCA plot for replicatesof each soil sample in distinguishing three doses of compostaddition It seem that PCA allow discriminating distinctlybetween soil conditions whether with or without compost(nutrient) addition indicated by separated blue zone evenwhen differentiating regardless of soil type (Figure 9(c))
It was only for sandy loam soil (Figure 9(a)) the levelof compost addition could be clustered clearly into threegroups as predefined previously while there was misiden-tification between soils with dose 20 tha and dose 30 tha
8 Journal of Sensors
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
SandSandy loam
12
14
16
18
2
22
Sens
itivi
ty
20 tha 30 tha0 tha 20 tha 30 tha0 tha
Figure 7 Response variances of TGS825 for five replicates betweensandy loam and sand soil in different dose of nutrient addition
containing sulfate acid due to bacterial activities in lowoxygen environment (like flooded soil) which depends onambient conditions such as temperature humidity and theconcentration of certainmetal ions [37]The result also showsthat the additional nutrient in sandy loam soil providedrelatively higher concentration than in sand soil and therewas a little cross-response in differentiating level of compostaddition between doses 20 tha and 30 tha
The operation of temperature modulation-SDP throughoscillating the heater voltage by square modulation does notonly cause altering the kinetics of both adsorption and reac-tion process at the surface of sensor (effect of the frequency)but also consequently lead the MOS gas sensor to run atlower effective temperature (effect of the duty cycle) as onthe TGS2444 which is driven by low duty cycle modulation[20] and shown to have high selectivity to ammonia gas [12]For particularmaterial the specificworking temperature pro-vides optimum sensitivity for sensing a certain gas evidently[38 39] Ou et al [39] found that under the low workingtemperature (ie 120∘C) a 2D metal disulfide-based gassensor has very high selectivity to NO
2in which the sensing
mechanism is dominated by charge transfer adsorptionbetween the surface-adsorbed NO
2gas molecules and metal
disulfide strongly due to paramagnetic behavior of NO2
Thus the combination of frequency and effective workingvoltage by duty cycle selection of temperature modulation-SDP had potential to sense sensitively the complex gas andorvolatile compounds of soil which then provide the uniquegaseous profiles
However like typical characteristic of the use of sensorarray in e-nose which does not allow individual sensor toidentify a specific or complex volatile compounds we foundthat there was no single sensor used which individuallyshowed a relation for characterization of the difference of soilconditions clearly and linearly with regard to soil type andnutrient additionThere was a cross-response on each sensorin differentiating the dose level of nutrient addition espe-cially between normal dose (20 tha) and high dose (30 tha)The complexity of soil gaseous compounds in potentiallyvarious kinds of gases especially volatile compounds [16 17]causes an inevitable cross-response onMOSgas sensor as also
PCA of sandy loam versus sand soil
Sandy loamSand
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus005minus02 minus01minus015 005 01 015 020Component 1
Figure 8 PCA plot showing discrimination between 2 soils withoutnutrient addition
founded by Rincon et al [40] who simulated a monitoringof VOC as soil contaminants through measuring 8 kindsof gases The cross-response of individual sensor may bereduced by projecting collectively into new dimension usingPCA as commonly used in e-nose
52 Performance of Discrimination of Soil under DifferentNutrient Addition Thepotential of nonparametric biologicalsystem for discriminating soil type as well as for differ-entiating between different nutrient additions treatmentsbased on its gaseous profile was tested Firstly the PCA asa nonsupervised technique was employed to find generalrelationships between samples while preserving most of thevariance within data PCA allow projecting variables ontofewer dimensions reflecting the most relevant analyticalinformation [41] This offers an advantage that the classifi-cation of unknowns is processed much faster thus reducingdetection time
Figure 8 shows the PCA plot of discrimination of twosoils both without addition of compost It shows a distinctzone of patterns volatile production between sandy loamsoil and sand soil where the principal component- (PC-) 1accounts for higher differentiation of cluster than PC-2 PC-1 and PC-2 cumulatively account for 7832 of the variancewithin the data set
Meanwhile Figure 9 shows the PCA plot for replicatesof each soil sample in distinguishing three doses of compostaddition It seem that PCA allow discriminating distinctlybetween soil conditions whether with or without compost(nutrient) addition indicated by separated blue zone evenwhen differentiating regardless of soil type (Figure 9(c))
It was only for sandy loam soil (Figure 9(a)) the levelof compost addition could be clustered clearly into threegroups as predefined previously while there was misiden-tification between soils with dose 20 tha and dose 30 tha
8 Journal of Sensors
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
PCA of sandy loam soil
No compost
minus025
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
minus03 minus01minus02minus04 01 02 030Component 1
Compost 20 tCompost 30 t
(a)
No compost
PCA of sand soil
minus015minus025 minus02 minus01 minus005 005 01 0150Component 1
minus008
minus006
minus004
minus002
0
002
004
006
008
Com
pone
nt 2
Compost 20 tCompost 30 t
(b)
No compost
PCA of sandy loam and sand
minus02 minus01minus03 01 02 03 04 050Component 1
minus02
minus015
minus01
minus005
0
005
01
015
02
Com
pone
nt 2
Compost 20 tCompost 30 t
(c)
Figure 9 PCA map for replicates of soil gaseous pattern projection for each soil sample in distinguishing three doses of compost addition(a) on sandy loam soil (b) on sand soil and (c) irrespective of soil type
in sand soil (Figure 9(b)) Interestingly irrespective of soiltype (Figure 9(c)) it seems to perform better in clusteringthe soil in different doses yet there is a half part of replicatesthat has no clear classification (black zone) when identifyingsoil with doses 20 tha and 30 tha Figure 9 shows that thesignificant discrimination on the clusters between the soilwithout nutrient addition (blue zone) and soil with nutrientaddition (yellow and red zones)was along the PC-1 while thatbetween normal dose (yellow zone) and high dose (red zone)was mainly along the PC-2
Finally we determined the performance of NN as deci-sion unit of e-nose to classify the level of nutrient additionin soil based on indicator the Mean Square Error (MSE)achieved resulting from the training process We put threeprincipal components (PCs) to distinguish the volatile com-pounds in the headspace released from soil samples as theinput of neural network since they represent more than90 of divergence samples data (Table 3) We designed thearchitecture ofMLPNN that comprises 3 layers (single hiddenlayer) We determined the optimum number of neurons
Journal of Sensors 9
Table 3 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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 Cumulative proportion of 3 PCs resulting from 6 sensorsused
PC PCs proportionSandy loam Sand Irrespective of soil type
PC-1 6427 7561 6653PC-2 8634 8896 8069PC-3 9373 9373 8918
Table 4 Target definition for learning the soil gaseous patterns
T2 T1 T0 Cluster category0 0 1 Soil without addition of compost0 1 0 Soil with compost doses of 20 tha1 0 0 Soil with compost doses of 30 tha
Table 5 MSE achieved by 6 neurons of hidden layer to discriminate3 levels of compost addition in soil
Soil type MSE of with PCA MSE of without PCASand 4204119890 minus 04 3490119890 minus 03
Sandy loam 1226119890 minus 04 5024119890 minus 04
Regardless of type 2678119890 minus 03 4080119890 minus 03
in hidden layer by Singular Value Decomposition (SVD)analysis of its output in each training dataset [42] By inputfrom 3 PCs and based on the SVD value obtained wechoose 6 neurons in hidden layer to differentiate between thepredescribed three categorized fertilizer levels in soil samplethus the neuron number architecture of MLPNN is 3-6-3 ofrespectively input hidden and output layer
In learning we took the learning parameters of BP asfollow maximum epoch is 104 error target is 10minus5 initiallearning rate is 08 and the constant of search time in search-then-converge annealing learning rate is 700 The target ofoutput layer was defined as shown in Table 4 We also trainedthe NN by input directly from sensors output (withoutpreprocessingPCA) with the same hidden layer (6-6-3 NNarchitecture) The achieved MSE of training results (Table 5)show that PCA helps in improving the NN classificationto discriminate the level of compost addition in soil Inaddition all the application of trained data was successful todiscriminate three levels of nutrient addition in soil
The e-nose approach with static headspace method waspotential for the aims of this work providing different soilvolatile profiles and allowing a discrimination between soiltype and among the several soil treatments to be obtainedThis supports previous study where the same samplingmethod was employed for sensing the headspace of a soilunder different condition and nutrient addition [15 43]which may overcome the overlapping between volatile pro-files Compared with the results of Bastos and Magan [43]it seems that the use of sensors that potentially can detectgasesvolatile compounds in complex compound providesbetter detection and economical value due to the smallnumber of sensors used and the less complexity of the patternidentification systemapplied rather thannonspecific sensors
6 Conclusions and Future Work
The 6 selected MOS gas sensors with temperature modula-tion-SDP in e-nose system were promising applied forindicating the presence of additional nutrients in soil sincethey could respond and have different sensitivity accordingto the samples They provided (unique) soil gaseous profileswhich accumulated in a static headspace optimized by ther-mostatting (60∘C) and stirring (200 rpm) in controlled envi-ronment condition The profiles show that the temperaturemodulation-SDP leads to distinguishing of the soils clearlyand to indicating the presence of nutrient addition in soilTheMLPNN in single hidden layer architecture (3-6-3) with PCAas prior data preprocessor performed optimum identificationin this study The gas sensors with this particular techniqueoffer a potential for replacing existing techniques in soilenvironmental fields for a quick and in situ application Italso suggests that it together with e-nose method could beused for monitoring microbial activity in soil and water aswell Depending on the applications and the type of sampleto be analyzed the choice of sensor array can be crucial forthe good performance of the system
Competing Interests
The authors declare that there are no competing interestsregarding the publication of this paper
Acknowledgments
Arief Sudarmaji is supported by Indonesian DirectorateGeneral of Higher Education (DIKTI) with Guarantee Letterno 672E44K2012 and Akio Kitagawa is supported byJapan Society for the Promotion of Science (JSPS) KAKENHIGrant nos 25286036 and 15K12504
References
[1] A P Lee and B J Reedy ldquoTemperaturemodulation in semicon-ductor gas sensingrdquo Sensors and Actuators B Chemical vol 60no 1 pp 35ndash42 1999
[2] R Chutia andM Bhuyan ldquoStudy of temperature modulated tinoxide gas sensor and identification of chemicalsrdquo in Proceedingsof the 2nd National Conference on Computational Intelligenceand Signal Processing (CISP rsquo12) pp 181ndash184 Guwahati IndiaMarch 2012
[3] X Huang F Meng Z Pi W Xu and J Liu ldquoGas sensing behav-ior of a single tin dioxide sensor under dynamic temperaturemodulationrdquo Sensors and Actuators B Chemical vol 99 no 2-3 pp 444ndash450 2004
[4] X Huang J Liu D Shao Z Pi and Z Yu ldquoRectangularmode ofoperation for detecting pesticide residue by using a single SnO
2-
based gas sensorrdquo Sensors andActuators B Chemical vol 96 no3 pp 630ndash635 2003
[5] E Martinelli D Polese A Catini A DrsquoAmico and C DiNatale ldquoSelf-adapted temperature modulation in metal-oxidesemiconductor gas sensorsrdquo Sensors and Actuators B Chemicalvol 161 no 1 pp 534ndash541 2012
[6] AVergara EMartinelli E Llobet ADrsquoamico andCDiNataleldquoOptimized feature extraction for temperature-modulated gas
10 Journal of Sensors
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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
sensorsrdquo Journal of Sensors vol 2009 Article ID 716316 10pages 2009
[7] E Brauns E Morsbach S Kunz M Baeumer and W LangldquoTemperature modulation of a catalytic gas sensorrdquo Sensors(Switzerland) vol 14 no 11 pp 20372ndash20381 2014
[8] S Nakata and K Kashima ldquoDistinguishing among gases with asemiconductor sensor depending on the frequency modulationof a cyclic temperaturerdquo Electroanalysis vol 22 no 14 pp 1573ndash1580 2010
[9] S Nakata HOkunishi and YNakashima ldquoDistinction of gaseswith a semiconductor sensor under a cyclic temperature mod-ulation with second-harmonic heatingrdquo Sensors and ActuatorsB Chemical vol 119 no 2 pp 556ndash561 2006
[10] K A Ngo P Lauque and K Aguir ldquoHigh performance of agas identification system using sensor array and temperaturemodulationrdquo Sensors and Actuators B Chemical vol 124 no1 pp 209ndash216 2007
[11] A Fort M Gregorkiewitz N Machetti et al ldquoSelectivityenhancement of SnO
2sensors by means of operating tempera-
ture modulationrdquoThin Solid Films vol 418 no 1 pp 2ndash8 2002[12] A Sudarmaji and A Kitagawa ldquoSensors amp transducers temper-
ature modulation with specified detection point on metal oxidesemiconductor gas sensors for E-nose applicationrdquo Sensors ampTransducers vol 186 no 3 pp 93ndash103 2015
[13] T Carson C M Bachmann and C Salvaggio ldquoSoil signaturesimulation of complex mixtures and particle size distributionsrdquoOptical Engineering vol 54 no 9 Article ID 094103 2015
[14] Soil Science Society of America ldquoSoilsmdashOverviewrdquo WaterResources 2010 httpswwwsoilsorgfilesabout-soilssoils-overviewpdf
[15] F De Cesare E Di Mattia S Pantalei et al ldquoUse of electronicnose technology to measure soil microbial activity throughbiogenic volatile organic compounds and gases releaserdquo SoilBiology and Biochemistry vol 43 no 10 pp 2094ndash2107 2011
[16] H Insam and M S A Seewald ldquoVolatile organic compounds(VOCs) in soilsrdquo Biology and Fertility of Soils vol 46 no 3 pp199ndash213 2010
[17] F Tassi S Venturi J Cabassi F Capecchiacci B Nisi andO Vaselli ldquoVolatile organic compounds (VOCs) in soil gasesfrom Solfatara crater (Campi Flegrei southern Italy) geogenicsource(s) vs biogeochemical processesrdquo Applied Geochemistryvol 56 pp 37ndash49 2015
[18] CMeiWang andD E Cane ldquoNIH public accessrdquo Journal of theAmerican Chemical Society vol 29 no 6 pp 997ndash1003 2008
[19] C-M Wang and D E Cane ldquoBiochemistry and moleculargenetics of the biosynthesis of the earthy odorantmethylisobor-neol in Streptomyces coelicolorrdquo Journal of the American Chem-ical Society vol 130 no 28 pp 8908ndash8909 2008
[20] Figaro Engineering Inc Data Sheet TGS 2444 for the Detectionof Ammonia 2011
[21] D Hercog and B Gergic ldquoA flexible microcontroller-based dataacquisition devicerdquo Sensors vol 14 no 6 pp 9755ndash9775 2014
[22] M A Naivar M E Wilder R C Habbersett et al ldquoDevelop-ment of small and inexpensive digital data acquisition systemsusing amicrocontroller-based approachrdquoCytometry Part A vol75 no 12 pp 979ndash989 2009
[23] R Gutierrez-Osuna H T Nagle B Kermani and S S Schiff-man ldquoIntroduction to chemosensorsrdquo inHandbook of MachineOlfaction T C Pearce S S Schiffman H T Nagle and J WGardner Eds pp 133ndash160 Wiley-VCH Verlag GmbH amp CoKGaA Weinheim Germany 2003
[24] A Sudarmaji A Kitagawa and J Akita ldquoDesign of wirelessmeasurement of soil gases and soil environment based onProgrammable System-on-Chip (PSOC)rdquo in Proceedings ofthe International Symposium on Agricultural and BiosystemEngineering (ISABE rsquo13) pp E5-1ndashE5-13 2013
[25] K-L Du and M N S Swamy Neural Networks and StatisticalLearning Springer London UK 2014
[26] N Haber B Deller H Flaig E Schulz and J ReinholdldquoSustainable compost application in agriculturerdquo ECN-INFO022010 European Compost Network 2008
[27] A R Conklin Introduction to Soil Chemistry Analysis andInstrumentation John Wiley amp Sons Hoboken NJ USA 2ndedition 2014
[28] K Malone and HWilliamsGrowing Season Definition and UseinWetland Delineation A Literature Review US Army EngineerResearch and Development Center Nacogdoches Tex USA2010
[29] M C Rabenhorst ldquoBiologic zero a soil temperature conceptrdquoWetlands vol 25 no 3 pp 616ndash621 2005
[30] C Yu J Cheng L Jones et al ldquoData collection handbook tosupport modeling the impacts of radioactive material in soilrdquoTech Rep Argonne National Laboratory Argonne Ill USA1993
[31] P R Chaudhari D V Ahire V D Ahire M Chkravarty andS Maity ldquoSoil bulk density as related to soil texture organicmatter content and available total nutrients of Coimbatore soilrdquoInternational Journal of Scientific and Research Publications vol3 no 2 pp 1ndash8 2013
[32] Corning Instruction Manual For All Hot Plates Stirrers andStirrerHot Plates with Digital Displays and for the 6795PRTemperature Controller Corning Lowell Mass USA 2007
[33] J A Amador and J A Atoyan ldquoStructure and composition ofleachfield bacterial communities role of soil texture depth andseptic tank effluent inputsrdquo Water vol 4 no 3 pp 707ndash7192012
[34] N H Hamarashid M A Othman and M-A H HussainldquoEffects of soil texture on chemical compositions microbialpopulations and carbon mineralization in soilrdquo The EgyptianJournal of Experimental Biology vol 6 no 1 pp 59ndash64 2010
[35] Figaro Engineering Inc General Information for TGS SensorsTechnical Information on Usage of TGS Sensors for Toxic andExplosive Gas Leak Detectors Figaro Engineering Inc 2005
[36] Figaro Engineering Inc Product Information TGS 825mdashSpecialSensor for Hydrogen Sulfide 2011
[37] S Chou JMOgdenH R Pohl et alDraft Toxicological Profilefor Hydrogen Sulfide and Carboxyl Sulfide Agency for ToxicSubstances and Disease Registry Atlanta Ga USA 2014
[38] D N Chavan G E Patil D D Kajale V B Gaikwad P KKhanna and G H Jain ldquoNano Ag-doped In
2O3thick film a
low-temperature H2S gas sensorrdquo Journal of Sensors vol 2011
Article ID 824215 8 pages 2011[39] J Z Ou W Ge B Carey et al ldquoPhysisorption-based charge
transfer in two-dimensional SnS2for selective and reversible
NO2gas sensingrdquo ACS Nano vol 9 no 10 pp 10313ndash10323
2015[40] M Rincon J M Getino J Robla G Hierro J Mochon and
I Bustinza ldquoGas sensor array for VOCrsquos monitoring in soilscontaminationrdquo Ingenierıa vol 14 no 1 pp 45ndash54 2010
[41] E LHines P Boilot JWGardner andMAGongora ldquoPatternanalysis for electronic nosesrdquo in Handbook of Machine Olfac-tion Electronic Nose Technology T C Pearce S S Schiffman
Journal of Sensors 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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 11
H T Nagle and J W Gardner Eds chapter 6 pp 133ndash160WILEY-VCH Weinheim Germany 2003
[42] JDA SantosGA Barreto andCM SMedeiros ldquoEstimatingthe number of hidden neurons of the MLP using singular valuedecomposition and principal components analysis a novelapproachrdquo in Proceedings of the 11th Brazilian Symposium onNeural Networks (SBRN rsquo10) pp 19ndash24 IEEE Sao Paulo BrazilOctober 2010
[43] A C Bastos and N Magan ldquoSoil volatile fingerprints use fordiscrimination between soil types under different environmen-tal conditionsrdquo Sensors and Actuators B Chemical vol 125 no2 pp 556ndash562 2007
[44] Figaro Engineering Inc TGS 2602mdashFor the Detection of AirContaminants 2005
[45] FIS Inc FIS GAS SENSOR SB-12A for Methane Detection 2006[46] FIS FIS Gas Sensor SB-30 for Alcohol Detection FIS 2008[47] FIS Inc FIS Gas Sensor SB-AQ1 for Air Quality Control (VOCs)
2010
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