sistemes multisensors aplicats al control de qualitat dels...
TRANSCRIPT
Sistemes multisensors aplicats al control de qualitatdels vins
Cecilia JiménezInstitut de Microelectrònica de Barcelona (IMB-CNM) CSIC
NATIONAL CENTER OF MICROELECTRONICS
(CNM)
D+T Microelectrónica, A.I.E.
BARCELONA INSTITUTE OF
MICROELECTRONICS (IMB)
MADRID INSTITUTE OF
MICROELECTRONICS (IMM)
SEVILLA INSTITUTE OF
MICROELECTRONICS (IMSE)
LARGE SCALE FACILITY
(Integrated Clean Room)
SPANISH COUNCIL FOR SCIENTIFIC RESEARCH (CSIC)
Barcelona Barcelona InstituteInstitute ofof Microelectronics (IMBMicroelectronics (IMB--CNM), CSICCNM), CSIC
2011 IMB Budget: 11.5 M€External funding: 50 %
IMB STAFF (2012)
• Researchers 66
• Ph.D. Students 57
• Clean room 43
• Support services 28
• Admin. & general serv. 20
• Visitors 5
TOTAL: 219
Research at IMB:Research at IMB:
µ/n-sensors
µ/n-actuators
µ/n-structures & passives
Signal processing
Powerdevices
Pac
kagi
ng &
in
terc
onne
ct
Embedded softwareCharacterization & Test
Micro/Nano-System
(µ)powersources
• Micro – Nanotechnologies
• Nanofabrication and functional properties of nanostructures
• Transducers for chemical and biochemical sensing
• Micro-Nano-Bio systems
• Integrated circuits and systems
• Power devices and systems
Technology focus : MICRO & Technology focus : MICRO & NANO INTEGRATED SYSTEMSNANO INTEGRATED SYSTEMS
Table of contents
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
Electronic Tongue: analogy to gustative organ
RecognitionElement Obtained Data TreatmentObtained Data Results
Nerve endings of the
tongue
Transmission ofnerve pulse
Brain
Signals (electrochemical,
absorbance, …)
Chemometric processing tools
(PCA, NNs, PLS)
Tastesense
• sweet
• salty
• bitter
• sour
• umami
QualitativeAnalysis
Classification
QuantitativeAnalysis
Nerve endings of the
tongue
Array of sensors
(crosselectivity)
Sabor: Sensación debida a conjunto de substanciasCategorías :
• Salado : NaCl• Dulce : glucosa, fructosa• Amargo : quinina, cafeína• Acido : HCl, ácido acético• “Umami ”: glutamato monosódico (MSG) Sweet Salt Sour Bitter
Neu
ral
Re
sp
on
se
Sucrose Test
Sweet Salt Sour Bitter
Neu
ral
Re
sp
on
se
HCl Test
Sweet Salt Sour Bitter
NaCl Test
Sweet Salt Sour Bitter
Quinine Test
Respuesta de sistema gustativo a distintos compuestos
Respuesta de sistema gustativo a distintos compuestos
Respuesta de conjunto de sensores a distintos compue stosRespuesta de conjunto de sensores a distintos compue stos
Toko, K., Taste sensor. Sensors and Actuators B-Chemical, 2000. 64(1-3): p. 205-215.
Electronic Tongue: analogy to gustative organ
Table of contents
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
ISFET
IDEs
Type of transducers:
• ISFET based sensors : pH, ions Na, K, Ca, Cl, NO3, etc
• Microelectrodes and UMEAs (Voltamperometry): Cl2and COD, heavy metals, molecules
• Interdigitated electrodes (IDEs): conductivity, dielectric properties.
• Multisensor chips (compatibilization of fabricationtechnologies)
Type of transducers:
• ISFET based sensors : pH, ions Na, K, Ca, Cl, NO3, etc
• Microelectrodes and UMEAs (Voltamperometry): Cl2and COD, heavy metals, molecules
• Interdigitated electrodes (IDEs): conductivity, dielectric properties.
• Multisensor chips (compatibilization of fabricationtechnologies)
Advantages
• Low cost (mass produced)
• Low energy consumption
• Reproducible
• Miniaturization (µTAS or LOC)
• Circuit integration
Advantages
• Low cost (mass produced)
• Low energy consumption
• Reproducible
• Miniaturization (µTAS or LOC)
• Circuit integration
UMEA
4-electrode Cell
Multisensor chip on BESOI
82% yielding perwafer
amperometric
Fabricated with microelectronic technology
Solution inlets/outlets
Sensors
Solution inlets/outlets
Microelectrodes used
ISFET and CHEMFETs
-0,6
-0,5
-0,4
-0,3
-0,2
-0,1
0
0,1
0,2
0,3
0,4
0 2 4 6 8 10 12 14
pH
Vo
(mV
)
Detection of:
• pH
• Cations Ca 2+, K+, Na+
• Anions: Cl-, CO32-
Detection of:
• pH
• Cations Ca 2+, K+, Na+
• Anions: Cl-, CO32-
Pt/Au Microelectrodes
Standard technology Si/SiO 2/metal
• Conductivity/impedance
• Redox Potential (ORP)
• Dissolved Oxygen (O2)
• Cloride (Cl2)
• Heavy Metals
• Electrochemical oxygen demand (EOD)
• Conductivity/impedance
• Redox Potential (ORP)
• Dissolved Oxygen (O2)
• Cloride (Cl2)
• Heavy Metals
• Electrochemical oxygen demand (EOD)
Au/Pt microelectrodesIDS
4B1
y = 0.1003x + 7.4826R2 = 0.9986
0
200
400
600
800
1000
1200
0 2000 4000 6000 8000 10000 12000Conductividad (mS/cm)
Pot
enc
ial,
mV
145 µm/cm - 11,2 mS/cm
4 bars
Conductivitycalibration curve
3-cell Au/Pt microelectrodes
Amperometric Microelectrodes
Ip = K x CEq. Randles
Oxidation- Reduction reactions: Fe2+
0 1 2 3 4 5 6 7 8 9 10
-3
-2
-1
0
i -i 0 (
µA)
Concentracion O2 (ppm)
Experimento 9 Experimento 7 Experimento 8
Electrodo 3360-02-03
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8
0
100
200
300
400
500
600
700 i (µA) = 0.256 * Gluc. conc (ppm O2) + 68.00
r = 0.998
0 400 800 12000
100
200
300
400
i / µ
A
Glucose concentration / ppm O2
i / µ
A
E / V vs Ag/AgCl Baseline 200 µl glucose 400 µl glucose 600 µl glucose
Disolved O2 sensor Glucose sensor
Peroxidase Biosensors
Biosensors based on nanomaterials
NS O
OO
O
N SO
OO
O
+
NS O
OO
O
NS O
OO
O
NS O
OO
O
S NH
O
S NH
O
S NH
O
NS O
OO
O
N SO
OO
O
+
Gold UMEA
4 mM DTSP/DMONS O
OO
O
NS O
OO
O
NS O
OO
O
HRP
S NH
O
S NH
O
S NH
O
S NH
O
S NH
O
S NH
O
Immunosensor approach for the detection of rabbit IgG
Carbon nanotube-polystyrene composite electrodes
O
NH
O
NH
O
HO
O
HO
O
NH
O
NH
O
HO
O
HO
Biological recogntion event usinganti-rabbit IgG peroxidase conjugateCovalent immobilization of rabbit IgG
-4.5
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
0 20 40 60 80
Time / s
Cur
ren
t / µ
A
031030100
[rIgG] (ng/mL)
a
b
c
de
abcde
-4.5
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
0 20 40 60 80
Time / s
Cur
ren
t / µ
A
031030100
[rIgG] (ng/mL)
a
b
c
de
abcde
Calibration:• I/µA = 0.44*log[rIgG] - 2.47
r= 0.99; n= 4Linear range:• 3-100 ng/mL IgG rabbitLimit of detection: • 3 ng/mL
UMEA modified with Au nanoparticles
10 times higher sensitivity/ Au microelectrode10 times higher sensitivity/ Au microelectrode
500
600
700
800
900
1000
1100
1200
1300
1400
1 3 5 7 9 11 13
pH
Vou
t (m
V)
2201-3-005
2201-3-006
2201-5-005
2201-5-006
Monolithic Integration on BESOI substrate6 NMOS ISFET1 IDEs1 Temperature Diode
I = 100 µA
0.525
0.550
0.575
0.600
0.625
0.650
0.675
20 30 40 50 60 70 80
PCB-1
PCB-2
PCB-3
PCB-4
PCB-5
PCB-6
PCB-7
PCB-8
PROMEDIO
T (ºC)
V (V)
Electrically Isolated byTrench (SOI Substrate)
pH ISFET
P-n diode
IDE
Multisensor chips
Optical integrated systems
Characteristics:
• Method of fabrication is fast and easy.
• Monolithic integration (µTAS).
• Compatible with aqueous media.
• Biocompatible, chemically stable and non-toxic.
• Absorbance between 200 and 1000 nm
Characteristics:
• Method of fabrication is fast and easy.
• Monolithic integration (µTAS).
• Compatible with aqueous media.
• Biocompatible, chemically stable and non-toxic.
• Absorbance between 200 and 1000 nmPDMS
Prism configuration MIR configuration
Polydimethylsiloxane
• Instrumento analítico para control de calidad de alimentos• Formado por:
• Sistema de muestreo• Conjunto de sensores químicos• Sistema de procesamiento de datos: reconocimiento de patrones o calibración multivariante (PCA, Redes Neuronales)
�AplicacionesAlimentación, bebidas, cosmética, farmacéutica y medio ambiente� Control de calidad y fraude� Selección de tipos de alimentos� Determinación de D.O. en vinos� Análisis cuantitativo
�AplicacionesAlimentación, bebidas, cosmética, farmacéutica y medio ambiente� Control de calidad y fraude� Selección de tipos de alimentos� Determinación de D.O. en vinos� Análisis cuantitativo
fluídica
Software para:control de bombas/válvulasVisualización señalTratamiento señal
Sistema multisensor
Table of contents
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
Graphic representationin the new axes
PC 1 (60 %)
PC 2 (30 %)
Group 1
Group 3Group 4
Group 2
Principal Component Analysis (PCA)
Y
XZ
Directions of maximumvariation, PCs
PCA
Original Data
Representation in twodimensions
Change of the axes
Reduction of variables
Method for pattern recognition
Partial-Least Squares (PLS)
∑=
+=k
k
kk xbby1
0
Method of multivariate calibration. Linear equation expressedgenerally like this:
Reduction of variables
RegressionComponents or factors
Standard Method in Chemometrics.
Wine Samples
Calibration set : 2/3 of the total number of samples
Prediction set : the rest 1/3 of samples. At least, one sampleof each variety
Table of contents
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
• Introduction to Multisensor systems or Electronic Tongues
•Microelectrodes used– ISFET based sensors
– Thin film metal microelectrodes
– Optical integrated systems
• Chemometric tools
• Results– Sensor signals & treatment of data
– Classification of wines: • Grape variety/ vintage year/ geographical origin
– Quantitative analysis
• Conclusions
-870
-860
-850
-840
-830
-820
-810
-8000 1000 2000 3000 4000 5000 6000
Time (s)E
(m
V)
ISF
ET
pH
pH ISFET forthe white wine set.
Sensor signals: Electrochemical Microsensors
0
10
20
30
40
50
60
70
80
0 100 200 300 400
Time (s)
Inte
nsity
(µA
)
MerlotCabernetGarnatxaTrepat
EOD sensor response from the red wine set.
-30
-20
-10
0
10
20
30
40
50
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
E (V) vs. Ag/AgCl
Inte
nsity
(µA
)
MacabeuParelladaChardonnayXarel·lo
I. +1.31 V
I. -0.38 V
I. +0.65 V
I. +1.01 V
-30
-20
-10
0
10
20
30
40
50
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
E (V) vs. Ag/AgCl
Inte
nsity
(µA
)
MacabeuParelladaChardonnayXarel·lo
I. +1.31 V
I. -0.38 V
I. +0.65 V
I. +1.01 V
Amperometric measurements for the white wine set.
Absorbance spectra obtained with the opticalsensor using DI water as reference
-0.1
0.1
0.3
0.5
0.7
0.9
300 350 400 450 500 550 600 650 700 750
Wavelength (nm)
Abs
orba
nce
Merlot 100%Cabernet 100%Pinot 100%
Abs. 420 nm
Abs. 520 nm
Abs. 620 nm
-0.1
0.1
0.3
0.5
0.7
0.9
300 350 400 450 500 550 600 650 700 750
Wavelength (nm)
Abs
orba
nce
Merlot 100%Cabernet 100%Pinot 100%
Abs. 420 nm
Abs. 520 nm
Abs. 620 nm
Sensor Signals: Optical sensors
-0.07
-0.06
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
202
249
296
343
389
435
481
526
571
616
661
705
750
793
837
880
923
966
1008
1051
1092
Mac 09/07 Abs Mitja
Par 78/07 Abs Mitja
Char 04/07 Abs Mitja
Xar 63/07 Abs Mitja
Red wines White wines
• Matrix of Data• Variables vs samples• We perform a PCA: all samples are
used for the training• Weight of samples: all are normalized to
1/sdv
• Matrix of Data• Variables vs samples• We perform a PCA: all samples are
used for the training• Weight of samples: all are normalized to
1/sdv
Data treatment
PCA analysis
• Scores distribution of all the samples all through the 2 C
• Loadings distribution of the variables (weight of each variable)
• Scores distribution of all the samples all through the 2 C
• Loadings distribution of the variables (weight of each variable)
•PLS 1. One model for each variable
• Comparison of our data with standard
method
•PLS 1. One model for each variable
• Comparison of our data with standard
method
Data treatment
PLS analysis
•One model for Ca detection
•Error ~ 0 for 2 PC
•Regression treatment
•One model for Ca detection
•Error ~ 0 for 2 PC
•Regression treatment
training set
test set
Classification of wines
-4
-3
-2
-1
0
1
2
3
4
-6 -4 -2 0 2 4 6
PC 1 (35%)
PC
2 (
17%
)
Chardonnay
Xarel·loParellada
MacabeuPenedes
MacabeuGarraf
-4
-3
-2
-1
0
1
2
3
4
-6 -4 -2 0 2 4 6
PC 1 (35%)
PC
2 (
17%
)
Chardonnay
Xarel·loParellada
MacabeuPenedes
MacabeuGarraf
white
-4
-3
-2
-1
0
1
2
3
4
-6 -4 -2 0 2 4
PC 1 (38%)
PC
2 (
27%
)
Merlot 08
Carbenet S.
Garnatxa
Trepat
Merlot 07
-4
-3
-2
-1
0
1
2
3
4
-6 -4 -2 0 2 4
PC 1 (38%)
PC
2 (
27%
)
Merlot 08
Carbenet S.
Garnatxa
Trepat
Merlot 07
red
Classification according to:• Grape variety• vintage year• geographical origin
Classification according to:• Grape variety• vintage year• geographical origin
PCA for monovarietal wines
Classification of wines
60 mixtures of
Cabernet Sauvignon, Pinot Noir and Merlot 0, 25, 50, 75, 85, 100%
Wine classification accordingto the percentage of grape
variety
Wine classification accordingto the percentage of grape
variety
Quantification ofeach variety
Quantification ofeach variety
0 20 40 60 80 100
0
20
40
60
80
100
Obt
aine
d %
Pin
ot N
.
Expected % Pinot N.
y = (0.96±0.08)x + (1.62±3.75)r2 = 0.976
0 20 40 60 80 100
0
20
40
60
80
100
Obt
aine
d %
Mer
lot
Expected % Merlot
y = (0.94±0.11)x + (3.01±4.68)r2 = 0.949
-4 -3 -2 -1 0 1 2 3
-3
-2
-1
0
1
2
3
4
PN >= 75 % CS >= 75 % MT >= 75 % Mescles 50:50 Mescles ternàries
PC
2 (
15 %
)
PC 1 (20 %)
(A)
-4 -3 -2 -1 0 1 2 3
-3
-2
-1
0
1
2
3
4
PN >= 75 % CS >= 75 % MT >= 75 % Mescles 50:50 Mescles ternàries
PC
2 (
15 %
)
PC 1 (20 %)
(A)
PLS for Chemical Wine Parameters
Sample VAD Total acidity pH Ca2+ Mg2+
Glycerol Macabeu 09/07 2.82 −3.7 −0.79 −0.7 −4.9 6.0
Parellada 77/07 0.07 9.5 −3.55 5.1 −0.1 2.0
Chardonnay 02/07 3.32 −2.7 1.16 5.7 −2.2 −4.3
Xarel·lo 44/07 −0.62 1.0 −4.76 4.0 −2.8 12.3
Sample VAD Total acidity pH Ca2+ K+
Glycerol 118/07 Trepat 4.45 0.6 -2.20 -5.1 -4.0 8.8
95/07 Garnatxa -4.05 -7.7 1.34 -6.1 6.7 2.3
49/07 Merlot -2.31 -2.6 -0.91 3.9 -0.7 3.8
45/08 Merlot -6.52 8.0 2.03 -2.4 -3.2 -5.6
105/07 Cabernet 0.99 -9.0 -4.01 8.7 -8.5 -2.7
Relative errors comparing with standard methods of analysis:
White wine
Red wineError < 5 % Error < 10 %
Quantitative analysis
VAD: volatil alcoholic degree
PLS for Optical Wine Paramenters
Sample Intensity of color Tonality Macabeu 09/07 4.67 1.09
Parellada 77/07 4.56 3.66
Chardonnay 02/07 0.87 2.48
Xarel·lo 44/07 7.29 −4.23
Sample Intensity of color Tonality CieLab a *
118/07 Trepat -89.74 2.20 -5.16
95/07 Garnatxa -44.57 -3.14 7.90
49/07 Merlot -32.31 4.30 8.98
45/08 Merlot 17.43 3.17 -7.16
105/07 Cabernet -10.22 -5.65 3.41
Relative errors comparing with standard methods of analysis:
Error < 10 %
Quantitative analysis
White wine
Red wine
• A multisensor system with different kind of sensors: potentiometric,
voltamperometric, conductimetric, optical � hybrid electronic tongue
• The system is capable of discriminating wines samples not only according to the
grape variety, but also to the origin and even the vintage.
• The system is able to quantify several parameters. The relative errors obtained are
below 10%.
• The ET could be applied in the cellars as a system for rapid detection, close to the
production and as an alarm system.
Conclusions
Acknowlegments:
Spanish R & D National Program (MICINN Project TEC2007-68012-C03-01/03)
ACC10 network GTQ-TECNIO
INCAVI: Santiago Minguez, Fina Capdevila
GTQ members specially: Manuel Gutierrez, Andreu Llobera
Tech. Univ. Braunschweig: S.Demming, S. Büttgenbach
Acknowledegements
Cecilia Jimé[email protected]
Sistemes multisensors aplicats al control de qualitatdels vins