Grasas y Aceites

96 Vol. 53. Fasc. 1 (2002), 96-114

Sensors: From Biosensors to the Electronic Nose

By Diego L. García-González and Ramón Aparicio

Instituto de la Grasa
Avda. Padre García Tejero, 4, 41012 Sevilla Spain
E-mail: [email protected]

CONTENTS 1. INTRODUCTION
1. Introduction.
2. Sensory quality.
For ages, there has been much interest in the
3. Sensors. design of devices for the sensing of food flavour. The
3.1. Metal oxide semiconductor sensors (MOS). devices, or instruments, are claimed to operate on
3.2. Metal oxide semiconductor field-effect transistor principles similar, in many aspects, to the human
(MOSFET) sensors.
3.3. Conducting polymer sensors. olfactory system. Today, the methodologies devoted
3.4. Acoustic sensors. to the sensory assessment of foodstuffs are based
3.5. Biosensors. on either the classical panels of trained human
4. The electronic nose. beings or the analysis of some chemical compounds by
5. The electronic tongue.
6. Data analysis. headspace-gas chromatography (-mass spectrometry)
6.1. Pre-processing data analysis. or the global evaluation of the odour intensity of
6.2. Multivariate data analysis. volatile compounds by the emergent sensor
6.2.1. Cluster analysis.
6.2.2. Factorial analysis.
technologies.
6.2.3. Discriminant analysis. The experts’ panel, usually the only official
6.2.4. Regression analysis. method for the sensory evaluation, allows the
6.3. Artificial neural networks. sensory qualification of foodstuffs by their own
References.
potential consumers, as well as the evaluation of all
the sensory perceptions in only one analysis. It has,
RESUMEN however, the great disadvantage of being a lengthy
and expensive methodology whose final result
Sensores: De los biosensores a la nariz electrónica. depends on many factors as, for example, the
panellists’ training and the specific vocabulary used
Los recientes avances en los sistemas de sensores han per- in the sensory analysis.
mitido el desarrollo de nuevas aplicaciones en muchos campos
tecnológicos. Este artículo de revisión describe el estado actual An alternative method is the identification and
de esta nueva tecnología, con especial énfasis en las aplicacio- quantification of the chemical compounds
nes alimentarias. El diseño, la tecnología y el mecanismo senso- responsible for the sensory perceptions. Non-volatile
rial de cada tipo de sensor son analizados en el artículo. También compounds are mainly focused on phenolic
se describen las principales características de la nariz y la lengua
electrónica (sensores de sabor). Finalmente, se describe breve- compounds and aglycons, responsible for tasting
mente el uso de algunos procedimientos estadísticos en sistemas perceptions (astringent, bitter and pungent) (Morales
de sensores. and Tsimidou 2000), while volatile compounds are
responsible for the aroma of foodstuffs (Aparicio et
PALABRAS-CLAVE: Athe nálisis sensorial – Biosensores -
Nariz electrónica – Sensores. al. 1996). The first set of compounds is quantified by
high performance liquid chromatography (HPLC)
SUMMARY
(Montedoro et al. 1992) while the quantification of
volatile compounds can be carried out by various
Sensors: From Biosensors to the Electronic Nose. analytical techniques (Morales et al. 1992). The latter
can be summarised as:
The recent advances in sensor devices have allowed the 1. those not involving concentration:
developing of new applications in many technological fields. This
review describes the current state-of-the-art of this sensor • direct injection.
technology, placing special emphasis on the food applications. • static headspace
The design, technology and sensing mechanism of each type of
sensor are analysed. A description of the main characteristics of 2. those with concentration-distillation.
the electronic nose and electronic tongue (taste sensors) is also
given. Finally, the applications of some statistical procedures in
• simultaneous distillation-extraction.
sensor systems are described briefly. • dynamic headspace/purge-and-trap.
KEY-WORDS: Biosensors – Electronic nose – Sensors –Sen- • supercritical fluid extraction.
sory analysis. • solid phase micro extraction
Vol. 53. Fasc. 1 (2002) 97

The usefulness of these techniques in sensory

analysis is based on the fact that volatile compounds Table I
are responsible for the foodstuff odours. Thus, an Application of the sensory panel in different
interesting approach has been carried out in virgin critical points of the food industry
olive oil to establish the relationships between
volatile compounds and flavour (Aparicio et al. 1996), Monitoring of the food ripening (e.g., wine, cheese)
and between phenols and taste (Morales and Quality assurance of the selected raw foods
Tsimidou 2000) using the mathematical algorithm of Monitoring of the cooking processes
the statistical sensory wheel (Aparicio et al. 1994). Monitoring of the fermentation processes
These techniques can assess the sensory quality
Monitoring other industrial processes (e.g., flavouring,
routine tests but cannot be used on-line because
blending, colouring).
they are time-consuming and they previously need of
Monitoring the storage of foods and food products
a sample pre-treatment or a volatile concentration
phase. (e.g., freshness and ageing control)
An alternative, recently developed, is the use of Quality assurance of the manufactured food products
sensors that have been widely used in many fields of Monitoring of the product-packaging interactions
analytical chemistry. Conceptually speaking, this kind Monitoring of the overall quality of the final food or food product.
of sensor is a device that is able to give a signal
proportional to the physical or chemical property to
which the device responds (Kress-Rogers 1997); for 2000). Colour and flavour are, however, the
example, the sensors transduce the signals sensations that mainly contribute to food
produced by magnetic, thermal and chemical fields. acceptability by consumers (Morales and Tsimidou
Based on these facts, the most common definition is 2000) or when monitoring food industry processes
that a sensor is a single transducer in which, due to a (Table I). Flavour is a complex sensation consisting
physical effect, a complete transduction process primarily of odour and taste, but is complemented by
takes place (Middelhoek 2000). The electronic tactile and kinesthetic sensations (Reineccious
integration of various sensors inside one set 1993). The importance of flavour in the global
constitutes an array of sensors, or the so-called sensory quality lies in the fact that it plays a crucial
electronic nose, that is the kind of low cost structure role in food recognition, selection, acceptance and
habitually used to analyse the aroma of the nutrition, as it is partly responsible for aiding the
foodstuffs. digestion of food humans (Ensor 1989).
The sensors have some advantages over other The main human sensor contributing to flavour is
cited techniques. They do not use reagents or need the olfaction system that detects the odorant
of any kind of pre-treatment, they have good levels of compounds. The basic characteristics of odorant
sensitivity and specificity, and besides they are rapid molecules are to have low relative molecular masses
supplying results to the analyst. This is a (up to approximately 300 Da), to be small and polar,
non-destructive technique that can be used in an on and also to be hydrophobic (Craven et al. 1996). In
line quality control during the manufacturing process. the nasal mucus, there are small odourant binding
But sensors have obviously disadvantages mainly proteins (OBPs) that accommodate the hydrophobic
focused on the reproducibility, recovery, and the molecules in solution (Breer 1997). The OBPs work
effect that humidity and temperature have on the as binders of the odourant compounds and starters
sensor responses. The analysis of these aspects, of the transduction reaction. After crossing the
advantages and disadvantages of the sensor mucus layer, the odourants reach the chemosensory
technologies, together with the applications of the ciliary membrane of the olfactory neuron in which the
different kinds of sensors in food science are the G-receptor proteins are located (Gardner and
objectives of this review. Finally, there is a schematic Bartlett 1994). More than 100 million olfactory cells
analysis of the mathematical procedures used to with these receptor proteins are inside the olfactory
analyse the sensor results as well as the application epithelium. The receptors interact with odourant
fields and the references. molecules and generate a second set of messengers
that cause excitation in the neurones. The signal is
processed in the olfactory bulb and then the
2. SENSORY QUALITY information is transmitted to the limbic system where
the associations between odour information and
Sensory quality consists on a set of quality recalled memories take place (Keller et al. 1999). The
attributes, that can be detected through sense information of the receptors is also transmitted to the
organs, like appearance, size and shape, colour, cerebral cortex where is combined with the taste to
viscosity, kinesthetic sensations, tactile sensations, produce the flavour sensation.
and chemesthetic sensations (stimuli elicited from On the other hand, the taste receptors are mainly
some chemical irritants), odour and taste (Angerosa situated on the tongue surface, over the entire oral
98 Grasas y Aceites

cavity, down the oesophagus and even on the main odour sensors while Table II shows the physical
trachea and larynx. When an appropriate soluble responses of these sensors to the concentration of
substance reaches the taste cell, the interaction of volatile compounds. When the analysts work with an
the molecule with the G-receptor protein leads to array of sensors of the same class (e.g., metal oxide
chemical changes that result in a neural impulse gas sensors made with doped plus undoped
(Winquist et al. 1999). The result is the tasting semiconductors) or combining different classes (e.g.,
perception that has four basic tastes (bitter, sweet, conducting polymer sensors plus piezoelectric
salt and sour) although the sensory assessors are crystal sensors), the resulting instrument is
able to distinguish between dozens of different habitually called “electronic nose”. Furthermore,
tastes, for instance qualifying virgin olive oil (Aparicio Winquist et al. (1999) have recently gone beyond
et al. 1994). describing the combination of an electronic nose and
an electronic tongue.
3. SENSORS
3.1. Metal oxide semiconductor sensors
Today, several commercial sensors are available (MOS)
on the market besides the great number of prototype
These sensors are the most commercially
sensors that are being designed by research
extended in the sensing systems. They consist of a
institutions year by year. All these types of sensors
ceramic former (tubular or flat) heated by wire and
exhibit physical and chemical interactions with the
coated with a metal oxide semiconducting film. For
chemical compounds when they flow over, or are in
this reason, they are usually called metal oxide, or
contact with, the sensors. The choice of sensors is
ceramic, gas sensors (Schaller et al. 1998) although
hence quite large, so that we have classified them
they can also be called Taguchi and Figaro because
into broad classes following an arborescent
of the inventor and the first company respectively.
structure. The odour sensors, tasting sensors and
The oxide coating may be either n-type or p-type
biosensors, constitute the first set. The latter two
semiconductors. The n-type semiconductor (SnO2,
groups are the result of a very recent research, and
ZnO, In2O3, WO3, Fe2O3, Ga2O3, TiO2) responds to
hence there is not a taxonomic structure for these
oxidising compounds and its thermal or photolytic
sensors yet. The sensors of the former group can
excitation results in an excess of electrons that
however be clustered into two great groups
increases the reactivity with oxidising molecules. The
according to their working temperature: hot sensors
p-type semiconductor (oxides of nickel or cobalt)
and cold sensors. The group of the so-called “hot
responds to reducing compounds and its excitation
sensors” mainly clusters the different kinds of metal
promotes reactions with reducing compounds due to
oxide gas sensors (based on doped or undoped
an electron deficiency in its valence band. These
semiconductors), while the group of the “cold
metal oxides (mostly SnO2, Ga2O3 and TiO2) can also
sensors” clusters the piezoelectric crystal sensors
be doped with metals such as Cu (Zhang and Liu
(bulk acoustic wave sensors, surface acoustic wave
2000), Pt plus Nb (Comini et al. 2000) and Sb, Bi, Fe,
sensors) and the conducting organic polymer
Au, Cd (Morimitsu et al. 2000), or metal oxides of Co,
sensors. Figure 1 shows schematic diagrams of the
Cr, La, Ce (Fukui and Katsuki 2000), Mg, Al (Xu et al.
2000), Ru, Rh and Ir (Lang et al. 2000) among
others. These doping metals help to diminish the
humidity and temperature dependence (Fukui and
Flat MOS sensor Tubular MOS sensor
Katsuki 2000; Morimitsu et al. 2000), and to improve
the sensitivity and the selectivity (Comini et al. 2000;
Mielle 1996).

Table II
MOSFET sensor CP sensor
Physical properties of the sensors described
in the review

Sensor Physical property

Metal-oxide sensors (MOS, MOSFET) Resistance and impedance

BAW sensor SAW sensor
Conducting polymer sensors Resistance and impedance
Electrochemical sensors Conductance, intensity and voltage
Figure 1 Acoustic sensors (SAW, BAW, QMB, Mass and frequency shift
Schematic diagrams of the main odour sensors: Flat and tubular Cantilever)
metal oxide semiconductor (MOS), metal oxide semiconductor
Calorimetric sensors Temperature
field effect transistor (MOSFET), conducting polymer (CP), bulk
acoustic wave (BAW) and surface acoustic wave (SAW) sensors. Optical sensors Optical properties
Vol. 53. Fasc. 1 (2002) 99

The metal film thickness further classifies sensors gases and surface oxygen ad-ions. They have a
into thin film (6-1000 nm) and thick film (10-300 µm) working temperature between 200oC and 400oC
sensors. The thickness is only related to the (Simon et al. 2001), and the organic volatiles are
technique used (physical or chemical vapour completely combusted to water and carbon dioxide
deposition for thin films versus screen printing or when settles the sensor surface.
painting depositions for thick films) but also with the These small size sensors are robust, they have a
response (faster for thin films), sensitivity (higher for quite good sensitivity, their signal processing is
thin films), and reproducibility (higher for thick films) simple, and they have a very low cost (Simon et al.
and cost (much more low for thick films). The screen- 2001). Their sensitivity is, however, affected by
printing techniques, based on a planar integrated humidity and poisoning materials. Although the
microelectronic technology, allow obtaining low size sensors are made of materials relatively resistant to
sensors with high performance in various application humidity and ageing, the humidity is still the main
fields. Nevertheless, screen-printed ceramic gas problem as the sensor sensitivity decreases when
sensors need improvements as lowing the power humidity increases. Another drawback is the
consumption, getting best selectivity and facilitating possibility of the sensor poisoning by sulphur
mounting technology. These aspects are reached compounds (e.g. H2S or SOx) that produces the total
with micromachined metal oxide gas sensors, where or partial irreversible destruction of the sensor.
sensitive layer is deposited in a thin dielectric film of Sulphur is known as an inhibitor for metal and oxide
low thermal conductivity, getting isolation between catalysts. On metals, sulphur chemisorbs an
substrate and gas sensitivity heated area (Simon et unoccupied orbital of the metal via its occupied lone
al. 2001). pair orbital, while, on oxides, sulphur is substituted
The mechanism of sensing of these sensors is for lattice oxygen atoms. This irreversible process
related to the semiconductor behaviour when can be caused by the presence of sulphur
exchanging oxygen between the volatiles and the compounds or some acids (e.g. acetic) in the
metal film. In non-operating, when only the carrier samples, e.g. cheeses, wines, vinegars, cruciferous
gas (e.g., synthetic air) flows in the sensor chamber, vegetables or lampant virgin olive oils. Ethanol also
the oxygen is adsorbed on the surface and inside the affects to sensor but only producing a temporary
metal coating, filling the sensor lattice vacancies, and blinding effect to any other analytes present in the
later oxygen attracts free electrons from the sample. Other undesirable characteristics of these
semiconductor material with which the sensor sensors are: their slow baseline recovery when
conductance increases. In operation, when the compounds with high molecular weight are
interactions between adsorbed volatile molecules analysed, their high working temperature that makes
and metal film take place, the conductance them inappropriate in environments containing
decreases due to the reaction with the adsorbed flammable chemicals, their poor specificity and
oxygen. Two mechanisms, that can occur selectivity, and the sensor drift mainly caused by
simultaneously, have been described in this phase sensitivity loss.
(Schaller et al. 1998). At relatively low temperatures, Most of the metal-oxide sensors show a
when reducing volatiles compounds are adsorbed on non-linear response to a given chemical compound
the surface, the charges are transferred between the or odour. Several attempts to find a satisfactory
volatiles and adsorbed oxygen. At high temperature,
however, the loaded oxygen reacts with the
adsorbed volatile compounds, and conductance
decrease in a non-linear manner. Zhang and Liu
(2000) have studied the particular cases of NO2 and
CO2 over various transition metals. Furthermore, the
authors have suggested some mechanisms that
produce more oxygen vacancies in the material,
leading to a decrease in the free electron
concentration and an increase in sensitivity of the
material (e.g., the substitution of tin ions for copper
ions).
Recently other measurement parameters have
been studied, in addition to the electrical resistance
change. The temperature drop during the sensing
reaction has been proposed (Liu et al. 2000; Takada
Figure 2
2000). The temperature decrease is caused by a Responses of a MOS sensor (Alpha MOS 4000) to virgin olive
change in the thermal conductivity of the sensors oils characterised by different levels of oxidation. Oxidized olive
and it is related to the reaction between reducing oil samples were prepared placing virgin olive oil samples in the
beam of an ultraviolet light for different days (0, 3, 8, 12 days).
100 Grasas y Aceites

relation between the sensor responses and the MOSFET are robust sensors with low sensitivity
analyte concentration or the odour intensity have to humidity but whose selectivity and sensitivity can
been done. The formula R=1/ACα (where, R is the be affected by the working temperature (75-200 oC),
response intensity, A and α are constant, and C is the kind of metal gate (as described) and microstructure
concentration) has been proposed to explain the of the catalytic metal. Their robust characteristic
sensor response at moderate concentration and makes them particularly useful in environmental
below the saturation level of the sensor (Wilson et al. applications (Spetz et al. 2000). They show good
2000). Khol et al. (2000) have also expressed the sensitivity to toxic and flammable substances (Zhao
sensor response as R-R0/R0=K(C)1/2 (where K is a et al. 2000) and because of their sensitivity to
sensor-dependent constant). Mielle et al. (2000) ethanol they are also used in food cooking,
have found another relation between the compound fermenting, and wine-making processes (Zhao et al.
concentration and the sensor response: 2000). Finally, these sensors have been used for the
R-R0/R0=10(a +b)logC, where “a” and “b” are constants ethylene measurement during fruit ripening process
for each compound. (Winquist et al. 1990).
The application of these sensors has been mainly
focused on the detection of gas leakage, combustible 3.3. Conducting polymer sensors
and toxic gases (Shurmer and Gardner 1992; Xu et
al. 2000; Ménil et al. 2000; Simon et al. 2001) but Since Diaz et al. (Persaud and Travers 1997)
their applications on food technology have always designed the first polypyrrole sensor, the interest in
been of relevance. Thus, these sensors have been conducting polymers (CPs) has increased year by
applied to detection of volatile compounds of year. These sensors are made of a broad range of
different foodstuffs as coffee (Gardner et al. 1992), monomers that are polymerised by chemical or
milk (Sberveglieri et al. 1998), strawberry electrochemical methods. A thin film of polymer is
(Hirschfelder et al. 1998), juices (Bazemore et al. deposited onto a sensing substrate formed by two
1997), meat (Braggins et al. 1999) and edible oils parallel platinum or gold plated electrodes with a
(Bazzo et al. 1998). Figure 2 shows the responses of insulating base such as oxidized silicon (Partridge et al.
a MOS sensor (AlphaMOS 4000) to virgin olive oils 1996). The most common polymers used in the
characterised by different levels of oxidation. synthesis of CPs are polypyrroles, polyanilines,
polytiophenes, butane to decane sulphonic acids,
3.2. Metal oxide semiconductor field-effect para-toluenesulphonic acid and its tetra-ethylammonium
transistor (MOSFET) sensors salt, sodium monohydrogenosulphate and
tetraethylammonium tetrafluoroborate (Mielle 1996)
These sensors comprise three layers, a silicon although new polymers are being synthesized to
semiconductor, a silicon oxide insulator and a open these sensors to new applications. Thus,
catalytic metal (Pd, Pt, Ir, Rh) called the gate polymers with a conjugated bonding system are
(Schaller et al. 1998). This design corresponds to the being now used due to the excellent electrical
transistors designed by Lundström in 1975 characteristics of these materials for microelectronic
(Middelhoek 2000). This kind of transistors operates device manufacturing (Rella et al. 2000). Sensors
by means of three contacts, two allow the current in based on aniline and derivates can be doped at
(source) and out (drain), and third acts as the gate specific ratios with other compounds, as sulphonic
and regulates the current through the transistor. acids (Koul et al. 2001) or HCl (Campos et al. 2000),
When a voltage is applied on the gate, an electric so enhancing their electrical-optical properties and
field is generated which affects the transistor solubility in organic solvents. The dopants provide
conductivity. If vapour (polar compounds) interacts extra charges in the chain of the polymer with the
with the gate, then the current flowing through the formation of charged defects, such as spinless
transistor (sensor) changes and a shift of the bipolarons (Rella et al. 2000) or small regions of
conductance is produced. The interaction between positive charge in the polymer chain that provide
the vapour and the gate depends on the gate mobile holes for electron transport.
structure; it can be thin, made of a porous metal film The CP sensing characteristics are based in a
(6-20 nm), or thick, built with a dense metal film measurable change in the electrical conductivity
(100-200 nm). The latter kind of gate (transistor) when CPs are exposed to volatile compounds. A
works well with vapours containing molecules that fixed voltage among the electrodes produces a
can dissociate hydrogen, because these atoms lead constant current that passes through the conducting
to a potential change in this kind of transistor. Almost polymer. When gas flows on the sensor, the volatile
all kind of compounds, however, respond well when compounds adhere to its surface, and an adsorption-
the gate is thin, due to the probable mechanisms of desorption process occurs on the polymer layer
voltage shifts described by Lundström et al. (1975; which alters the electron flow in the system and
1990) and Spetz et al. (1992). hence the sensor conductivity. Therefore these
Vol. 53. Fasc. 1 (2002) 101

changes in conductivity are based in the mobility of results (Partridge et al. 1996). The sensors have also
charges along the conjugation of the π-backbone of other drawbacks as their too long response time
the polymer. Counterions (stabilised ions that are put (20-40 s), in comparison with MOS (Mielle 1996),
inside the polymer by the manufacturer in order to and their inherent drift over time or with changes in
balance other charges inside the microstructure of temperature. The main problem is, however, that they
polymer layer) can produce some modifications such are extremely sensitive to moisture.
as changes in morphology of polymer-dopant Because CPs are sensitive to many types of
complex, in the concentration of chemical defects volatiles, many applications are described in the
(polarons or bipolarons) or the oxidation state of the literature. They are sensitive to alcohols, acetone,
polymer (Tan and Blackwood 2000) although the benzene and other polar volatile chemicals (Hatfield
operating mechanism is not yet well understood. et al. 1994). References show applications of CPs in
Spectral studies, as UV-VIS spectroscopy, FTIR foodstuffs like beer and other alcoholic beverages
spectroscopy, and electron spin resonance (Pearce et al. 1993; Slater et al. 1993), olive oil
spectroscopy (ESR), are being carried out to know (Aparicio et al. 2000), orange juices (Bazemore et al.
more about the interactions occurring during the 1998), fishes (Luzuriaga et al. 1999), milk (Magan et
sensing process (Tan and Blackwood 2000; Koul et al. 2001). Figure 3 shows the responses of polymer
al. 2001). sensors (AromaScan A20S) to virgin olive oils
The sensor characteristics explain why they are characterised by different levels of oxidation.
ones of the most used in sensor arrays. They are Conducting polymers are often associated to
easily fabricated, their power consumption is low, other transducer mechanisms as sensing materials.
and they are robust and can work at room This is the case of the optical sensors based on CPs,
temperature. As it is said above, there are many types since these polymers show spectral changes in the
of polymers, including a broad range of derivates, UV-visible region when the doping/dedoping process
and more types of polymers appear each year. The happens (Kondratowicz et al. 2001). Optical sensors
nature of the dopants also modifies their initial incorporating CPs sensitive to ammonia, humidity,
properties (Koul et al. 2001), and it means the pH and acetic acid have been recently developed
possibility of designing sensors tailored to specific (Kondratowicz et al. 2001; de Marcos et al. 2000).
series of volatiles (e.g. ketons, aldehydes or Sensors based on field effect transistor (FET)
alcohols) or towards a particular application (e.g. incorporating a CP as the gate material (Hatfield et
characterisation or authentication). The oxidation al. 2000) or quartz-crystal microbalance associated
state of the polymer can also be changed after to CP (Slater et al. 1992) are other examples. Finally,
deposition to tailor the sensing characteristics CPs provide a surface to immobilise biological
(Partridge et al. 1996). CPs are resilient to poisoning elements like enzymes or antibodies, in order to
by the compounds that inactivate the described prepare biosensors. Recently, a dielectric monitoring
inorganic semiconductor sensors but surprisingly has been proposed to study the adsorption process
their lifetime, only about 9-18 months, is shorter. of biomolecules onto CPs, such as DNA onto
The random nature of the deposition process is, polypyrroles (Saoudi et al. 2000).
however, a problem for the production of
reproducible sensors, although the use of four 3.4. Acoustic sensors
electrodes, instead of two, seems to improve the Electroacoustic devices based on the
piezoelectrical properties of quartz material have
been successfully used to implement sensors since
1964 (King 1964). These sensors are based on the
propagation of acoustic waves produced by
piezoelectrical materials (e.g. quartz or LiNbO3 or
Si-SiO2-ZnO) in a multilayer structure (Lucklum and
Hauptmann 2000). Surface acoustic wave (SAW)
and bulk acoustic wave (BAW) are the most common
of those acoustic sensors. The operation mode of
these sensors is based on the physical changes
produced on their membranes made of chemically
interactive materials (CIM). The physical changes
that can be detected are related to the piezoelectric
materials used for the acoustic transduction, or the
mass density and (visco-)elastic properties of CIM.
Figure 3 The operation principles of SAW and BAW
Mean of the raw response of conducting polymer sensors to sensors are very similar (D’Amico et al. 1997).
a pure virgin olive oil and samples spiked with different
percentages of a virgin olive oil rancid standard. However, SAW-type sensor exploits the propagation
102 Grasas y Aceites

of radiofrequency (102-103MHz) oscillations (e.g. layer that interacts with the analyte (Battiston et al.
Rayleigh waves) along a layer consisting at least of a 2001). The cantilever-based sensor oscillates at a
substrate covered by the CIM. The acoustic waves fixed resonance frequency, and any change of the
are transmitted by the electrical field generated by mechanical properties is detected by measuring the
electrodes deposited at the surface of the solid. SAW shift of the resonance frequency (dynamic mode) or
devices produce bi-dimensional waves travelling the static deflection (static mode). The mechanism of
through the surface of the crystal. Depending on the sensing of microcantilever is similar to SAW and
geometry of the acoustic structure and on the BAW sensors. The molecules are absorbed in the
frequency scaling, SAW sensors can produce active layer, increasing the mass of the cantilever
different types of waves: Rayleigh waves, surface (Porter et al. 2001). This change of mass loading
transverse waves, Bleustein-Gulyaev waves, and results in a change of the vibration frequency or
Lamb and Love waves. The choice of one of these amplitude. Thus, the response of a cantilever is
waves or the piezoelectric material (ZnO, AIN, characterized by its geometrical dimensions and the
Bi12GeO20, α-SiO2, LiTaO3, LiNbO3) is influenced by vibration frequency among others (Battiston et al.
the characteristics of the membrane and by the 2001).
operation in gaseous or liquid environments. From Acoustic sensors have various advantages over
the sensing point of view, the presence of volatile other sensors, as high sensitivity and short response
compounds produces the adsorption of molecules time, low power consumption and size, and
on the sensor and hence changes on the properties robustness (Gutierrez et al. 2000). The sensitivity of
of the CIM that affect both the phase velocity and the the acoustic sensor is related to the operating
propagation loss of the acoustic wave. The resulting frequency (SAW: 100MHz-1GHz; BAW: 10-30 MHz),
is a frequency shift whose magnitude is related to the and it increases as the square of the fundamental
amount of material adsorbed onto the layer sensor. frequency does (Ballantine and Wohltjen 1989).
A BAW sensor is a piezoelectric resonator, often Furthermore, the frequency shift is directly
made of quartz, with one or both surfaces covered proportional to the sensing layer thickness for a
with a sensitive coating (membrane) of a few µm specific compound (Li and Ma 2000).
(1µm-10nm) thickness (D’Amico et al. 1997), whose Disadvantages of acoustic sensors are the
chemical and physical (thermal) characteristics are temperature and humidity dependence, the difficulty
similar to the gas-chromatography stationary of replacing sensors, the poor reproducibility in the
phases. The structure is connected to an amplifier to deposition of the coating material (Mielle 1996) and
form an oscillator whose three-dimensional waves certain level of noise (overall in SAW) because of
travel through the crystal (Schaller et al. 1998) at a their high frequency (Schaller et al. 1998). In the
frequency of 10-30MHz. In the presence of volatile case of QCM sensors, their main disadvantage is
compounds there is a process of adsorption or that they need of a thin crystal to increase the
absorption of the chemical species that alter the sensitivity that prevents their inclusion in multisensor
physical properties of the membrane and hence arrays (Ferrari et al. 2000).
affects the resonant frequency of the structure. The The acoustic sensor applications can be divided
changes are detected by measuring the shift in the between physical and chemical applications
resonance frequency; the magnitude of the shift is (Ballantine and Wohltjen 1989). Some physical
related to the mass of the compound although the properties can be measured by acoustic sensors, as
response can be no-linear due to viscoelastic temperature, pressure, electric field, displacement
behaviour of the membrane. Because of the and flow (Cai and Grimes 2000). The common
response is proportional to the mass adsorbed, this characteristic of these sensors is that they have
device (a quartz disk sandwiched between two magnetoelastic thick-film as coating material. The
electrodes) is also called quartz crystal microbalance chemical applications require two kinds of sensors,
(QMB or QCM) (Ferrari et al. 2000) and it was liquid-phase sensors and vapour-phase sensors,
already proposed as a promising sensor for gas according to the medium in which they operate
detection in 1964 (King 1964). (Ballantine and Wohltjen 1989). The vapour-phase
Other kinds of acoustic sensors are the flexural acoustic sensors have been used to detect gases
plate wave (FPW) sensors and cantilever-based and volatiles as NO2 (Müller et al. 2000), NH3
sensors. FPW sensors are similar to SAW sensors (Ballantine and Wohltjen 1989), xylene, toluene,
although they are coated with a membrane whose tetrachloroethylene (Dickert et al. 1999), H2, toxic
thickness is much smaller than the acoustic gases (McGill et al. 2000), organophosphorous
wavelength (Cai et al. 2000) and, in consequence, compounds (Zimmermann et al. 2001) and SO2 (Qin
the mechanical flexure occurs in the entire et al. 2000). Acoustic sensors have also been used
membrane. Cantilever-based sensors, which derive to analyse the headspace of foodstuffs as olive oil
from the atomic force microscopy, consist of samples (Di Natale 2001). SAW sensors with a
free-standing cantilever beams coated with a sensor sensitive material containing a high degree of
Vol. 53. Fasc. 1 (2002) 103

hydroxyl groups is used to detect the humidity sensors, and the current generated by the oxidation
because it produces changes in the polarity of the or reduction of electroactive species at the surface of
coating material (Dickert et al. 1999). Humidity the working electrode is measured (Griffiths and Hall
sensors have been manufactured using polymer 1993). In amperometric biosensors, redox enzymes
sensitive to wet air, such as polyimid (Bernou et al. (oxidoreductases) have a special importance, since
2000). Although SAW sensors are no suitable to be the transformation of the substrate takes place with a
used in liquid-phase because of their propagation transfer of electron between redox enzymes and
loss inside the liquid interface (Tong and Zhang electrodes. These redox enzymes are oxidases,
1999), some authors have used them to measure the peroxidases and dehydrogenases (Reviejo and
flow of liquid samples (Nomura et al. 2001), or to Pingarrón 2000). There are four main categories of
determine the liquid density and viscosity with a amperometric sensors (Palleschi and Cubadda
Love-mode acoustic sensor (Herrmann et al. 1999). 2001): Oxygen electrodes, hydrogen peroxide
Finally, SAW sensors are also used in scientific areas electrodes, carbon-base mediated electrodes and
as different as biochemistry, electrochemistry NADH electrochemical sensors.
(Ballantine and Wohltjen 1989) and immunosensing The potentiometric biosensors measure the
(Freudenberg et al. 2001). changes of potential at constant current (usually
zero). Three kinds of transducer are used in the
3.5. Biosensors potentiometric biosensors: ion selective electrodes
A biosensor consists of an immobilised biologic (ISEs), gas sensing electrodes, and field effect
molecule (enzymes, cellules or antibodies) next to a transistors (FETs) (Griffiths and Hall 1993).
transducer, which transforms chemical signal into an In the case of conductimetric biosensors changes
electric signal or into other kind of output as optical, of conductivity between two electrodes are
acoustic and heat signal when an analyte reaches to measured. This kind of transduction technology is
it. (Arroyo 1998). Arnold and Meyerhoff (Kress- not widely used due to low selectivity.
Rogers 1997) defined biosensors as “a self- Nowadays, lots of studies have been done to
contained analytical device that responds selectively fabricate electron-transferring interfaces between
and reversibly to the concentration or activity of redox enzymes and electrodes in order to perform an
chemical species in biological samples”. Biosensors enhanced electrochemical biosensor (Shinohara et al.
then consist of a biological sensing element and a 2000). These mediators are able to replace oxygen as
transduction element. an electron acceptor and to operate at a potential that
The biological element (Griffiths and Hall 1993) greatly reduces the effects of other electrochemically
can be made of catalytic (enzymes, microorganisms active species (White and Turner 1997). Some of
and tissues) or non-catalytic components these mediators are ferrocene and its derivatives,
(antibodies, receptors and nucleic acids). The cobaltphtalocyanine, tetrathiafulvalene, electropolymers
characteristics of biological elements allow clustering (Albareda-Sirvent 2000), and proteins (Shinohara et
the biosensors in three great groups: metabolism al. 2000).
biosensors (enzymes, combinations of enzymes and Optical biosensors are based on optical changes
cofactors, and cells like algae and bacteria), affinity while the biochemical reaction takes place. If a light
biosensors (antibodies, “immunosensors”) and beam is propagated through a waveguide, then the
recombinant biosensors (DNA probes). internally reflected light generates an
The transduction element allows the electromagnetic “evanescent wave” that can be used
transformation of the chemical signal, obtained from a to excite fluorescent molecules at the surface of the
biological process, into another kind of signal. According waveguide. The “evanescent wave” can couple with
to the kind of the transduced signal, biosensors can also the electron plasma of a metal when the waveguide
be grouped in electrochemical (amperometric, is coated with a metal thin layer. It causes the
potentiometric, conductimetric), optical, calorimetric electrons to oscillate and thus generating a surface
and acoustic biosensors (Griffiths and Hall 1993). plasmon wave (Griffiths and Hall 1993) that is
The biosensor sensing mechanisms depend on characterised by high sensitivity and specificity. The
the transduction technology. In the case of surface plasmon is an electromagnetic wave
electrochemical biosensors for example, the associated with the longitudinal oscillation of the free
transduction system is based on an electrochemical electron gas on the interface of the metal and the
nature process, and so the transduced signal is dielectric (Caide and Sui 2000). Recently, surface
electric. Most of the commercial biosensors are plasmon resonance interferometry (SPRI) has been
electrochemical; amperometric biosensors being the established as a novel method (Nikitin et al. 2000)
most widespread because the electrochemical that gains in sensitivity and resolution for the
instrumentation is relatively simple and inexpensive. detection of a multiplicity of analytes.
The current flows at constant potential with respect Finally, there are calorimetric and acoustic
to a reference electrode in the amperometric biosensors. The calorimetric biosensors are based
104 Grasas y Aceites

Table III
Methods for immobilising the enzymes of biosensors

Method for enzyme immobilisation Brief description References

Entrapment Immobilisation in matrices like gels, polymers or pastes. Ortega et al. 1998
Microencapsulation The enzymes are enclosed within semipermeable Ramos et al. 2001
polymer membranes.
Physical adsorption van der Waals attraction forces between an enzyme Lüth et al. 2000
and a solid support surface.
Covalent binding Enzymes and carriers are bound by covalent bond. Senillou et al. 1999
Cross-linking Formation of three-dimensional linkings between Dzyadevych et al. 2001
the biological material and bi- or
multifunctional reagents.

on the fact that almost all biochemical reactions are In military defence, biosensors could be used to
exothermic. Their commercial success is very limited detect nerve gases and other products used in
to the date, thermopiles and enzyme thermistors chemical warfare (Griffiths and Hall 1993).
being the most used. The acoustic biosensors Biosensors are applied to the control processes in
measure changes in the acoustic properties of the pharmaceutical, chemical and petroleum industries,
sensors. There are two types: surface acoustic wave mining and construction (Griffiths and Hall 1993).
or bulk wave (Griffiths and Hall 1993). The environmental control of toxic compounds and
The principal advantage of enzyme biosensors is pollutant gases (Köhler et al. 2000) and wastewater
related with their high specificity (stereoselectivity toxicity (Liao et al. 2001) can also be done by
and regiospecificity) and catalytic activity. However, biosensors. Finally, biosensors have been also used
they have the disadvantages of low stability and in Food Technology to monitor the quality control of
possibility of being reusable (Arroyo 1998). These raw materials and on-line or off-line manufacturing
inconveniences can be partially avoided by the processes. Thus, glucose and sucrose biosensors
immobilisation of the enzymes, which consists of have been used to control the process of alcohol
fixing enzymes onto a surface by physical or fermentation (Ferreira et al. 2001), and to detect
chemical methods in order to obtain an insoluble enterotoxins (Tempelman et al. 1996) and bacteria
by-product with catalytic activity. The immobilisation (Su et al. 2001) in many foodstuffs. Recently, a
provides advantageous effects over enzymes such biosensor based on catalase enzyme has been
as an activity increment, high reusability and developed for hydroperoxide determination in extra
feasibility of handling and control. But the enzyme virgin olive oil samples (Campanella et al. 2001).
biosensors also have inherent disadvantages such as
heterogeneity of the system, high cost, and changes in 4. THE ELECTRONIC NOSE
the conformation of the enzymes (Arroyo 1998). The
conventional methods of the enzyme immobilisation Recently, sensors with sensitivity to a wide range
include covalent binding, encapsulation in a polymer of volatiles compounds have been integrated in a
or cross-linking to a suitable supporting matrix (Lillis system called electronic nose. The name of
et al. 2000). Table III shows different methods to ‘electronic nose’ is due to the similarities between
immobilise the enzymes. this instrument and the physiological system.
The applications of biosensors spread over a lot Gardner and Bartlett (1994) defined electronic nose
of fields like clinical, agriculture, military, industrial as an instrument, which comprises an array of
and food areas although their commercial set up is electronic chemical sensors with partial specificity
being carried out slowly. In medicine, biosensors are and an appropriate pattern-recognition system,
used in the assessment of drug or biomolecule capable of recognising simple or complex odours.
concentrations in blood, serum or other corporal fluid Using this definition it can be though that electronic
in-vivo or in-vitro measurements, and data are used noses are composed of parts such as arrays of
in diagnostic or on-line monitoring (Liu and Li 2000). non-specific solid state gas sensor, a variety of
The agricultural applications of biosensors are transducers, data collectors and data analysis tools,
focused on the detection of viral, bacteria and fungal all oriented to the classification and quantification of
infections, the use of fertilizers and pesticides chemical clusters of volatile compounds, including, in
(Griffiths and Hall 1993), and volatiles released by particular, odours (D’Amico et al. 2000). Nevertheless,
damaged plants in glasshouses (Schütz et al. 2000). the term “electronic nose” is being incorrectly used to
Vol. 53. Fasc. 1 (2002) 105

name different systems, as mass-spectra based

system. Other authors use “sensor array technology”
(Mielle 2000), “electronic olfactometry” (García et al.
2001), “active odour sensing system” (Nakamoto et al.
2001) and “artificial olfaction system” (Di Natale et al.
2000), between other synonyms, to refer to a gas
sensor system.
When an analysis is carried out with an electronic
nose, the issues that should be taken into account
are: the sampling method, the sensor array and the
sensor cleaning. The first of them is the sampling
method. Previously, the sample is put in a sample
chamber at a controlled temperature. In the odour
analysis, the headspace is analysed rather than the
sample itself. There are two methods for sampling Figure 4
the headspace of the material to be analysed: static Typical response of a MOS sensor (Alpha MOS 4000).
headspace analysis and dynamic headspace Previously, data were preprocessed using the fractional
analysis. Static headspace consists of analysing an difference equation.
aliquot from the vapour phase, in equilibrium with the
sample, in a sealed vial subjected to a determined with a slight change in selectivity is to provide an
temperature over certain time. The so-called overall pattern or “fingerprint” as a single response,
headspace generation time allows a generation of which is characteristic of the odour. The power of
volatile compounds until a constant concentration of discrimination of a sensing system increases with the
them in the headspace. The volatile compounds are number of sensors in the sensor array (Shurmer and
transferred into the sensor chamber once the Gardner 1992), since each sensor provides information
equilibrium between vapour and liquid/solid phase is about a broad profile of compounds. However, a large
reached. number of sensors can provide useless redundant
Dynamic headspace, on the contrary, consists of information and noise. Furthermore, data from an
sweeping the volatile compounds of the sample, array composed by sensors with different chemical
subjected to a determined temperature, with an inert principles are more effective than those provided by
gas at a controlled flow and passing them through sensors with the same principle (Stetter et al. 2000).
the sensor chamber. In liquid samples, stirring the The interaction between the volatile compounds
sample or bubbling an inert gas through the sample and the sensitive sensor surface provides
usually increases the generation of volatile physical/chemical changes in the sensor. These
compounds. changes result in an electrical measurable signal that
Many variables affect the sampling method, such is recorded along the measuring time. The typical
as the temperature, the surface area of the sample, response of sensors (e.g., MOS and CP sensors)
and the volume of the headspace (Hodgins 1997). (Figure 4) consists of an initial slope related to an
Furthermore, the concentration of a volatile adsorption process of volatiles onto the sensor
compound in the liquid/solid sample and its partial surface. After a while, a dynamic equilibrium is
pressure are decisive in the concentration of the reached and the volatiles are adsorbed and
same volatile compound in the vapour phase (Mielle desorbed at the same speed (Schaller et al. 1998).
1996). Schaak et al. (1999) carried out a study of the At this stage, the sensor signal is constant. Finally,
correlation between metal-oxide sensor response the volatile desorption occurs and the sensor signal
and the vapour pressure of pure chemical shows a negative slight slope; the sensor tends to
compounds. Other variables are the humidity and recover its initial value.
temperature of the carrier gas used in the sampling After the measuring time, the sensors must be
process. Both factors affect the sensor signal, so that cleaned to get the baseline again and avoid the
they must be precisely controlled. On the other hand, cross-contamination between samples. The cleaning
the material of vessels and tubes should be inert as method depends on the kind of sensor. In the case of
glass, stainless steel, and PTFE (Hodgins 1997). conducting polymers a clean air, previously passed
After the sampling method, the volatile over a 2% water/butanol solution, can be blown over
compounds are located in the sensor chamber the sensor array (Aparicio et al. 2000). In metal-oxide
where the sensor array is placed. The sensor array sensors clean carrier gas is blown continually over
can be constituted by cold, hot sensors or both kinds the sensor during the non-measuring time (González
of sensors (Lucas et al. 1998). Sensors usually have Martín et al. 1999; Bazzo et al. 1999), since a
a poor selectivity, for instance metal oxide gas dynamic cleaning provides a faster recovery of the
sensors (Mielle 1996). The aim of combining sensors baseline (Bazzo et al. 1999).
106 Grasas y Aceites

The sensor responses depend on many variables solid-phase microextraction (SPME) technique
(Gardner and Bartlett 1994): the flow of the carrier (Guadarrama et al. 2001).
gas, the nature of the odour, interaction kinetics, the The commercial electronic noses have allowed
sensing material (nature and substrate), the odour the development of new applications of this kind of
diffusion within the sensing material, and the ambient system. There are many applications related to toxic
conditions (temperature, humidity, pressure, etc). gases, environment and medicine. However, the
During the analysis, the odour sensing system main applications are found in food industry. The
should have an electronic control of sensor applications in foodstuff include quality assessment
temperature, valves, sample temperature, frequency (Sinesio et al. 2000), control of manufacturing
of the acoustic sensors, etc. Furthermore, data from process (Zondervan et al. 1999), checking the aroma
sensor responses must be digitised and acquired in changes during the storage (Luzuriaga and Balaban
a PC. After the analysis, data are pre-processed and 1999), and study of packaging materials (Werlein
included in a multivariate data analysis. 2001). Table IV summarises applications of the
Although electronic noses are useful for rapid electronic nose in Food Science.
on-line screening, many inconveniences need to be
overcome before. The main limitations of an e-nose
are related to drift, noise, repeatability, environmental 5. THE ELECTRONIC TONGUE
influence (sensitivity to temperature and humidity),
poisoning and non-linearity in sensor response Recently, the same concept of artificial sense
(Craven et al. 1996). Many studies are being carried applied to electronic nose has been applied to the
out to enhance the performance of the electronic sense of taste. So, new taste sensors have been
nose. The improvements of the electronic nose can developed to be integrated in an ‘electronic tongue’.
be focused on the sampling system, new sensing D’Amico et al. (2000) defined electronic tongues as
materials and transducers combining different sensors, instruments that include arrays of non-specific solid
and multivariate statistical procedures (Stetter et al. state ion sensors, transducers even of different types,
2000; Göpel 2000; Craven et al. 1996). Recently, the data collectors and data analysis tools, all oriented to
electronic nose has been coupled with other the classification of liquid samples and the
techniques. The coupling of gas chromatography (GC) quantification of chemical species presented into them.
with a sensor system has been proposed (Kohl et al. The scheme of the system is very similar to that
2000; Masila et al. 1999). This coupling shows an of the electronic nose, except the conditioning of the
improvement in reproducibility. Sensor arrays have also sample and the measurement (D’Amico et al. 2000).
been coupled with a thermal desorption unit, with a Moreover, the transducer mechanisms of the taste
tenax trap (Muenchmeyer et al. 2000) or with sensors are different. Winquist et al. (1999) used an
electronic tongue based on pulsed voltammetry.
These authors used metallic sensors that are
Table IV affected by an onset of a voltage pulse. Authors
Main applications of the Electronic nose in Food measured the current transients and obtained a
Science voltammogram based on large amplitude pulsed
voltammetry (LAPV). On the other hand, Toko (2000)
Foodstuffs References
manufactured taste sensors that consisted of
polymer/lipid membranes and Ag/AgCl electrodes
Beer Shurmer and Gardner, 1992; Wilson that produce electrical signals in response to taste
et al. 2000; Pearce et al. 1993 substances.
Wine Guadarrama et al. 2001; Di Natale et al. 2000
Many applications of the electronic tongue have
been suggested. Electronic tongue has been applied
Meat products Braggins et al. 1999; Vernat-Rossi 1996
in tea (Ivarsson et al. 2001), soy sauce (Iiyama et al.
Fish Luzuriaga and Balaban 1999
2000), wine and mineral water (Legin et al. 2000)
Fruit Winquist et al. 1990; Bazemore et al. 1997;
between others. Besides, the combination of
Hines et al. 1999; Sinesio et al. 2000 electronic nose and electronic tongue data provides
Coffee Aishima, 1991; Singh et al. 1996; Gardner an improvement in the classification rate (Winquist et
et al. 1992; Pardo et al. 2000b al. 1999; Rong et al. 2000).
Tea Lucas et al. 1998
Olive oil Aparicio et al. 2000; Stella et al. 2000;
6. DATA ANALYSIS
Fat Pardo et al. 2000a
Milk Carrapiso et al. 2001; Muhl et al. 2000
The quantity and complexity of the data collected
Cereals Magan et al. 2001; Zondervan et al. 1999
by an electronic nose do not allow the use of
Chocolate Philip et al. 2000; Schnürer et al. 1999 conventional data analysis. The electronic nose
Werlein 2001 produces a tri-dimensional (n, k, t) output matrix,
Vol. 53. Fasc. 1 (2002) 107

Table V 6.1. Pre-processing data analysis

Equations suggested for the normalization of the
sensor baselines Gutiérrez-Osuna and Nagle (1999) suggest three
steps for the data pre-processing: baseline
manipulation, compression of data and normalization,
Normalization Equation
although the order of these operations is variable. The
Difference X= R-R0 baseline manipulation is a transformation based on the
Relative X= R/R0
initial value of the sensors; Table V shows the most
common transformations.
Fractional difference X= (R- R0)/R0 or X= (R0- R)/R0
Logarithm X=log (R-R0) Data compression is necessary to obtain a matrix
with a reasonable size; for example, a typical
electronic nose (18 sensors, a sampling frequency of
where “n” is the number of samples, “k” is the number 0.5s and sampling time of 600s) will produce 21600
of sensors and “t” is the number of points supplied by data per sample. Data compression consists in
each sensor. Such an amount of information requires selecting the regions of the sensor response with the
a data pre-processing step to remove redundant and most differential properties without removing the
irrelevant information but retaining the original essential information. There are many algorithms for
information as much as possible. extracting the information from the time dependant
sensor response. Hermle et al. (1999) have

Table VI
The most applied algorithms to the data compression

Model Description/equation
1,2,3,4
Steady-state or maximum Y = R(tmax), maximum response
1,4
Transient integral Y = ΣR(t)∆t
1
W = ΣR(t)∆tK(t)∆t ; K(t) = 1/(1+((t-c)/a) )
2b
Windowed Time Slicing
5
Window averaging The average of the neighbour 100 data.
5
Response Time A datum per time step.
2,3,4
Maximum derivation Y = Max (R’(t))
2
Time for the signal to reach its maximum Y = Max{t / R(t)=Max}
6
Time for the signal to reach 75% of the full response Y = Max{t / R(t)=0.75Max}
2
Area below the signal Y = Sum(R-R0)

1Gutiérrez-Osuna and Nagle 1999; 2 Hermle et al. 1999; 3 Pardo et al. 2000a; 4 Paulsson et al. 2000; 5 Wilson et al. 2000; 6 Brown et al.
2000;.

Table VII
Methods for the data normalization

Method Description
a ,b, c 2 1/2
Vector normalization Y = R/(ΣR )
a, c
Vector auto-scaling Y = (R- R )/δR; mean ( R ) & standard deviation (δR) per sample.
a, c
Dimension auto-scaling Y = (R- R )/δR; mean ( R ) & standard deviation (δR) per dimension.
c min max min
Sensor normalization Y = (R-R )/(R -R )
d
Maximum normalization Y = R/Max(R)
d
Sum normalization Y = R/ΣR

a Gutierrez-Osuna and Nagle, 1999; b Romain et al. 2000; c Gardner et al. 1998; d Amine et al. 1999.
108 Grasas y Aceites

evaluated 22 methods to reduce the number of data as input versus different time steps, but they did not
that are grouped into three classes: methods that found significant differences between the temporal
consider the extremes values, methods that filtering methods. We have however found that the
consider the transient behaviour, and methods that fractional difference method (baseline) and the
consider the shape of the signal. Table VI summarises window averaging (data compression) are the best
the models used by various authors although many for clustering and classification applications.
commercial electronic noses use only a single data
per sensor such as the steady state or its
maximum. 6.2. Multivariate data analysis
Data normalization is the third step. The scaling
(also so-called standardization) is able to readjust Multivariate statistical procedures can be applied
the values of the sensor responses to an equal to pre-processed and normalized sensor responses,
basis, avoiding that some variables weight more since it is able to perform simultaneous analyses of
than others in the results. For instance, the vector numerous variables. These procedures can be
auto- scaling normalization or the sample wise clustered into descriptive and explanatory analysis,
normalization normalizes the dimensions of each although chemometricians prefer the terms non-
sample to mean zero, and standard deviation 1. supervised and supervised (Aparicio 2000). The
Normalization methods are able to avoid non-supervised procedures are characterized by
experimental variation, and to reduce the non-having previous hypothesis classifying the
computational errors when a pattern recognition samples while the supervised procedures have the
technique is applied (Gardner and Bartlett 1994). objective of checking a previous hypothesis. The
Table VII shows some normalization methods. procedures can also be classified as parametric and
Data pre-processing techniques can be nonparametric, or linear and non-linear. The first group
evaluated by setting a score based on the prediction concerns to the probability distribution of variables, and
of the classification of a training set of samples. the non-parametric procedures are suggested when
Combinations of feature extractions can be used, the sample sizes are small (Statistica 2000). The
but only the best parameters should be selected. non-linear relation between the response matrix and
This selection can be performed using principal the compound concentration makes the non-linear
components analysis (Pardo et al. 2000b; Wilson et procedures attractive for sensor analyses.
al. 2000; Amine et al. 1999), linear discriminant The most applied multivariate procedures are
analysis (Gutierrez-Osuna and Nagle 1999) or cluster analysis, factorial analysis, multidimensional
neural network (Gardner et al. 1998) for evaluating scaling, discriminant analysis, regression analysis
the results. Thus, Wilson et al. (2000) found and artificial neural networks. Figure 5 shows an
differences in the discrimination studies applying the analysis with a sensor array using multivariate
fractional difference method (baseline) and raw data statistical analysis.

Figure 5
Scheme of the process to analyse the information of a sensor array.
Vol. 53. Fasc. 1 (2002) 109

6.2.1. Cluster Analysis non-linear property of sensors, for example, is due to

the saturation during the adsorption process, the
Cluster analysis is a generic name used to call inhibition or competition between gas species, and
the methods designed to understand the information the hysteresis or ageing of the sensors (Kato et al.
of data matrices, to describe the similarities and 2000).
dissimilarities between objects, and to single out Partial Least squares (PLS), Multiple linear
categories grouping similar objects. The different regression (MLR), principal component regression
types of cluster analysis are: K-means clustering, (PCR) and ridge regression (RR) are among the
block clustering and tree clustering (Aparicio 2000). most used regression procedures. Edible oils have
Aishima (1991) used the cluster analysis to classify been analysed by PLS and MLR (Amine et al. 1999),
four types of coffee while Pearce et al. (1993) used it PLS showing the best results. PCR and RR have
to distinguish two kinds of lagers. Cluster analysis also been applied to correlate the sensor responses
has also been used to study the similarities between to the percentages of rancidity in virgin olive oils
sensors in order to select the sensors with the (Aparicio et al. 2000).
highest sensitivity from each batch of sensors and so
avoiding redundancy (Chaudry et al. 2000; Penza et 6.3. Artificial neural networks
al. 2000).
Artificial neural networks (ANNs) have been used
6.2.2. Factorial Analysis to analyse high complex and non-linear data with
satisfactory results. ANNs are useful analytical
This kind of analysis consists in obtaining methods with advantageous properties as adaptability
unobservable factors from the data matrix in order to (learning, self-organisation, generalisation and
get a smaller matrix but retaining most of the original training) and noise tolerance (Gardner et al. 1990)
information (percentage of the variance explained). that make them appropriate for e-nose applications.
Principal component analysis (PCA) is one of the Furthermore, ANNs have the ability to handle
most known factorial analysis. It is easy to interpret non-linear data and to compensate for the drift of the
and it is not necessary to invert the matrix when sensor array (Lazzerini and Maggiore 1998). ANNs
carrying out this analysis. PCA has been used have, however, some disadvantages as the
(Capone et al. 2000) to remove the variable subjectivity on the choice of the way in which the
redundancy, to detect the outliers (Amine et al. inputs are pre-processed and the difficulty in
1999), and as pre-treatment of data previously to use selecting an optimum architecture and training
neural network (Pardo et al. 2000a). procedure (Singh et al. 1996). Another drawback is
that it is difficult to gain insight into the knowledge
6.2.3. Discriminant Analysis
that ANN analysis holds. Besides, the parameters at
This supervised procedure allows classifying the beginning of the analysis are usually set to
samples into pre-established groups. The procedure random initial values, since no physical meaning can
selects the independent variables (sensor be associated with them (Lazzerini and Maggiore
responses) that better classify the samples. 1998) with the negative consequence of an
Stepwise linear discriminant analysis (SLDA) or unpredictable training time to achieve the optimal
discriminant function analysis (DFA) are the most result. Thus, previous mathematical analyses can
used. Thus, DFA has been applied by Gardner et al. help to get a good knowledge of the data. In fact,
(1992) to classify coffee aromas while Aparicio et al. principal components of a PCA can be used as
(2000) applied SLDA to classify olive oil samples by inputs of the neural network model. On the other
the data obtained with conducting polymers.

6.2.4. Regression analysis Table VIII

Applications of artificial neural networks
Multiple regression is one of the most applied
regression methods. The general purpose of the ANN method References
multiple regression is to study the relationship
between several independent or predictor variable Multilayer perceptron (MLP) Pardo et al. 2000c; Gardner et al. 1998;
(e.g., pre-processed sensor data) and a dependent Wilson et al. 2000

or criterion variable (e.g., the known concentrations Radial basis function (RBF) Evans et al. 2000; Wilson et al. 2000
Probabilistic neural networks (PNN) McGill et al. 2000
of a chemical compound or the scores of a foodstuff
Backpropagation neural networks Magan et al. 2001; Haugen et al. 1997;
obtained by a panel test). The redundancy, the
Romain et al. 1997; Tao and Seals 1993;
multicollinearity and the assumption of linearity of the Dickert et al. 1999; Zondervan et al. 1999
sensor responses are the main problems of the Feed-forward networks Paulsson et al. 2000; Reibel et al. 2000;
regression models, and they should be taken into Qin and Wu 2001; Schweizer-Berberich et al. 2000
account before applying regression procedures. The Self organizing map (SOM) Ortega et al. 2000; Sinesio et al. 2000
110 Grasas y Aceites

hand, ANN needs, like the human brain, a previous Ballantine, D.S., Wohltjen, H. (1989). Surface acoustic
training stage with a known data set in order to wave. Devices for chemical analysis. Anal. Chem., 61,
11, 704-715.
reduce the error of the model. The training process Battiston, F.M., Ramseyer, J.-P., Lang, H.P., Baller, M.K.,
allows determining the accuracy of the model, but a Gerber, Ch., Gimzewski, J.K., Meyer, E., Güntherodt,
re-training process should be applied periodically H.-J. (2001). A chemical sensor based on a
when a change is done in the system, such as the microfabricated cantilever array with simultaneous
replacement of the set of sensors (Hodings 1997). resonance-frequency and bending readout. Sens.
Some neural network applications in electronic nose Actuators B, 77, 122-131.
Bazemore, R., Rouseff, R. (1998). Discrimination of
and sensor array are showed in Table VIII. thermally treated orange juices by an electronic nose
The prediction and discrimination of odours can equipped with organic polymer sensors. Semin. Food
also be done with exact mathematical procedures, Anal., 3, 59-66.
but always there is inexactness as part of the original Bazemore, R., Rouseff, R., Nordby, H., Goodner, K., Jella,
model, which escapes of the mathematical model. P. (1997). Discrimination of grapefruit juice varietal
differences by an electronic nose equipped with metal
Thus, fuzzy logic has been integrated in neural oxide sensors. Semin. Food Anal., 2, 239-246.
network models in order to join the advantages of Bazzo, S., Loubet, F., Labreche, S., Tan, T.T. (1999).
both methods. Singh et al. (1996) carried out a Optimization of Fox sensor array system for QC
comparative study of fuzzy neural networks and packaging in factory environment. Robustness, sample
back-propagation neural networks, the former throughput and transferability In: Electronic Noses &
Sensor Array Based System, Design and Applications,
showing the best performance. Fuzzy logic can be p. 225-234. W.J. Hurst (Ed.), Technomic Co, Inc.,
introduced in neural networks by different ways, so Lancaster, PA.
there are many kinds of fuzzy neural networks Bazzo, S., Loubet, F., Tan, T.T., Hewitt-Jones, J.D.,
(FNNs). For instance, data can be fuzzified and be Engelen-Cornax, C.E.M., Quadt, J.F.A. (1998). Quality
used as inputs in ANN. Other authors have applied a control of edible oils using an electronic nose. Semin.
linguistic fuzzy model to classify signals produced by Food Anal., 3, 15-25.
Bernou, C., Rebière, D., Pistré, J. (2000). Microwave
a sensor array (Lazzerini and Maggiore 1998). Hong sensors: a new sensing principle. Application to
(1999) also applied fuzzy logic to compensate the humidity detection. Sens. Actuators B, 68, 88-93.
non-linear bias drift of sensors. Braggins, T.J., Frost, D.A., Agnew, M.P., Farouk, M.M.
(1999). Evaluation of an electronic nose for use in
the meat industry. In: Electronic Noses & Sensor
REFERENCES Array Based System, Design and Applications, p.
51-82. W.J. Hurst (Ed.), Technomic Publishing:Lancaster,
Aishima, T. (1991). Aroma discrimination by pattern PA.
recognition analysis of responses from semiconductor Breer, H. (1997). Sense of smell: Signal recognition and
gas sensor array. J. Agric. Food Chem., 39, 752-756. transduction in olfactory receptor neurons In:
Albareda-Sirvent, M., Merkoci, A., Alegret, S. (2000). Handbook of Biosensors and Electronic Noses.
Configurations used in the design of screen-printed Medicine, Food and Environment, p. 521-532. E.
enzymatic biosensors. A review. Sens. Actuators B, 69, Kress-Rogers (Ed). CRC Press: Boca Raton, FL.
153-163. Brown, J.R., Haycock, P.W., Smith, L.M., Jones, A.C.,
Amine, H., Bazzo, S., Labreche, S. (1999). Intensity and Williams, E.W. (2000). Response behaviour of tin oxide
quality discrimination using the Fox4000 gas sensor thin film gas sensors grown by MOCVD. Sens.
array system In: Electronic Noses & Sensor Array Actuators B, 63, 109-114.
Based System, Design and Applications, p. 235-248. Cai, Q.Y., Grimes, C.A. (2000). A remote query
W.J. Hurst (Ed.), Technomic Co., Inc., Lancaster, PA. magnetoelastic pH sensor. Sens. Actuators B, 71,
Angerosa, F. (2000). Sensory quality of olive oils. In: 112-117.
Handbook of Olive Oil: Analysis and Properties, p. Cai, Q.Y., Park, J., Heldsinger, D., Hsieh, M.D., Zellers, E.T.
355-392. J. Harwood and R. Aparicio (Eds.), (2000). Vapor recognition with an integrated array of
Aspen:Gaithersburg, MA. polymer-coated flexural plate wave sensors. Sens.
Aparicio, R. (2000). Characterization: Mathematical Actuators B, 62, 121-130.
procedures for chemical analysis In: Handbook of Olive Caide, X., Sui, S F. (2000). Characterization of surface
Oil. Analysis and Properties, p. 285-353. J. Harwood plasmon resonance biosensor. Sens. Actuators B, 66,
and R. Aparicio (Eds.), Aspen: Gaithersburg, MA. 174-177.
Aparicio, R., Alonso, M.V., Morales, M.T., Calvente J.J. Campanella, L., Sammartino, M.P., Tomasseti, M.,
(1994). Relationship between COI Test and other Zannella, S. (2001). Hydroperoxide determination by a
sensory profiles by statistical procedures. Grasas catalase OPPE: Application to the study of extra virgin
Aceites 45, 26-41. olive oil rancidification process. Sens. Actuators B, 76,
Aparicio, R., Morales, M.T., Alonso, M.V. (1996). 158-165.
Relationship between volatile compounds and sensory Campos, M., Bulhoes, L.O.S., Lindino, C.A. (2000).
attributes by statistical sensory wheel. J. Am. Oil Chem. Gas-sensitive characteristics of metal/semiconductor
Soc. 73, 1253-1264. polymer Schottky device. Sens. Actuators A, 87, 67-71.
Aparicio, R., Rocha, S.M., Delgadillo, I., Morales, M.T. Capone, S., Siciliano, P., Quaranta, F., Rella, R., Epifani,
(2000). Detection of rancid defect virgin olive oil by M., Vasanelli, L. (2000). Analysis of vapours and foods
electronic nose. J. Agri. Food Chem., 48, 3, 855-860. by means of an electronic nose based on a sol-gel
Arroyo, M. (1998). Inmobilized enzymes: Theory, methods metal oxide sensors array. Sens. Actuators B, 69,
of study and applications. Ars Pharmaceutica, 39, 2, 230-235.
23-39.
Vol. 53. Fasc. 1 (2002) 111

Chaudry, A.N., Hawkins, T.M., Travers, P.J. (2000). A Gardner, J.W., Craven, M., Dow, C. Hines, E.L. (1998). The
method for selecting an optimum sensor array. Sens. prediction of bacteria type and culture growth phase by
Actuators B, 69, 236-242. an electronic nose with a multi-layer perceptron
Comini, E, Faglia, G., Sberveglieri, G., Li, Y.X., Wlodarski, network. Meas. Sci. Technol., 9, 120-127.
W., Ghantasala, M.K. (2000). Sensitivity enhancement Gardner, J.W., Hines, E.L., Wilkinson, M. (1990).
towards ethanol and methanol of TiO2 films doped with Application of artificial neural networks to an electronic
Pt and Nb, Sens. Actuators B, 64, 169-174. olfactory system. Meas. Sci. Technol., 1, 446-451.
Craven, M.A., Gardner, J.W., Bartlett, P.N. (1996). Gardner, J.W., Shurmer, H.V., Tan, T.T. (1992). Application
Electronic noses-development and future prospects. of an electronic nose to the discrimination of coffees.
Trends in Anal. Chem., 15, 9, 486-493. Sens. Actuators B, 6, 71-75.
D’Amico, A., Di Natale, C., Paolesse, R. (2000). Portraits of González Martín, Y., Pérez Pavón, J.L., Moreno Cordero,
gases and liquids by arrays of non-specific chemical B., García Pinto, C. (1999). Classification of vegetable
sensors: trends and perspectives. Sens. Actuators B, oil by linear discriminant analysis of Electronic Nose
68, 324-330. data. Anal. Chim. Acta, 384, 83-94.
D’Amico, A., Di Natale, C., Verona, E. (1997). Acoustic Göpel, W. (2000). From electronic to bioelectronic olfaction,
devices. In: Handbook of Biosensors and Electronic or: from artificial “moses” to real noses. Sens.
Noses. Medicine, Food and Environment, p. Actuators B, 65, 70-72.
197-223. E. Kress-Rogers (Ed). CRC Press: Boca Griffiths, D., Hall, G. (1993). Biosensors-What real
Raton, FL. progress is being made? Trends Biotechnol., 11,
De Marcos, S., Asensio, C., Uruñuela, I., Gallarta, F., 122-130.
Galbán, J., Castillo, J.R. (2000). New approaches to Guadarrama, A., Fernández, J.A., Íñiguez, M., Souto, J.,
polyaniline optical sensors: pH, acetic acid and amonia de Saja, J.A. (2001). Discrimination of wine aroma
determination. Química Analítica, 19, 1, 99-104. using an array of conducting polymer sensors in
Di Natale, C., Macagnano, A., Nardis, S., Paolesse, R., conjunction with solid-phase micro-extraction (SPME)
Falconi, C., Proietti, E., Siciliano, P., Rella, R., Taurino, technique. Sens. Actuators B, 77, 401-408.
A., D’Amico, A. (2001). Comparison and integration of Gutiérrez, F.J., Fontecha, J.L., Horrillo, M.C., Sayago, I.,
arrays of quartz resonators and metal-oxide Fernández, M.J., Gómez-Espinosa, R., Cané, C.,
semiconductor chemoresistors in the quality evaluation Grácia, I. (2000). Microsensores de gases basados en
of olive oil. Sens. Actuators B, 78, 303-309. ondas acústicas superficiales. Proceeding Congreso
Di Natale, C., Paolesse, R., Macagnano, A., Mantini, A., TEDEA, Almagro (Ciudad Real, Spain) 26-9-2000.
D’Amico, A., Ubigli, M., Legin, A., Lvova, L., Gutiérrez-Osuna, R., Nagle, H.T. (1999). A method for
Rudnitskaya, A., Vlasov, Y. (2000). Application of a evaluating data-preprocessing techniques for odor
combined artificial olfaction and taste system to the classification with an array of gas sensors. IEEE Trans.
quantification of relevant compounds in red wine. Sens. Syst. Man Cyber., 29, 626-632.
Actuators B, 69, 342-347. Hatfield, J.V., Covington, J.A., Gardner, J.W. (2000).
Dickert, F.L., Hayden, O., Zenkel, M.E. (1999). Detection of GasFETs incorporating conducting polymers as gate
volatile compounds with mass-sensitive sensor arrays materials. Sens. Actuators B, 65, 253-256.
in the presence of variable ambient humidity. Anal. Hatfield, J.V., Neaves, P., Hicks, P.J., Persaud, K., Travers,
Chem., 71, 1338-1341. P. (1994). Towards an integrated electronic nose using
Dzyadevych, S., Arkhypova, V.N., Korpan, Y.I., El’skaya, conducting polymer sensors. Sens. Actuators B, 18-19,
A.V., Soldatkin, A.P., Jaffrezic-Renault, N., Martelet, C. 221-228.
(2001). Conductometry formaldehyde sensitive Hermle, T, Weimar, U., Mitrovics, J., Rosenstiel, W., Göpel,
biosensor with specifically adapted analytical W. (1999). Feature extraction for a hybrid gas sensing
characteristics. Anal. Chim. Acta, 445, 47-55. system In: Electronic Noses & Sensor Array Based
Ensor, D.R. (1989). The contribution of flavour chemistry to System, Design and Applications, p. 97-102. W.J. Hurst
the food industry In: Flavour Chemistry of Lipid Foods, (Ed.), Technomic Co, Inc., Lancaster, PA.
p. 1-12. D.B. Min and T.H. Smouse (Eds.). The Herrmann, F., Hahn, D., Büttgenbach, S. (1999). Separate
American Oil Chemists’ Society, Champaign. determination of liquid density and viscosity with
Ferrari, V., Marioli, D., Taroni, A., Ranucci, E. (2000). sagittally corrugated Love-mode sensors. Sens.
Multisensor array of mass microbalances for chemical Actuators A, 78, 99-107.
detection based on resonant piezo-layers of Hirschfelder, M., Hanrieder, D., Ulrich, D. (1998).
screen-printed PZT. Sens. Actuators B, 68, 81-87. Discrimination of strawberry varieties by a gas sensor
Ferreira, L.S., De Souza, M.B., Folly, R.O.M. (2001). array in correlation with human sensory evaluation.
Development of an alcohol fermentation control Semin. Food Anal., 3, 27-36.
system based on biosensor measurements interpreted Hodgins, D. (1997). The electronic nose: Sensor
by neural networks. Sens. Actuators B, 75, 166-171. array-based instruments that emulate the human nose
Freudenberg, J., von Schickfus, M., Hunklinger, S. (2001). In: Techniques for Analyzing Food Aroma, p. 331-371.
A SAW immunosensor for operation in liquid using a R. Marsili (Ed.), Marcel Dekker Inc., New York.
SiO2 protective layer. Sens. Actuators B, 76, 147-151. Hong, S.K. (1999). Compensation of nonlinear thermal
Fukui, K., Katsuki, A. (2000). Improvement of humidity bias drift of Resonant Rate Sensor using fuzzy logic.
dependence in gas sensor based on SnO2. Sens. Sens. Actuators A, 78, 143-148.
Actuators B, 65, 316-318. Iiyama, S., Yahiro, M., Toko, K. (2000). Measurements of
García Pinto, C., Fernández Laespada, M.E., Pérez soy sauce using taste sensor. Sens. Actuators B, 66,
Pavón, J.L., Moreno Cordero, B. (2001). Electronic 205-206.
olfactometry. A new tool in analytical chemistry. Ivarsson, P, Holmin, S., Höjer, N.E., Krantz-Rülcker, C.,
Química Analítica, 20, 3-11. Winquist, F. (2001). Discrimination of tea by means of a
Gardner, J.W., Bartlett, P.N. (1994). A brief history of voltammetric electronic tongue and different applied
electronic noses. Sens. Actuators B, 18-19, 211-220. waveforms. Sens. Actuators B, 76, 449-454.
112 Grasas y Aceites

Kato, K., Kato, Y., Takamatsu, K., Udaka, T., Nakahara, T., Lundström, I., Spetz, A., Winquist, F., Ackelid, U.,
Matsuura, Y., Yoshikawa, K. (2000). Toward the Sundgren, H. (1990). Catalytic metals and field-effect
realization of an intelligent gas sensing system utilizing devices – a useful combination. Sens. Actuators B, 1,
a non-linear dynamic response. Sens. Actuators B, 71, 15-20.
192-196. Lüth, H., Thust, M., Steffen, A., Kordos, P., Schöning, M.J.
Keller, P.E. (1999). Physiologically inspired pattern (2000). Biochemical sensors with structured and
recognition for electronic noses. Proceedings of the porous silicon capacitors. Sens. Actuators B, 69,
SPIE, 3722, No. 13. 104-108.
King, W.H. JR. (1964). Piezoelectric sorption detector. Anal. Luzuriaga, D.A., Balaban, M.O. (1999). Electronic nose
Chem., 36, 1735-1739. odor evaluation of salmon fillets stored at different
Kohl, D., Heinert, L., Bock, J., Hofmann, Th., Schieberle, P. temperatures. In: Electronic Noses & Sensor Array
(2000). Systematic studies on responses of Based System, Design and Applications. W. J. Hurst
metal-oxide sensor surfaces to straight chain alkanes, (Ed.), Technomic:Lancaster, PA, pp. 219-279.
alcohols, aldehydes, ketones, acids and esters using Magan, N., Pvlou, A., Chrysanthakis, I. (2001). Milk-Sense:
the SOMMSA approach. Sens. Actuators B, 70, A volatile sensing system recognises spoilage bacteria
43-50. and yeast in milk. Sens. Actuators B, 72, 28-34.
Köhler, S., Bachmann, T.T., Schmitt, J., Belkin, S., Schmid, Masila, M., Breimer, M., Sadik, O. A. (1999). Strategies for
R.D. (2000). Detection of 4-chlorobenzoate using improving the analysis of volatile organic compounds
immobilized recombinant Escherichia coli reporter using GC-based electronic nose In: Electronic Noses &
strains. Sens. Actuators B, 70, 139-144. Sensor Array Based System, Design and Applications, p.
Kondratowicz, B., Narayanaswamy, R, Persaud, K.C. 26-42. W.J. Hurst (Ed.), Technomic Co, Inc., Lancaster, PA.
(2001). An investigation into the use of electrochromic McGill, R.A., Nguyen, V.K., Chung, R., Shaffer, R.E.,
polymers in optical fibre gas sensors. Sens. Actuators DiLella, D., Stepnowski, J.L., Mlsna, T.E., Venezky,
B, 74, 138-144. D.L., Dominguez, D. (2000). The ‘‘NRL-SAWRHINO’’: a
Koul, S., Chandra, R., Dhawan, S.K. (2001). Conducting nose for toxic gases. Sens. Actuators B, 65, 10-13.
polyaniline composite: A reusable sensor material for Ménil, F., Coillard, V., Lucat, C. (2000). Critical review of
aqueous ammonia. Sens. Actuators B, 75, 151-159. nitrogen monoxide sensors for exhaust gases of lean
Kress-Rogers, E. (1997). Biosensors and electronic noses burn engines. Sens. Actuators B, 67, 1-23.
for practical application In: Handbook of Biosensors Middelhoek, S. (2000). Celebration of the tenth
and Electronic Noses. Medicine, Food and transducers conference: The past, present and future
Environment, p. 3-39. E. Kress-Rogers (ed). CRC of transducer research and development. Sens.
Press: Boca Raton, FL. Actuators A, 82, 2-23.
Lang, A.C., Fleischer, M., Meixner, H. (2000). Surface Mielle, P. (1996). Electronic noses: Towards the objective
modifications of Ga2O3 thin film sensors with Rh, Ru instrumental characterization of food aroma. Trends
and Ir clusters. Sens. Actuators B, 66, 80-84. Food Sci. Technol., 7, 432-438.
Lazzerini, B., Maggiore, A. Linguistic fuzzy classification of Mielle, P., Marquis, F., Latrasse, C. (2000). Electronic
olfactive signals. 1998 IEEE International Conference noses: specify or disappear. Sens. Actuators B, 69,
on fuzzy systems proceedings. IEEE World Congress 287-294.
on Computational Intelligence (CAT No. 98CH36228), Montedoro, G. F., Servili, M., Baldioli, M., Miniati, E. (1992).
p. 1518-1523. Vol. 2. Simple and hydrolyzable phenolic compounds in olive
Legin, A., Rudnitskaya, A., Vlasov, Y., Di Natale, C., oil: 1. Their extraction, separation, and quantitative and
Mazzone, E., D’Amico, A. (2000). Application of semiquantitative evaluation by HPLC. J. Agric. Food
electronic tongue for qualitative and quantitative Chem. 40, 1571-1576.
analysis of complex liquid media. Sens. Actuators B, Morales, M.T., Aparicio, R., Gutiérrez, F. (1992). Técnicas
65, 232-234. de aislamiento y concentración de volátiles de aceites
Li, D., Ma, M. (2000). Surface acoustic wave microsensors vegetales. Grasas Aceites, 43, 164-173.
based on cyclodextrin coatings. Sens. Actuator B, 69, Morales, M.T., Tsimidou, M. (2000). The role of volatile
75-84. compounds and polyphenols in olive oil sensory
Liao, J.D., Wang, S-H., Hsu, D-J. (2001). Studies on the quality. In: Handbook of Olive Oil: Analysis and
early detection of wastewater’s toxicity using a Properties, p. 393-458. J. Harwood and R. Aparicio
microbial sensing system. Sens. Actuators B, 72, (Eds.), Aspen:Gaithersburg, MA.
167-173. Morimitsu, M., Ozaki, Y., Suzuki, S., Matsunaga, M. (2000).
Lillis, B., Grogan, C., Berney, H., Lane, W.A. (2000). Effects of surface modification with platinum and
Investigation into immobilisation of lactate oxidase to ruthenium on temperature and humidity dependence
improve stability. Sens. Actuators B, 68, 109-114. of SnO2-based CO gas sensors. Sens. Actuators B,
Liu, J., Li, G. (2000). Applications of biosensors for 67, 184-188.
diagnostic analysis and bioprocess monitoring. Sens. Muenchmeyer, W., Walte, A., Matz, G. (2000). Improving
Actuators B, 65, 26-31. electronic noses using a trap and thermal desorption
Liu, W., Ma, W., Li, L., Li, J. (2000). The temperature unit. Sens. Actuators B, 69, 379-383.
change in SnO2- based gas sensors during sensing Müller, C., Nirmaier, T., Rügemer, A., Schickfus, M.v.
reaction. Sens. Actuators B, 66, 222-224. (2000). Sensitive NO2 detection with surface acoustic
Lucas, Q., Poling, J., Benincasa, V. (1998). Analysis with a wave devices using a cyclic measuring technique.
hybrid system QCM/CP/MOS. Semin. Food Anal., 3, Sens. Actuators B, 68, 69-73.
53-58. Nakamoto, T., Nakahira, Y., Hiramatsu, H., Moriizumi, T.
Lucklum, R., Hauptmann, P. (2000). The quartz crystal (2001). Odor recorder using active odor sensing
microbalance: mass sensitivity, viscoelasticity and system. Sens. Actuators B, 76, 465-469.
acoustic amplification. Sens. Actuators B, 70, 30-36. Nikitin, P.I., Grigorenko, A.N., Beloglazov, A.A., Valeiko,
Lundström, I., Shivaraman, M.S., Svensson, C., Lundkvist, M.V., Savchuk, A.I., Savchuck, O.A., Steiner, G.,
L (1975). Applied Physics Letters, 26, 55. Kuhne, C., Huebner, A., Salzer, R. (2000). Surface
Vol. 53. Fasc. 1 (2002) 113

plasmon resonance interferometry for micro-array of DNA adsorption on polypyrrole: Interest of dielectric
biosensing. Sens. Actuators A, 85, 189-193. monitoring. Sens. Actuators B, 62, 35-42.
Nomura, T., Saitoh, A., Horikoshi, Y. (2001). Measurement Sberveglieri, G., Comini, E., Faglia, G., Niederjaufner, G.,
of acoustic properties of liquid using liquid flow Benussi, G., Contarini, G., Povolo, M. (1998). A novel
SH-SAW sensor system. Sens. Actuators B, 76, 69-73. electronic nose based on semiconductor thin film gas
Ortega, N., Busto, M.D., Perez-Mateos, M. (1998) sensor to distinguish different heat treatments of milk.
Optimisation of -glucosidase entrapment in alginate Semin. Food Anal., 3, 67-76.
and polyacrilamide gels. Bioresour. Technol., 64, Schaak, R.E., Dahlberg, D.B., Miller, K.B. (1999). The
105-111. electronic nose: studies on the fundamental response
Palleschi, G., Cubadda, R. (2001). Electrochemical and discriminative power of metal-oxide sensors In:
biosensors for food analysis and the food industry. Electronic Noses & Sensor Array Based System,
Italian J. Food Science, 13, 137-157. Design and Applications, p. 43-50. W. J. Hurst (Ed.),
Pardo, M., Niederjaufner, G., Benussi, G., Comini, E., Faglia, Technomic Co, Inc., Lancaster, PA.
G., Sberveglieri, G., Holmberg, M., Lundstrom, I. (2000b). Schaller, E., Bosset, J. O., Escher, F. (1998). Electronic
Data preprocessing enhances the classification of nose and their application to food. Lebensm.-Wiss.
different brands of Espresso coffee with an electronic u.-Technol., 31, 305-316.
nose. Sens. Actuators B, 69, 397-403. Schütz, S, Schöning, M.J., Schroth, P., Malkoc, Ü,
Pardo, M., Sberveglieri, G., Gardini, S., Dalcanale, E. Weibecker, B., Kordos, P., Lüth, H., Hummel, H.E.
(2000a). A hierarchical classification scheme for an (2000). An insect-based BioFET as a bioelectronic
Electronic Nose. Sens. Actuators B, 69, 359-365. nose. Sens. Actuators B, 65, 291-295.
Partridge, A.C., Harris, P., Andrews, M.K. (1996). High Senillou, A., Jaffrezic-Renault, N., Martelet, C., Cosnier, S.
sensitivity conducting polymers sensors. Analyst, 121, (1999). A miniaturized urea sensor based on the
1349-1353. integration of both ammonium based urea enzyme field
Paulsson, N, Larsson, E., Winquist, F. (2000). Extraction effect transistor in a single chip. Talanta, 50, 219-226.
and selection of parameters for evaluation of breath Shinohara, H., Kusaka, T., Yokota, E., Monden, R., Sisido,
alcohol measurement with an electronic nose. Sens. M. (2000). Electron transfer between redox enzymes
Actuators A, 84, 187-197. and electrodes through the artificial redox proteins and
Pearce, T.C., Gardner, J.W., Friel, S., Barlett, P.N., Blair, N. its application for biosensors. Sens. Actuators B, 65,
(1993). Electronic nose for monitoring the flavour of 144-146.
beers. Analyst, 118, 371-377. Shurmer, H.V., Gardner, J.W. (1992). Odour discrimination
Persaud, K.C., Travers, P.J. Arrays of broad specificity films with an electronic nose. Sens. Actuators B, 8, 1-11.
for sensing volatile chemicals In: Handbook of Simon, I, Bârsan, N., Bauer, M., Weimar, U. (2001).
Biosensors and Electronic Noses. Medicine, Food And Micromachined metal oxide gas sensors: opportunities
Environment, p. 563-592. E. Kress-Rogers (Ed.) CRC to improve sensor performance. Sens. Actuators B, 73,
Press, Inc. Boca Raton, FL. 1-26.
Porter, T.L., Eastman, M.P., Pace, D.L., Bradley, M. (2001). Sinesio, F., Di Natale, C., Quaglia, G.B., Bucarelli, F.M.,
Sensor based on piezoresistive microcantilever Moneta, E., Macagnano, A., Paolesse, R., D’Amico, A.
technology. Sens. Actuators A, 88, 47-51. (2000). Use of electronic nose and trained sensory
Qin, S., Wu, Z., Tang, Z., Song, Y., Zeng, F., Zhao, D. (2000). panel in the evaluation of tomato quality. J. Sci. Food
The sensitivity to SO2 of the SAW gas sensor with Agric., 80, 63-71.
triethanolamine modified with boric acid. Sens. Singh, S., Hines, E.L., Gardner, J.W. (1996). Fuzzy neural
Actuators B, 66, 240-242. computing of coffee and tainted water data from an
Ramos, M.C., Torijas, M.C., Navas Díaz, A. (2001) electronic nose. Sens. Actuators B, 30, 3, 190-195.
Enhanced chemiluminiscence biosensor for the Slater, J.M., Paynter, J., Watt, E.J. (1993). Multi-layer
determination of phenolic compounds and hydrogen conducting polymer gas sensor arrays for olfactory
peroxide. Sens. Actuators B, 73, 71-75. sensing. Analyst, 118, 379-384.
Reineccius, G. (1993). Biases in analytical flavor profiles Slater, J.M., Watt, E.J., Freeman, N.J., May, I.P., Weir, D.J.
introduced by isolation method In: Flavor (1992). Gas and vapour detection with poly(Pyrrole)
Measurement, p. 61-67. C.T. Ho and C.H. Manley gas sensors. Analyst, 117, 1265-1270.
(Eds.), Marcel Dekker, New York. Spetz, A., Winquist, F., Sundgren, H., Lundström, I. (1992).
Rella, R., Siciliano, P., Quaranta, F., Primo, T., Valli, L., Field effect gas sensors In: Gas sensors, p. 219-279.
Schenetti, L., Mucci, A., Iarossi, D. (2000). Gas sensing G. Sberveglieri (Ed.), Kluwer Academic:Dordrecht,
measurements and analysis of the optical properties of Germany.
poly[3-(Butylthio) Thiophene] Langmuir-Blodgett films. Spetz, A.L., Tobias, P., Unéus, L., Svenningstorp, H.,
Sens. Actuators B, 68, 203-209. Ekedahl, L-G, Lundström, I. (2000). High temperature
Reviejo, A.J., Pingarrón, J.M. (2000). Biosensores catalytic metal field effect transistors for industrial
electroquímicos. Una Herramienta útil para el applications. Sens. Actuators B, 70, 67-76.
análisis medioambiental, alimentario y clínico. STATISTICA. Users Guide, release 5.5; Statsoft: Tulsa, OK,
Anales de la Real Sociedad Española de Química, 2000.
96, 2, 5-15. Stetter, J.R., Strathmann, S., McEntegart, C., Decastro,
Romain, A-C., Nicolas, J., Wiertz, V., Maternova, J., André, M., Penrose, W.R. (2000). New sensor arrays and
Ph. (2000). Use of a simple tin oxide sensor array to sampling systems for a modular electronic nose. Sens.
identify five malodours collected in the field. Sens. Actuators B, 69, 410-419.
Actuators B, 62, 73-79. Su, X., Low, S., Kwang, J., Chew, V.H.T., Li, S.F.Y. (2001).
Rong, L., Ping, W., Wenlei, H. (2000). A novel method for Piezoelectric quartz crystal based veterinary diagnosis
wine analysis based on sensor fusion technique. Sens. for Salmonella enteritidis infection in chicken and egg.
Actuators B, 66, 246-250. Sens. Actuators B, 75, 1-2, 29-35.
Saoudi, B., Despas, C., Chehimi, M.M., Jammul, N., Takada, T. (2000). Temperature drop of semiconductor gas
Delamar, M., Bessiére, J., Walcarius, A. (2000). Study sensor when exposed to reducing gases-simultaneous
114 Grasas y Aceites

measurement of changes in sensor temperature and in structure and compared with a gas chromatographic
resistance. Sens. Actuators B, 66, 1-3. method. Anal. Chim. Acta, 231, 93-100.
Tan, C.K., Blackwood, D.J. (2000). Interactions between Winquist, F., Lundström, I., Wide, P. (1999). The
polyaniline and methanol vapour. Sens. Actuators B, combination of an electronic tongue and an electronic
71, 184-191. nose. Sens. Actuators B, 58, 512-517.
Tempelman, L.A., King, K.D., Anderson, G.P., Ligler, F.S. Xu, Y., Zhou, X., Sorensen, T. (2000). Oxygen sensors
(1996). Quantitating Staphylococcal enterotoxin B in based on semiconducting metal oxides: an overview.
diverse media using a portable fiber-optic biosensor. Sens. Actuators B, 65, 2-4.
Anal. Biochem., 233, 50-57. Zhang, G., Liu, M. (2000) Effect of particle size and dopant
Toko, K. (2000). Taste sensor. Sens. Actuators B, 64, 205-215. on properties of SnO2-based gas sensors. Sens.
Tong, X., Zhang, D. (1999). Novel propagation direction of Actuators B, 69, 144-152.
quasi-longitudinal leaky surface acoustic wave on Zhao, S., Sin, J.K. O., Xu, B., Zhao, M., Peng, Z., Cai, H.
quartz and its potential as liquid sensors. Sens. (2000). A high performance ethanol sensor based on
Actuators A, 78, 160-162. field-efecct transistor using a LaFeO3 nano-crystalline
Werlein, H-D. (2001). Discrimination of chocolates and thin-Film as a gate electrode. Sens. Actuators B, 64,
packaging materials by an electronic nose. Eur. Food 83-87.
Res. Technol., 212, 529-533. Zimmermann, C., Rebière, D., Déjous, C., Pistré, J.,
White, S.F., Turner, A.P.F. (1997). Enzymes, cofactors, and Chastaing, E., Planade, R. (2001). A love-wave gas
mediators In: Handbook of Biosensors and Electronic sensor coated with functionalized polysiloxane for
Noses. Medicine, Food and Environment, p. 43-57. E. sensing organophosphorus compounds. Sens.
Kress-Rogers (Ed). CRC Press: Boca Raton, FL. Actuators B, 76, 86-94.
Wilson, D.M., Dunman, K., Roppel, T., Kalim, R. (2000). Zondervan, C., Muresan, S., de Jonge, H.G., van Velzen,
Rank extraction in tin-oxide sensor arrays. Sens. E.U.T., Wilkinson, C., Nijhuis, H.H., Leguijt, T. (1999).
Actuators B, 62, 199-210. Controlling Maillard reactions in the heating process of
Winquist, F., Lundström, I., Bergkvist, H. (1990). Ethylene blockmilk using an electronic nose. J. Agric. Food
production from fruits measured by a simple field-effect Chem., 47, 4746-4749.