Electronic Nose.pdf (Size: 289.22 KB / Downloads: 548)
In an ever-developing world, where electronic devices are duplicating every other sense of perception, the sense of smell is lagging behind. Yet, recently, there has been an urgent increase in the need for detecting odours, to replace the human job of sensing and quantification. Some of the most important applications fall in the category where human beings cannot afford to risk smelling the substance. Other important applications are continuous monitoring, medical applications, etc. These applications allow man to perform tasks that were once considered impossible. The fast paced technology has helped develop sophisticated devices that have brought the electronic nose to miniature sizes and advanced capabilities. The trend is such that there will be accurate, qualitative and quantitative measurements of odour in the near future.
Mimicking the nose is a challenging task. The human nose can smell 10,000 different odour molecules mixed in air. Odour in a substance is due to certain volatile organic compounds (VOCs), which easily evaporate and get carried by an air stream. An e-nose can smell and estimate odours quickly though it has little or no resemblance to the human nose. A human nose has receptors, which serve as binding sites for VOCs. A receptor is just a molecular structure on the surface of the nerve cell to which an odorous molecule with the right shape binds. The receptor and the binding molecule fit exactly as in a key and lock arrangement. These odour-sensing nerve cells line the upper part of the cavity in the human nose. Once an odour molecule binds to a receptor, a chain reaction follows which ultimately transmits an electrical signal to the brain. A specific odour of coffee or wine is usually caused not by one, but a mixture of hundreds of organic compounds. So, the brain has a mammoth task of processing signals received from the nerve cells originating from the nose, to identify the nature of smell. The exact working of the brain in processing these signals is yet to be fully understood.
An electronic nose can be defined as Ëœan instrument which is comprised of an array of electronic chemical sensors with partial specificity and an appropriate pattern recognition system, capable of recognizing simple or complex odours (and other gaseous mixtures). The ability of an electronic nose to rapidly discriminate between slight variations in complex mixtures makes the techniques ideal for on-line process diagnostics and screening across a wide range of application areas. An electronic nose is a machine that is designed to detect and discriminate among complex odours using a sensor array. The sensor array of consists of broadly tuned (non-specific) sensors that are treated with a variety of odour-sensitive biological or chemical materials. An odour stimulus generates a characteristic fingerprint (or smell-print) from the sensor array. Patterns or fingerprints from known odours are used to construct a database and train a pattern recognition system so that unknown odours can subsequently be classified and identified. Thus, electronic nose instruments are comprised of hardware components to collect and transport odours to the sensor array â€œ as well as electronic circuitry to digitise and stored the sensor responses for signal processing. The two main components of an electronic nose are the sensing system and the automated pattern recognition system. The sensing system can be an array of several different sensing elements (e.g., chemical sensors), where each element measures a different property of the sensed chemical, or it can be a single sensing device (e.g., spectrometer) that produces an array of measurements for each chemical, or it can be a combination. Each chemical vapour presented to the sensor array produces a signature or pattern characteristic of the vapour. By presenting many different chemicals to the sensor array, a database of signatures is built up. This database of labelled signatures is used to train the pattern recognition system. The goal of this training process is to configure the recognition system to produce unique classifications of each chemical so that an automated identification can be implemented. The quantity and complexity of the data collected by sensors array can make conventional chemical analysis of data in an automated fashion difficult. One approach to chemical vapour identification is to build an array of sensors, where each sensor in the array is designed to respond to a specific chemical. With this approach, the number of unique sensors must be at least as great as the number of chemicals being monitored. It is both expensive and difficult to build highly selective chemical sensors.
Artificial neural networks (ANNs), which have been used to analyse complex data and to recognize patterns, are showing promising results in chemical vapour recognition. When an ANN is combined with a sensor array, the number of detectable chemicals is generally greater than the number of sensors. Also, less selective sensors, which are generally less expensive, can be used with this approach. Once the ANN is trained for chemical vapour recognition, operation consists of propagating the sensor data through the network. Since this is simply a series of vector-matrix multiplications, unknown chemicals can be rapidly identified in the field. Electronic noses that incorporate ANNs have been demonstrated in various applications. Some of these applications will be discussed later in the paper. Many ANN configurations and training algorithms have been used to build electronic noses including back propagationtrained, feed-forward networks; fuzzy ART maps; Cohuneâ„¢s self-organizing maps (SOMs); learning vector quantizers (LVQs); Hamming networks; Boltzmann machines; and Hopfield networks. Figure 1 illustrates the basic schematic of an electronic nose. 6 One of our prototype electronic noses, shown in Figure is composed of an array of nine tinoxide vapor sensors, a humidity sensor, and a temperature sensor coupled with an ANN. Two types of ANNs were constructed for this prototype:the standard multilayer feedforward network trained with the backpropagation algorithm and the fuzzy ARTmap algorithm . During operation a chemical vapor is blown across the array, the sensor signals are digitized and fed into the computer, and the ANN (implemented in software) then identifies the chemical. This identification time is limited only by the response time of the chemical sensors, which is on the order of seconds. This prototype nose has been used to identify common household chemicals by their odor.
Although each sensor is designed for a specific chemical, each responds to a wide variety of chemicals. Collectively, these sensors respond with unique signatures (patterns) to different chemicals. During the training process, various chemicals with known mixtures are presented to the system. By training on samples of various chemicals, the ANN learns to recognize the different chemicals. This prototype nose has been tested on a variety of household and office supply chemicals including acetone, ammonia, ethanol, glass cleaner, contact cement, correction fluid, iso-propanol,lighter fluid, methanol, rubber cement and vinegar.For the results shown in the paper, five of these chemicals were used: acetone, ammonia,isopropanol, lighter fluid, and vinegar. Another category, none was used to denote the absence of all chemicals except those normally found in the air which resulted in six output categories from the ANN.
Both networks were trained using randomly selected sample sensor inputs. The ANNs used here were not trained to quantify the concentration level of the identified analytes, but were trained with samples with different concentrations of the analytes. This allowed the ANN to generalize well on the test data set.Performance levels of the two networks were basically equivalent ranging from 89.7% to 98.2% correct identification on the test set depending on the random selection of training patterns. Figures 4 and 5 illustrate the responses of the sensors and the ANN classification for a variety of test chemicals presented to the ANNs. The ANN was able to correctly classify the test samples with only small residual errors. While the ANN used here was not trained to quantify the concentration level of the identified analytes, it was trained with samples with different concentrations of the analytes. This allowed the ANN to generalize well on the test data set. From the responses of the sensors to the analytes, one can easily see that the individual sensors in the array are not selective . In addition, when a mixture of two or more chemicals is presented to the sensor array, the resultant pattern (sensor values) may be even harder to analyze (see Figure 5: c, d, and e). Thus, analyzing the sensor responses separately may not be adequate to yield the classification accuracy achieved by analyzing the data in parallel.
2.2. Sensing an Odorant
In a typical e-nose, an air sample is pulled by a vacuum pump through a tube into a small chamber housing the electronic sensor array. The tube may be of plastic or stainless steel. A sample-handling unit exposes the sensors to the odorant, producing a transient response as the VOCs interact with the active material. The sensor response is recorded and delivered to the signal-processing unit. Then a washing gas such as alcohol is applied to the array for a few seconds or a minute, so as to remove the odorant mixture from the active material.
Finally, the reference gas is again applied to the array, to prepare it for a new measurement cycle. The odorant is applied for a period equal to the response time of the input output Array of gas sensors Sample Handler Signal Processing System sensor array. The washing and reference gases are applied for the sensor array to recover and come to a reference point. This duration is termed the recovery time. The main steps of odor recognition can be briefly explained as follows:
- Heating the sample for a certain time generates the smell.
- The gas phase is sampled and transferred to a detection device which reacts to the
presence of various molecules.
- The difference in the sensor reactions is revealed using different statistical pattern recognition techniques to classify the odors. From this pattern and from previous human input (human training from sensory panels), the system predicts the mostly likely human response to the new pattern.
The electronic nose gives either a simple answer like recognized, good, or bad or a more sophisticated response such as an odor intensity or a molecule concentration The terminology can be simple and qualitative or more specific and quantitative.
2.3. Gas Sensors
The main advantages of the gas sensors are as follows.
1. High Speed
3. Continuous real time monitoring of sites, etc.
The problems associated with human panels are individual variation, adaptation, fatigue, infectious mental state, subjectivity and exposure to hazardous compounds. So, the enose can create an odour profile that extends beyond the capabilities of the human panel or GC/MS measurement techniques. The output of the e-nose can be the identity of the odorant, an estimate concentration of the odorant, or the characteristic properties of the odour as might be perceived by a human.
Fundamental to the artificial nose is the idea that each sensor in the array has different sensitivity. Also, the pattern of responses across all sensors in the array is used to identify and/or characterize the odour.
INTRODUCTION TO SENSORS
A sensor is a device which can respond to some properties of the environment and transform the response into an electric signal. The general working mechanism of a sensor is illustrated by the following scheme :
In the field of sensors, the correct definition of parameters is of paramount importance because these parameters:
allow the diffusion of more reliable information among researchers or sensor operators, allow a better comprehension of the intrinsic behaviour of the sensors help to propose new standards, give fundamental criteria for a sound evaluation of different sensor performances. Response curves and sensitivity
Output signal :
The output signal is the response of the sensor when the sensitive material undergoes modifications, in the following pattern :
SENSITIVE MATERIAL (M) TRANSDUCER SENSOR (Vout)
Different types of response curves exist. The linear response is the easiest one; it is characterized by the following equation Vout = aM + b. The other response is a non linear one; its equation is Vout = f(M).
Output noise :
Noise measurement must be done if one wants to have an accurate definition of the sensor. The noise is the output signal when the sensor does not measure any variations of the sensitive material. The noise depends on the frequency (cf. graph). If we consider two sensors with the same output noise but a different sensitivity (cf. graph), we can underline two statements :
statement 1 : the sensor which has the greatest sensitivity allows the detection of a lower M
statement 2 : the sensor which has the greatest sensitivity allows a better resolution with
respect to the other.
The resolution is the measurement level which gives, at the output, a signal to noise ratio S/N)
equal to 1.
In practice, (S/N) = 3 or (S/N) = 9.
We must distinguish between the resolution at the minimun M value and the resolution elsewhere on the response curve. Moreover, it is essential to consider that the resolution value follows the working point along the response curve and the boundary conditions. Selectivity / Contents The selectivity is the capacity of a sensor to be sensitive to a specific compound. The artificial sensing techniques are often based on sensor arrays (electronic tongue and electronic nose, for instance). In those cases, using less selective sensors is more interesting because one can detect a larger field of compounds. Reversibility / Contents The reversibility is the aptitude of the sensing mechanism to follow (of course with a given delay) the variation of the environment. It means that initial conditions must be obtained when the input reaches initial values.
In pratice, reversibility is a requirement for continuous monitoring applications (e.g. in environmental applications). However reversibility, as it requires week interactions between sensors and analytes, can not be compatible with high selectivity which needs strong interactions. When the sensors are non reversible,we can distinguish between : Regenerable sensors : the initial conditions can be ripristinated through an additional chemical process Disposable sensors which can be used only one time (e.g. medical sensors)
TYPES OF SENSORS
E-nose is classified based on the type of sensors used.
1. Conductivity Sensors
2. Piezoelectric Sensors
3. FET gas Sensors
4. Optical Sensors
5. Spectrometry based sensing methods
3.1. Conductivity Sensors
There are two types of conductivity sensors: 1) metal oxide 2) polymer, both of whichexhibit a change in resistance when exposed to volatile organic compounds.
3.1.1. Metal Oxide Type
These sensors are made of a ceramic former heated by a heating wire and coated by a semiconducting film. These semiconductor sensors can sense gases by monitoring changes in the conductance during the interaction of a chemically sensitive material with molecules that need to be detected in the gas phase.
They are used to detect toxic and flammable gases in domestic and environmental applications and for food aromas.
How to increase selectivity
The metal oxides are generally less selective than many other sensor technologies. Selectivity may be obtained using several methods:
- use of filters
- deposition of a suitable catalyst layer
- pulsing of sensor temperature in working conditions
- use of other semiconducting metal oxides
- control of the grain size Preparation techniques for gas sensors Metal oxide gas sensors can be subdivised into:
- Thick film devices (depositing a paste of material between two electrodes)
easier to produce
depend on ambient
temperatures and relative
long stabilizing times after
large power consumption
-Thin film devices: they use vapor deposition technologies in order to obtain a very thin film of metal oxide between two electrodes.
lower power consuption
more difficult to produce
Different deposition methods like PVD ( sputtering, thermal evaporation ... ), spray and solgel
techniques can be used for the preparation of thin film gas sensors.
A new method called RGTO enables to prepare mixed oxide thin films with high surface area and nanosized crystallites.
These sensors are manufactured by, among others, the Japonese company FIGARO.
Advantages and disadvantages of metal oxide semiconductor
- they are available because they are commercially produced
- they have high sensitivities to a range of organic vapors
- a variety of different types are available with broadly different sensitivities so that an array can be constructed
- they are characterized by a relatively fast response, typically less than 10 seconds
- their size
- they operate at elevated temperatures
- they are highly sensitive to compounds such as ethanol, CO2 or humidity Typical offerings include oxides of tin, zinc, titanium, tungsten and iridium, doped with a noble metal catalyst such as platinum or palladium, which operates at 200Ã‚Â°C to 400Ã‚Â°C.
As a VOC passes over the doped oxide material, the resistance between the two metal contacts changes in proportion to the concentration of the VOC.
Advantages: Wide availability and low cost.
Disadvantages: These are prone to drift over periods of hours to days. So, signal-processing algorithm should be employed to counteract this property. These sensors are also susceptible to irreversible binding by sulphur compounds.
Applications: The sensitivity ranges from 5 to 500 ppm. Used for sensing CO, NH3, or H2O.
3.1.2. Polymer Sensors
Here the active material is a conducting polymer from such families as the polypyroles, thiophenes, indoles or furans. Changes in the conductivity of these materials occur as they are exposed to various types of chemicals, which bond with the polymer backbone. A given compoundâ„¢s affinity for a polymer and its effect on the polymerâ„¢s conductivity are strongly influenced by the counter ions and functional group attached to the polymer backbone. Here the response time is inversely proportional to the polymerâ„¢s thickness, which is usually in the range of 10 to 20Ã‚Âµm.
Advantages: The sensitivity varies from 0.1 ppm to 100 ppm. No need of heaters, as they will operate in the ambient temperatures. High portability.
Disadvantages: They are difficult to make. Their responses also drift over time. More susceptible to humidity.
3.2. Piezoelectric Sensors
These are of two types â€œ QCM (Quartz Crystal Microbalance) and SAW (Surface Acoustic Wave devices). Here they are configured as mass change sensing devices.
3.2.1. QCM Type
It consists of resonating disc with metal electrodes on each side connected to read wire. The device resonates at a characteristic frequency (10 to 30 MHz), when excited with an oscillating signal. During manufacturing, a polymer coating is applied to the disc to serve as the active sensing material. In operation, a gas sample is adsorbed at the surface of the polymer, increasing the mass of the disk polymer device and thereby reducing the resonance
frequency. The reduction is inversely proportional to the odorant mass adsorbed by the polymer when the sensor is exposed to a reference gas. The resonance frequency returns to the baseline value.
Advantages: High sensitivity. Remarkably, linear over a wide dynamic range. Sensitivity does not change with temperature. Humidity response will depend upon the type of adsorbent material used. Batch to batch variability is not a problem as a differential change measurement of frequency change will remove the common mode noise. Disadvantages: When dimensions are scaled down to micrometer level the surface to volume ration will increase and so the noise to signal ratio also will increase.
3.2.2. SAW Type
Here a surface wave travels over the surface of the device. So sensors operate at much higher frequency and so can generate a large change in frequency. A typical SAW device operates in hundreds of Megahertz, while 10 MHz is more typical for a QCM. But SAW devices can measure changes in mass to the same order of magnitude as QCMs.
3.3. FET gas Sensors
The FET is a " metal "/ insulator / semiconductor structure in which the gate (the " metal ") can be any conducting layer or medium. The FET is a semiconducting device which acts as an amplifier (like a transistor). There are different FET configurations:
-the MOSFET: metal oxide semiconductor FET
-the SGFET: fabrication of a Suspended Gate on a metal-oxide-semiconductor
-the ISFET: ion sensitive FET
Advantages and disadvantages of the MOSFET
The reproducibility and the sensibility of the sensor are not sufficient enough to use it in a real measuring system, particularly for a multiple-component gas mixture. These are based on the principle that VOCs in contact with a catalytic metal can produce a reaction in the metal. The reaction products can diffuse through the gate of a MOSFET to change the electrical properties of the device. The sensitivity and selectivity of the device can be optimised by ranging the type and thickness of the metal catalyst and operating them at different temperatures.
Advantages: They can be made by IC fabrication processes. So the batch variations can be minimised.
Disadvantages: There should be a window in the IC chip to permit the flow of odorants to cause the catalytic product above the gate structure. So there is a necessity for sealing off all other areas. These sensors are also susceptible to baseline drift.
3.4. Optical Fibre Sensors
These utilise glass fibres with a thin chemically active material coating on their sides or ends. A light source at a single frequency is used to interrogate the active materials, which responds with the change in colour to the presence of VOCs. The active material contains chemically active fluorescent dyes immobilized in an organic polymer matrix. As VOCs interact with it, the polarity of the fluorescent emission spectrum.
Advantages: Cheap and easy to fabricate. Arrays of fibre sensors have wide range of sensitivities. Differential measurement is possible to avoid common mode noise. Disadvantages: Complexity of the measuring system. Limited lifetime to photo bleaching.
3.5. Spectrometry Based Sensors
Here a vapour trap is used to concentrate the VOCs and then it being injected into a spectrometer that generates a spectral response characteristic of the vapour. Then the efficient signal processing technique can be used for finding out the odorant. Here the disadvantage is that is the use of highly complex electronic measuring device.
Potentiometric Chemical Sensors Potentiometric Chemical Sensors are based on the measurement of a potential under no current flow. The measured potential may then be used to determine the analytical concentration of some components of the analyte solution. For useful definitions please go to Electrochemical terms and concepts. There exist different types of potentiometric chemical sensors. A classification shows the binding between them. This web-page will only develop the ion selective sensors (ISE) and the biosensors.
Ion selective sensors (ISE)
An ISE produces a potential which is proportional to the concentration of an analyte. Making measurements with an ISE is therefore a form of potentiometry. The most common ISE is the pH electrode, which contains a thin glass membrane that responds to the H+ concentration in a solution. Ion selective sensors are susceptible to several interferences. Samples and standards are therefore diluted 1:1 with total ionic strength adjuster and buffer (TISAB). The instrumentation of an ISE consists of the ion-selective membrane, an internal reference electrode, an external reference electrode, and a voltmeter. Different sorts of ion selective membranes exist : the glass, the chalcogenide and the crystal membrane. Research currently focuses on chalcogenide membranes.
The principle : coupling enzymes and electrode reactions. The enzyme is used as a bioelectrocatalyst of the oxydation or reduction of a substrate. There are 2 different ways of coupling enzymes and electrode reactions : with mediator or mediatorless
Example : measurement of the concentration of glucose
The electrochemical kinetic of enzymatic catalysis
Xi + ni e- Xi n- K(j) = K0 exp (-a ni F j /RT) avec a =0.5
Xj + n2 e- Xj n2-
Xj n- / Xj = K0 exp (-n2 F j /RT) =K(j)
In the first step, the substrate is absorbed. Then there is an equilibrium of the e- into the reacting site. The last step is the biochemical reaction. The e- tunneling in the bioelectrocatalysis is the process which transfers the e- from the electronic matrix to the enzyme site.
Two steps : E E* E + P
w® = exp (-r /A)
a = w® / (Kcat + w®)
Each step can be the limiting one.
The importance of the deposition of the enzymes on the surface.
The effectivity to electrons transfer depends on the distance between the enzymes and the electrode. Therefore the deposition on the enzymes of the surface must be made with care. Advantages and disadvantages of potentiometric chemical sensors Advantages
- A wide range of available sensing materials and sensors.
- Wide variations of sensor properties, some unique features
- A wide knowledge about composition / properties relationship
- Simple installation. Easy, direct measurements.
- Different configurations (static, flow, bulk, micro).
- Easy applicability for automatic and / or industrial analysis.
- Low cost.
- Insufficient selectivity of many sensors.
- The number of available sensors is still smaller than the number of analytes.