WO2006126011A2 - Spectral nose - Google Patents

Abstract

The application is directed to an artificial nose comprising an array of solvatochromic dyes, eg. Reichardt's dye, Nile red, 4-nitro-N,N' -diethylaniline, Cu(II) bis-acetoacetonate etc. and pattern recognition means, eg. a neural network. The device finds use in identifying volatile vapours, eg. from explosives, perfumes, drugs and diagnostic markers of disease, eg. diabetes.

Description

Spectral Nose

Field of the Invention

The present invention relates to a method of using a sensor-array to identify a vapour. The invention also relates to apparatus comprising a sensor-array to identify a vapour.

Background of the Invention Of all the senses, smell is arguably the most valuable and impressive technically. Smell can be used to track and find, e.g. a mate or potential food, and to identify, e.g. a location, friend from foe and good from bad. The large diversity in areas of smell application is due to the wide range of smells, which can be individually identified, and the high sensitivity- exhibited by the olfactory system. Consequently, there is increasing interest in the development of artificial noses . ^l A number of different types of artificial noses have been developed to date, with most using an array of nonspecific and generally similar sensor elements. For any particular artificial nose, all the array elements are usually based on the same sensing process, such as surface-acoustic-wave (mass change),² piezoelectric (frequency change),³ electrochemical (current or potential change)⁴ or conducting polymer (conductivity change) sensing.⁵ Thus, each sensor array element measures a single parameter, such as mass, frequency or conductivity. Most artificial noses then use artificial neural networks to process and interpret the data. Neural networks attempt to mimic some of the unique characteristics of the brain, such as the ability to learn general mechanisms from a small set of examples and to extract out useful information from a complex series of inter-related inputs.⁶ Such neural networks can learn the patterns of response associated with different odours and thus be trained to recognise certain smells .

Artificial noses can be trained to recognise a particular body breath or odour.^7'9 Diabetes is characterised by a rotten apple smell, which is due to high levels of acetone in the blood. This is passed from the blood to the lungs, and subsequently exhaled on the breath.¹⁰

Most artificial noses developed to date are electrical devices and are generally termed ^Λelectronic noses'¹¹ such as that reported by Walt and co-workers.¹²' ¹³ In Walt and co-workers ¹²'¹³ they encapsulated 1 fluorescent dye, Nile Red, into a series of different polymers to create a bundle of 19 individual optical sensor elements, each measuring a single parameter, i.e. overall luminescence intensity. The responses of the individual sensor elements are different because each employs a different polymer for encapsulating the dye. The different polymers interact differently with the same vapour; the vapour partitioning and swelling each polymer to a different degree. Rakow and co-workers¹⁴ also report a sensor array based on colour changes. Their method utilises the colour change induced in an array of metalloporphyrin dyes. The dyes complex with vapours, inducing shifts in wavelength (colour) . The shifts are, however, small. Most optical sensors for carbon dioxide or oxygen rely on a dye-containing transducer that responds, via a change in colour or fluorescence intensity, in a reproducible, quantitative, reversible and, often, selective manner when exposed to the analyte under test.¹⁵ Optical sensors for carbon dioxide usually employ a pH- sensitive dye, such as m-cresol purple, to measure the change in the pH' of its surroundings produced by a change in the ambient level of carbon dioxide.¹⁶ Oxygen sensors are usually based on fluorescent dyes, the luminescence of which is quenched by oxygen.¹⁷

It is an object of at least one aspect of the present invention to obviate or mitigate at least one or more of the aforementioned problems.

It is a further object of the present invention to provide a method for spectroscopically identifying a vapour.

Summary of the Invention

According to a first aspect of the present invention there is provided a method of identifying a vapour of a volatile compound comprising: collecting a vapour of a volatile compound; spectroscopically analysing the vapour and obtaining a vapour fingerprint of the vapour using a sensor array comprising at least two solvatochromic dyes in a host medium; and performing a pattern recognition analysis on the vapour fingerprint to thereby identify the vapour.

The present invention relates to a spectral nose (sometimes called a photonic nose) based on a minimum of two different solvatochromic dyes dissolved or encapsulated in a common host medium. The at least two solvatochromic dyes may make up sensor elements forming the sensor array. The response of the sensor elements to organic or inorganic vapours may be analysed using an artificial neural network, or similar, such that spectroscopic information such as absorbance, wavelength and related temporal information may be collected in order to identify a vapour fingerprint (i.e. a molecular signature) of an associated vapour. The method may be capable of analysing qualitatively and quantitatively both single and multi- component (i.e. greater than or equal to 2 components) vapours .

In a first step of the method of identifying the vapour, the vapour may be collected using any suitable type of gas collection device such as any form of sniffing device known in the art. A sniffing device may collect a finite sample of vapour and allow it to come into contact with the sensors. There are many variations that may be used. For example, a relatively sophisticated sniffing device may have a purge of carrier gas carrying the vapour to the sample. This type of system may be particularly useful for temporal/fast response information as the vapour is only momentarily in contact with the sensors. Another type of device may hold the vapour in contact with the sensors for a defined time, i.e., vapour headpsace analysis. Thus, in this case the response is more of an equilibrium nature.¹¹

The collected vapour may be held in a storage chamber to await spectral analysis. Alternatively, the collected vapour may be transported to a remote site where the analysis may be conducted.

The sensor array may have a rapid response, such as in seconds (e.g. 1 to 10 seconds), and have a broad range of detection and high sensitivity, for example less than lOOOppm, less than 500 ppm or less than 100 ppm. The sensor array may have a detection range of 1100 nm to 350 nm. The spectral nose may be simple to train to recognise the vapour which in some instances may be a complex vapour with many separate components . Characteristic odours associated with many diseases as well as commercial products, such as soaps, perfumes, food and beverages may be analysed. The method may also be used as a medical diagnostic device and/or quality control tool .

In the current invention, solvatochromic dyes are integral to the sensor elements because of their sensitivity to organic solvents, hence vapours.¹⁸' ¹⁹

It has long been known that solvent may have an effect on the rate of chemical reactions and these effects may be due to the interactions of the solvent molecules with the solute, which result in varying degrees of bonding and energetic stabilisation. Solvatochromism occurs because of similar effects.¹⁸' ¹⁹

Typically, solvatochromic dyes may be compounds that have UV/vis/NIR absorption spectra that may be changed in position, intensity and shape when placed in different solvents. Solvatochromism may be caused by differential solvation of the ground and first excited electronic states of a light absorbing molecule, or its chromophore.

A hypsochromic (or blue) shift of the UV/vis/NIR absorption band with increasing polarity may be called ^ynegative solvatochromism' and the corresponding bathochromic (or red) shift may be called ^positive solvatochromism' . One tangible effect of this solvent effect may be that the (JV/vis absorption spectrum of the dye can change, depending on the solvent used. This change can be in the position, intensity, or even shape of the absorption band. A classic example of a solvatochromic dye may be that of the ΛT-phenolate betaine dyes such as (I) shown below - 2, 6-Diphenyl-4- (2, 4, 6-triphenylpyridinio) - phenolate.¹⁹

R = -H R = -C(CH₃J₃ R = -CF₃

(I-Reichardt's dye, [R = -H])

Using the above dye, commonly known as Reichardt's dye, as a standard, a solvent polarity scale, £_τ(30), may be developed based on the shift in the UV/vis spectrum of the dye.¹⁹ It has a large spectral range, from approx 450 to 900nm (all of the visible region) . The primary dye (R = -H) , may be only slightly soluble in water and other less polar solvents. It may also be insoluble in non- polar solvents. This can be overcome by using differently substituted analogues, e.g. R = -C(CH₃)₃ or -CF₃ as secondary standards because there may be an excellent correlation between the dyes. R = -CF₃ may allow determination of values for more acidic solvents whereas R = -C (CH₃) ₃ may be more soluble in non-polar solvents than the original dye, R = -H. The simple scale of solvent polarity, E_τ(30) (kcal mol^"1) , may be constructed based on the wavelength maximum (λ_max) values for Reichardt's dye in a wide variety of different solvents. Thus, the value of E_τ(30) for any solvent can be obtained from the values of λ_max or v_max (v_raax = l/λ_max) through the following expression:

E_τ (30) = 2.8591. v_max (in 1000 cm^"1) = 28591/λ_max (in nm)

Illustrated in Figure 1 are examples of visible absorption spectra of Reichardt's dye when dissolved in a selection of solvents of different polarity. Figure 1 helps enforce the striking solvatochromism exhibited by Reichardt's dye. E_τ(30) may be a single parameter solvent scale and, as a result, the amount of information it can provide about a solvent is limited because all the features of a solvent, such as hydrogen bond donor ability (HBD) , hydrogen bond acceptor ability (HBA) , polarity and polarisability are represented by just one parameter, E_τ(30) . Most importantly, Reichardt's dye does not interact specifically and significantly with HBA solvents, such as organic amines but does with HBD solvents, like alcohols and weak acids. This drawback of the E_τ (30) scale has led to the development of a number of multi-parameter solvent scales, the most well known is that developed by Kamlet and Taft in 1976²⁰ based on a linear solvation energy relationship. In it simplest form this relationship can be expressed as follows:

XYZ = XYZ₀ + a.α + b.β + s.π* where XYZ₀₁ a, b and s are solvent independent coefficients characteristic of the process under study and indicative of its susceptibility to the solvents parameters α, β and π* . In work with solvatochromic dyes, the parameter XYZ is equal or, directly related, to the inverse of λ_max of the dye in the solvent.

The properties, α, β and π* are solvatochromic properties of the solvent, derived from the absorption spectra of a series of carefully selected probe solvatochromic dyes. The parameter, α, provides a measure of the HBD ability of the solvent, β, the HBA ability of the solvent, and π*, the polarity/polarisability of the solvent. These parameters may be essentially orthogonal to each other and, as a result, in most cases their values effectively may characterise any and every solvent.²¹ Thus, if these three parameters are measured for a single component vapour, they may effectively characterise and identify that vapour. This feature may be exploited in this invention, the spectral nose, which may comprise two or more sensor array elements, each containing a different, specially chosen solvatochromic dye, may be used to create a spectral nose.

The Kamlet-Taft relationship for Reichardt's dye is detailed below: Name: 2, 6-diphenyl-4- (2, 4, 6- triphenylpyridinio) phenolate) ; Reichardt's dye

v_max (in 1000 cm^"1) : 12 . 3 (ether) - 122 . 1 (water) Δv_max = -9730 cm^"1 α = 0 . 186v_max - 2 . 03 - 0 . 72π* One element in this invention may be the use of Reichardt's dye, since its value of v_max in many solvents is related to both α and π* .

Reichardt's dye has previously been used in optical sensing for a number of vapours including, water (humidity),²²' ²³ ethanol,²⁴ methanol²⁴ and ammonia,²⁵ among others. The types of encapsulation media for Reichardt's dye have generally been polymer, such as polyvinyl acetate²² and polymethylmethacrylate.²³ The novelty of this invention, as opposed to the sensors referred to above, is that an array of sensor elements with different solvatochromic dye sensor elements may be used to give an array of shifts, rather than just one element shifting. With just one sensor, vapours with similar characteristics would be very difficult to distinguish between, e.g. the homologous series of alcohols will have similar shifts and it may be difficult to distinguish between two similar vapours, e.g. methanol and ethanol. In the present invention the shifts generated from more than two sensors, such as 3 - 10 or 3 - 5 sensors sensors cannot be the same and the combination of shifts generated for each vapour will be unique, i.e. a series of spectral coordinates which define a vapour fingerprint . As well as Reichardt's dye, a number of other solvatochromic dyes which can extract values of α, β and π* may be used. Some of those used in this invention are detailed in Table 1. The solvatochromic dyes in Table 1 may be combined into any suitable combination. For example, 2 - 10, 3 - 5 or 3 solvatochromic dyes may be combined to form the sensor array. Table 1 Table of solvatochromic dyes. NB. Abbreviation in parenthesis C].

* (tmen) (acac) = tetramethylethylenediaminoacetylacetonato **Z scale = 4-carbomethoxy-l-ethylpyridinium iodide

As can be seen from Table 1, the dyes can be classified into one of three categories, α, β and %* , depending on which parameter they are most suitable for probing. This is not to say that the dye measures exclusively that particular property. For example, as stated above, due to its molecular structure Reichardt' s dye can measure both the α and, π* properties of a solvent. Another example of a more specific probe dye may^¬ be 4-Nitro-N, N-diethylaniline, (II). It's value of v_max may provide a direct measure of π*,²¹ i.e.

(ID

Name: 4-nitro-IV_/I\J^r-diethylaniline

v_max (in 1000 cm^"1): 27.5 (cyclohexane) - 24.3 (DMSO) = +3185 cm^"1

0.314(27.52 -

A third example may be that of tetramethylethylenediaminoacetylacetonato-copper ( II ) perchlorate, (III) , since its value of may provide a

direct measure o ²⁶' ²⁷ i.e.

(III)

Name : tetramethylethylenediaminoacetylacetonato- copper(II) prechlorate; [Cu¹¹ (acac) ( tmen) ] CIO₄

v_max (in 1000 cm^"1): 18.5 (1, 2-dichloroethane) - 17.4 (water)

Δv_max = +1100 cm^"1 β = 0.358(18.76 - v_max)

The list of dyes above is not meant to be restrictive; rather it is used to illustrate the ideal feature of orthogonality in optical response the solvatochromic dyes of different analysis ^Λclass' should exhibit. Other solvatochromic dyes can be used depending on other factors such as, lack of solubility. Shown below are the molecular structures of the dyes identified in Table 1. Any form of combination of the dyes may be used. The combination of dyes may be such that the solvent parameter values associated with α, β and π* for the vapour under test can be extracted from the spectral data and so be used to identify the vapour. Ideally, a combination of only 3 dyes would be needed, each one specific to only one of the solvent parameters, i.e. one for α, β and π* . In practice, most solvatochromic dyes show some cross-sensitivity and so, in order to define/identify a vapour by its α, β and π* values, 5 or 6 dyes may be needed, depending upon the dyes chosen. The dyes will provide spectral information which relates to any vapour and so define a vapour set of spectral coordinates. These co-ordinates can be used to calculate α, β and π* values, but this is not required in order to operate the spectral nose.

Some dyes, such as Reichardfs dye, are "better" than others, in the sense that they are incredibly responsive. Concentrations are variable depending predominantly on the molar absorptivity (ε) of the dye. For example a dye with a large ε value (4-nitroaniline) has a concentration value of 1-5 mg/3g polymer solution; medium ε value (Reichardfs dye), 50-75 mg/3g polymer solution; low ε value (copper [tmen] [acac] perchlorate) , 100-300 mg/3g polymer solution. Note, these concentrations are chosen specifically to give an absorbance value of between 1 and 2.

Reichardt ' s dye , moreso than any of the other solvatochromic dyes detailed above , has been incorporated into a variety of matrices and solid supports for possible use as an optical sensor for volatile organic compounds. The basic premise is that a dye in a solid support will undergo a noticeable/detectable colour change when exposed to solvent vapour, much as it would when dissolved in different solvents. However, the containment of the dye within a solid support should prevent its dissolution by the solvent vapour and so minimises the risk and cost of losing the dye.

Typically, the sensor array may be exposed to a number of common inorganic or organic vapours selected from any one of or any combination of the following: aromatics, alkenes, alkynes, alkanes, alcohols, ketones, aldehydes, ethers, halogenated solvents, nitriles and carboxylic acids. Examples include: pyridine, xylene, benzene, toluene, tetrahydrofuran, dioxane, ethyl acetate, ethene, ethylene, ethane, pentane, hexane, cyclohexane, methanol, ethanol, isopropanol, butanol, acetone, butanone, methyl ethyl ketone, formaldehyde, diethyl ether, carbon tetrachloride, chloroform, dichloromethane, dimethyl sulphoxide, acetonitrile, carbon disulphide, acetic acid, water, among many others.

The spectral responses such as changes in wavelength, absorbance, spectral shifts may be recorded on exposure to the at least two solvatochromic dyes . The vapour may be generated by bubbling dry air through the solvent of interest. It may be observed that the position and intensity of the spectra of the at least two solvatochromic dyes may undergo a substantial change when the sensor array may be exposed to the various vapours. It may be found that each of the various vapours may form a vapour fingerprint thereby allowing a pattern recognition analysis to be conducted.

For example, a neuro network analysis may be conducted to obtain as an output Kamlet-Taft parameters thereby obtaining Kamlet-Taft parameters determining a set of spectral coordinates for the vapour. Typically, the neuro network analysis may be carried out using parameters obtained by determining the spectral changes and associated Kamlet-Taft parameters associated with a set of predetermined calibration vapour samples such as methanol, ethanol, isopropanol (IPA), dichloromethane

(DCM), acetone and water.

Conveniently, computer software may then be used to perform a pattern recognition analysis on the obtained spectral data wherein the neuro network may be used against a database of calibration samples to identify the vapour .

Alternatively, any suitable type of pattern recognition analysis may be conducted using spectral changes in wavelength, absorbance and spectral shifts.

Alternative pattern recognition approaches include: hierarchical cluster analysis, principal component analysis, K-nearest neighbour and soft independent modelling.^28"30

The host medium for the at least two solvatochromic dyes may be any suitable organic or inorganic material (e.g. solvent) .

Typically, the host medium may be a polymer matrix prepared by dissolving an appropriate amount of polymer in an appropriate organic solvent. Polymer concentration may depend very much upon the polymer (solubility, molecular weight etc.) . Typical examples are ethyl cellulose (46%), l-5g per 20 cm³ solvent (17.5:2.5 toluene : ethanol) ; cellulose butyrate acetate l-5g per 20 cm³ solvent (chloroform or acetone) . These are very dependent on the thickness of polymer film required.

Also, the solubility of dye has an influence on the polymer/solvent ratio. If the dye is not particularly soluble a higher solvent : polymer ratio may be used to facilitate dissolution of the dye. The table below shows more examples.

Table 3 - Examples of Suitable Polymers

For example, the polymer may be any combination of the following: ethyl cellulose; polyvinylchloride; hydroxethyl cellulose; nitrocellulose; polyethylene oxide; polyethylene glycol; polystyrene; biopol; polyvinyl alcohol; polymethylmethacrylate; polyvinylpyrollidone; poly (2, β-dimethyl-1, 4-phenylene oxide) ; poly [2- (4-benzoyl-3-hydroxy-phenoxy) ethyl acrylate] ; poly (acrylonitrile-co-butadiene-co-styrene) ; polymethacrylsaeuremethylester poly (methyl methacrylate) ; poly (ethyl methacrylate-co-methyl acrylate); poly (ethylene-co-acrylic acid); poly (tetrafluoroethylene- co-vinylidene fluoride-co-propylene; poly (vinyl butyral- co-vinyl alcohol co-vinyl acetate) and cellulose butyrate acetate.

The solvent may be any combination of the following: ethanol/Toluene; tetrahydrofuran; water; ethanol; chloroform; dichloromethane; methanol; toluene.

Alternatively, the host material may be formed from any form of silica such as fumed silica, or aerogel formed from silica-based sol-gels. Any suitable solvent such as methanol may be used.

In further alternatives, the host material may be formed from titania based material such as nanocrystalline titania thin films.

Conveniently, it may also be found that in addition to a qualitative analysis, a quantitative analysis may also be conducted to determine the quantity of each type of vapour present. This may be performed by integrating under the obtained spectral curves. Concentration may also be measured as a function of wavelength shift or, more likely, a combination of absorbance and wavelength shift. The method may be used for, for example, analysing packaging 'materials, pharmaceuticals, perfumery, petrochemicals, food and beverages, household products, raw materials, monitoring indoor air quality, monitoring gaseous effluents from industrial waste heaps, the detection of explosives and biological agents, drugs and the diagnosis of diseases. For example, diseases which may be identified include any of the following: cirrhosis of the liver, lung cancer and diabetes which are often characterised by a particular body breath or odour. Lung cancer has no characteristic "smell" as such, although there are vapours associated with it. In fact there are many conditions with charcteristic vapours but not necessarily "smells" as such. Other diseases/conditions with characteristic smells include: lung abcess or bronchiectasis, renal failure, nasal cancer.

According to a second aspect of the present invention there is provided apparatus for identifying a vapour of a volatile compound comprising: a sensor array including sensor elements comprising at least two solvatochromic dyes in a host medium wherein the sensor array is capable of interacting with a vapour; a spectroscopic device for analysing the vapour by utilising the interaction between the vapour and the sensor array to obtain a vapour fingerprint; and means for performing a pattern recognition analysis of the vapour fingerprint and thereby identifying the vapour .

The colour change may due to a change in the polarity of the environment of the dye which is brought about by the vapour.

Typically, the apparatus may comprise a device for collecting a vapour. Typically, any form of device may be used for collecting the vapour such as any suitable sniffing device well known in the art.

The apparatus may also comprise a sampling chamber for containing the vapour.

Typically, any form of sampling chamber may be used to contain and hold the vapour. The sampling chamber may have a volume of about 50-100 cm³.

Conveniently, the sensor array may comprise any suitable number of solvatochromic dyes more than 2. For example, 2 - 10, 3 - 4 or 3 sensors may be used.

The sensor array may consist of a mounting "stage" for the sensors. The array may be enclosed in a sealed sample chamber (where the vapour is held) such that the sensor elements are in intimate contact with the test vapour. Typically, the means for performing the pattern recognition may comprise computer software for comparing the obtained vapour fingerprint with a database of calibration samples prepared from, for example, any of the following: methanol, ethanol, isopropanol (IPA) , dichloromethane (DCM), acetone and water.

Conveniently, the apparatus may be portable. In the situation when the apparatus is portable, this allows point-of-care analysis to be conducted such as in a general practitioner's surgery.

Conveniently, the vapour may be collected and then transported to a remote site for spectral analysis.

A non-portable device may comprise all necessary components such as a spectrophotometer and a control/data processing P. C. and the necessary pattern recognition software. Neural network analysis, principle component analysis etc. may be used. The apparatus may also comprise: a light source; a photo detector; spectral filters, a dedicated chip with vapour profiles; comparison algorithms; and vapour profile storage facilities; a fibre optic splitter; and a device to split/recombine as necessary various light beams into components for each sensor element.

According to a third aspect of the present invention there is provided the use of the method according to the first aspect in analysis of vapours in any of the following: packaging materials; pharmaceuticals; perfumery; petrochemicals; food and beverages; household products; raw materials; monitoring indoor air quality and gaseous effluent from industrial waste heaps; the detection of explosives, biological agents and drugs; and the diagnosis of diseases.

For example, the diagnosis of diseases may include cirrhosis of the liver, lung cancer and diabetes.

According to a fourth aspect of the present invention there is provided a sensor array including sensor elements comprising at least two solvatochromic dyes in a host medium wherein the sensor array is capable of interacting with a vapour and identifying a vapour using a vapour fingerprint analysis.

Brief Description of the Drawings

Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings in which:

Figure 1 is a visible absorption spectra of Reichardt's dye in ethanol, MeCN, acetone and 1,4- dioxane; Figure 2 is a UV/Vis spectra of a Reichardt's dye/ethyl cellulose sensor element exposed to a number of common organic vapours and water;

Figure 3 shows wavelength/absorbance shifts of a Reichardt's dye/ethyl cellulose sensor element exposed to various concentrations of ethanol vapour;

Figure 4 shows a calibration curve of absorbance versus percentage ethanol vapour for a Reichardt's dye/ethyl cellulose sensor element exposed to varying concentrations of ethanol vapour;

Figure 5a shows a plot of UV/Vis spectra of a Reichardt's dye/ethyl cellulose sensor element in air and ethyl vapour;

Figure 5b shows the change in absorbance at 680 nm in response to ethanol vapour (decrease in absorbance) and on recovery from the vapour (increase in absorbance) ;

Figure 6 shows a UV/Vis spectra of a Reichardt's dye/cellulose butyrate acetate sensor element after exposure to a number of common organic vapours and water; Figure 7 shows a UV/Vis spectra of a Reichardt's dye/fumed silica sensor element exposed to a number of common organic vapours;

Figure 8 shows a plot in the change in absorbance at 550 nm in response to ethanol vapour (increase in absorbance) and on recovery from the vapour (decrease in absorbance) ;

Figure 9 shows a UV/Vis spectra of a Reichardt's dye/fumed silica sensor element exposed to various concentrations of methanol vapour; Figure 10 shows a calibration curve of absorbance versus percentage methanol vapour for a Reichardt's dye/silica sensor element exposed to various concentrations of methanol vapour; Figure 11 shows a UV/Vis spectra of a Reichardt's dye/TiO₂ sensor element exposed to a number of common organic vapours;

Figure 12 shows UV/Vis spectra of various ethyl cellulose-based sensor elements in dry air;

Figure 13 shows histograms of spectral shifts for six organic vapours in a sensor array containing sensor elements consisting of Reichardt's dye, 1-ethyl, 2,6-D and copper; and Figure 14 is a schematic representation of apparatus used to identify vapours according to the present invention .

Detailed Description

1) POLYMER/DYE SENSOR ELEMENTS

(i) Preparation of polymer substrate:

The polymer solution (A) was prepared by dissolving the appropriate amount of polymer in an appropriate organic solvent (typically l-5g in 20-lOOml) . Some examples are shown in Table 3.

(ϋ) Preparation of dye/polymer solution

A defined amount of, for example, 3g, of polymer solution (A) was then added to the required amount of the dye (B) and stirred at room temperature until all of the dye was fully dissolved in the polymer matrix. This produced a dye/polymer solution of known dye concentration (C) .

|iii) Preparation of sensor element A small amount (typically 1-2 drops) of dye/polymer solution (C) was deposited on to a small glass coverslip which was then spun at 1200rpm for 30s to give a thin film ca. lOOμm in thickness. The sensor element was then placed in a vented oven for 2 hours at 70°C to remove the solvent .

2) SILICA/DYE SENSOR ELEMENTS

(i) Preparation of dye/silica powder

A defined amount of dye (typically 2-25mg) was added to a defined amount of various grades of fumed silica powder (typically Ig) in methanol, or another appropriate organic solvent; chosen for solubility of the dye. The mixture was then stirred at room temperature for 4 hours after which time the solvent was removed under vacuum/heat. A homogeneous dye/silica powder was thus obtained.

(ii) Preparation of the dye/silica sensor element

A glass coverslip, precoated in a thin (lOOμm) coating of ethyl cellulose, or another suitable polymer, was placed in the dye of a standard press unit. 30-50mg of dye/silica powder was then weighed out and placed on the coverslip (polymer coated side) and spread as evenly as possible. The powder was then pressed at approximately 2 tonnes for 30 minutes, and then removed from the press. 3) NANOCRYSTALLINE TITANIA THIN FILM/DYE SENSOR ELEMENTS

(i) Preparation and deposition of Titania precursor

The titania precursor was prepared according to the sol-gel method. A full preparation is detailed in Example 4.

Figure 14 is a representation of apparatus, generally designated 10, used to identify a vapour according to the present invention. The apparatus 10 may have a number of sensor elements, generally designated 12. The sensor elements 12 are housed in a sensor chamber 14 which is mounted within a disposable head 16. First of all, vapours enter through a mouthpiece 20 and is then passed through a vapour dehumidifier 22. The vapours on passing through the vapour dehumidifier 22 then enter the sensor chamber 14 housing the sensor elements 12. As shown in Figure 14, a fibre optics splitter 18 is mounted on either side of the disposable head 16. Gases may be expunged through vent 24. As shown in Figure 14, light exiting the fibre optics splitter 18a passes through a UV/visible spectrophotometer 26. Light then passes into the fibre optics splitter 18b. A control/data processing PC 30 is used to monitor and control the operation of the apparatus 10. In alternative embodiments, in place of a UV/visible spectrophotometer, a light source photo detector and filters may be used. In addition, in alternatives to a control/data processing PC, a dedicated chip with vapour profiles, simple comparison algorithms and a vapour profile storage facility may be used.

Examples of Different Sensor Elements The present invention is further illustrated by the following non-limiting examples. The general methods of preparation are detailed in the examples . These methods are general and many variations on these have been used:

Example 1: A sensor element based on Reichardt' s dye encapsulated in ethyl cellulose — exposed to a number of common organic vapours and water

In this example, one sensor element used in the array of sensor elements, is demonstrated. The sensor element is based on Reichardt' s dye encapsulated in a polymer matrix, specifically ethyl cellulose. The sensor was exposed to a number of common organic vapours, namely methanol, ethanol, isopropanol (IPA), dichloromethane (DCM), acetone, and water. The spectral responses were recorded.

The sensors were prepared by dissolving 25mg of Reichardt' s dye in an ethyl cellulose solution (2g ethyl cellulose in toluene/ethanol, v/v 17.5:2.5ml) under stirring (30 minutes) . The dye solution was then cast onto a 22mm glass coverslip at 1200rpm, 30s. The films were dried in the oven at 7O⁰C for 2 hours prior to testing . The vapours were generated by bubbling dry air through the solvent of interest, contained in a Dreschel bottle, forcing the solvent vapour to pass over the sensor, housed in a sealed cell.

The results are illustrated in Fig. 2 and collated in Table 5. As can be seen there is a substantial change in the position and intensity of the spectra when the sensor edement was exposed to the various vapours. Table 5 Wavelength maxima (λ_maκ) and wavelength shift (Δλ) values for a Reichardt' s dye/ethyl cellulose sensor element exposed to a number of common organic vapours and water .

Following the qualitative identification of the various vapours, the possibility of quantitative analysis was tested using ethanol as a test volatile. The concentration of the vapour was varied by mixing ethanol vapour with a flow of air in a gas blender; increasing the flow from 0 to 100% solvent vapour in 10% increments. 100% vapour was taken as the concentration at saturation vapour pressure, P_satf at 2O⁰C. The sensor was exposed to various vapour/air mixtures and the spectra recorded (Fig. 3) . The data show that it is possible to use a plot of λ_max vs % ethanol as an empirical calibration curve for the quantitative analysis of ethanol in air using the RD/EC films. It was found that by monitoring the absorbance of such a film at a fixed wavelength (associated with the maximum change in absorbance as a function of % ethanol) a calibration graph such as illustrated in Fig. 4 could be generated.

The calibration graph is very linear, demonstrating that the absorbance of a Reichardt 's dye/ethyl cellulose film at 680nm is directly related to the % ethanol in the ethanol/air mixture. An important characteristic of any sensor element is response time, i.e. how quickly it responds to/recovers from exposure to the vapour. The response and recovery times of the sensor in ethanol vapour were recorded. The response of the sensor was measured at a fixed wavelength (680nm) . It was exposed to ethanol vapour for 10 minutes and then to air for a further 10 minutes. This process was repeated and the cyclic change in the absorbance of the film recorded, as shown in Fig. 5.

In Fig. 5 it can be seen that the change in absorbance is reversible over an extended period, allowing for multiple uses of one sensor. The data presented in these plots was used to calculate the tso and tgo values for response and recovery. These values represent the time taken to achieve 50% and 90% of the total change in absorbance in the system and are shown in Table 6.

Table 6 Response and recovery times of a Reichardt' s dye/ethyl cellulose sensor element on cyclic exposure to ethanol vapour and air.

This, and other work, shows that for equilibrium work, the sensor elements need « 2 minutes exposure to respond to most vapours.

Example 2: A sensor element based on Reichardt' s dye encapsulated in cellulose butyrate acetate - exposed to a number of organic vapours In this example, one sensor element used in the array of sensors is demonstrated. The sensor element is based on Reichardt's dye encapsulated in a polymer matrix, specifically cellulose butyrate acetate. This example was chosen as the sensor performs well in humid as well as dry environments. The sensor was then exposed to a number of common organic vapours, namely methanol, ethanol, isopropanol (IPA) , dichloromethane (DCM) , acetone, and water-and the spectral responses recorded. The sensors were prepared by dissolving 25mg of Reichardt's dye in an cellulose butyrate acetate solution (5g celluose butyrate acetate in 10ml chloroform) under stirring (30 minutes) . The dye solution was then cast onto a 22mm glass coverslip at 1200rpm, 30s. The films were dried in the oven at 70⁰C for 2 hours prior to testing.

The results are illustrated in Fig. 6 and collated in Table 7. There is a substantial change in the position and intensity of the spectra when exposed to the various vapours.

Table 7 Wavelength maxim nd wavelength shif alues for a

Reichardt' s dye/cellulos u yra e acetate sensor element exposed to a number of common organic vapours and water.

Example 3 : A sensor element Reichardt' s dye encapsulated in hydrophilic silica - exposed to a number of organic vapours

Example 3 relates to a sensor element prepared by encapsulating Reichardt' s Dye in Fumed Silica.

A typical film was prepared by mixing 1.9g of fumed silica into approximately 125ml of continuously stirred methanol in a round-bottomed flask. To this was added 0. Ig of Reichardt' s dye pre-dissolved in small volume of methanol - typically 10ml. This solution was slowly added to the silica suspension. The resultant mixture was stirred for 30 minutes before the solvent was removed by rotary evaporation, leaving a homogeneous, dry blue/violet coloured powder. The sensor element was prepared by pressing the powder into a thin opaque disk. This was carried out by first preparing a glass cover slip layered with ethyl cellulose (using a spin-coating technique) . This glass disk was then placed on top of the first stainless steel pellet in a 25mm diameter infra red

(IR) die. An infra red press may be used i.e. the type used to make KBr disks. 50 mg of the dye/silica powder were then placed on top of the first pellet in the IR die and the second pellet then placed on top to form a sandwich. Once the die was assembled approximately 2000kg of pressure was then slowly applied to the system, using an IR press, and then left for 30 minutes. After this time, the pressure was slowly relieved and the glass cover slip, now bonded with a thin layer film of the impregnated silica, was removed from the press. The final discs were a violet/blue colour. Fig. 7 shows the spectra of the discs exposed to various organic vapours, as well as dry air. The solvents chosen for study were the same as the previous examples; methanol, ethanol, isopropanol (IPA), dichloromethane

(DCM), and acetone.

Table 8 Wavelength shifts of a Reichardt' s dye/fumed silica sensor element exposed to a number of common organic vapours .

Again, as with the ethyl cellulose sensors, the response and recovery times were measured. Ethanol was also chosen as the test vapour once more. All the procedures described previously were followed and the following plots were obtained: The data in Fig. 8 clearly show that, like the polymer-based systems, the change in absorbance with repeated exposure to air and then solvent vapour is reversible over an extended period. The change in absorbance occurs over a short time period. The tso and tgo values calculated from the data in Fig. 8 are shown in Table 9 below.

Table 9 Response and recovery times for a Reichardt' s dye/fumed silica sensor element exposed to ethanol vapour.

Having noted the fast response of the silica-based sensor in ethanol vapour, the response times of the homologous series of alcohols was tested. The alcohol vapours tested were: methanol, ethanol (results above), 1-propanol, 1-butanol, 1-pentanol.

As expected, the response and recovery times increase with increasing alcohol chain length (and decreasing vapour pressure) . The results of the tests are detailed in Table 10.

Table 10 Response and recovery times for a Reiσhardt' s dye/fumed silica sensor element exposed to various alcohol vapours .

The increase in response and recovery times shown above is to be expected. The higher alcohols have a lower vapour pressure and, consequently, are less volatile. As they are also larger molecules they will diffuse through the matrix at a slower rate than methanol or ethanol, thus slowing the time it takes to enter and exit the system. Similarly to the ethyl cellulose-based sensors, tests on the quantitative analysis of vapours was carried out. The same procedure was followed as outlined for the ethyl cellulose discs;^' a gas blender was used to mix the solvent vapour with a flow of air at fixed percentages. In this instance the experiment was repeated for the five alcohols used previously: methanol, ethanol, propanol, butanol, and pentanol . Shown below in Fig. 9 are the data for methanol at various concentrations . Fig. 9 clearly shows that a silica disc impregnated with Reichardt' s dye can distinguish between different concentrations of methanol, likewise with the other alcohols (results not shown) . Thus, a calibration graph can be constructed, as shown below (Fig. 10) . It is clear from the results that the fumed silica discs are able to distinguish between different concentrations - and allow calculation of calibration curves - for a variety of solvents and as such ^■ can be used as quantitative, as well as qualitative, analytical sensors.

Example 4: Reichardt' s dye encapsulated in titania film on glass - exposed to a number of organic vapours

The following example relates to a sensor element comprising glass coated with a film of titania prepared with by a sol-gel method. The titania film is infused with Reichardt' s dye.

A homogenised TiO₂ paste was prepared as follows: 1. Ig of nitric acid was added to 120ml of deionised water. Separately, to 4.65g of glacial acetic acid was added 20ml of titanium isopropoxide . The two mixtures were added together and stirred at 8O⁰C for 8 hours. After the 8h period the contents were filtered through a cellulose nitrate membrane syringe filter (0.45μm). The filtered solution was then placed in an autoclave at 220°C for 12 hours. When removed from the autoclave the material was sonicated for 3 minutes until a milky white liquid was obtained. Most of the water was then removed under vacuum to leave a thick white paste. Carbowax was then added to the paste under stirring overnight. The carbowax, when removed, left a porous ceramic matrix. The deposition of the titania films was on to the glass substrate was carried out using the ^doctor-blade' method. The film (~60μm thick), was allowed to dry at room temperature for Ih after which a colourless film approximately lOμm thick was obtained. Finally, after drying in air, the film was annealed at 45O⁰C for 30 minutes .

The films were then coated in a soaked of Reichardt' s dye in acetone and the solvent was allowed to evaporate. This process was repeated several times until the Tiθ₂ film was a uniform green colour.

The results of subjecting the films to various vapours are detailed below (Fig. 11 and Table 11) . The spectra show the response of the sensor in a number of solvents, including methanol, ethanol, isopropanol (IPA), dichloromethane (DCM), acetone and water.

Table 11 Wavelength shifts of a Reichardt' s dye/TiO₂ sensor element exposed to a number of common organic vapours .

Example 5 : 'Spectral Nose' ; a combination of sensor elements prepared by encapsulating solvatochromic dyes in ethyl cellulose - exposed to a number of organic vapours

The following data show the response of a number of sensor elements prepared using various solvatochromic dyes impregnated in an ethyl cellulose matrix.

A typical dye/polymer solution was prepared as in example 1, with varying amounts of dye used, depending on the solubility and/or absorbtion coefficient of the particular dye.

Initially, the UV spectra of the films were measured under a purge of flowing, dry air as shown in Fig. 12. The λ_max of the film after equilibration in the air is detailed in Table 11. All tests were carried out in a purge of dry air for at least 5 minutes before being exposed to organic vapour for 2 minutes and then allowed to recover for 2 minutes. This time is substantially longer than required^' as dynamic (temporal) measurements show that the response time is within seconds. However, to ensure complete equilibration the films were always exposed for this time. This allowed comparisons to be made with slower reacting films in terms of the magnitude of response (Δλ_max, ΔA) , rather than speed of response.

When the film was exposed to various organic vapours, the wavelength, hence colour, shifts were easily identifiable . Table 12 of ethyl cellulose-based sensor elements,

The sensor elements were then subjected to a number of carefully chosen organic vapours. The vapours were chosen according to their properties in terms of α, β and π* , as shown in Table 13. For the study, 6 solvents (2 from predominantly each class, α, β, and π*) were chosen. The basis of choice for each of the solvents is that they are predominantly from one class, and show as little behaviour associated with the other two classes as possible. For example, in the HBA class solvents the β value should be significantly higher than either α or π 21

Table 13. List of CC, β and π* for the organic test vapours.

Table 14 Shows the wavelength (λ_π,_ax) of the sensor elements in the chosen vapours, and the wavelength shifts (Δλ) , in parenthesis. The arrows indicate the shift direction, i.e. <— denotes a hypsochromic (blue) shift and -> denotes a bathochromic (red) shift.

As it can be seen from the above data, any vapour will have a characteristic fingerprint, determined by the combination of spectral shifts of the chosen sensor elements . This is illustrated by the histograms shown below (Fig. 13), in which the data from 4 sensor elements are chosen from Table 14. These sensor elements consist of dyes whch cover at least one each of the three types of solvatochromic probe type, i.e. α, β and π* : Reichardt (α,π*), 1-Ethyl (α) , 2,6-D (α, π*), and Copper (β) . The plots indicate the wavelength shifts of each of the 4 sensor elements in the 6 test vapours; a positive shift value on the histograms indicates a bathochromic (red) shift and a negative value indicates a hypsochromic (blue) shift.

From the plots it is clear that the chosen 4 sensors define a sensor array and the response of the array to each of the 6 vapours is unique. Each vapour has a characteristic pattern, or fingerprint, that defines that vapour, and only that vapour. The wavelength shift of each sensor element is due to the particular chemical interaction the vapour has with the solvatochromic dye (encapsulated in the sensor) and can be quantified using the relationship between the wavelength (or v_max) shift and the solvatochromic property the dye predominantly probes, i.e. α, β or π* . For example, in the case of the copper dye (Copper), the relationship is:

β = 0.358. (18.76-v_max)

Therefore, in any Copper sensor element a value for β can be extracted. To extend this idea, in the case of any vapour, the value α, β or π* can be extracted and each value, being characteristic of any vapour, defines a ^spectral coordinate' which, when collated define a vapour fingerprint. Other, more complex information, such as temporal and absorbance can also be extracted to reinforce the wavelength shift data.

Depending on the simplicity or complexity of the vapour, or vapour mixture, being analysed more or less sensor elements could be employed in any array. The wealth of information obtained and the flexibility underline the value of Spectral Nose and why it is a very- successful, yet simple, vapour detection system. In particular, the valuable information obtained from each sensor element (shift, temporal and absorbance data) , used in combination with similar information gained in the other sensor elements in the array, separates Spectral Nose from other devices of a similar type.

References

1. Pendick D., Earth Mag., 1991, Feb, 25.

2. Gate J. and Klusty M., Anal. Chem. , 1991, 63, 1719. 3. Thundat W. T., Chen G., Warmack R., Allison D. and

Wachter E., Anal. Chem., 1995, 67, 519.

4. Stetter J., Jurs P. and Rose S., Anal. Chem., 1986, 58, 860.

5. Freund M. and Lewis N., Proc. Natl. Acad. Sci. U.S.A., 1995, 92, 2652.

6. Despagne F. and Massart D. L., Analyst, 1998, 123, 157R.

7. Chen S., Zieve L. and Mahadevan V., J. Lab. Clin. Med., 1970, 75(4), 628. 8. Phillips M., Cataneo R. N., Cummin A. R. C,

Gagliardi A. J., Gleeson K., Greenberg J., Maxfield R. A., Rom W. N., Chest, 2003, 123(6), 2115.

9. Sulway M. J. and Malins J. M., Lancet, 1970, 1, 736.

10. Rooth G. and ϋstenson S., Lancet, 1966, Nov, 1102. 11. Gardner J. W. and Bartlett P. N., Electronic Noses:

Principles and Applications, Oxford University

Press, Oxford, 1999. 12. Dickinson T. A., White J., Kauer J. S. and Walt D.

R., Nature, 1996, 382, 697. 13. White J., Kauer J. S., Dickinson T. A., and Walt D.

R., Anal. Chem., 1996, 68, 2191.

14. Rakow N. A., Suslick K. S., Nature, 2000, 406, 710.

15. Wolfbeis (Ed.), Fiber Optic Chemical Sensors and Biosensors, VoIs. I and II, CRC Press, Boca Rafton, FL, 1991.

16. Mills A. and Chang Q., Anal. Chim. Acta., 1994, 285, 113.

17. Mills A. and Thomas M., Analyst, 1997, 122, 63. 18. Suppan P. and Ghoneim N., Solvatochromism, The Royal Society of Chemistry, 1997.

19. Reichardt C, Chem. Rev., 1994, 94, 2319.

20. Kamlet M. J. and Taft R. W., J. Am. Chem. Soc. , 1976, 98, 377.

21. Marcus Y., Chem. Soc. Rev., 1993, 409.

22. Sadaoka Y., Matsugushi M., Sakai Y. and Muranta Y- u., Chem. Lett., 1992, 53.

23. McGiIl R. A., Paley M. S. and Harris J. M., Macromolecules, 1992, 25, 3015.

24. Krech J. H., Rose-Pehrsson S. L., Anal. Chim. Acta., 1997, 341, 53.

25. Sadaoka Y., Sakai Y. and Muranta Y-u., Talanta, 1992, 39(12), 1675. 26. Migron Y. and Marcus Y., J. Phys . Org. Chem., 1991, 4, 310.

27. Fukuda Y. and Sone K., Bull. Chem. Soc. Jpn. , 1972, 45. 465.

28. Zupan Z. and Gasteiger J., Neural Networks for Chemists, VCH, New York, 1993.

29. Beebe K. R., Pell R. J., and Seasholtz M. B., Chemometries, Wiley Interscience, New York, 1998.

30. Johnson S. R., Sutter J. M., Engelhardt H. L., Jurs P. C, White J., Kauer J. S., Dickinson T. A., Walt D. R., Anal. Chem., 1997, 69(22), 4641.

Claims

1. A method of identifying a vapour of a volatile compound comprising: collecting a vapour of a volatile compound; spectroscopically analysing the vapour and obtaining a vapour fingerprint of the vapour using a sensor array comprising at least two solvatochromic dyes in a host medium; and performing a pattern recognition analysis on the vapour fingerprint to thereby identify the vapour.

2. A method according to claim 1, wherein the response of sensor elements to organic or inorganic vapours is analysed using an artificial neural network such that spectroscopic information such as absorbance, wavelength and related temporal information is collected in order to identify a vapour fingerprint (i.e. a molecular signature) of an associated vapour.

3. A method according to any of claims 1 or 2, wherein the method is capable of analysing qualitatively and quantitatively both single and multi-component vapours .

4. A method according to any preceding claim, wherein in a first step of the method of identifying the vapour, the vapour is collected using a sniffing device.

5. A method according to claim 4, wherein the sniffing device collects a finite sample of vapour and allows it to come into contact with the sensors.

6. A method according to any preceding claim, wherein the sensor array has rapid response and a broad range of detection and high sensitivity such as less than 1000 ppm, less than 500 ppm or less than 100 ppm.

7. A method according to claim 6, wherein the sensor array has a detection range of 1100 nm to 350 nm.

8. A method according to any preceding claim, wherein the solvatochromic dyes are compounds that have UV/vis/NIR absorption spectra that may be changed in position, intensity and shape when placed in different solvents .

9. A method according to any preceding claim, wherein the solvatochromic dye is that of the N-phenolate betaine dyes .

10. A method according to any preceding claim, wherein the solvatochromic dye is represented by compound (I) shown below - 2, 6-Diphenyl-4- (2, 4 , β-triphenylpyridinio) - phenolate .

R = H

R = -C(CH₃I₃

R = -CF₃

11. A method according to any preceding claim, wherein any combination of the following dyes are used: Reichardt's Dyes, 4-nitro-aniline, 4-nitro-N N- diethylaniline, 4-carbomethoxyl-l-ethylpyridinium iodide, 4-nitrophenol, 3-nitro-N N-diethylaniline, Nile red, tetramethylethylenediaminoacetylacetonato, 4- nitroanisole, bis (1, 10-phenanthroline) -dicyano-iron (II), Copper (II) acetylacetonate, 4-nitro-l-ethylbenzene, 2,6- dichloro-4- (2, 4, β-triphenyl-1-pyridinio) phenolate, Effenberger ' s dye, Michler's ketone, Phenol blue, MGH-I, and 4-aminobenzophenone .

12. A method according to any preceding claim, wherein at least one of the solvatochromic dyes is 4-Nitro- A/diethylaniline .

13. A method according to any preceding claim, wherein at least one of the solvatochromic dyes is tetramethylethylenediaminoacetylacetonato-copper (II) perchlorate.

14. A method according to any preceding claim, wherein the solvatochromic dyes are in a solid support which undergo a noticeable/detectable colour change when exposed to solvent vapour.

15. A method according go any preceding claim, wherein the sensor array is exposed to a number of common inorganic or organic vapours selected from any one of or any combination of the following: aromatics; alkenes; alkynes; alkanes; alcohols; ketones; aldehydes; ethers; alogenated solvents; nitriles and carboxylic acids; pyridine; xylene; benzene; toluene; tetrahydrofuran; dioxane; ethyl acetate; ethane; ethylene; ethane; pentane; hexane; cyclohexane; methanol; ethanol; isopropanol; butanol; acetone; butanone; methyl ethyl ketone; formaldehyde; diethyl ether; carbon tetrachloride; chloroform; dichloromethane; dimethyl sulphoxide; acetonitrile; carbon disulphide; acetic acid and water.

16. A method according to any preceding claim, wherein spectral responses such as changes in wavelength, absorbance, spectral shits are recorded on exposure to the at least two solvatochromic dyes.

17. A method according to any preceding claim, wherein the vapour is generated by bubbling dry air through a solvent .

18. A method according to any preceding claim, wherein the position and intensity of the spectra of the at least two solvatochromic dyes undergoes a substantial change when the sensor array is exposed to various vapours.

19. A method according to any preceding claim, wherein the various vapours form a vapour fingerprint thereby allowing a pattern recognition analysis to be conducted.

20. A method according to any preceding claim, wherein a neuro network analysis is conducted to obtain as an output Kamiet-Taft parameters thereby obtaining Kamlet- Taft parameters determining a set of spectral coordinates for the vapour.

21. A method according to claim 20, wherein the neuro network analysis is carried out using parameters obtained by determining the spectral changes and associated Kamlet-Taft parameters associated with a set of predetermined calibration vapour samples such as methanol, ethanol, isopropanol (IPA) , dichloromethane (DCM), acetone and water.

22. A method according to any preceding claim, wherein computer software is used to perform a pattern recognition analysis on the obtained spectral data wherein a neuro network is used against a database of calibration samples to identify the vapour.

23. A method according to claim 19, wherein the pattern recognition approaches include: hierarchical cluster analysis, principal component analysis, K-nearest neighbour and soft independent modelling.

24. A method according to any preceding claim, wherein the host medium of the at least two solvatrochromic dyes is a polymer matrix prepared by dissolving an appropriate amount of polymer in an appropriate organic solvent.

25. A method according to claim 24, wherein the polymer matrix is selected from any of the following: ethyl cellulose; polyvinylchloride; hydroxethyl cellulose; nitrocellulose; polyethylene oxide; polyethylene glycol; polystyrene; biopol; polyvinyl alcohol; polymethyl methacrylate; polyvinyl pyrollidone; poly (2, 6-dimethyl- 1 , 4-phenylene oxide); poly [2- (4-benzoyl-3-hydroxy- phenoxy) ethyl acrylate; poly (acrylonitrile-co-butadiene- co-styrene) ; polymethacrylsaeuremethylester, poly (methyl methacrylate) / poly (ethyl methacrylate-co-methyl acrylate) ; poly (ethylene-co-acrylic acid); poly ( tetrafluoroethylene-co-vinylidene fluoride-co- propylene; poly (vinyl butyral-co-vinyl alcohol co-vinyl acetate); and cellulose butyrate acetate.

26. A method according to claim 24, wherein the polymer is any combination of the following: ethyl cellulose; polyvinylchloride; hydroxethyl cellulose; nitrocellulose; polyethylene oxide; polyethylene glycol; polystyrene; biopol; polyvinyl alcohol; polymethylmethacrylate; polyvinylpyrollidone; poly (2, β-dimethyl-1-, 4-phenylene oxide) ; poly [2- ( 4-benzoyl-3-hydroxy-phenoxy) ethyl acrylate]; poly (acrylonitrile-co-butadiene-co-styrene) ; polymethacrylsaeuremethylester poly (methyl methacrylate); pσly(ethyl methacrylate-co-methyl acrylate); poly (ethylene-co-acrylic acid); poly (tetrafluoroethylene- co-vinylidene fluoride-co-propylene; poly (vinyl butyral- co-vinyl alcohol co-vinyl acetate) and cellulose butyrate acetate.

27. A method according to any preceding claim, wherein the host medium is formed from any form of silica including that of fumed silica, or aerogel formed from silica-based sol-gels.

28. A method according to any preceding claim, wherein the host medium is formed from' titania based material such as nanocrystalline titania thin films.

29. A method according to any preceding claim, wherein a quantitative analysis is conducted to determine the quantity of each type of vapour present.

30. A method according to any preceding claim, wherein the method is used for any of the following: analysing packaging materials; pharmaceuticals; perfumery; petrochemicals; food and beverages; household products; raw materials; monitoring indoor air quality; monitoring gaseous effluents from industrial waste heaps; the detection of explosives and biological agents; drugs and the diagnosis of diseases.

31. A method according to claim 30, wherein the diseases which can be identified include any of the following: cirrhosis of the liver, lung cancer and diabetes.

32. Apparatus for identifying a vapour of a volatile compound comprising: a sensor array including sensor elements comprising at least two solvatochromic dyes in a host medium wherein the sensor array is capable of interacting with a vapour; a spectroscopic device for analysing the vapour by utilising the interaction between the vapour and the sensor array to obtain a vapour fingerprint; and means for performing a pattern recognition analysis of the vapour fingerprint and thereby identifying the vapour .

33. Apparatus according to claim 32, wherein a colour change is due to a change in the polarity of the environment of the dye which is brought about by the vapour.

34. Apparatus according to any of claims 32 or 33, wherein the apparatus comprises a device for collecting a vapour .

35. Apparatus according to any of claims 32 to 34, wherein a device is used for collecting the vapour.

36. Apparatus according to any of claims 32 to 35, wherein the apparatus comprises a sampling chamber for containing the vapour.

37. Apparatus according to any of claims 32 to 36, wherein the sensor array comprises any suitable number of solvatochromic dyes more than 2.

38. Apparatus according to any of claims 32 to 37, wherein the sensor array comprises 2 - 10, 3 - 4 or 3 sensors .

39. Apparatus according to any of claims 32 to 38, wherein the array is enclosed in a sealed sample chamber

(where the vapour is held) such that the sensor elements are in intimate contact with the test vapour.

40. Apparatus according to any of claims 32 to 39, wherein the means for performing the pattern recognition comprises computer software for comparing the obtained vapour fingerprint with a database of calibration samples.

41. Apparatus according to claim 40, wherein the database of calibration samples is prepared from any of the following: methanol, ethanol, isopropanol (IPA) , dichloromethane (DCN) , acetone and water.

42. Apparatus according to any of claims 32 to 41, wherein the apparatus is portable, allowing point-of-care analysis to be conducted such as in a general practitioner's surgery.

43. Apparatus according to any of claims 32 to 42, wherein the vapour is collected and then transported to a remote site for spectral analysis.

44. Apparatus according to any of claims 32 to 42, wherein the apparatus is non-portable and comprises all necessary components such as a spectrophotometer and a control/data processing P. C. and the necessary pattern recognition software.

45. Apparatus according to any of claims 32 to 44, wherein the apparatus comprises: a light source; a photo detector; spectral filters, a dedicated chip with vapour profiles; comparison algorithms; and vapour profile storage facilities; a fibre optic splitter; and a device to split/recombine as necessary various light beams into components for each sensor element.

46. Use of the method according to any of claims 1 to 31 in analysis of vapours in any of the following: packaging materials; pharmaceuticals; perfumery; petrochemicals; food and beverages; household products; raw materials; monitoring indoor air quality and gaseous effluent from industrial waste heaps; the detection of explosives, biological agents and drugs; and the diagnosis of diseases .

47. Use according to claim 46, wherein the diagnosis of diseases includes cirrhosis of the liver, lung cancer and diabetes .

48. A sensor array including sensor elements comprising at least two solvatochromic dyes in a host medium wherein the sensor array is capable of interacting with a vapour and identifying a vapour using a vapour fingerprint analysis .