WO1999032878A1

WO1999032878A1 - Liquid crystal sensors

Info

Publication number: WO1999032878A1
Application number: PCT/GB1998/003841
Authority: WO
Inventors: Paul Nicholas; Tony Mccollum
Original assignee: Paul Nicholas; Tony Mccollum
Priority date: 1997-12-19
Filing date: 1998-12-18
Publication date: 1999-07-01
Also published as: GB9726923D0; EP1047930A1; GB2332511A; AU1770399A; CA2314366A1

Abstract

A device for sensing one or more analyte molecules, comprising: a membrane (1) whose two surfaces are rendered electrically conducting; a liquid crystal material or a liquid crystal blend impregnated within the membrane; a light source; a light sensor; a light guide (8, 9) between the second surface of the membrane (1) and the light sensor, whereby the light sensor senses light from the light source which has been transmitted through the membrane (1) via the light guides (8, 9).

Description

LIQUID CRYSTAL SENSORS

The present invention relates to a device for sensing and identifying one or more analyte molecules using optical properties of liquid crystals. It is known that liquid crystal membranes can have a high degree of specificity towards molecules permeating through them. This selective behaviour or permselectivity can be controlled so as to enhance the transmission of species of interest. Such systems exploit the spontaneous or induced molecular ordering exhibited by liquid crystal molecules which encourages anisotropic molecular interactions. Differences in the steric (size and shape) and dipole-dipole (solubility) interactions between different permeants and the liquid crystal are thus enhanced, altering membrane selectivity. The present invention further exploits these interactions to detect the presence of permeant species by monitoring the dynamic optical properties of liquid crystals.

Liquid Crystal, LC, sensors are known which rely on measuring a colour or refractive index change to detect the species of interest. Such systems can be based around optical fibres to convey light to and from a detecting region at the end of the fibre. They tend to be specific to one compound and so arrays of different sensors are required to sense more than one species.

EP 0441120 (Yeda Research and Development Co. Ltd.) is directed towards a biosensor for qualitative and quantitative analysis comprising (1) a reference electrode at its upper part, (2) a recording electrode at its bottom, and (3) an amphiphilic liquid crystalline membrane composed of a lipid bilayer doped with synthetic or biological ion channels. The sensor works by the opening of ion channels activated by the interaction of the ion channel or ion channel site with a chemical entity. This results in an increased conductance which, if calibrated appropriately, can be related to the concentration of the desired species. However, the system has to be respectively calibrated for each separate antibody.

Devices are known which comprise thin, ceramic membranes which may or may not be surface treated and are impregnated with electroactive liquid crystals. However, these membranes have been used to measure electrical and optical (not dynamic optical) parameters and the dielectric characteristics of the liquid crystal. Various studies have been undertaken to measure dielectric properties and liquid crystal NMR spectra which measure relatively rapid (MHz range) liquid crystal molecular motions directly. This type of configuration has not been used as a chemical sensor.

It is an object of the present invention to provide a device which is capable of sensing the presence of and identifying a plurality of molecular species without the need to resort to an array of sensors, each specific to one analyte, or having to recalibrate or in any way modify the system.

The invention consists of a device for sensing one or more analyte molecules, comprising: a membrane whose two surfaces are rendered electrically conducting; a liquid crystal material or a liquid crystal blend impregnated within the membrane; a light source; a light sensor; a light guide between the light source and a first surface of the membrane; and a light guide between the second surface of the membrane and the light sensor, whereby the light sensor senses light from the light source which has been transmitted through the membrane via the light guides. Any one or more analyte molecule causes a distinctive and detectable change in the transmitted optical beam.

Thus, one advantage of the present system is that a number of molecules can be sensed simultaneously by the same sensor without separate calibration and measurement being necessary, whereas, the sensors of the prior art are specific to one species and therefore for complete identification of a mixture of compounds, an array of sensors is necessary. This would be both expensive and time consuming.

The use of the optical properties of the liquid crystal in response to an applied field allows a number of different molecules to be identified in one sensor. The optical response may be converted into an electrical signal which is split into 3 channels. The combination of high and low frequency response is characteristic of the analyte molecule being sensed and the DC offset is dependent upon concentration. For more than one analyte, the DC offset is a composite response, i.e. it is a signal representative of a combined concentration. In a preferred embodiment, the membrane is constructed of a material containing uniform channels connecting both surfaces. Preferably the membrane is constructed of a ceramic material; each surface is sputter coated with a thin layer of conducting material to render it electrically conducting; and the surface of the membrane is treated with a solution of alkyl acid prior to the addition of the liquid crystal. Preferably the conducting material is gold, another metal or a conducting polymer such as polypyrrole.

Preferably, the liquid crystal in this embodiment is K15 or CB5, supplied by Merck UK Ltd. The interaction between the liquid crystal and the alkyl acid on the surface of the membrane causes the liquid crystal molecules to lie parallel to the surface of the membrane (across the channels of the membrane) resulting in low optical transmission. When a voltage is applied, the membrane becomes more transparent therefore a periodic waveform about 0 V will produce a corresponding transmission signal of varying intensity.

In a second embodiment, the membrane is constructed of a ceramic material or material containing uniform channels connecting both surfaces; each surface is sputter coated with a thin layer of conducting material to render it electrically conducting; and the membrane is impregnated with a blend of liquid crystals. Again, the conducting material is preferably gold, another metal or a conducting polymer such as polypyrrole.

Preferably, in this embodiment, the blend of liquid crystals is a combination of K15 and 8 % by weight of CE5. The liquid crystal molecules in the blend naturally adopt an helical orientation induced by the presence of chiral CE5 molecules. Therefore the molecular long axes are tilted with respect to the channel long axis and again the result is low optical transmission. As before, a periodic waveform about 0 V will produce a similar transmission signal of varying intensity as in the first embodiment.

The invention also extends to a process for the sensing of one or more molecules comprising the application of a periodic waveform about 0 V across a device as described above; and simultaneous interrogation of the optical transmission of the device containing the analyte, the transmission output signal being converted into an electrical signal.

Preferably the electrical signal is split into three channels. The first channel being the composite optical response. The second channel is a low frequency response with a cut-off frequency of at most 200 Hz, more preferably 150 Hz and most preferably 100 Hz. The third channel is a high frequency response with a cut-off frequency of at least 800 Hz, more preferably 900 Hz and most preferably 1 kHz. As explained above, the low and high frequency responses taken together are characteristic of the solvent molecule being sensed and the DC offset response is dependent upon concentration.

The invention further extends to a process for the sensing of one or more molecules on the channel surface of the membrane. Preferably, the sensing comprises observation of the dynamics of the liquid crystal. This identification of molecules on the channel surface of the membrane is in addition to the detection and identification of molecules which have dissolved in the liquid crystal. The surface treatment of the membranes described above uses solutions of straight chain organic acids such as hexanoic, decanoic acid etc. In some cases, these may not bond sufficiently strongly to the membrane surface and may be dislodged causing sensor failure. As an alternative, the carbonyl chloride analogues of these compounds, such as decanoyl chloride, could be used. These should chemically react with the surface and remain there. As with the straight chain organic acids, a surface array of molecules is produced, but the chemical reaction of the carbonyl chloride end group binds the compound to the surface of the membrane.

The liquid crystals referred to above are all thermotropic materials, i.e. single component, pure compounds which exhibit liquid crystalline properties (these may be blended together in some cases). Sensors based on so called lyotropic liquid crystals are also possible. These are based on a compound dissolved in a solvent. Individually, neither exhibit liquid crystalline phases. However, when mixed together in the correct proportions liquid crystalline behaviour can be seen. These systems may be important since lyotropic systems can be water based. A soap solution is an example, as is the correct concentration of strands of DNA in water. Water based systems can be used for the detection of biological molecules and other water soluble compounds.

While a few of the applications of the liquid crystal sensing device have been described above and will be explained in more detail below, the use of it is not limited to these applications. Further use and development of the sensor may be made in the following areas:

(i) altering surface treatment to enhance sensitivity for specific analyte molecules e.g. chiral specific analysis. (ϋ) liquid crystals bonded directly to the surface.

(ϋi) liquid crystal blends.

(iv) blends of liquid crystals and non-liquid crystal materials. The addition of dopants to the sensor can modify and, in some cases, enhance performance. In addition to coating the surface of the channels, surfactant also dispenses throughout the liquid crystal and modifies its behaviour. Consequently, asymmetrical optical responses are generated on the application of symmetrical voltage waveforms, such as sine waves. Such asymmetry may improve the ability of the sensor to discriminate between different molecules. This effect has been shown by using low concentrations of surfactant solution in the preparation of the sensors and/or by washing the sensor to remove unbound material. Under these conditions, unbound surfactant is reduced or eliminated and the optical behaviour is then symmetrical. Discrimination between different compounds may then be impaired or prevented completely. Deliberate controlled doping of the sensors with low concentrations of materials such as methylene blue induces asymmetrical optical responses. Such behaviour may improve optical discrimination, (v) the use of plane and rotationally polarised light to enhance the system sensitivity to liquid crystal molecular orientation and, thus to the presence of analytes.

(vi) the use of fluorescent surface coatings and/or liquid crystals to yield information concerning the chemical interaction between molecules, (vii) using reflected light to investigate surface interactions; using interferometric and holographic measurements to measure bulk and surface liquid crystal dynamics. (vϋi) the deposition of liquid crystal monolayers on channel surfaces to enhance detection sensitivity. The use of extremely small volumes or thin layers (including mono-layers) in porous media or on impervious surfaces may give rise both to rapid response times and high sensitivity, (ix) dielectric measurements in addition to or instead of optical interrogation. The analysis of both optical and electrical properties could provide more comprehensive analysis of molecules and mixtures, (x) two dimensional mapping of the sample to yield kinetic data assisting identification of the analyte molecules, (xi) the use of membrane stacks to separate molecules as they migrate down the stack as well as detecting their presence in each membrane element.

(xϋ) alternative methods of molecular alignment e.g. magnetic field, temperature change, mechanical forces as well as the applied voltage described in detail in this document.

The data obtained from the sensors may be interpreted by signal processing. The periodic waveforms generated by the sensor represent several constituents in the analyte mixture, interactions between the constituents, effects due to the instrumentation such as random noise, and changing environmental conditions that affect the signal baseline.

However, in combination with this complex of data, there will be a number of independent variations present in the signal, the largest of which represent the analyte mix. To extract meaningful information from the sensor waveforms, they are preprocessed and subjected to statistical analysis as described below.

The preprocessing procedure is used to reduce random variations in the signal and may consist of one or more of the following:

1. low pass filtering or autocorrelation - noise in the original signal can be reduced either by low pass filtering or autocorrelation. Low pass filtering smooths the data using local averaging over time. Autocorrelation reduces noise by summing subsequent instances over the entire period of the waveform. Because the signal is repetitive, the underlying signal is enhanced, while random noise tends to cancel out.

2. DC offset correction or extraction of first derivative -the repetitive waveform is superimposed on a low frequency or DC component. This can be extracted out by subtraction of a constant value, or by high pass filtering. The first derivative of the signal represents the slope at every point of the waveform, and is unaffected by low frequency or constant components.

3. Scale normalisation or variance scaling - as the concentration of the analyte changes, the peak-to-peak range of the waveforms vary. To compensate for this, the dynamic range of the signal can be normalised to a constant interval. Variance scaling is calculated by dividing the signal at each point by the standard deviation of the overall data. 4. Mean centring - for subsequent statistical analysis it is useful to subtract the mean value from the signal.

The characteristics of the signal can be extracted by a range of statistical and analytical methods including use of: mean, median and mode values; moments; variance; number of peaks; normal, F and T distribution; chi square distribution; auto power density spectrum; cross-power density spectrum; autocorrelation; fourier analysis; and principal component decomposition.

The data generated by this analysis are classified by examining their distribution and labelling them as due to an appropriate analyte. The method involves an off-line training phase where the system is exposed to a number of samples. During actual use, the incoming data is compared to the training set. There are several appropriate pattern classification, or cluster analysis methods, including adaptive vector quantisation, K-means clustering and principle value decomposition. The last of these will be explained in more detail here to explain how the information that corresponds to the analyte is extracted.

The waveforms generated by the sensor are digitised and represented by a vector of elements x. The mean value of this vector is x. The data is then transformed as follows:

(y-∑) = A(^χ-^χ) where the rows of A are the eigenvectors of Cx such that

Cx = (l/n)(x-x)(x-x)^τ Cx is the covariance matrix of the variables x, centralised about their means.

An important property of the transformation is that the terms (y-χ) have associated eigenvalues that are also the value of their variance in the transformed domain. The linear combination with the highest eigenvalue is therefore the linear combination with the highest variance. The principal component has the largest possible variance of any possible linear function. The second component has the largest variance under the condition that it is uncorrelated with the first. The third component will have the smallest variance subject to the constraint that it is uncorrelated with the others. Because the terms (y-y) are uncorrelated, they can be separated. The transformed waveform can be built up as a linear combination of the components.

The principal components can be used for classification of the analyte. The process starts with training phase, carried out off-line as a one-time procedure. A set of waveforms corresponding to known analytes is generated, and a matrix A is produced for this set. Each row of A (the eigenvectors) can be considered as a breakdown of the variation in the waveforms. The eigenvectors with the lowest eigenvalues contribute least, and tend to represent random effects such as noise, and can be discarded. Only the rows (the eigenvectors) of A that correspond to the principal components need to be retained, reducing the computational demands. The resultant A matrix represents the space of the training set. For each known analyte, its distribution in this space is calculated, by projecting the waveform on to the matrix, and a weighting value is computed that represents this distribution.

For analyte classification, the data is projected onto the A matrix and a weighting value is calculated. The new weight is compared to the set of weights generated during the training stage. If the difference between the weights is less than a tolerance value, the analyte is classified according to the minimum of the differences.

If any of the elements in the training set corresponds to the unknown analyte waveform, a close match will be found. If the analyte is contaminated, for example, the fit may be out of range to qualify as an acceptable match, but the method will still be able to classify the signal according to the training sample it is closest to.

Although a specific excitation/interrogation regime is described, there are many possible ways to enhance the data being recorded. These include: multifrequency modulation - a modulation signal consisting of a combination of several sinusoids of different frequencies (excitation with the same or different waveforms at several frequencies, either simultaneously or sequentially, yields optical transmission signals that, collectively, are unique for a given compound at a given concentration); and frequency sweep modulation - a modulation signal consisting of a linearly increasing modulation frequency. The application of voltage pulses may also be advantageous.

Excitation with a simple low frequency (1 to 20Hz) sine (or square or almost any periodic) waveform will yield an optical transmission signal that is unique for a given compound at a given concentration. The sensor can also be excited with a frequency swept, typically 1Hz to 20kHz sine (or any other periodic) waveform. The amplitude of the resulting response varies with the excitation frequency. The "envelope" of the optical response versus frequency is unique for a given compound at a given concentration.

In specific cases the applied frequency may have a direct influence on the bulk orientation of the liquid crystals. Some electroactive liquid crystals undergo a change in the sign of their dielectric anisotropy when subjected to electric fields exceeding a certain threshold frequency. This is observed as a change in the orientation of the molecular long axis from parallel to the electric field to perpendicular to it or visa versa. This is known as two frequency switching and is sometimes exploited to reorientate liquid crystal molecules in display devices. The judicious selection of liquid crystals or liquid crystal blends could exploit this phenomenon for sensor and separation processes. Different liquid crystal components in a blend could be switched independently so highlighting specific analyte interactions.

The data handling may also be extended to permit multichannel, multifrequency analysis rather than just low and high frequency channels used currently. Data processing/analysis will almost certainly involve techniques such as Fast Fourier Transform (FFT) to measure frequency and power response spectra.

To date, interrogation of the sensor has been accomplished using relatively simple optical techniques. Apart from making direct electrical measurements of dielectric properties there is also the possibility that techniques such as nuclear magnetic resonance (NMR), electron spin resonance (ESR) and infrared, UV, visible and raman spectroscopy could also be used to monitor liquid crystal dynamics.

The phase relationships between current and voltage during excitation of the sensor can be measured by connecting a resistor (typically lOOkΩ to 1MΩ) in series with the sensor. The voltage drop across the resistor is proportional to the current flowing through the sensor. A plot of current against applied voltage reveals changes in phase relationships which are unique for a given compound at a given concentration. This is caused by changes in the dielectric properties of the sensor. Taking this a step further, direct measurements of sensor dielectric permittivity at several frequencies during exposure to various analytes have been made. The measurements were made using a dielectric analyser manufactured by Polymer Laboratories Ltd. Results show that the biggest changes occur at the lowest frequencies (this agrees with optical transmission behaviour). The observed changes at different frequencies are unique for a given compound. This behaviour is also seen for Anodiscs (Anopore ceramic membranes) with no surface pre-treatment which are impregnated with liquid crystal. Unlike the optical measurements, pre-orientation of the liquid crystal parallel to the membrane surfaces (i.e. across the Anodisc channels) is not necessary.

The membrane itself could also take many forms. To date, the only matrix used to contain the liquid crystal has been an Anopore ceramic membrane made from alumina (duminium oxide). Other materials with different microstructures may also be suitable. For example, the manufacture of polydisperse liquid crystal (PDLC) films could provide an alternative means by which small droplets of liquid crystal could be suspended in a matrix; in this case a polymeric matrix. Thin layers of silica (silicon dioxide), other oxides, nitrides or carbides impregnated with liquid crystal may also perform the same function. The microstructure of these layers can be finely controlled for optimum results and are readily formed on silicon wafers.

A well defined layer or layers of liquid crystal on a suitable substrate could also be manipulated by electric/magnetic fields and operate as a sensor. It has been shown that changes in dielectric permittivity on exposure to analytes can be detected in a range of porous media impregnated with a range of liquid crystals either singly or blended together as well as on Anopore ceramic membranes. Systems as simple as a disc of impregnated filter paper behave as sensors when exposed to analytes. Furthermore, by simply coating rather than impregnating an untreated Anopore ceramic membrane (Anodisc) with liquid crystal, sensing behaviour is also apparent as a change in dielectric properties. Anodiscs are coated by dissolving the liquid crystal in a solvent such as chloroform and then immersing the Anodisc in the solution. The chloroform is removed by evaporation leaving a thin film of liquid crystal on the channel surfaces of the Anodisc. The same procedure could be used for any porous medium or for coating a non porous surface. This technique can be further refined to produce mono layers of liquid crystal. This is normally accomplished by using a Langmuir Trough which creates molecular layers under controlled conditions. The method and apparatus are also useful in the detection of protein binding which is extremely relevant to biotechnology and medical applications. Untreated Anodiscs have been impregnated with K15 liquid crystal and then exposed to aqueous solutions of one of a complimentary pair of proteins. For clarity, they will be identified as proteins A and B. A and B will normally bind together in aqueous solution. After exposure to protein A or B, the treated systems (system A or B) were dried and each then exposed to aqueous solutions of A or B. In theory, system A should respond to solutions of B but not solutions of A and visa versa. By measuring dielectric properties it could be seen that there were dramatic differences in dielectric permittivity between system A exposed to A and system A exposed to B. Similarly, system B exposed to B was very different to system B exposed to A. These changes correlate with protein binding and form the basis of very simple, cheap devices to detect biological molecules with very high specificity. The proteins used were biotin labelled mouse IgG2a immunogen and streptavidin. A further application of the present invention is in the area of gas chromatography. By connecting a simple sensor (i.e. as described in the present application) to the output of a gas chromatograph, the elution of each analyte from the chromatographic column is detected as a change in optical transmission through the sensor. The analytes are carried over the surface of the sensor by the carrier gas. The sensor is behaving purely passively; with millisecond response times; no voltage is applied at all. This is the "DC" behaviour described in the present application where a shift in overall transmission is seen as well as a change in dynamic electro-optic transmission on exposure to an analyte. The DC offset is proportional to analyte concentration.

In addition to the interrogation methods referred to above and described in detail below, there are a number of further means of determining analytes using liquid crystal sensor. A few of these are mentioned here:

(a) X-ray of gamma ray diffraction and/or scattering. Changes in the behaviour/alignment of the liquid crystal molecules could be detected by changes in diffraction/scattering patterns.

(b) Electron, neutron or other subatomic particle diffraction and/or scattering.

(c) Light absoφtion by evanescent wave propagation rather than by transmission. An example of this would be ATR (attenuated total reflectance) as used for infrared (IR) absorption measurements (although this is not restricted to IR) on samples which are strongly absorbing. Instead of attempting to transmit, for example IR radiation through the sample, this is held against a slab of suitable IR transparent crystalline material. The LR radiation is then introduced at one end of the slab and is contained within it by internal reflection, emerging at the far end where it is detected as normal. However, some radiation interacts with the sample at the sample-crystal interface producing an absoφtion spectrum. Thus the sample's IR spectrum can be measured without the transmission of radiation.

(d) Fluorescence and/or phosphorescence measurements where light, e.g. ultra-violet or even X-rays stimulate light emission. Light is not transmitted but emission of light is induced. (e) Microwave absoφtion/excitation: measurement of microwave spectral changes.

(f) Measurement of electrical properties such as resistance, capacitance, impedance, inductance and dielectric strength.

(g) Heat flow through the liquid crystalline system, determined by the thermal conductivity. This will change depending on the macroscopic orientation of the liquid crystals.

(h) Transmission of sound or vibration (especially ultrasonic frequencies) will be dependent upon the liquid crystalline orientation and macroscopic structure. This could be used in the deterrnination of analytes. Ultrasound could additionally be used to enhance mixing between the liquid crystals and the analyte. Of course, any combination of the methods of determination mentioned above can be employed for either interrogation or excitation.

The invention may be put into practice in various ways and two specific embodiments will be described by way of example to illustrate the invention with reference to the accompanying drawings, in which:

Figure 1 is a schematic block diagram of the excitation and interrogation regime used for the membrane sensors; Figure 2 is the chemical structure of the liquid crystal identified as K15 or CB5 as supplied by Merck UK Ltd.;

Figure 3 is the chemical structure of the liquid crystal identified as CE5 as supplied by Merck UK Ltd.;

Figure 4 is an enlarged, expanded cross-section of the membrane holder; Figure 5 is a typical excitation waveform applied to the membrane sensor;

Figure 6 is an example of the high and low frequency response raw signal output for 10μl toluene in air;

Figure 7 is an example of the high and low frequency response raw signal output for 10μl benzene in air; Figure 8 is an example of the high and low frequency response raw signal output for a mixture of 5μl toluene and 5μl benzene in air;

Figure 9 shows the raw optical response and the applied low frequency (about 2 Hz) sine wave for a membrane which has been surface treated with acetic acid;

Figure 10 shows the raw optical response and the applied low frequency (about 2 Hz) sine wave for a membrane which has been surface treated with hexanoic acid;

Figure 11 shows the raw optical response and the applied low frequency (about 2 Hz) sine wave for a membrane which has been surface treated with decanoic acid;

Figure 12 shows the raw optical response and the applied low frequency (about 2 Hz) sine wave for a membrane which has been surface treated with octadecanoic (stearic) acid;

Figure 13 is a Fourier Transform of the raw response of figure 9;

Figure 14 is a Fourier Transform of the raw response of figure 10;

Figure 15 is a Fourier Transform of the raw response of figure 11;

Figure 16 is a Fourier Transform of the raw response of figure 12; Figures 17 (a) and (b) are schematic representations of an alternative sensor of the present invention;

Figures 18 (a), (b) and (c) are schematic representations of alternative interrogation techniques of the sensor shown in figure 17; and

Figure 19 is a schematic representation of liquid crystals in the sensor before and after the application of an electric field.

The system, as indicated in figure 1, consists of an Anopore ceramic membrane 1 which is illurninated on one side by a stabilised light source. The membrane has a liquid crystal impregnated within it and on the application of a periodic waveform about 0 V a transmission signal of varying intensity is measured. The transmitted light is detected by a photodiode 2 and converted into an electrical signal which is amplified and split into 3 channels A, B and C. The identity of the analyte molecules can be determined from the change in transmission signal after the analyte is added.

The actual sensor comprises an Anopore (Trade Mark) ceramic membrane 1, for example sold by Whatman UK. In this example the membrane 1 is approximately 26 mm in diameter and 60 microns thick. The membrane 1 has a sieve like structure with well defined channels peφendicular to and connecting its two faces. The channel diameters are 0.2 microns in this specific example although other Anopore membranes are available with 0.1 and 0.02 micron channel diameters.

The two faces of the membrane 1 are rendered conducting by sputter coating a layer of gold (or any other conducting material, e.g. a metal, polypyrrole) onto the surfaces. This is followed by a chemical treatment to modify the surface of the channels and involves immersion of the sputter coated membrane 1 in a solution of an alkyl acid such as hexanoic, decanoic or stearic (octodecanoic) acid. The preferred alkyl acid is stearic acid because the long hydrocarbon chain length decouples the liquid crystals from the surface most effectively. The acid group, which is at one end of the hydrocarbon chain, becomes anchored to the channel surface and the chain extends peφendicularly from the surface into the channel volume. Excess solvent is removed by pressing the membrane 1 between two pieces of filter paper and any residual solvent is removed in a vacuum oven.

The membrane 1 is prepared by placing it on a piece of filter paper on a hot plate. The entire exposed membrane surface is then covered with a thin film of a nematic electroactive liquid crystal and left for a time (typically of the order of 24 hours). This ensures complete impregnation of the membrane 1.

The temperature of the hot plate is such that the liquid crystal is in its isotropic liquid phase which ensures minimum viscosity. For example, K15 (or CB5) is in its nematic liquid crystal phase between 25 and 35°C. Beyond 35°C it is an isotropic liquid and below 25°C it is a crystalline solid. CE5 is a crystalline solid below 18.7°C, exists in a chiral smectic-A phase between 18.7 and 42°C and is an isotropic liquid beyond 47.5°C.

The liquid crystal used in this case is identified as K15, which exhibits a nematic phase between 25 and 35°C as explained above, and is supplied by Merck UK Ltd. The full chemical structure is shown in figure 2. The chemical name for K15 or CB5 is 4-cyano-4'- pentyl biphenyl.

An alternative system can also be employed whereby the ceramic membrane 1 is sputter coated with a conducting material as before but it is not subjected to any surface treatment prior to impregnation. The liquid crystal used is a blend of K15 and 8% by weight of an optically active, chiral nematic (or cholesteric) liquid crystal identified as CE5 and supplied by Merck UK Ltd. The full chemical structure of CE5 is shown in figure 3. The chiral centre of the liquid crystal is indicated by the asterisk.

The use of the device as a sensor involves the application of voltages across the membrane 1 using the sputter deposited electrodes, and simultaneous interrogation of its optical transmission. A membrane holder 3 was devised in which the membrane 1 could be clamped, electrically connected and optically interrogated using optical fibres. One of the faces of the membrane 1 is exposed to the analyte permitting its diffusion into the system as shown in figure 4. This configuration prevents any current leakage when used in aqueous media.

Figure 4 shows an enlarged, expanded cross-section of the membrane holder 3. The holder 3 consists of two components, one each on either side of the membrane 1. One component 4 is electrically insulating and the other component 5 is made of metal to form an electrical contact with one side of the membrane 1. The electrical contact on the other side is made through an insulated electrical contact 6. As stated above, the Anopore porous ceramic membrane 1 is sputter coated with a metal e.g. gold on both sides.

The metal component 5 has holes 7 to allow the sample analyte liquid or gas access to one side of the coated membrane 1. Two sets of optical fibres 8, 9 are aligned on either side of the surfaces of the coated membrane 1. One pair of optical fibres 8 are aligned to pass through the sample cavity and they are referred to as the "sample" optical fibres. The second pair of optical fibres 9 are "reference" optical fibres used to detect small differences between the exposed "sample" area of the membrane and the unexposed regions. One of each pair of optical fibres is connected to light emitting diodes, LEDs, to direct infrared and red light to the membrane. The light transmitted through the membrane is captured by the second optical fibre of each pair on the opposite side of the holder. The transmitted light is detected by a photodiode connected to an oscilloscope and computer data capture interface. This configuration of optical fibres minimises stray light interference and ensures a fixed geometry, thus optimising the signal to noise ratio of the captured data.

The two components 4, 5 are joined together around the coated membrane 1 by means of fixing screws 10 threaded into fixing holes 11. The membrane is sealed within the holder with the aid of a peripheral sealing ring 12. The membrane holder 3 is then placed in the system as shown in figure 1.

Prior to the application of a voltage across the surface treated K15 membrane, the liquid crystal molecules lie parallel to the membrane surfaces (across the channels) and optical transmission is relatively low. A sufficiently high electric field will reorientate the liquid crystal molecules peφendicular to the membrane surfaces increasing light transmission. Thus a periodic zero crossing waveform will induce a corresponding periodic optical transmission signal of varying intensity.

The liquid crystal molecules in the untreated K15/CE5 impregnated membrane of the second embodiment naturally adopt an helical orientation induced by the presence of chiral CE5 molecules. Thus, the molecular long axes are tilted with respect to the channel long axis and the application of a sufficiently high electric field reorients the liquid crystals so that they lie parallel with the channel axis. A periodic zero crossing waveform will induce a corresponding optical response similar to that seen for the treated K15 membrane.

The electrical stimulus applied across the membrane 1 is an amplitude modulated signal of approximately 400 V p-p (peak to peak), symmetrical about 0 V as shown in figure 5. The modulation is sinusoidal with a frequency in the range 1 kHz to 15 kHz. An unusual feature of the carrier signal is that it is much lower in frequency than the modulation, and is a triangle waveform in the range 0.5 Hz to 10 Hz. A modulation level of approximately 25% was used experimentally.

The optical fibres are used to couple the input and output light to and from the membrane. The optical interrogation beam is supplied by an 800 nm LED driven from a stabilised power source. A photodiode 2 is used to sense the transmitted output, converting it into an electrical signal that is fed to a balanced preamplifier as shown in figure 1. The output of the preamplifier is split into three channels A, B, and C. The first channel A is the composite optical response, consisting of an overall DC offset upon which the response to the modulation signal is imposed. The second channel B is fed to an electrical low pass filter with cut-off frequency of 100 Hz, the output of which indicates the response of the membrane to the low-frequency triangle carrier signal. The third channel C is fed to a high pass filter with cut-off frequency of 1 kHz, the output of which indicates the response of the membrane to the modulation frequency. The signal is then demodulated with a precision RMS detector.

In uncontaminated air, the overall DC offset is constant for a given temperature. The low-pass signal shows a complex but periodic form at the same frequency as the modulation carrier. The detected high pass signal gives two peaks, corresponding to the rising and falling gradients of the triangular carrier.

With the introduction of the analyte, the "sample" DC offset will change, indicating an increase in light transmission through the membrane. The complex form of the low pass signal also changes, as do the number of peaks detected from the high pass signal. The precise nature of the changes in the observed waveforms are specific to the analyte concerned and its concentration.

As stated above, the "reference" optical fibre can be used to detect small differences between the exposed "sample" area of the membrane and the unexposed regions. However, over time, analyte may diffuse into the reference region and the holder will have to be reset.

An example of the high and low frequency response raw signal output for 10μl toluene in air is shown in figure 6. This was carried out at 38°C and using a mixture of K15 and 0.2μm decanoic acid. The low frequency response is indicated by LI and the high frequency response by HI . A further example of the high and low frequency response raw signal output for 10μl benzene in air is shown in figure 7. Again, this was carried out at 38°C and using a mixture of K15 and 0.2μm decanoic acid. The low frequency response is indicated by L2 and the high frequency response by H2.

As stated in the object of this invention, it is possible to detect more than one analyte species at one time and an example of this multiple detection is included in figure 8. This shows the high and low frequency response raw signal output for a mixture of 5μl toluene and 5μl benzene. The conditions are as for the previous 2 examples. The low frequency response is indicated by L3 and the high frequency response by H3. These 2 compounds are difficult to distinguish using simple discrete sensors or even on more advanced devices; preseparation of these compounds is normally required.

As mentioned above, it is a further aspect of the present invention that the dynamics of the liquid crystal can be exploited to identify molecules on the channel surface of the membranes. This is in addition to detecting and identifying molecules which are dissolved in the liquid crystal. Figures 9 to 12 show the raw optical response for a variety of surface treated membranes.

The membranes were surface treated with alkyl acids having different chain lengths - acetic acid (2 carbons), hexanoic acid (6 carbons), decanoic acid (10 carbons) and octadecanoic or stearic acid (18 carbons) for the membranes in figures 9 to 12 respectively. The liquid crystal employed is K15 and the electrical stimulus applied is a low frequency sine wave which is not modulated.

Figures 13 to 16 show the Fourier Transforms of the raw responses of figures 9 to 12 respectively. It is clear from both the raw response and the corresponding Fourier Transform that the behaviour of the liquid crystal under the influence of the applied voltage is highly sensitive to the surface layer. This sensitivity can be used to detect and identify molecules on the surface of the membrane, in addition to the molecules dissolved in the liquid crystal.

In addition to the above-mentioned sensor device using a porous membrane, it is also possible to use a sensor bearing some resemblance to conventional liquid crystal displays. The active liquid crystal is sandwiched between two films; these can be glass, polymer or even metal (if light transmission is unimportant). The sample to be analysed is introduced either by diffusion through one or both of the films or by capillary action/forced flow between the films, so making direct contact with the liquid crystal. Figures 17 (a) and (b) show schematically examples of these sensors. In figure 17(a), the sample to be analysed diffuses through the film(s) and in figure 17 (b), the sample is introduced by capillary action or forced flow.

Electrodes on the inside surfaces can be used to excite/interrogate the liquid crystal and optical properties can also be monitored. The optical interrogation can be accomplished in three ways as shown in figures 18(a), (b) and (c). Optical transmission can be measured across the device by simply positioning a light source on one side and a detector on the other (via optical fibres if required) (see figure 18(a). The barriers themselves can act as light guides (see figure 18(b)); the amount of light coupled through the system is then dependent on liquid crystal behaviour/orientation. Lastly, optical fibres can be introduced directly into the device as shown in figure 18(c).

As for the Anopore membranes, preferred orientations can be induced by treatments applied to the inside surfaces of the containment barriers. If the barriers are made of a polymer, or have a polymer coating, this can be vigorously rubbed in one direction to induce bulk liquid crystalline alignment parallel with the barrier surface. A special case of surface induced parallel alignment is realised when the rubbing directions for the two inside surfaces are peφendicular. Thus, the alignment of the nematic phase liquid crystal molecules at one surface is at right angles to that at the opposite surface, as shown in figure 19. The orientation in the bulk is parallel to the two surfaces but describes a 90° helical twist on going from one surface to the other. The same technique is used in twisted nematic displays which can rotate plane polarised light. When an electric field is applied, the helical structure is replaced with a simple nematic order where the liquid crystal molecules are now oriented 90° to the containment barriers and do not rotate plane polarised light.

By inducing particular alignments as described above, a passive response can be observed in a similar way to that seen with surface treated Anopore membranes. The introduction of analyte molecules interferes with intermolecular liquid crystal forces, changing light transmission or the intensity of plane polarised light in the case of a twisted nematic configuration. Application of an excitation voltage waveform will produce analyte specific optical behaviours as seen for the Anopore systems.

Claims

1. A device for sensing one or more analyte molecules using liquid crystal material.

2. A device as claimed in Claim 1 comprising: a liquid crystal material or a liquid crystal blend; an energy transmission source; and a receiver arranged to receive energy of a similar characteristic to that transmitted, whereby the receiver detects energy which has been transmitted through the liquid crystal material.

3. A device as claimed in claim 2, comprising: a membrane whose two surfaces are rendered electrically conducting; a liquid crystal material or a liquid crystal blend impregnated within the membrane; a light source; a light sensor; a light guide between the light source and a first surface of the membrane; and a light guide between the second surface of the membrane and the light sensor, whereby the light sensor senses light from the light source which has been transmitted through the membrane via the light guides.

4. A device as claimed in Claim 3, in which the membrane in constructed of a material containing uniform channels connecting both surfaces.

5. A device as claimed in Claim 3 or Claim 4, in which the membrane is constructed of a ceramic material.

6. A device as claimed in any of claims 3 to 5, in which each surface is sputter coated with a layer of conducting material to render it electrically conducting.

7. A device as claimed in Claim 6, in which the conducting material is gold, another metal, a conducting polymer such as polypyrrole, or a semi-conducting material such as n or p doped silicon.

A device as claimed in any of claims 3 to 7, in which the surface of the membrane is treated with a solution of alkyl acid prior to the addition of the liquid crystal.

9. A device as claimed in any of claims 3 to 8, in which the liquid crystal is Kl 5 or CB5.

10. A device as claimed in any of Claims 3 to 7, in which the membrane is impregnated with a blend of liquid crystals.

11. A device as claimed in Claim 10, in which the blend of liquid crystals is a combination of K15 and 8 % by weight CE5.

12. A device as claimed in any of claims 3 to 8, in which the liquid crystal is doped.

13. A device as claimed in Claim 12, in which the liquid crystal is doped with methylene blue.

14. A device as claimed in any of claims 3 to 13, in which the light sensor is a photodiode.

15. A device as claimed in any of claims 3 to 14, including a plurality of light sources and/or a plurality of light sensors.

16. A device as claimed in claim 2, comprising: two films; an active liquid crystal sandwiched between the two films; means for exciting the liquid crystal; means for optical interrogation; and means for introducing the sample to be analysed to the liquid crystal, whereby the optical transmission through the liquid crystal is measured.

17. A device as claimed in claim 2, comprising: a membrane whose two surfaces are rendered electrically conducting; a liquid crystal material or a liquid crystal blend impregnated within the membrane; an electromagnetic radiation source; an electromagnetic radiation sensor; an electromagnetic radiation guide between the electromagnetic radiation source and a first surface of the membrane; and an electromagnetic radiation guide between the second surface of the membrane and the electromagnetic radiation sensor, whereby the electromagnetic radiation sensor senses electromagnetic radiation from the electromagnetic radiation source which has been transmitted through the membrane via the electromagnetic radiation guides.

18. A process for the sensing of one or more molecules which comprises applying a periodic waveform across a device as claimed in any preceding claim, simultaneously interrogating the transmission of the device containing the analyte, and converting the transmitted output into an electrical signal.

19. A process as claimed in Claim 18, in which the transmission is optical transmission.

20. A process as claimed in Claim 18 or Claim 19, in which the electrical signal is split into three channels.

21. A process as claimed in Claim 20, in which one channel records the low frequency response with a cut-off frequency of at most 200 Hz, preferably 150 Hz and most preferably 100 Hz.

22. A process as claimed in Claim 20 or Claim 21, in which one channel records the high frequency response with a cut-off frequency of at least 800 Hz, preferably 900 Hz and most preferably 1 kHz.

23. A process for the sensing of one or more molecules on the channel surface of the membrane.

24. A process as claimed in Claim 23 in which the sensing comprises observation of the dynamics of the liquid crystal.

25. A process for the sensing of one or more molecules which comprises applying a periodic waveform over a range of frequencies across a device as claimed in any of Claims 1 to 17.

26. A device and process for sensing one or more analyte molecules constructed and arranged substantially as herein specifically described with respect to and as shown in figures 1 and 3 or figures 17(a), (b), 18(a), (b), (c) and 19 of the accompanying drawings.