GB2418491A - Device and method for analysing a sample plate - Google Patents

Abstract

The invention relates to a device and a method for analysing a sample plate on which at least two material samples are arranged. According to said method, an impedance spectrum is measured for each material sample, and a structure of a switching circuit equivalent comprising at least one electronic component is then determined according to the respectively measured impedance spectrum. Starting values for the components of the respective switching circuit equivalent are determined for an error minimising calculation. During the error minimising calculation, a theoretical impedance spectrum for at least one of the material samples is calculated on the basis of the impedance spectrum measured for the material sample, the starting values for the components of the corresponding switching circuit equivalent, and fit values for the components of the corresponding switching circuit equivalent. A validation variable is then determined for the calculated theoretical impedance spectrum, and an evaluation variable is determined by comparing at least one of the fit values for the components with reference value.

Description

Device and process for analysing a sample panel State of the Art The

invention relates to a device for analysing a sample panel, according to the type defined more precisely in the precharacterising clause of Claim 1, and also to a process for analysing a sample panel.

In particular in the field of the material sciences, chemistry and pharmacy, discovering and developing substances and materials that are optimized with respect to the particular application is of considerable importance.

A special field of application in this connection is constituted by sensor technology, which is an important key technology with a constantly growing number of applications both in the industrial field and in the private domain.

For instance, sensors are employed in technical process-

monitoring systems, in the field of environmental

protection, in the medical field or even in the automotive field. A considerable amount of development work is currently being devoted, in particular, to the development of fast-acting and highly sensitive sensors having little cross-sensitivity.

The developments hitherto have, as a rule, been limited to an optimization or modification of known materials.

However, there is a problem that for certain fields of application of sensor technology there is a high demand for new materials, which cannot be adequately met by conventional methods which are distinguished by production of individual sensors and subsequent sequential characterization.

In particular in the course of the development of novel sensitive materials or combinations of materials it can be

expedient to employ processes from the field of

combinatorial chemistry, or, to be more exact, so-called high-throughput methods. With these processes it is a question of processes for synthesis and screening, running in parallel, by virtue of which new materials or combinations of materials can be revealed, or processes already known for synthesizing existing materials can be

optimised within a broad field of parameters.

A general presentation of high-throughput processes is known from US 5, 985,356, the application of combinatorial chemistry, which is essentially known from the field of pharmacy, to chemical and material-science fields of application being proposed in this document in particular.

A device of the type mentioned in the introduction for the purpose of analysing a sample panel is known from DE 101 31 581 Al, for example. This device comprises a sample panel, on which there have been applied, in the manner of a matrix, 64 material samples which are each connected to two electrodes which in turn are provided with contact points with which a means for reversible and addressable contacting can be brought into contact.

Advantages of the Invention The device according to the invention for analysing a sample panel, having the features according to the precharacterising clause of Claim 1 and having a measuring head that is capable of being inserted into a housing support - which for the purpose of electrical connection to the contacting means comprises two measuring wires per material sample, which bear with preloading against contact areas of the sample panel and are connected to a measuring and evaluating unit - has the advantage of simple handling, since in the case where use is made of a standardized sample panel the contact between the material samples and the measuring and evaluating unit is capable of being established by simple insertion of the measuring head into the housing support.

The device according to the invention is suitable, in particular, for the development and discovery of materials and combinations of materials that are capable of being employed as sensor materials and capable of being characterized by way of their electrical properties. For example, the device according to the invention can be employed for the purpose of developing an optimised sensor material of a gas sensor.

With the device according to the invention it is possible to investigate a large number of potential sensor materials - which are arranged on the sample panel under various test gases at varying temperatures which may amount, for example, to up to 800 C - almost simultaneously. The investigation may be undertaken potentiometrically, resistively, capacitively, or even with the aid of complex impedance spectroscopy.

In the present case the term 'spectroscopy' is to be understood in such a way that it is a question of frequency-dependent measurements - i.e. the impedance of a sample is examined at varying measuring frequencies. For example, the individual material samples are each examined within a frequency range between 10 Hz and 107 Hz with a measured-data density of 15 measuring-points per decade.

At such a measured-data density this means that 180 items of measured data are ascertained per material sample. An extensive reduction of the data can be obtained, for example, by an adaptation of a suitable equivalent circuit to the measured data.

In order to keep the measuring wires, which in particular bear against the contact areas of the sample panel via fusible bulbs, under the preloading, the measuring wires may each be connected to a spring-loaded contact, in particular to a gold-plated spring-loaded contact, which guarantees a constant bearing pressure of the respective measuring wire on the respective contact area.

With a view to developing a gas sensor, in order to be able to expose the material samples to a certain test gas, the measuring head may be connected to a gas-supply unit.

In order to be able to subject the material samples to varying test-gas or reference-gas atmospheres, the gas- supply unit, which is expediently connected to a data- processing unit pertaining to the measuring and evaluating unit, comprises a gas-mixing apparatus. Furthermore, for the purpose of humidifying the test-gas or reference-gas atmosphere the gas-supply unit may comprise a water reservoir.

The measuring head may, in addition, be designed in such a way that it comprises, as an integrated component, a gas space which is arranged above the material samples of the sample panel and is preferably constituted by a substantially bell-shaped distributor apparatus. The gas space is connected to the gas-supply unit.

In order to obtain a homogeneous distribution of the test gas or reference gas in the gas space, in an advantageous embodiment of the device according to the invention a diffuser is arranged in the gas space.

In order, in the case of a plurality of material samples on the sample panel, to be able to gauge the individual material samples in straightforward manner, the measuring and evaluating unit advantageously comprises two relay plugboards which are connected to the measuring wires and, in the case of 64 material samples on the sample panel for example, preferably each exhibit a 3 x 64 matrix consisting of high-frequencysuitable relays. In this case the 64 material samples on the sample panel are capable of being gauged in one measuring cycle, with at least three measured quantities being accessible - to be specific, the impedance with the aid of an impedance analyser, for example, and the d.c. resistances of the material samples and the current/voltage characteristics thereof with the aid of appropriate further measuring instruments.

The measuring and evaluating unit is expediently equipped with measuring and control software which, on the one hand, controls the measuring procedure and, on the other hand, transfers acquired measured data to an appropriate file or even to a relational database which can be read in from evaluating software.

The evaluating software preferably operates in such a manner that it comprises a fit functionality for the purpose of computing theoretical impedance spectra in respect of the individual samples, the computation preferably being undertaken by taking as a basis an equivalent circuit that comprises at least one virtual or real electronic component. A virtual component is, for example, a constant-phase element (CPE). On the basis of the measured data the fit functionality accordingly computes for an equivalent circuit - consisting, for example, of a serial RC element a theoretical impedance spectrum that is optimally approximated to a measured impedance spectrum, whereby for the purpose of adapting the theoretical spectrum to the measured spectrum a variation of the capacitance and/or of the resistance of the components of the RC element is implemented. Accordingly, in this case a resistance value and a capacitance value for the RC element are assigned to the optimally adapted theoretical impedance spectrum. The sensitivity of the respective material sample can be inferred from the resistance value, for example by comparison with a reference value.

In order, in the case of a plurality of material samples which are being gauged under various measurement conditions, to make the output values acquired in respect of the individual material samples, representing the sensitivities for example, accessible to a simple evaluation, the evaluating software advantageously comprises a data-mining functionality. The data-mining functionality ascertains the optimal material sample for the respective application by numerical methods, for example by applying preferably multidimensional target functions.

Alternatively, or in addition, the data-mining functionality may also comprise a visualization functionality. In this case a user is able to ascertain the optimal material sample for the respective application with the assistance of a display screen. For example, the visualization functionality operates with a colour spectrum, a high sensitivity of the material sample being assigned to a certain colour, and a low sensitivity being assigned to a different colour.

In order also to be able to discover sensor materials that are suitable for applications in which - such as in the case of an exhaust gas, for example - high temperatures prevail, the device according to the invention preferably exhibits a heating arrangement into which the sample panel is preferably capable of being plunged. In this case the device exhibits, in particular, a high-temperature reactor which is delimited by the heating arrangement and in which the sample panel comprising the material samples can be exposed to varying test gases or reference gases.

The invention also provides a process for analysing a sample panel, having the features according to Claim 19.

By application of this process it is possible to gauge electrically a sample panel on which there are arranged a large number of material samples, for example 64 material samples, under varying conditions and to select, in fully automatic manner, a material sample that is best suited for the application in question. This is effected by automatic selection of starting values for the components of the respective equivalent circuit and for the subsequent error- minimising computation. The starting values are employed in the error- minimising computation, whereby a theoretical impedance spectrum adapted to the impedance spectrum measured in the given case in respect of the material sample in question is computed under the respective measurement conditions, proceeding from the starting values. By virtue of this procedure an extensive reduction of the data is possible, since, starting from the impedance spectra which have a plurality of measuring-points, only a few fit values representing derived quantities are ascertained that describe the individual components of the equivalent circuit. The fit values accordingly represent dimensional designs of the components of the equivalent circuit with which the measured impedance spectrum can be optimally simulated.

In the course of the determination of the equivalent circuit, a serial connection of four RC elements, for example, may be selected, the higher RC elements being optionally set to a starting value that has no influence on the computation of the theoretical impedance spectrum.

The starting values of an RC element that are required for the errorminimising computation are preferably computed from the maximally measured imaginary impedance Z"_MAX and from the corresponding measuring frequency f_Z"_MAX in accordance with the following formulae: R1_START = - 2 Z"_ MAX Cl START= - 27r f _ Z"_ MAX - R_ START If the equivalent circuit comprises several RC elements, the starting values for the error-minimising computation that are required for the higher RC elements are preferably ascertained from the difference spectra between the measured data and data that are computed or simulated on the basis of the starting values computed for the first RC element.

In particular, the selection of so-called 'igood" starting values which lie close to the actual quantities of the components of the equivalent circuit crucially shortens the duration of the following error-minimising computation; "poor" starting values, on the other hand, may have the consequence that the error-minimising computation carried out on the basis of the starting values provides meaningless values.

In a preferred embodiment of the process according to the invention the equivalent circuit for simulating impedance spectra on the basis of fit values acquired in the course of an error-minimising computation consists of a serial connection of four RC elements. By ascertaining starting values for the individual components, it is then determined on the basis of a threshold value how many RC elements are taken into account in the simulation computations. The threshold value is preferably a value that is capable of being preset by a user. The percentage ratio of the resistance of the first RC element to the resistance of the current RC element n is examined in accordance with the formula: RCn_ START > valuer%] RC 1_START The quantity "value" is the quantity that is capable of being changed by the user, which is preset to 10%, for example. If the argument is not satisfied, the starting values for the components of the RC element in question are set to values that have no influence on a simulation of the impedance spectrum. These values are kept constant in the course of the error-minimising computation.

Furthermore, in the course of the process the validation quantity is determined that evaluates the concordance between the computed, theoretical impedance spectrum and the measured impedance spectrum assigned in the given case.

The evaluation quantity represents the output value that is relevant for the respective analysis and that reproduces a sensitivity of the respective material sample, for example in the course of the ascertainment of a sensor material for a gas sensor. The sensitivity is a measure of the quality of a sensor.

The sensitivity of a resistive gas sensor may be defined in various ways. If, for example, the direction of the change in the resistances in the course of an exposure to gas is taken into account, a sensitivity S can be expressed as the quotient of the resistance R_TEST under a test-gas atmosphere and of the resistance R_O under reference conditions, as follows: S R TEST; for oxidising gases S = OR TEST; for reducing gases.

The algebraic sign of the sensitivity S provides information as regards the change in resistance.

Alternatively, the sensitivity may be described as a change in the resistances, to be specific in accordance with the formulae:

R TEST- R O

S A=- - - ; for oxldlslng gases

- R_ TEST

+ R_O- R_TEST f d i In these formulae, sensitivities S_A between -1 and 1 arise, i.e. normalized sensitivities.

It is also possible to convert the sensitivities of the two definitions into one another. The sensitivity expressed by way of the value S is particularly meaningful in the event of a large change in the resistance of a material sample as a consequence of an exposure to a test gas. The sensitivity expressed by way of S_A is particularly meaningful if the resistance of a material sample changes only slightly as a consequence of an exposure to a test gas. The sensitivity expressed by S_A, however, has the advantage of a large tolerance in relation to inaccuracies of measurement, a precise illumination of a region of small changes of resistance and hence a better evaluation of cross-sensitivities, and also the possibility of automated visualization and data processing.

In a particularly simple embodiment of the process according to the invention the equivalent circuit consists of a virtual arrangement of real electronic components such as capacitors and resistors. In its simplest version the equivalent circuit consists of one resistor.

By virtue of the error-minimising computation, in the course of which an adaptation of the quantities of the components to the measured data is undertaken, a simulated impedance spectrum is ascertained in respect of the equivalent circuit for the respective material sample.

From the adapted quantities of the components it is

possible to draw conclusions as to the electrical

properties of the processes in the material sample that are described by the equivalent circuit. If the electrical behaviour of the material sample is determined by several processes with variable relaxation-times, it is necessary to make use of more complicated equivalent circuits which consist, for example, of several series-connected RC elements. If the processes react variably to a variation of the measurement conditions, an assignment of individual processes to components or component groups can be performed, in order in this way to be able to analyse the individual processes separately. By virtue of the error minimising computation, the volume of data for the purpose of describing the impedance measurement is reduced to the arrangement and the quantities of the components.

In the case of a sample panel with, for example, 64 material samples and in the case of a measurement under eleven different gas atmospheres at, in each instance, four different temperatures, the yield of measured data amounts to 2816 impedance spectra. The individual impedance spectra are simulated in accordance with the process according to the invention and are reduced in each instance to the derived measured quantities that are reproduced by the quantities of the components of the equivalent circuit.

The computation of a theoretical impedance spectrum of an equivalent circuit is undertaken in a preferred embodiment of the process according to the invention in such a manner that a complex admittance Y* and a phase shift of the individual components of the RC elements at a given angular frequency (20 measuring frequency) is determined in accordance with the following formulae: l Y*= ;(-R) +(C) =arctan(wRC) Y'= cos Y"= sin (p By means of a transformation, the respective impedances can be ascertained: Y' yt' Z = 2 + y, ,2 ' Z = y,2 + y, ,2 where Z' is the real part of the impedance and Z'' is the imaginary part of the impedance.

In the case of a serial connection of the RC elements, the impedances of the individual RC elements can be summed directly. A computation of the impedances in respect of frequencies that correspond to the measuring frequencies yields a data record corresponding to the measured data, so that an estimation of the error between the measured data record and the data record ascertained theoretically is possible.

In an advantageous version, the error-minimising computation that is carried out in the process according to the invention is carried out by variation of the quantities of the individual components by 1%. By analysis of the differences of the theoretically computed spectrum and the measured spectrum, an error can be determined. If the error diminishes after a variation, a renewed variation in respect of the same component of the equivalent circuit is carried out, and on the basis of this variation a theoretical impedance spectrum is computed with which a new error computation is carried out. If the error does not diminish, the variation that has been effected is cancelled, and the variation of the components is varied with reversed algebraic signs, or a different component is varied.

The error computation is preferably carried out in such a way that the function for determining the error takes account of the starting value of the resistance of the first RC element. If the value is greater than a nominal measuring resistance of the impedance analyser that is employed in the course of the measurement, for example greater than 3107 Q. the error is determined only from the imaginary part of the computed impedance. In order to be able to weight the high-frequency region of the ascertained spectrum, the ascertained deviations are preferably multiplied by the logarithm of the measuring frequency.

This weighting enables physically meaningless or even faulty measured data in the low-frequency region to be suppressed.

For the purpose of computing the validation quantity - i.e. for the purpose of validating the quality of the impedance spectrum computed theoretically in respect of the respective material sample - a corridor around the theoretical impedance spectrum is preferably ascertained which includes, for example, 90% of the measured data. For the purpose of minimizing the time required for computing the validation quantity, a successive approximation algorithm can be utilised, in which case the limiting values of the algorithm may be O Q and twice the value of the real part of the summed impedance at the lowest measuring frequency.

The process according to the invention is preferably employed for the purpose of analysing material samples under various test-gas atmospheres and, in particular, at various temperatures. In this case, impedance spectra for all the material samples arranged on the sample panel are measured under the various measurement conditions.

The spectra measured in the given case are then simulated by theoretical means in the process according to the invention, in each instance by means of an error-minimising computation taking an equivalent circuit as a basis. This results in a large set of evaluation quantities which represent the target quantities in the process.

In the process according to the invention the target quantities and evaluation quantities are preferably written to a database and evaluated by means of a data-mining functionality.

In the case of the database that serves for accepting data records and making them available, it is expediently a question of a relational database in which information is stored - ordered by interrelated subjects - in the form of tables. Relationships between the individual data records in the tables are established by so-called identifying keys.

The database may, for example, contain further properties of the material samples, such as conditions of synthesis of the educts thereof, the sample history thereof, and such like. These properties are linked to one another via tabular relationships.

The data mining may be carried out by means of an optionally multidimensional target function and/or by means of a visual data-mining functionality. The data mining carried out by means of a target function is a numerical process that is based on the individual evaluation quantities stored in the database, for example classified by measuring temperature and test gas. In this connection, the properties that are demanded of the material being sought are firstly defined. In the search for a sensor material for a gas sensor, for example, the gas in respect of which the sensor material is to be sensitive and the cross- sensitivities that might constitute interference may be specified. A requirement profile with respect to a fingerprint of the sensitivities accordingly results therefrom.

The visual data-mining functionality advantageously operates in such a manner that the evaluation quantities of the material samples of the sample panel are represented, classified by test gases and temperatures, for example.

When the device according to the invention and the process according to the invention are employed, a preferably fully automatic high-throughput impedance system is available with which new materials can be developed with a high sample throughput and a low expenditure in terms of time and cost. For instance, by using the system on two days, 64 varying material samples at four different temperatures and under eleven different test-gas atmospheres can be investigated and also evaluated with respect to their sensory properties.

In particular, the system can be employed for the general development of materials and especially for the development of sensors in the automotive field and in the field of safety technology.

The invention also provides a data-processing system with a dataprocessing program for implementing the process according to the invention.

Further advantages and advantageous configurations of the subject-matter according to the invention will become apparent from the description, the drawing and the claims.

Drawing An exemplary embodiment of the device according to the invention is represented in the drawing in schematically simplified manner and will be elucidated in detail in the

following description in connection with a process

according to the invention. Shown are: Figure 1: a measuring arrangement of a device according to the invention; Figure 2: a measuring apparatus pertaining to the device according to Figure 1; Figure 3: a measuring head pertaining to the measuring apparatus according to Figure 2; Figure 4: a gas-distribution apparatus pertaining to the measuring head according to Figure 3; Figure 5: a measuring sequence on the basis of a flow chart; Figure 6: a diagram that represents a temporal measuring sequence for material samples at various temperatures and under various gas atmospheres; Figure 7: a diagram in which a running-in behaviour of four differently surface-doped material samples is represented; Figure 8: a diagram in which a response-time behaviour of three different material samples during a pulse of test gas is represented; Figure 9: an equivalent circuit; Figure 10: a flow chart of an error-minimising computation; Figure 11: an exemplary curve of a sensitivity of a material sample; Figure 12: an exemplary fingerprint of sensitivities of a material sample with respect to various gas atmospheres; and Figure 13: a greatly schematised representation of a visual data-mining functionality.

Description of the Exemplary Embodiment

In Figures 1 to 4 a measuring arrangement of a device 10 for analysing a sample panel 12 is represented on which 64 material samples 13 are arranged. The device 10 comprises a high-temperature reactor 14, a gassupply unit 16 and a measuring and evaluating unit 18.

The high-temperature reactor 14, which is represented in greater detail in Figures 2 to 4 in particular, comprises a frame or housing 20 in which a vertically adjustable heating block 22 is arranged which is supported on a guide rod 23 and on a threaded rod 24 which is provided with a crank 25.

The heating block 22 - which consists, for example, of four heating panels delimiting a heating chamber, each with a heating power of 1100 W - is capable of being vertically adjusted in such a way that the panel support 28 and hence the sample panel 12 is capable of being plunged into the heating chamber.

In addition, the high-temperature reactor 14 comprises a measuring head 26 which is inserted into a measuring-head support 27 connected to the frame 20 and serves for bringing the 64 material samples 13 arranged on the sample panel 12 into contact with the measuring and evaluating unit 18. The sample panel 12 is arranged on a panel support 28 connected to the measuring-head support 27.

The measuring head 26, which is represented in Figure 3 in particular, comprises a baseplate 29 which is formed, for example, from a workable glass ceramic and serves for retaining 128 measuring wires 30A, 30B which each consist of platinum and are jacketed by an aluminium-oxide tube. At their lower ends the measuring wires 3OA, SOB each have a fusible bulb

31A, 31B which serves for contacting the respective measuring wire on a contact area of the sample panel 12. The sample panel 12 has two contact areas per material sample - i.e. in the present case a total of 128 contact areas - which each interact with one of the measuring wires 30A, SOB. At their upper ends in the region of the baseplate 29 the measuring wires 30A, 30B are each provided with a gold-plated spring-loaded contact 32A, 32B which, with fixed measuring head 26, guarantees a constant bearing pressure of the measuring wires 30A, SOB or, to be more exact, of the fusible bulb 31A, 31B on the respectively assigned contact areas of the sample panel 12.

The jackets of the measuring wires 30A, 30B manufactured from platinum are, in addition, fixed to retaining plates 33 and 34 which are fastened to a system of rods 35 and oriented parallel to the baseplate 29.

The spring-loaded contacts 32A, 32B are connected via lines 36A, 36B to SMB sockets 37A, 37B integrated within the measuring head 26, which in turn are connected to the measuring and evaluating unit 18 of the device 10 via shielded SMB lines 38A and 38B, respectively. The measuring head 26 has a total of 128 SMB sockets 37A, 37B, each of which is connected to a measuring wire 30A, 30B and to which an SMB line 38A, 38B leading to the measuring and evaluating unit 18 is respectively linked. But for the sake of clarity only two SMB sockets 37A, 37B and two SMB lines 38A, 38B and also the respectively assigned measuring wires 30A, 30B are represented in each instance in Figures 1 to 3.

The measuring head 26 has, in addition, a substantially bell-shaped gasdistribution apparatus 39, produced from quartz glass for example, which is represented in detail in Figure 4 and which is connected to the gassupply unit 18 via a gas-supply line 40 made of stainless steel. The bellshaped gas-distribution apparatus 39 delimits a gas space which is arranged above the 64 material samples 13 arranged on the sample panel 12.

The material samples 13 are formed from, for example, tin(IV) oxide SnO2 and exhibit varying dopants which, for example, are formed from lanthanoids. The material samples 13 are distributed on the sample panel 12 in the manner of a matrix, in eight rows and eight columns.

The gas-distribution apparatus 39 which is represented in detail in Figure 4 has in its gas space 41 in addition a diffuser insert 42 which is formed from a quartz sphere and exhibits a plurality of bores 43 which have a diameter of, in each instance, approximately 1 mm.

In order to guarantee a uniform escape of gas from the gas space 41, at its edges the bell-shaped gas-distribution apparatus 39 has spacers 43 which define a gap of 0.8 mm width between the gas-distribution apparatus 39 and the sample panel 12.

For the purpose of exposing the gas space 41 and hence the material samples 13 to varying test-gas atmospheres, the gas-supply unit 16 exhibits two gas-cylinder cabinets, not represented here in any detail, each with four gas cylinders which are each connected to a gas-flow regulator 44A, 44B, 44C, 44D, 44E, 44F, 44G and 44H, respectively, humid synthetic air being contained in one gas cylinder, hydrogen being contained in the second gas cylinder, methane being contained in the third gas cylinder, synthetic air being contained in the fourth gas cylinder, nitrogen dioxide being contained in the fifth gas cylinder, nitrogen monoxide being contained in the sixth gas cylinder, propene being contained in the seventh gas cylinder, and carbon monoxide being contained in the eighth gas cylinder. The capacities of the flow regulators 44A, 44B, 44C, 44D, 44E and 44F are each between O scam and scam. The capacities of the flow regulators 44G and 44H are each between O scam and 10 scam. Via appropriate control of the volumetric flow of the various gases, test gases of varying compositions can be fed by means of the battery of the eight gas-flow regulators 44A, 44B, 44C, 44D, 44E, 44F, 44G and 44H into a collecting line 45 which is connected to the gas-supply line 40.

In order to adjust a relative humidity of the respective test gas, a humid carrier gas, which consists of synthetic air for example, may be admixed to the test gas. The humidity of the carrier gas is adjusted by said gas being conveyed through a water reservoir 46. A measurement of the humidified carrier gas by means of a humidity sensor which is not represented here in any detail yields, for example, a relative humidity of approximately 90% at room temperature.

The measuring and evaluating unit 18 comprises two relay plugboards 50 and 51, to which in each instance 64 lines 38A and 38B, respectively, are linked which lead to the measuring head 26 of the high-temperature reactor 14. The relay plugboards 50 and 51 each form a 3 x 64 matrix consisting of high-frequency-suitable relays.

The relay plugboards 50 and 51 are connected to a measuring and evaluating computer 53 via a digital control line 52.

In addition, the relay plugboards 50 and 51 are connected via measuring lines 54 to an impedance analyser 64 and to a so-called source meter 55. These two measuring instruments are likewise connected to the measuring and evaluating computer 53 via the digital control line 52. The addressing of the impedance analyser 64 and of the source meter 55 is undertaken via the two relay plugboards 50 and 51.

The measuring and evaluating computer 53 is furthermore connected via a further digital control line 56 to a D/A- A/D converter 57 which via an analogue control line 58 is connected, on the one hand, to the heating arrangement 22 of the high-temperature reactor 14 and, on the other hand, to the gas-flow regulators 44A, 44B, 44C, 44D, 44E, 44F, 44G and 44H of the gas-supply unit or gas-mixing battery 16.

In the measuring and evaluating computer 53 there is stored modular measuring and control software which, via a script control system, enables a complete automation of measurements carried out by means of the device 10. By virtue of the modular structure of the measuring and control software, problem-free upgrading of the measuring system is possible.

D.C. resistances, V/I characteristics or voltages of the individual material samples 13 can be measured by means of the source meter 55. These values, and also the measured values ascertained by means of the impedance analyser 64, can be transferred via the digital control line 52 to the measuring and evaluating computer 53 which comprises a database for the measured data.

For the purpose of gauging the material samples 13 arranged on the sample panel 12 in a high-throughput mode or in accordance with a highthroughput process, a scripting language is employed, whereby via a script file a list of tasks which consist of key words and parameters is transferred to the software by means of which the script file is processed and which controls all the functions of the system. The script files guarantee a continuous examination of control parameters, so that measurements under incorrect measurement conditions are excluded by further processing of the script file being abandoned. The script file is not limited as regards its length.

A measuring sequence that is capable of being implemented by means of the device according to Figure 1 is represented in Figure 5 on the basis of a flow chart. For the purpose of analysing resistive-gas sensory properties, the material samples 13 of the sample panel 12 are characterized electrically, on the one hand under a reference-gas atmosphere, which, for example, is formed from synthetic air with a relative humidity of, for example, 45%, and under various test-gas atmospheres. By virtue of a modulation of the measuring temperature, information can be obtained concerning the influence of the respective operating temperature on the material samples that are suitable as sensor material and also concerning the activation energy of conductivity processes.

In order to guarantee comparability of measured data pertaining to several sample panels, the script file that is employed is constructed as a standard script which controls the complete high-throughput screening, the chronological sequence of which is described in the flow chart represented in Figure 5. In this connection, in a first process step M1 a measuring temperature T_MEAS of the sample panel 12 is adjusted. Measurements may, for example, be carried out within a temperature range between 400 C and 250 C in steps of 50 C. During the cooling phases, the sample panel 12 is exposed to a reference gas consisting of synthetic air with a relative humidity of 45.

In the event of a change in the temperature of the sample, a stable fundamental resistance of the material samples appears only after a certain time, so that a conditioning of the material samples in a step M2 is required. Changes of temperature of the material samples result in a metastable state of intrinsic defects which may represent oxygen vacancies, the establishment of thermodynamic equilibrium of which requires infinite time. In Figure 7 the conditioning behaviour or runningin behaviour in respect of four variably surface-doped In203 samples upon attaining a target temperature of 300 C is represented in exemplary manner. The resistances R of the samples approximate asymptotically with time to a limiting value which represents the so-called fundamental resistance or reference resistance. This limiting value is attained, in extrapolated manner, after approximately 90 minutes. For the purpose of guaranteeing a constant reference resistance, it is advantageous to choose a conditioning time of 120 minutes.

Subsequently the test gas that is required for the measurement is prepared in a process step M3 and introduced into the gas space 41 arranged above the sample panel 12.

The first test gas comprises, for example, hydrogen with a concentration of 25 ppm. Synthetic air serves as carrier gas. The stream of gas is adjusted to 100 scam with humid synthetic air.

In order to guarantee that the individual material samples have attained their fundamental resistance independently of their relative position on the sample panel, subsequently in a step M4 an initial-feed phase that was investigated previously for the test gas takes place. In Figure 8 the resistance curve is represented of three material samples situated on a diagonal of the sample panel during a pulse of test gas lasting 40 minutes, the pulse gas comprising propene with a concentration of 50 ppm. In the example represented in Figure 8 the basic material of the material samples is tin(IV) oxide SnO2, on account of its high sensitivity to hydrocarbons. Irrespective of the relative position on the sample panel, the resistance of the material samples falls to a constant value within approximately 6 minutes by reason of the pulse of test gas.

The response-time accordingly amounts to 6 minutes in each instance, the response behaviour being substantially independent of the position of the respective material sample on the sample panel. After the pulse of test gas has finished, the resistances approximate to the fundamental resistances asymptotically within approximately minutes. The response behaviour is accordingly substantially independent of the position of the material sample on the sample panel. The initial-feed time of each test gas or reference gas that is chosen for the script file amounts to approximately 15 minutes prior to a corresponding measurement.

In a subsequent process step M5 the measurement of impedance spectra in respect of the 64 material samples arranged on the sample panel takes place. In this connection, in the script file for the measurements of the impedance spectra the following are defined as parameters: amplitude of the measuring voltage [V] 0.1 starting frequency [Hz] 10 final frequency [Hz] 107 measuring-points per frequency decade 15 bias [V] 0 mode [HS: high-speed; NO: normal; AV: average] HS The measured data acquired are ascertained by means of the impedance analyser 64 and transferred to the measuring and evaluating computer 53 or, to be more exact, to the database stored therein.

Subsequent to this, the measurement can be carried out under a different test-gas atmosphere, in which case the steps Ma, M4 and M5 are also passed through again. For example, carbon monoxide with a concentration of 50 ppm, nitrogen monoxide with a concentration of 5 ppm, nitrogen dioxide with a concentration of 5 ppm or even propene with a concentration of 25 ppm may be employed as further test gases.

The entire screening can then be carried out by returning to step M1 at a different measuring temperature.

A measuring sequence determined by a standard script is visualised in Figure 6, measurements under reference conditions and during conditioning phases not represented in this diagram, for the sake of clarity. Measurements under reference conditions - i.e. under a reference-gas atmosphere - are always carried out prior to introducing a new test gas H2, CO, NO, NO2 or propene at the temperature T in question. A so-called test-gas initial-feed phase X and also a measuring phase Y are always present in each instance. Furthermore, the respective concentrations C of the test gas are represented in Figure 6.

The measuring and evaluating software may have a functionality with which the impedance spectrum measured in the given case is represented graphically on a monitor.

The measured data, which are preferably stored in ASCII format in the database or in the directories in the measuring and evaluating computer that are assigned to the sample panels, are consequently directly open to a visual inspection.

In addition, the measuring and evaluating software may be provided with a functionality that represents impedance spectra of the sample materials of a sample panel in positionally dependent manner in an image matrix. In this connection, raw data and/or derived data may be represented. The functionality may also comprise an evaluation window, into which data pertaining to a certain material sample may be transmitted. The functionality may also be combined with further imaging systems for measurement or evaluation. Image data from these systems may then likewise be read into the image matrix in positionally dependent manner, in order in this way to obtain further information about the samples.

After measurement of the impedance spectra for the individual material samples, a theoretical impedance spectrum is computed on the basis of an equivalent circuit in respect of each material sample. In the present case the equivalent circuit consists of a serial connection of four RC elements of the type represented in Figure 9. In this connection the theoretically computed impedance spectra are each optimally adapted to the correspondingly measured impedance spectrum, to be specific by the quantities or dimensional designs of the individual components of the equivalent circuit being varied.

The error-minimising computation that is carried out will be elucidated in the following with reference to the flow chart represented in Figure 10.

For the purpose of implementing the error-minimising computation it is necessary to determine starting values for the individual components of the equivalent circuit.

At the same time, it is necessary to determine the number of RC elements to be taken into account in the error- minimising computation. For these purposes, in a step S1 the maximal imaginary impedance Z"_MAX is ascertained from the measured data in respect of the material sample in question. In the course of the selection an assessment is also made as to whether in the case of the maximally measured imaginary impedance it is a question of a faulty measurement, this being undertaken by an examination of the imaginary impedances measured at adjacent measuring-points - i.e. by an examination of the local maximum.

Subsequently, in a step S2, starting values R1_START for the resistance and C1_START for the capacitance of the first RC element are computed on the basis of Z"_MAX and the corresponding measuring frequency in accordance with the formulae: R1_ START = -2 Z"_ MAX Cl START= - Or f_Z"_ MAX R_START where f_Z"_MAX is the measuring frequency at the maximally measured imaginary impedance.

Subsequently, starting values for the resistances and capacitances of the further three RC elements of the equivalent circuit are ascertained in a loop S3.

To this end, firstly in a step S4 a theoretical impedance spectrum is computed on the basis of the starting values R1_START and C1_START, and in a step S5 a difference spectrum between the theoretical impedance spectrum and the measured impedance spectrum is computed, and from this difference spectrum the maximum of the imaginary impedance Z"_MAX is ascertained once again. Proceeding from this maximum, in a step S6 the starting values Rn_START and Cn_START (n = 2 to 4) for the RC element being considered in the given case are then computed in accordance with the formulae elucidated in conjunction with step S2.

Subsequently, in a step S7 a threshold-value inspection for the computed starting resistance value Rn_START is carried out, by means of which it is determined whether the respective RC element is to be taken into account in the subsequent simulation computations. The threshold value is a value that is capable of being changed by the user. In the course of the comparison the percentage ratio of the resistance of the first RC element to the resistance of the current component is examined.

If the computed starting value Rn_START is greater than the threshold value, the RC element is taken into account, the program returns to step S3, and a computation of the starting values Rn_START and Cn_START for the next RC element is performed. If the argument is not satisfied, the quantities of the components of the higher RC elements are defined at values that have no influence on the simulation computation for an impedance spectrum. These values remain constant in the course of the following adaptation steps and amount to, for example, Rn_START = 1 and Cn_START = 10-1s. This definition is undertaken in a step S8.

The threshold-value query accordingly determines the number of RC elements that are taken into account in the following steps. If the threshold value is fallen short of, the program passes directly over to the error-minimising computation on the basis of the number of RC elements determined, for example on the basis of one RC element, in which case a loop S9 via all m RC members taken into account is implemented.

Within the loop S9 a further loop S10 run = 1 to 3 is implemented, in which connection it is a question of an empirical number of passes which are implemented for the purpose of increasing the precision of the errorminimising computation.

Within the loop S10, firstly in a step Sll the starting resistance value R1_START of the first RC element is drawn upon by way of variable VAR. The variation of the individual components in the error-minimising computation amounts to 1%, which is expressed in a value FAK = 0.01 in step Sll. The error is preset to 1099. The variation of the respective component is undertaken in a step S12 in accordance with the formula VAR = V.FAK. A computation of a theoretical impedance spectrum is then undertaken on the basis of the varied value in a step S13.

Subsequently an error is computed which is based on a comparison between the impedance spectrum computed for the adapted component quantities in question and the measured impedance spectrum. A function for determining the error is dependent on the starting value of the resistance of the first RC element. For the purpose of computing the error, in a step S14 a check is made as to whether the respective starting value is greater than the nominal measuring resistance of the impedance analyser being employed, for example greater than 3 x 107 Q. If this is the case, the error is determined in a step S15 only by inspection of the imaginary parts of the impedances in accordance with the following formula: error = |Z''_fit Z"_meas| log f whereby for the purpose of weighting the high-frequency region of the spectrum a multiplication by the logarithm of the measuring frequency is effected, as a result of which physically meaningless or faulty measured data in the low- frequency range are suppressed. Otherwise, in a step S16 an error inspection on the basis both of the real parts and of the imaginary parts of the impedances is undertaken in accordance with the formula: error = |Z''_fit - Z''_meas| + |Z'_fit Z'_meas| . In a step S17 a check is made as to whether the error has become smaller by virtue of the variation of the component in step S12. If this is the case, the component in question, for example the resistance of the RC element in question, is varied again by returning to step S12. If the error has not diminished, the variation undertaken in step S12 is cancelled in a step S18, and in a step S19 it is defined whether the algebraic sign of the variation is changed in a step S20 or whether in a step S21 and in a following step S22 the next component, in this case the capacitance C, is chosen as variable component. In this case a return to step S12 is then effected once again, the quantity pertaining to the component then again being varied for as long as the error diminishes.

The variation of the resistance and the variation of the capacitance is repeated twice by returning to step S10 for each RC element. Then, depending on the number of RC elements to be taken into account, an efold return to step S9 is effected.

If no further minimising of the error is possible by variation of the resistance and of the capacitance of the RC element, the values of the resistance and the capacitance yielding the smallest error are output as evaluation quantities of the error-minimising computation.

After the error-minimising computations taking place in steps S9 to S22 have been completed, in a step S23 a validation function is ascertained for the theoretically ascertained impedance spectrum, by means of which the quality of the theoretically computed spectrum can be estimated. In the course of the determination of the validation function, for the purpose of estimating a quality of the error-minimising computation a corridor around the computed impedance spectrum is ascertained that contains 90% of the measured data. The error ascertained in this way lies between O and 1, an error of 0 expressing an ideal concordance between the measurement and the simulation computation, and an error of 1 expressing no concordance between the measurement and the simulation computation.

In the case where the process is employed for the purpose of determining a sensor material of a gas sensor, sensitivities S_A are computed on the basis of the resistances of the equivalent circuits acquired in the course of the error-minimising computation. The sensitivities S_A acquired in such a way are normalised and lie between -1 and +1, as can be gathered from the diagram in Figure 11. Sensitivities for an assumed reference resistance of 100 Q with varying test resistance R_test are represented in Figure 11. The sensitivities S_A are stored in a database.

Subsequently, on the basis of the ascertained sensitivities S_A a numerical data-mining is carried out, in order to select an optimal gas sensor for a special application.

The data mining is carried out, in particular, by numerical methods, whereby the individual sensitivities S_A that are stored in the database, classified by measuring temperature and test gas, are accessed. Firstly, the properties that are demanded of the desired sensor material are defined.

In the simplest case, it is specified for which test gas the sensor is to be employed and which cross-sensitivities constitute interference. This results in a requirement profile with respect to the fingerprint of the sensitivities, as represented in Figure 12. The requirement profile is converted into so-called ">" or "<" requirements, by means of which data records having the desired properties are characterized. Further conditions, such as the type of the basic material of the sensor material or the surface doping thereof, may be included in the target function. In the case of the example in Figure 12, the requirement profile is configured in such a way that all the test gases A, B. C, E, apart from test gas D, have a sensitivity less than 0.2 and greater than -0.2 and test gas D has a sensitivity greater than 0.9.

The reference is specified by X in the given case. The requirement profile is converted by means of an evaluating functionality in the form of an SQL filter query and transferred directly to the database. The results of the SQL instruction may then be represented in tabular form on a display screen.

Alternatively, or in addition, the sensitivities ascertained in respect of the various material samples under the various measurement conditions may be evaluated by means of a visual data-mining functionality. With this functionality, which is represented in Figure 13, sensitivities of so-called library panels are represented on a display screen, classified by test gases and temperatures. In the exemplary representation according to Figure 13, library panels 101 to 112 for four temperatures A, B. C and D and for three different test gases I, II and III are represented. Each library panel 101 to 112 is assigned to a temperature A, B. C or D and to a test gas I, II or III. The sensitivities S_A of the individual materials of the library panels 101 to 112 are each represented, in a manner corresponding to their position on the sample panel, as circles 120 and in off-colours. In Figure 13, for the sake of clarity, only four of 64 material samples are represented for each library panel 101 to 112. Positive sensitivities S_A are, for example, represented in colours that range from black via red tones to yellow, whereas negative sensitivities are, for example, represented in colours that range from black via blue tones to turquoise. The diameter of the individual circles 120 is determined by the validation of the error-minimising computation, a small circle being represented in the case of a large error, and a comparatively large circle being represented in the case of a small error. Hence an intuitive assessment of large amounts of data is possible for a user.

The requisite data records relating to the functionality represented on the basis of Figure 13 are extracted directly from the database, so that the measurements to be represented can be selected via filter functions. The number of data records displayed is limited only by the main memory of the evaluating computer that is employed.

In addition, in the case of the visual data-mining functionality it is possible for additional information relating to the individual material samples to be displayed by display-screen-assisted selection of a particular circle.

Claims (29)

1. Claims 1. A device for analysing a sample panel ( 12) on which at least

   two material samples (13) are arranged, comprising a support (28) for the sample panel (12) and contacting means for electrical contacting of the material samples (13), characterized by a measuring head (26) which is capable of being inserted in a housing support (27) and which for the purpose of electrical connection to the contacting means comprises for each material sample (13) two measuring wires (30A, SOB) which bear with preloading against contact areas of the sample panel ( 12) and are connected to a measuring and evaluating unit ( 18) .
2. 2. Device according to Claim 1, characterized in that the measuring wires (30A, BOB) bear against the contact areas of the sample panel ( 12) via fusible bulbs (31A, 31B) .
3. 3. Device according to Claim 1 or 2, characterized in that the measuring wires (30A, BOB) are each connected to a spring-loaded contact (32A, 32B) which guarantees a constant bearing pressure of the respective measuring wire (30A, BOB) on the respective contact area.
4. 4. Device according to one of Claims 1 to 3, characterized in that the measuring head (26) is connected to a gas- supply unit (16).
5. 5. Device according Claim 4, characterized in that the gas-supply unit (16) is connected to a data-processing unit (53) pertaining to the measuring and evaluating unit (18).
6. 6. Device according to Claim 4 or 5, characterized in that the gas-supply unit (16) comprises a gas-mixing apparatus.
7. Device according to one of Claims 4 to 6, characterized in that the gassupply unit (16) comprises a water reservoir (46) .
8. 8. Device according to one of Claims 4 to 7, characterized in that for the purpose of exposing the sample panel (12) to gas the measuring head (26) comprises a gas space which is preferably constituted by a substantially bell-shaped distributor apparatus (39) and which is connected to the gas-supply unit (16).
9. 9. Device according Claim 8, characterized in that a diffuser (42) is arranged in the gas space.
10. 10. Device according Claim 8 or 9, characterized in that the gas space is provided with a gas outlet which is preferably constituted by at least one spacer (43) which is arranged between the sample panel ( 12) and the distributor apparatus (39).
11. 11. Device according to one of Claims 1 to 10, characterized in that the measuring and evaluating unit (18) comprises two relay plugboards (50, 51) which are connected to the measuring wires (30A, SOB) and preferably each exhibit a 3 x 64 matrix consisting of high-frequency-suitable relays.
12. 12. Device according to one of Claims 1 to 11, characterised in that the measuring and evaluating unit (18) comprises an impedance analyser (64).
13. 13. Device according to one of Claims 1 to 12, characterised in that the measuring and evaluating unit (18) is equipped with measuring and control software which transfers acquired and/or derived measured data to a relational database which is preferably linked to evaluating software.
14. 14. Device according to Claim 13, characterised in that the evaluating software comprises a fit functionality for the purpose of computing theoretical impedance spectra in respect of the individual material samples, the computation preferably being effected by taking as a basis an equivalent circuit (90) that comprises at least one electronic component.
15. 15. Device according to Claim 13 or 14, characterised in that the evaluating software comprises a data-mining functionality.
16. 16. Device according to Claim 15, characterised in that the data-mining functionality operates by applying preferably multidimensional target functions.
17. 17. Device according to one of Claims 14 to 16, characterised in that the data-mining functionality comprises a visualization functionality.
18. 18. Device according to one of Claims 1 or 17, characterised by a heating arrangement (22) into which the sample panel (12) is preferably capable of being plunged.
19. 19. A process for analysing a sample panel (12) on which at least two material samples are arranged, comprising the following steps: - measuring an impedance spectrum in respect of each of the material samples; storing the measured impedance spectra in a file or in a database; determining a structure of an equivalent circuit depending on the impedance spectrum measured in the given case in respect of each of the material samples, the respective equivalent circuit comprising at least one electronic component, in particular at least one resistor and/or at least one RC element; - determining starting values for the components of the respective equivalent circuit that are required for an error-minlmlsing computation; - computing a theoretical impedance spectrum in respect of at least one of the material samples by means of the error-minimising computation by taking as a basis the impedance spectrum measured in respect of this material sample and also the starting values for the components of the equivalent circuit in question; - determining fit values for the components of the equivalent circuit in question; determining a validation quantity for the computed theoretical impedance spectrum; - determining an evaluation quantity by comparison of at least one of the fit values for the components with a reference value.
20. 20. Process according to Claim 19, characterized in that a number of serially connected RC elements is determined by taking account of a preferably preselectable threshold value, whereby preferably a maximum of four RC elements are selected.
21. 21. Process according to Claim 19 or 20, characterized in that the starting values for the components of a first RC element of the equivalent circuit are ascertained in a manner depending on the maximally measured imaginary impedance Z"_MAX, whereby a starting resistance R1_START and a starting capacitance C1_ START are computed.
22. 22. Process according to one of Claims 19 to 21, characterized in that the error-minimising computation is carried out by variation of the dimensional design of the individual components of the equivalent circuit preferably by 1%.
23. 23. Process according to one of Claims 19 to 22, characterized in that in the course of the error minimising computation an error of the theoretical impedance spectrum is ascertained by analysis of the difference from the measured impedance spectrum.
24. 24. Process according to one of Claims 19 to 23, characterized in that impedance spectra are measured under various test-gas atmospheres and preferably at various temperatures in respect of each of the material samples.
25. 25. Process according to one of Claims 19 to 24, characterized in that the evaluation quantity for each material sample is written to a database and a data mining is carried out on the basis of the evaluation quantities stored in the database.
26. 26. Process according to Claim 25, characterized in that the data mining is carried out by means of a target function.
27. 27. Process according to Claim 26, characterized in that the data mining is carried out by means of a visual data-mining functionality.
28. 28. Process according to one of Claims 19 to 27, characterized in that the measured impedance spectra are examined and/or evaluated visually by means of a monitoring functionality.
29. 29. A data-processing system with a program for implementing a process according to one of Claims 19 to 27.