EP0769137A1 - A method for monitoring performance of an incubator module, said incubator module being comprised in an automated system for assaying multiple samples and a kit suitable for use in said method - Google Patents

Abstract

A method and a kit for monitoring performance of an incubator, siad incubator module being suitable for incubating multiple samples simultaneously at a defined temperature (e.g. for incubating microtitre plates), said incubator module being comprised in an automated system for assaying multiple samples optionally in multiple simultaneous assays e.g. in multiple microtitre plates.

Description

A METHOD FOR MONITORING PERFORMANCE OF AN INCUBATOR MODULE SAID INCUBATOR MODULE BEING COMPRISED IN AN AUTOMATED SYSTEM FOR ASSAYING MULTIPLE SAMPLES AND A KIT SUITABLE FOR USE IN SAID METHOD.

Automated systems for carrying out multiple assays are well known in the art. Such automated systems can incorporate a number of functions for processing assays; incubation, reagent addition, washing, absorbance measurement, shaking and sample transport. The latter implies that the automated system will basically execute complete assay procedures, without operator intervention, according to assay protocols present in the systems software. These assay protocols are, of course, based on the instructions in the test kit package inserts. In particular in the field of biotechnology and e.g. in the area of immunotechnology large numbers of samples are to be processed. In general in the field of biotechnology microplates are the preferred containers for carrying out reactions. Good Laboratory Practice (GLP), includes the periodic checking of laboratory instrumentation for proper functioning. For the majority of the functions of the automated systems for carrying out multiple assays, suitable means for checking their performance exist. An exception is formed by the incubator modules, said incubator modules being used to heat the samples e.g. present in microplates to a defined temperature and to maintain that temperature over a defined period of time.

To date the means available entail indirect measurement of the temperature of the samples at the site of the incubator module. Such indirect measurement for example comprises measurement of the temperature of the heating elements of the incubator module which obviously is a very crude measurement of the actual sample temperature and does not take into account the fact that the module, the sample container and the samples themselves have to warm up to the temperature set for the incubator module. Neither does it enable an accurate analysis of temperature dispersion over a microplate. Another method comprises measurement of the temperature of the environment in which the microplate is present, but also has the same shortcomings with regard to warming up and temperature dispersion determination. Another alternative is to directly measure the temperature of a sample being incubated using a thermosensor immersed in the sample. This is however undesirable due to the fact that for large numbers of samples such as are present on a microplate an equivalent large number of thermosensors is required, which would require expensive adaptation of the systems and involve unacceptable increased costs of the system. The impracticality of use of thermosensors on small samples such are often used in assays of biotechnological nature stems also from the interference with the incubation temperature and the assay itself due to the presence of the sensor itself. The sample sizes are so small that heat transport through the metal sensor itself will influence the temperature of the sample. Thus even determination of the temperature in one sample present in a microplate could not be considered accurate.

A performance check for an incubator comprised in an automated system for assaying multiple samples simultaneously at a defined temperature in particular suitable for asaying small samples requires a medium and means, compatible with the microplate format, that will allow quantification of the temperature and in particular the temperature distribution during incubation at the site of incubation. In order to be able to detect inadequate performance of the incubator in the system the method should have an accuracy better than 1 °C and a precision better than 0.1 °C.

A liquid with thermochromic properties would be most suitable as a solution as this would allow the photometer of an automated system for carrying out assays such as the Microplate Processor 3000 (MPP) to be used as a means of quantification linking absorbance to temperature. As a literature search, aimed at finding formulations for such a liquid, did not render any usable option, an alternative was investigated; A buffer solution with a large acidity (pH) dependence on temperature (T) and an acid-base indicator with a large absorbance (A) dependence on acidity (pH). The result is what was aimed for: a solution of which the absorbance spectrum is temperature dependent.

The subject invention is directed at a method for obtaining the desired accuracy and precision with regard to the actual temperature of a sample in an incubator in an automated assay system and even with regard to the temperature dispersion over a number of samples being simultaneously incubated. In particular the invention is directed at obtaining this information in systems directed at automated processing of microplates. The subject invention has the additional advantages that not only the previous inadequacies regarding the technical aspects are overcome but that the solution is simple and relatively low in cost, simple to carry out and requires little adjustment to existing automated systems.

The subject invention is directed at a method for monitoring performance of an incubator module said incubator module being suitable for incubating multiple samples simultaneously at a defined temperature (e.g. for incubating microtitre plates), said incubator module being comprised in an automated system for assaying multiple samples optionally in multiple simultaneous assays e.g. in multiple microtitre plates, said method comprising

1) measuring the absorbance of a temperature dependent solution in a photometer (phot 2) at at least two different wavelengths λl and λ2, wherein the absorbance of the temperature dependent solution is determined at a location different to the location of the incubation and at a moment in time subsequent to the incubation, said temperature dependent solution comprising a buffer with a temperature dependent pH and an acid-base indicator with an absorbance spectrum linked to pH and said wavelengths being selected such that δA/δT is positive for one wavelength and negative for the other, said measuring providing absorbance values A_test.phot ₂,λi and Aι-_st,_Phot ι,xι, wherein ₂,xι indicates the absorbance

(A) determined for the temperature dependent solution (test), using the photometer (phot 2) at wavelength λl and A,_est,_phot ₂,λ₂ indicates the absorbance (A) determined for the temperature dependent solution (test), using the photometer (phot 2) at wavelength λ2,

2) saving data of step 1) in a data file 3) measuring the absorbance of a temperature independent calibration solution for the two wavelengths λl and λ2 of step 1) in the photometer (phot 2), said temperature independent calibration solution exhibiting the same absorbance spectrum as the temperature dependent solution at the defined temperature, said measuring providing absorbance values and Ac_aι._Phot 2x2. wherein Acai_>Phot 2./.1 indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the photometer (phot 2) at wavelength λl and ₂,λ2 indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the photometer (phot 2) at wavelength λ2 and 4) . saving the data of step 3) in a data file 5) calculating the actual temperature T of the temperature dependent solution using the saved data and the formula

In (Atest,phot2,λl/Atest.phot ₂.λ₂) ⁼ Ct"+ β'. T, wherein α"= α'- In (A_cal._photl._λl/A_caι._Photi,λ₂) + In (A_caι_jphot2._λι/A_caι,_pnot ₂,u) with the values of α'- In (A_caι_>p_h0tι, i Acai.photi,j.₂) being either predetermined or else being obtainable by subjecting the temperature independent calibration solution to absorbance measurement on a further photometer (photl) for the two wavelengths λl and λ2 of step 1) thereby obtaining A_caι._n„_0t i and ι._λ2. wherein A_∞Lphot ι.λi indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the further photometer (phot 1) at wavelength λl and Ac_aι,_ph₀t indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the further photometer (phot 1) at wavelength λ2 and saving the data of step 5 in a data file and subjecting the temperature dependent solution to absorbance and temperature measurement on the further photometer (phot 1) for the two different wavelengths λl and λ2 of step 1) thereby obtaining absorbance values At_est,_Phot i and At_est,_phot u₂, at known temperatures from which α' and β' can be calculated using the formula In (A_λι/A_λ2) = α' + β'.T and saving the data in a data file 6) comparing the amended temperature or amended temperature distribution of a temperature dependent solution incubated in the module with the defined temperature the module is set to ascertain.

Using the ratio of absorbance measurements at two different wavelengths instead of using absorbance measurements at a single wavelength offers some interesting advantages. The relation between the signal and the temperature can be expressed as the formula 2:

ln(A_λl/A_λ2) = (α - α_k2) + (β_λl - β_λ2) . T = ' + β\ T (2)

When one wavelength lies in the range with a positive δA/δT value and the other in the range with a negative δA/δT value, using the ratio of two absorbance measurements has the following advantages:

Increased sensitivity of the method.

Increased accuracy of the method since errors caused by different light-path lengths of samples assayed in photometer (phot 1) and photometer (phot 2), and therefore errors caused by volumetric differences in case of a microplate, are eliminated.

Increased accuracy of the system since errors caused by differences in chromophore concentrations are eliminated.

In Clinical Chemistry 39/2, 251-256 (1993) Schilling et al describe the use of thermochromic indicator solutions such as TRIS/Cresol Red in determining the temperature of a multicuvette photometer in combination with a thermically independent solution of Cresol Red in

HEPES and phosphate buffer in a manner to eliminate errors of optical measurements caused by variations of pathlength and blank transmittance of the wells. However they were not faced with the subject problem of not being able to determine the absorbance at the location of the heated sample. They carried out their absorbance measurements at the location of the temperature measurement i.e. the sample is measured by the photometer at the site of the temperature determination. They used an electronic probe in one well in order to calibrate the photometer. Furthermore a third wavelength was required to eliminate errors produced by different blank transmittances in addition to the quotient of two wavelengths to eliminate the errors due to different light path lengths or chromophore concentrations. The temperature independent buffer is used in the cited literature to simply study the influence of photometric noise on the accuracy and precision of the method and not for calibration purposes as in the subject method. No problem with regard to cooling down effect is present in the cited method as it is not directed at a process similar to the subject method. Simply carrying out the described method of Schilling et al in the subject situation will not solve the problem the subject method solves.

Once the system has been properly calibrated with the method according to the invention, it allows determination of the temperatures of the liquid in the wells of a microplate e.g. while the microplate is in an incubator module of an automated assay system comprising the incubator module. A suitable example of a system in which the method according to the invention can be carried out is the Microplate Processor 3000.

Based on a literature search, Tris(hydroxymethyl)-aminomethane (TRIS) in combination with hydrochloric acid (HC1) was selected as an extremely suitable example of a buffer system to be used in a method according to te invention. This is mainly because of its large δpH/δT value of approx. -0.03 pH units/°C. Preferably the buffer system of the temperature dependent solution has a large δpH/δT, preferably at least an absolute value of 0,02 pH units /°C. The larger the absolute value the better any small change in temperature will register as a noticeable change in pH leading to a change in the absorbance which is measured, thereby increasing the sensitivity of the method. In line with this a temperature dependent solution exhibiting a large δA/δpH value is preferred. Large can suitably be quantified as a value sufficient to result in at least a precision of 0,1 °C and an accuracy of at least 0,5 °C, preferably of at least 0,3 °C. A number of buffers other than TRIS are also suitable for use in a method according to the invention. Suitable examples comprise an aqueous solution selected from citrate, tartrate, phtalate, phosphate, tris(hydroxymethyl)aminomethane (also known commonly as TRIS), Borax and sodium bicarbonate. A number of acid-base indicators can be used in temperature dependent and temperature independent solutions for the subject method. The selection of an appropriate indicator will be obvious to a person skilled in the art after considering the invention in the light of this description. The solubility of the acid-base indicator in the temperature dependent and temperature independent calibration solution must be sufficient for measurable and reliable absorbance values. In general current apparatus can reliably measure absorbances higher than 0,5 AU, so an indicator concentration leading to such an absorbance at the defined temperature and at wavelengths λl and λ2 is acceptable. A preference for a concentration leading to an absorbance between 0,5 and 1,5 AU is preferred. Out of the acid-base indicators applicable for the pH buffering range of TRIS, Cresol Red was selected, mainly because of its good solubility resulting in large δA/δpH values. Earlier investigations [1], using phenolphthalein as a pH indicator e.g. gave poor results because of the poor solubility of phenolphthalein in TRIS buffer.

Calibration solutions that have temperature independent spectra being identical to the spectra of the temperature dependent solution at certain temperatures offer some interesting possibilities. Firstly, such calibration solutions offer the possibility to experimentally determine the precision of the method. Secondly such calibration solutions offer the possibility to use different photometers for calibrating the method (determination of α' and β' using measured temperatures and absorbance values) and for actual temperature measurements (calculate T with known ' and β' and measured absorbance values). Using a temperature independent calibration solution to compensate for the differences between photometers works as follows [4]:

The temperature to be measured must have a certain expected value. A calibration solution must then be available that has a spectrum approximately identical to that of the temperature dependent solution at that expected temperature. Now the temperature dependent solution is calibrated (α' and β' are determined) using photometer (phot 1). Also the absorbance values of the temperature independent calibration solution at the two wavelengths of interest are measured on photometer 1. (Ac_aι._photι,λi and Ac_aι._phoU,_λ2)

The second photometer is used to measure the absorbance values at the two wavelengths of interest of both the temperature dependent (At_est,_Phot2,λi and A,est,_Phot2,λ2) and temperature independent calibration solution and Ac_aι._phot2,λ2)- Now the measurements of the temperature dependent solution from photometer 2 can be expressed in absorbance units of photometer 1, i.e. the corrected absorbance values, for both wavelengths: Acorr -" (Acal,_phou Acal,_phot₂) • At_est,phot2 (3)

By using the corrected absorbance values the actual temperature can now be calculated. Combining equations (2) and (3) gives:

ln(A_test,phot2,λi/A_test,_phot₂,λ₂) = α" + ^β'.T (4)

with α." = A' - ln(Acai,_Photi.λi/ c_aι_ιPhoiι, 2) + ln(A_caι_>phot₂.λi A_caι,phoi₂,λ₂) (5)

So, in fact, α' is re-calibrated into α".

This method compensates for: - Small differences in blank media, e.g. water versus air.

Differences in light-path lengths.

Small spectral differences in optical interference filters. (Dependent on the value of δβ/δλ at the wavelengths used).

As described in literature [2], calibration solutions as described above exist, in the case of a TRIS/Cresol Red temperature dependent solution the TRIS in the TRIS/Cresol Red solution is replaced by a mixture of HEPES and phosphate. By setting the pH of the calibration solution, its spectrum can be made approximately identical to that of the TRIS/Cresol Red solution at any temperature in the range of interest, thereby enabling the production of temperature independent calibration solutions. As a container comprising temperature dependent solution e.g. a microplate filled with the temperature dependent solution such as TRIS/Cresol Red solution cools down during the transport from the incubator module to the photometer (phot 2), (also referred to as the reader module in the following text) the obtained absorbance values do not accurately represent the temperatures as they were when the microplate was actually still in the incubator module. Therefore in a method according to he invention calculating means are used to correct the absorbance deteremined in the photometer (phot 2) with regard to the cooling down effect that occurs between the locations of incubation and absorbance measurement. A way to minimise the inaccuracy is to have the photometer as close to the incubator as possible and preferably in an atmosphere as close to the temperature ofthe incubator as possible. A way to overcome inaccuracy due to the transport is to registrate the cooling down curve for the sample in the container, preferably for each well of a microplate if a microplate is used and to use these curves to estimate the temperatures as they were in the incubator module.

The subject invention is therefore also directed at a method as disclosed above, wherein a first microplate filled with temperature independent calibration solution is transported to the photometer (phot 2) and the absorbance is read at the two wavelengths λl and λ2, measurement data is saved in a data file and the microplate is taken to the output module to be removed, a second microplate comprising temperature dependent solution is taken to the incubator module and incubated at the defined temperature according to the setting of the incubator module and after completion of incubation is transported to the photometer (phot 2), where the absorbance is read a number of times at the two wavelengths λl and λ2 at time intervals preferably controlled by a software timer and all data is saved in a data file. In such an embodiment the time interval between incubation termination and the first reading in the photometer (phot 2) is preferably as short as possible. It is limited by the time required for transport of the microplate from incubator module to the photometer (phot 2), said time interval preferably being less than 45 seconds, suitably being 25-35 seconds. A suitable embodiment of a method where the absorbance is read a number of times at the two wavelengths after the container with the temperature deependent solution has been removed from the incubator is a method wherein the time interval between absorbance measurements in the photometer (phot 2) is determined by the length of time required for the measurement of the absorbance at the two wavelengths λl and λ2 and the length of time required to save the data in a file, said time interval in general being longer than 25 seconds, preferably being less than 45 seconds and suitably being 30 seconds.

In a method according to the invention, wherein the temperature determined on the basis of absorbance data from the photometer (phot 2) is corrected for the cooling down that occurs between transport from incubator module to said photometer calculating means on the basis of a mathematical function fitting the pattern of cooling down such that the regression coefficient R² is equal to or larger than 0.98 can suitably be used. The calculating means employed can for example enable a function for absorbance against time to be fitted in the least squares sense on the absorbance measurement data obtained in the photometer (phot 2) enabling calculation of actual absorbance in the incubator module at the moment of taking the temperature dependent solution from the incubator module i.e. at time zero thereby enabling subsequent calculation of the actual temperature in the incubator module at time zero.

More specifically for the cooling down of microplates the following was ascertained. The temperature T of an object with an initial temperature T_ini_t that is placed in an environment with constant temperature T_env as a function of time t is theoretically given by:

T(t) = T_env + (Tini, - T_en ) ■ e^{" τ} (τ is a time constant) (6)

However, for a microplate that cools down, things are much more complicated since the environment is not constant and differs per well. In fact it is almost impossible to have a theoretical model for this. In practice, absorbance values instead of temperatures are measured as a function of time. It was decided to use a second or third order polynomial as a model for the absorbance - time relation instead of applying a model for the temperature-time relation. This was accomplished using a function like:

A(t) = a + b . t + c . t² (+ d . t³) (a, b, c and d are constants) (7)

By fitting this model in the least squares sense on the measurement data, a is an estimate for the absorbance value at time zero.

When multiple samples are simultaneously incubated as is the case for microplates the temperature of sample in each well is influenced by neighbouring wells. The influence will vary with location of the well. In order to increase the accuracy of the method according to the invention data filtering can be applied in the calculations. The incubator modules of the automated assaying systems such as the Microplate

Processor 3000 have low frequency characteristics, i.e. the temperature distribution over the microplate can show temperature gradients, cold areas, warm areas or an edge effect. Temperature values jumping up and down from well to well (high frequency behaviour) is not possible. This implicates that any high frequency components in the temperature distribution obtained by the method in this report, may be filtered out by means of a so-called "two dimensional low pass" filter. Many commonly used data filtering techniques are based on multiplication with an appropriate function in the frequency domain (with the Fourier transform of the signal) or taking the convolution of the signal with an appropriate function. These techniques require endless signals or at least signals that are defined over a relative large area. Since for the temperature only a limited number (8 x 12) values are available in the case of a microtitre plate, this method of filtering cannot be applied especially because reliable filtered values for the temperatures at the edges ofthe microplate, that are most interestingly, cannot be calculated.

An alternative was investigated namely to define the temperature of each well as a function ofthe temperatures ofthe well itself and the temperatures ofthe 8 wells closest to that well, e.g.:

^Tfilt_D6 = fl ■ T_C5 + f₂ . Tee + f₃ • Tc7 + t\ . T_D5 + fj . T_D6 + + f₆ . T_D7 + f7 . T_E5 + f₈ . T_E6 + f9. T_E7 (8)

Where Tf is the calculated temperature after filtering and fι through f₉ are constants. The constants are calculated by fitting a linear model in the sense of the least squares to the temperature data of a square block of 9 wells, (two dimensional linear regression)

Model: Tπu = a.rownumber + b.columnnumber + c (9) (a, b and c are constants)

Minimizing in the sense ofthe least squares results in minimizing function F:

F = Σ (Tf_lh(welli) - T(welli))² (i = 1 .. 9) (10)

Which means that δF/δa = δF/δb = δF/δc = 0. Now a, b and c can be expressed in terms of T(welli) and the factors fi through f₉ can be calculated. For a well not at the edge of the microplate, as in (8), all factors fj are equal to 1/9. For a well at the edge of the microplate but not at a corner, the following is valid: e.g.

^Tfllt_A6 = fl.TAS + ₂-T_A6 + f₃.T_A7 + f₄.T_B5 + f₅.T_B6 + + f₆.T_B7 + f₇.Tc₅ + f₈.T_C6 + fo.T_C7 (1 1) With fi = f₂ = f₃ = 5/18, f₄ = fj = f₆ = 2/18 and f₇ = f₈ = f₉ = -1/18.

It can be noted that the linear regression used to calculate ^τfilt_A6, as in equation (11), is also used for calculating ^τfilt_B6.

For a well at the corner of a microplate, the following is valid: e.g.

^Tfllt_Al = fl.T_Al + f₂.TA2 + f₃.T_Bl + f .TA3 + f₅.T_B2 +

+ f₆.Tcι + f₇.T_B3 + f₈.T_C2 + f₉.T_C3 (12)

With fi = 8/18, f₂ = f₃ = 5/18, f₄ = f_s = f₆ = 2/18, f₇ = ft = -1/18 and f₉ = -4/18

It can be noted that the linear regression used to calculate ^τfilt_Aι, as in equation (12), is also used for calculating ^τfiltA2, ^τfilt_Bι and ^Tfilt_B2.

This spatial low pass filtering technique is illustrated in Figure 1 in which the temperature distribution is represented by a well defined function to which Gaussian noise is added, it will be appreciated by a person skilled in the art that the calculation ofthe temperature in the incubator module can be ensured by fitting a 2 dimensional linear model in the sense ofthe least squares to the temperature data of a matrix of samples with the above illustrated square of nine wells for a microtitre plate merely being an example.

The subject invention is also directed at a test kit comprising components necessary for carrying out the invention as described above and in the Example. Such a test kit comprising at least a container comprising a temperature dependent solution exhibiting at least an absorbance higher than 0,5 AU at wavelengths λl and λ2 at the defined temperature, said temperature dependent solution comprising a buffer system with a temperature dependent pH and an acid- base indicator with an absorbance spectrum linked to pH and a container comprising at least one temperature independent calibration solution exhibiting an absorbance spectrum identical to that of the temperature dependent solution at the defined temperature, said temperature independent calibration solution preferably comprising a composition as close to that of the temperature dependent solution as possible. Optimally the solutions in such a kit have a pH a) in the working range of the buffer system of the temperature dependent solution at the defined temperature at which measurement is to take place, said defined temperature being within a desired temperature range, preferably being within the range 20-60 °C, b) in the indication range of the acid-base indicator at the defined temperature at which measurement is to take place, c) such that the absorbances at the two wavelengths λl and λ2 are as close to eachother as possible at a temperature in the middle ofthe specified range. d) preferably as low as possible to prevent CO2 absorption.

In particular when a number of defined temperatures are to be checked on the incubator a kit will be used as described above comprising multiple temperature independent calibration solutions for a number of defined temperatures, wherein the pH of each temperature independent calibration solution is selected such that the absorbance spectrum ofthe temperature independent solution is equal to the absorbance spectrum of the temperature dependent solution at the defined temperature. It will be clear from the description of the method that the temperature dependent buffer can comprises an aqueous solution of citrate, tartrate, phtalate, phosphate, Tris(hydroxymethyl)aminomethane, Borax or sodium bicarbonate, preferably of Tris(hydroxymethyl)aminomethane. It will also be apparent from the description of the method according to the invention that a kit wherein the acid base indicator is Cresol Red and is present in soluble form is a suitable embodiment. In such a kit the temperature independent calibration solution can comprise HEPES, phosphate and a Cresol Red solution and is preferably further identical to the temperature dependent solution which comprises Tris(hydroxymethyl)- aminomethane as buffer and a Cresol Red solution as acid-base indicator. In order to prevent microbial degradation the solutions in the kit are preferably provided with anti microbial agents generally used inthe art such as azide. Preferably cinnamaldehyde and gentamicin sulphate will be used with a view to regulations in particular countries. The Example provides further precise details ofthe solutions that can be present in a kit and an embodiment of how they can be used. Preferably a kit according to the invention will comprise the calibration data required using photometer (phot 1), thereby rendering the practical application extremely simple and userfriendly.

In general the method and kit according to the invention in the various embodiments disclosed can be used to check the temperature dispersion of the incubator module at a number of different defined temperatures, said method or kit requiring a number of temperature independent calibration solutions equivalent to the number of different temperatures. In particular an automated assay in an automated system can be carried out whilst the method according to the invention is carried out.

EXAMPLE

The method according to the invention has been carried out using a Microplate Processor 3000 as automated assaying system comprising an incubator. The following reagents were used:

1. Tris(hydroxymethyl)-aminomethane (TRIS)

The working range of a TRIS buffer system is from pH 7 to pH 9. As reported in literature [4,5], the temperature - pH relation is almost linear in this range with dpH/dT = -0.03 pH units/°C for a 0.1 mol/L solution. A stock of a 0.1 mol/L (= 12.114 g/L) TRIS buffer was prepared by dissolving TRIS in NEN class 1 quality water.

Setting the pH of the solution was done by adding small amounts of a high concentration HC1 solution (4-10 mol/L). Because of the temperature dependency, this has to be done at a controlled temperature, or the desired pH has to be calculated for the actual temperature.

2. Cresol Red

One ofthe indication ranges of Cresol Red is reported to be from pH 7.2 to pH 8.8 with a yellow to red colour transition [6]. This means that in the lower region of the visible spectrum the absorbance increases with increasing acidity (decrease of pH), whereas in the higher region ofthe visible spectrum the absorbance decreases with increasing acidity. A high concentration solution in ethanol was thought to be the most convenient dosage form for Cresol Red. For this purpose a stock solution was prepared by dissolving Cresol Red (Kodak) in ethanol (Baker, 96%) in a concentration of 10.0 g/L.

3. HEPES/phosphate solution For experimentally determining the precision of the method and for calibration purposes, the need was felt to have a solution with a temperature independent absorbance spectrum, but identical to that ofthe TRIS/Cresol Red solution at a certain temperature. As described in literature [2], such a solution can be prepared by replacing the TRIS in the TRIS/Cresol Red solution by a mixture of HEPES and phosphate.

A stock of HEPES/phosphate solution was prepared by dissolving HEPES (Organon) and sodium phosphate, monobasic (Baker) in NEN class 1 quality water in concentrations of 26 mmol/L (= 6.196 g/L) and 76.2 mmol/L (= 11.289 g/L) respectively.

4. Preservatives

As the method is intended to be used in a commercial product, the TRIS/Cresol Red and calibration solutions have to be stable for quite some time. For this purpose, preservatives have to be added to prevent microbial degradation. Although commonly used, sodium azide was not selected because of restrictions in some countries. Instead, cinnamaldehyde and gentamicin sulphate, as used in some components of the latest Organon Teknika Microelisa assays, was used.

A stock solution of cinnamaldehyde was prepared by diluting cinnamaldehyde in 96% ethanol + 5% methanol (Baker) in a concentration of 200 ml/L.

Cinnamaldehyde (Merck) and gentamicin sulphate (USBC) were added to the TRIS and HEPES/phosphate solutions up to final concentrations of 0.2 ml/L and 0.1 g/L respectively.

The TRIS/Cresol Red temperature dependent and temperature independent calibration solutions were prepared in the following manner:

For the Cresol Red concentration and the pH setting of the TRIS/Cresol Red solution, the following was taken into consideration: pH should be in the working range of TRIS buffer in the temperature range of interest. pH should be in the indication range of Cresol Red in the temperature range of interest. - pH should be as low as possible to minimize C0₂ absorption. pH setting should be such that in the middle of the temperature range of interest (at approx. 37 °C) the absorbances at the two wavelengths being used, are approximately equal.

Cresol Red concentration should be such that the absorbance values at the two wavelengths being used, are in the range 0.5 - 1.5 AU (optimal working range of the reader module ofthe MPP) in the temperature range of interest. The pH of the stock solution of TRIS buffer, with preservatives added, was set to 7.80 at 37 °C. To a portion of this stock solution Cresol Red (from the stock solution) was added up to a final concentration of 30 mg/L to be used for measurements on the spectrophotometer. To another portion, Cresol Red was added up to a final concentration of 75 mg/L to be used for measurements on the MPP. It was verified that the addition of Cresol Red did have no measurable effect on the pH.

Two calibration solutions had to be prepared, i.e. one to be used for temperature measurements with the incubator module set for 37 °C and the other for 50 °C incubations. For the preparation ofthe calibration solutions, two portions were taken from the HEPES/phosphate stock solution (with preservatives added) of which the pH was set to 7.80 and 7.46 respectively.

These pH values correspond to the pH values of the TRIS/Cresol Red solution at 37 °C and 50 °C respectively. The two portions were each divided into two portions to which Cresol Red (stock solution) was added up to final concentrations of 30 mg/L and 75 mg/L respectively. It was verified that the addition of Cresol Red did have no measurable effect on the pH.

The measurement procedures were as follows: 1 Calibration procedure

Spectra were recorded in the visible region, 400 through 700 nm, using a Pye Unicam Model PU8700 spectrophotometer and polystyrene cuvettes (1 cm optical pathlength) with an internal width of approx. 1 cm. The double walled cell-holder of this instrument is connected to a temperature controlled waterbath via tubings and a pump. The temperature in the cuvette is measured by means of a thermocouple, positioned just above the lightbeam generated by the spectrophotometer. The thermocouple is connected to a Fluke Model 27 multimeter equipped with a Model 80TK Thermocouple Module. The amount of fluid in the cuvette is such that the insertion depth ofthe thermocouple is approx. 3 mm.

Spectra for the calibration solutions were recorded at room temperature and saved in data files for further processing. Just before every measurement, the spectrophotometer was blanked against a cuvette containing water.

Spectra for the TRIS/Cresol Red solutions were recorded in the range room temperature to 52 °C. For practical reasons, spectra were recorded during warming up or cooling down of the liquid in the cuvettes, instead of stabilizing the temperature in the cuvette for every measurement. Heating of the waterbath was set in such a way that a temperature raise of approx. 1 °C per 6 minutes in the cuvette was achieved. By doing so, the error of the measured temperature because ofthe fact that it takes a certain time to record a spectrum (approx. 20 sec), is less than 0.1 °C. The measurement accuracy of the thermocouple system is 0.1 °C. Spectra were recorded at approx. 2 °C intervals and saved in data files for further processing. For each recorded spectrum, the temperature of the fluid in the cuvette was noted. Just before every measurement the spectrophotometer was blanked against a cuvette containing water.

Cooling ofthe waterbath is achieved by heat conduction to a coiled tube positioned in the waterbath through which (cold) tap water runs. The flow of tap water was set in such a way to achieve a temperature drop of approximately 1 °C per 6 minutes. Spectra were recorded in a similar way as during heating up.

2. Measurement in the Microplate Processor 3000.

These measurements were performed in microplates ofthe type Greiner, 12 well stripplate, flat bottom with curved edges. Each well was filled with 100 ml of fluid that was pipetted with a calibrated single channel pipette. All fluids were allowed to reach room temperature before pipetting and pipetted plates were checked for the absence of air bubbles. The optical pathlength for this volume and type of plate is approx. 4 mm. In order to obtain approximately the same absorbance values as for the spectrophotometer with a 1 cm optical pathlength, the Cresol Red concentration in calibration fluids and TRIS/Cresol Red fluid as used for the MPP had a 2.5 times higher Cresol Red concentration. As mentioned above the signal is independent of the chromophore concentration.

The used MPP was one of the 5 prototypes conFigured with prototype software, i.e. created with Turbo C and running under MS-DOS. For the experiments, two protocols were programmed for the MPP that essentially only differ for the incubation temperature, i.e. one protocol for 37 °C incubation and one for 50 °C incubation. 37 °C and 50 °C are the two incubation temperatures at which the incubator modules of the MPP are to be tested for accuracy and temperature distribution over the microplate.

Each protocol performed the following processing steps:

The first microplate filled with calibration fluid(s) is transported to the reader module and is read at two wavelengths (endpoint readings). Measurement data is saved in data files and the microplate is taken to the output module for removal by the operator.

The second microplate is taken to the reader module and is read at two wavelengths (endpoint readings). These readings are related to the temperature outside the MPP and the temperature inside the instrument.

The microplate was then taken to the incubator module #3 and incubated for 20 minutes at 37 °C setpoint or for 40 minutes at 50 °C setpoint.

After completion ofthe incubation, the microplate was taken to the reader module where the plate was read 10 times at two wavelengths at time intervals controlled by a software timer. Time zero was defined as the moment the microplate was picked up by the transport module from the incubator plate carriage. Exact measurement times for the first wavelength are 30, 60, 90, 120, 150, 180, 210, 240, 270 and 300 sec. For the second wavelength these times are 35, 65, 95, 125, 155, 185, 215, 245, 275 and 305 sec. All measurement data was saved in data files for further processing. - After completion of all measurements, the microplate was taken to the output module for removal by the operator.

The time interval of 30 sec. was chosen because it takes approx. this time to transport a microplate from an incubator module to the reader module. The time needed by the reader module to read at two wavelengths and to save the measurement data in a data file takes approximately 30 sec. as well. Via the RS-232 connection and a LAN, data files were transferred to a PC for data processing and data reduction using Lotus Symphony.

The following results were obtained: 1. Temperature - pH relation of TRIS buffer The 0.1 mol/L TRIS buffer with pH = 7.80 at 37 °C was used. The pH values of this solution were measured in the temperature range from approx. 20 to 53 °C. On the measurement data a linear regression in the sense ofthe least squares was performed resulting in:

pH(T) = 8.83 - 0.0277.T (T in °C) (13) With R² = 0.999 and SterrY_est = 0.0087 which is well below the measurement accuracy of the pH. The value δpH/δT = - 0.0277 (standard error is 0.00016) is well in accordance with the values reported in literature [4,5]. Results are visualized in Figure 2.

2. ABSORBANCE SPECTRA OF TRIS/CRESOL RED SOLUTION AS A FUNCTION OF TEMPERATURE Absorbance spectra of the TRIS/Cresol Red solution, prepared as above were recorded in the temperature range from 26 - 52 °C. The results are visualized in Figure 3. What can be seen is that in the range 400 - 473 nm, the absorbance increases with an increasing temperature (β is positive) and in the range 473 - 600 nm, the absorbance decreases with an increasing temperature (β is negative). An isobestic (temperature independent) point exists for 473 nm and local maxima are located at 576 nm and 435 nm. To get an impression of the values for α and β from equation (1) as a function of the wavelength, these values were calculated using only the absorbance spectra at 26 °C and 52 °C. (For each wavelength two equations are then available from which α and β can be calculated) The results are visualized in Figure 4. The relevant wavelengths, for which optical interference filters are available in the reader module ofthe MPP, are drawn as vertical lines in Figure 4 (405, 450, 492 and 540 nm). With the limitation of only being able to use these wavelengths, it is clear that the best temperature sensitivity, corresponding to the greatest absolute value for β' in equation (2), is obtained when using 405 and 540 nm.

The ln(A₅4o/A ₀5) values as a function of the temperature were calculated and α' and β' from equation (2) were calculated by means of a linear regression in the sense of the least squares. The results are visualized in Figure 5. and the regression data are:

α' = 1.4939 (standard error is 0.0108) β' = - 0.03958 (standard error is 0.00026)

R² = 0.999 SterrY« = 0.0093.

Absorbance values of the two calibration solutions were read at 540 nm and 405 nm on the spectrophotometer. The found values were then used to calculate virtual temperatures using equation (2) and the values calculated for α' and β'. Surprisingly the calculated virtual temperatures were a few degrees higher than expected, i.e. 40.4 °C where 37.0 °C was expected and 56.6 °C where 50.0 °C was expected. Based on the estimated δpH/δT value for the calibration solution; (7.80 - 7.46)/(40.4 - 56.6) = -0.021, new calibration solutions were prepared with pH settings of 7.87 and 7.60 respectively. Absorbance measurements on the spectrophotometer for these new calibration solutions yielded the following values:

A₅₄₀ = 0.695 AU and A-røs = 1.138 AU for the 50 °C calibration solution.

A540 = 0.997 AU and A405 = 0.943 AU for the 37 °C calibration solution.

Calculating the virtual temperatures of these solutions using equation (2) and the values calculated for α' and β' yielded values of 50.2 °C and 36.3 °C respectively. The spectra of the solutions corresponded very well with the spectra of the TRIS/Cresol Red solution at these temperatures.

Apparently the absorbance spectrum is, beside by the pH, also slightly influenced by whether the solution contains TRIS or HEPES/phosphate.

3. MEASUREMENTS IN THE MICROPLATE PROCESSOR 3000

3.1. Cooling down behaviour

Four experiments were performed with the incubator module of the MPP set for 37 °C incubations and one experiment with the incubator module set for 50 °C incubations. Absorbance values at the time the microplate was still in the incubator module of the MPP were estimated by fitting a second order polynomial in the sense of the least squares on the measurement data. Regression data were excellent indicating a second order polynomial to be a good model for the cooling down behaviour. Using a third order polynomial did not show any improvement. SterrYest values in the regression analysis are well below the measurement accuracy of the reader module ofthe MPP.

Regression data for all experiments can be found in Table 1 and some typical examples of the absorbance values as a function of time (measurements and model) are shown in Figures 6 and 7. 3.2. Accuracy and precision

The accuracy of the method is defined as the error in the calculated average temperature versus the actual average temperature ofthe fluid in the wells of the microplate in the incubator module. Precision is defined as the obtained variation in successive determinations of the temperature.

The accuracy depends mainly on the following: The accuracy at which α' and β' are determined.

Degradation ofthe TRIS/Cresol Red solution and the calibration solutions. Definition ofthe time zero moment for registration ofthe absorbance - time curve (cooling down).

When the temperature T is calculated from equation (4), the variance of the calculated temperature can be calculated according to:

Var(T) = [Var(ln(A₅4o/A₄₀₅)) + Var(α") + T².Var(^β') + 2.T.Cov(α",β')]/β'² (14)

Because the accuracy is based on the average of a high number of absorbance values, Var(ln(A₅₄o/A405)) in equation (14) may be neglected. When it is furthermore assumed that the inaccuracy of the absorbance measurements of the calibration fluids on the spectrophotometer may be neglected, Var(α") will be equal to Var(α') and Cov(α",β') will be equal to Cov( ',β').

Equation (14) then reduces to:

Var(T) = [Var(α') + T².Var(β') + 2.T.Cov( ',β')]/β'² (15)

Using equation (15) for determining the accuracy of the method gives standard errors in the range 0.3 - 0.4 °C dependant on the temperature. Standard errors are 0.31 °C at 20 °C, 0.37 °C at 37 °C and 0.43 °C at 50 °C.

Degradation of the solutions has not been investigated It can only be stated that no obvious degradation effects could be observed in the MPP experiments over a period of two months.

Inaccuracies in the determination of time zero of the cool down curve would lead to a systematic error that can be corrected by redefining time zero based on a number of experiments. Thanks to the, temperature independent, calibration solutions, the precision can very well be experimentally determined. A 1 : 1 mixture of the two calibration solutions was prepared and two microplates were filled with 100 μl in each well. Both plates were read at 540 nm and 405 nm in the reader module of the MPP. The virtual temperatures of the wells of one plate were calculated using the other plate as a calibration plate. This experiment was performed twice. In first instance, α" was calculated using the average ofthe absorbance values of all 96 wells ofthe calibration plate. Not being satisfied with the results, the effect of calculating " per row of twelve wells and per individual well has been studied, as well as the effect of applying the spatial low pass filter on the calculated virtual temperatures. Results can be found in Table 2. From the results in Table 2 it is clear that there is a dramatic difference in the precision

(standard error) for α" being calculated based on the average absorbance of all 96 wells and based on the average absorbance of each row of twelve wells. The largest relative effect of applying spatial low pass filtering is seen for α" being calculated based on the absorbance values of each individual well. This makes sense since in that case the calculated temperature for each well is based on four individual absorbance values, resulting in the relative largest influence of photometric and electronic noise from the reader module ofthe MPP.

The best results for the precision (standard error) of respectively 0.03 °C and 0.06 °C are obtained for calculation of A" per individual well and applying spatial low pass filtering.

The reason for the differences in precision based on the calculation method of α" was found to lie within the reader module of the MPP. Studying the absorbance values of the calibration plates used in the various experiments learns that there is a relation to the well location. The average absorbance value of a row of twelve wells differ from row to row and the average absorbance value of a column of 8 wells differ from column to column although all differences seemed to lie within the measurement accuracy of the reader module. What is more important however, is that the row dependency seems to be wavelength dependent whereas the column dependency seems to be wavelength independent. When taking the ratio of the absorbance values at two wavelengths, as is being done to calculate the temperature, the column dependency is automatically compensated for but the row dependency could theoretically even become worse. Table 3 provides data indicating this effect. Only when calculating α" per row of twelve wells or per individual well, the row dependency is compensated for as well. 3.3. Temperature determinations

Four experiments were performed with the incubator module of the MPP set for 37 °C incubations and one experiment with the incubator module set for 50 °C incubations.

In all 5 experiments, 48 wells of the calibration plate (columns 1 through 6) were filled with the 37 °C calibration fluid and the other 48 wells (columns 7 through 12) were filled with the 50 °C calibration fluid.

Unfortunately temperature calculations with " calculated per individual well can only be performed for 48 wells because of the experiment set-up. At the time the experiments were performed however, the calculation method for α" was not expected to play an important role, as the experiments for determining the precision were carried out at a later stage.

Results can be found in Tables 4 through 8. In Figures 8 through 13, the temperature distribution over the plates in the 5 experiments has been visualized. These Figures correspond to α" being calculated per row. From the data in Tables 4 through 8 the effect of spatial low pass filtering is not clear. By comparing Figure 8 (experiment 1 without spatial low pass filtering) and Figure 9 (same experiment with spatial low pass filtering) the effect is evident. Figures 9 through

13 show the results with spatial low pass filtering applied.

From the results in Tables 6 through 8 it is clear that the best results are obtained by calculating " per individual well and applying spatial low pass filtering which is in accordance with the experiments in determining the precision of the method. This implies the need of a full microplate for each calibration solution.

4. DISCUSSION

All temperature determination experiments showed a somewhat lower temperature than the average in the Al well region which can very well be explained by the mechanical construction of the heating elements. Also an edge effect is particularly visible in the 50 °C experiment. Also this is likely to be caused by the mechanical construction of the heating elements.

Taking the accuracy of the method into account, it can be stated that the calculated average temperatures in all five experiments are very well in accordance with the set incubation temperatures ofthe incubator module. The variation ofthe calculated average temperature in the four 37 °C experiments is in accordance with the accuracy of the method. As far as the temperature distribution is concerned, there is no reason to suspect any insufficient performance for 37 °C incubations, taking the precision of the method into account. For 50 °C incubations the performance could be considered questionable, especially due to the rather low temperatures found in the Al well region. Thus illustrating the effectiveness of hte use of a method according to the invention. As the spectrophotometer is blanked against a cuvette filled with water and the reader module of the MPP is blanked against air, an error is introduced in the method because of the absorbance of the microplate itself. Correcting all absorbance measurements in the reader module of the MPP with absorbance measurements at a third (reference) wavelength would be the fully correct way of working. However, at the absorbance levels being used the effect is only marginal and the introduction of extra measurements would influence the precision of the method in a negative way. Besides, by using α" instead of ot' in the temperature calculations, the error is already partly corrected.

5. CONCLUSIONS The objectives of an accuracy better than 1 °C and a precision better than 0.1 °C are met.

The accuracy and precision of the method are only just not good enough to test an incubator module against its technical specifications, but are by all means sufficient to detect serious incubator defects.

The method is easy to perform. Especially when integrated in a commercial product, including the availability of processing and calculation routines in the Microplate Processor

3000, an easy to perform and unique method for checking the performance of incubator modules will be available to customers.

REFERENCES

1. "A temperature indicative liquid for use in Micro ELISA incubator validation; a feasibility study.", Bally, R.W., Subject memo 8760/S AH 10004, (May 7, 1991) 2. "Multiwavelength Photometry of Thermochromic Indicator Solutions for Temperature

Determination in Multicuvettes", Schilling K. et al, Clin Chem 39/2, 251-256 (1993)

3. "Optical methods for monitoring temperature in spectrophotometric analysers", O'Leary T.D. et al, Ann Clin Biochem 1983;20:153-157

4 "Development of an Aqueous Temperature-Indicating Technique and its Application to Clinical Laboratory Instrumentation", Bowie L. et al, Clin Chem 22/4, 449-455 (1976)

5 "Buffers for pH and Metal Ion Control, Perrin D.D. and Dempsey", B., Chapman and Hall Ltd., London, GB, 143, (1974)

6 "The Merck Index", Merck & Co. Inc., Rahway, NJ, USA, 11^th ed., monograph 2583, (1989)

TABLES AND FIGURES

Table 1. Regression data for absorbance values during cooling down.

Table 2. Virtual temperature determinations of calibration plates.

Table 3. Row and column dependency of the reader module measured with calibration plates.

Table 4. Temperature determination for a full plate with α" calculated per plate.

Table 5. Temperature determination for a full plate with α" calculated per row.

Table 6. Temperature determination for half a plate with α" calculated per plate.

Table 7. Temperature determination for half a plate with " calculated per row.

Table 8. Temperature determination for half a plate with α" calculated per well.

Figure 1A B. Spatial low pass filtering by means of two dimensional linear regression. Figure 2. pH - temperature relation of 0.1 mol/L TRIS buffer. Figure 3. Absorbance spectra of TRIS/Cresol Red solution at various temperatures. Figure 4. Wavelength dependency of α and β. Figure 5. L^Asαo/^os) as a function ofthe temperature for the TRIS/Cresol Red solution. Figure 6. Absorbance as a function ofthe time for 37 °C incubation. Figure 7. Absorbance as a function ofthe time for 50 °C incubation. Figure 8. Temperature distribution of experiment 1 without spatial low pass filtering. Figure 9. Temperature distribution of experiment 1 with spatial low pass filtering. Figure 10. Temperature distribution of experiment 2 with spatial low pass filtering. Figure 11. Temperature distribution of experiment 3 with spatial low pass filtering. Figure 12. Temperature distribution of experiment 4 with spatial low pass filtering.

Figure 13. Temperature distribution of experiment 5 with spatial low pass filtering.

Abreviations

A Absorbance

AU Absorbance Unit

CovO Covariance δA/δT Sensitivity; change of absorbance with temperature GLP Good Laboratory Practice HC1 Hydrochloric acid λ wavelength

LAN Local Area Network

MPP Microplate Processor 3000

NEN Nederiandse Eenheids Norm (= Dutch Standard Norm)

PC Personal Computer pH Acidity; -Log(H₃0⁺ activity)

R² Correlation coefficient

RS-232 Serial communication standard sec Second

Sterr Standard error

SterrY_est Standard error ofthe estimated Y

T Temperature in °C

TRIS Tris(hydroxymethyl)-aminomethane

VarO Variance

Table 1. Regression data for absorbance values during cooling down.

Experiment 1 2 4 5

Set temperature 37 37 37 37 50

Average A_t=o 0.900 0.871 0.890 0.869 0.631

Average R² 0.999 0.999 0.998 0.999 0.999

Minimal R² 0.984 0.997 0.991 0.995 0.999

540 nm Average SterrY_esl 0.0018 0.0014 0.0018 0.0017 0.0013

Maximal SterrY_est 0.0072 0.0026 0.0042 0.0032 0.0033

Average 0.912 0.891 0.913 0.924 1.047

Average R² 0.998 0.998 0.998 0.998 0.999

405 nm Minimal R² 0.979 0.990 0.992 0.994 0.995

Average SterrY_esl 0.0016 0.0015 0.0015 0.0013 0.0014

Maximal SterrY_est 0.0055 0.0036 0.0033 0.0024 0.0035

Table 2. Virtual temperature determinations of calibration plates

Experiment 1 Experiment 2 without with spatial without with spatial spatial low low pass spatial low low pass pass filtering pass filtering filtering filtering

Avera ige T 44.6 44.6 43.1 43.1

a" maximal DT 0.9 0.7 0.7 0.7 calculated SD_T 0.20 0.19 0.16 0.15 per plate CV_T (%) 0.4 0.4 0.4 0.3

a" maximal DT 0.6 0.3 0.3 0.3 calculated SD_T 0.09 0.07 0.07 0.05 per row CV_T (%) 0.2 0.1 0.2 0.1

a" maximal DT 0.6 0.3 0.4 0.1 calculated SD_T 0.09 0.06 0.08 0.03 per well CV_T (%) 0.2 0.1 0.2 0.1

Table 3. Row and column dependency ofthe reader module measured with calibration plates.

n = 9 540 nm 405 nm 540/405

Row Aro 'plate SD "row plate SD Factor_row/ SD Factor_piate

A 1.001 0.006 1.009 0.006 0.993 0.002

B 0.989 0.006 0.993 0.006 0.996 0.002

C 0.990 0.006 0.990 0.006 1.000 0.002

D 1.000 0.003 1.002 0.003 0.998 0.001

E 1.000 0.007 1.001 0.007 0.999 0.003

F 1.002 0.003 1.004 0.003 0.998 0.001

G 1.005 0.004 1.002 0.005 1.004 0.001

H 1.013 0.004 1.001 0.005 1.013 0.003

n = 6 540 nm 405 nm 540/405

Row o ' _p|_at_e SD SD Factor_coiumn/ SD Factor_P]_,_e

1 1.020 0.005 1.019 0.005 1.001 0.001

2 1.006 0.008 1.004 0.007 1.002 0.001

3 0.998 0.007 0.999 0.007 0.999 0.001

4 0.995 0.006 0.997 0.005 0.999 0.001

. 5 0.988 0.009 0.989 0.009 0.998 o.obi

6 0.999 0.009 1.000 0.009 0.999 0.001

7 1.000 0.006 1.002 0.007 0.998 0.001

8 1.003 0.005 1.004 0.005 0.999 0.002

9 0.998 0.005 0.998 0.005 1.001 0.001

10 0.992 0.004 0.992 0.004 1.000 0.002

11 0.998 0.010 0.996 0.011 1.002 0.002

12 1.001 0.008 0.999 0.008 1.002 0.001 Table 4 Temperature determination for a full plate with a" calculated per plate

Experiment 1 2 3 4 5

n=96 without with spatial without with without with without with spatial without with spatial low low pass spatial spatial spatial spatial spatial low pass spatial spatial pass filtering low pass low pass low pass low low pass filtering low pass low pas O O filtering filtering filtering filtering pass filtering filtering filterin filtering

average T 37.6 37.6 37.1 37.1 37.0 37.0 37.5 37.5 49.8 49.8 maximal DT 1.2 0.8 1.0 0.7 1.1 0.9 1.3 1.1 2.3 2.1

SD_T 0.23 0.20 0.23 0.18 0.24 0.20 0.27 0.24 0.49 0.46 ro . CV_T (%) 0.6 0.5 0.6 0.5 0.6 0.5 0.7 0.6 1.0 0.9

Table 5 Temperature determination for a full plate with a" calculated per row

Experiment 1 2 3 4 5

n=96 without with spatial without with without with without with spatial without with spatial low low pass spatial spatial spatial spatial spatial low pass spatial spatial O pass filtering low pass low pass low pass low low pass filtering low pass low pas o filtering filtering filtering filtering pass filtering filtering filtering filtering m average T 37.6 37.6 37.1 37.1 37.0 37.0 37.5 37.5 49.8 49.8 maximal ΔT 1.1 0.9 0.8 0.8 1.3 1.1 1.3 1.1 2.3 2.1

30 cr

SD_T 0.21 0.18 0.19 0.15 0.22 0.19 0.26 0.24 0.45 0.42 ro

CV_T (%) 0.5 0.5 0.5 0.4 0.66 0.5 0.7 0.6 0.9 0.9

Table 6 Temperature deterrnination for half a plate with α" calculated per plate.

Experiment 1 2 3 4 5

n=48 without with spatial without with without with without with spatial without with spatial low low pass spatial spatial spatial spatial spatial low pass spatial spatial co

Γ B pass filtering low pass low pass low pass low low pass filtering low pass low pas co filtering filtering filtering filtering pass filtering filtering filterin filtering

average T 37.5 37.5 37.1 37.1 37.0 37.0 37.4 37.4 50.0 50.0 z maximal ΔT 0.8 0.7 0.9 0.7 0.8 0.7 0.9 0.8 1.7 1.5

TO SD_T 0.21 0. 18 0.21 0. 17 0.22 0.19 0.25 0.21 0.39 0.35

CV_T (%) 0.6 0.5 0.6 0.5 0.6 0.5 0.7 0.6 0.8 0.7

Table 7 Temperature determination for half a plate with α" calculated per row.

Table 8 Temperature determination for half a plate with α" calculated per well.

Experiment 1 2 3 4 5

n=48 without with spatial without with without with without with spatial without with spatial low low pass spatial spatial spatial spatial spatial low pass spatial spatial pass filtering low pass low pass low pass low low pass filtering low pass low pas filtering filtering filtering filtering pass filtering filtering filterin filtering

average T 37.5 37.5 37.1 37.1 37.0 37.0 37.4 37.4 50.0 50.0 maximal ΔT 0.8 0.7 0.8 0.7 0.9 0.8 1.0 0.9 1.3 1.0 . SD_τ 0.19 0.17 0.18 0.16 0.22 0.20 0.24 0.21 0.31 0.27

CV_T (%) 0.5 0.5 0.5 0.4 0.6 0.5 0.7 0.6 0.6 0.5

Claims

1. A method for monitoring performance of an incubator module said incubator module being suitable for incubating multiple samples simultaneously at a defined temperature (e.g. for incubating microtitre plates), said incubator module being comprised in an automated system for assaying multiple samples optionally in multiple simultaneous assays e.g. in multiple microtitre plates, said method comprising

1) measuring the absorbance of a temperature dependent solution in a photometer (phot 2) at at least two different wavelengths λl and λ2, wherein the absorbance of the temperature dependent solution is determined at a location different to the location of the incubation and at a moment in time subsequent to the incubation, said temperature dependent solution comprising a buffer with a temperature dependent pH and an acid- base indicator with an absorbance spectrum linked to pH and said wavelengths being selected such that δA/δT is positive for one wavelength and negative for the other, said measuring providing absorbance values A_test._Phot ₂.λi and A_lest,_Ph_<,t λi, wherein A_les,,_phot ₂,λi indicates the absorbance (A) determined for the temperature dependent solution (test), using the photometer (phot 2) at wavelength λl and A_test,_Phot .λ2 indicates the absorbance (A) determined for the temperature dependent solution (test), using the photometer (phot 2) at wavelength λ2, 2) saving data of step 1) in a data file

3) measuring the absorbance of a temperature independent calibration solution for the two wavelengths λl and λ2 of step 1) in the photometer (phot 2), said temperature independent calibration solution exhibiting the same absorbance spectrum as the temperature dependent solution at the defined temperature, said measuring providing absorbance values Acai._Phot zλi and Acai. hot ,λ2, wherein Aoai.phot ₂.λi indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the photometer (phot 2) at wavelength λl and Ac_aι._Ph₀t ₂.λ2 indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the photometer (phot 2) at wavelength λ2 and 4) saving the data of step 3) in a data file

5) calculating the actual temperature T of the temperature dependent solution using the saved data and the formula In (Atesl,_Phot2.λl/Atest.phot 2.λ2) ⁼ 0 "+ β'. T, wherein α"= α'- In (A_c._aι,_photι.λi/A_caι.photι.λ2) + In (A_caι.phot₂.λi/A_caι,_Phot 2x2) with the values of ¹- In (A_caι_jPhotι.λi/A_caι,_Photi,λ₂) being either predetermined or else being obtainable by subjecting the temperature independent calibration solution to absorbance measurement on a further photometer (photl) for the two wavelengths λl and λ2 of step 1) thereby obtaining Ac_aι,_Phot ui and Ac_aι_!phot z wherein "Acai.phot ι.λi indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the further photometer (phot 1) at wavelength λl and ₂ indicates the absorbance (A) determined for the temperature independent calibration solution (cal), using the further photometer (phot 1) at wavelength λ2 and saving the data of step 5 in a data file and subjecting the temperature dependent solution to absorbance and temperature measurement on the further photometer (phot 1) for the two different wavelengths λl and λ2 of step 1) thereby obtaining absorbance values A_tesι._phot ι,λi and A_test.phot ι.λ₂, at known temperatures from which ' and β' can be calculated using the formula In (A_λι/A_λ2) = α' + β'.T and saving the data in a data file

6) comparing the amended temperature or amended temperature distribution of a temperature dependent solution incubated in the module with the defined temperature the module is set to ascertain.

2. A method according to claim 1, wherein calculating means are used to correct the absorbance determined in the photometer (phot 2) with regard to the cooling down effect that occurs between the locations of incubation and absorbance measurement.

3. . A method according to claim 1 or 2 wherein a first microplate filled with temperature independent calibration solution is transported to the photometer (phot 2) and the absorbance is read at the two wavelengths λl and λ2, measurement data is saved in a data file and the microplate is taken to the output module to be removed, a second microplate comprising temperature dependent solution is taken to the incubator module and incubated at the defined temperature according to the setting of the incubator module and after completion of incubation is transported to the photometer (phot 2), where the absorbance is read a number of times at the two wavelengths λl and λ2 at time intervals preferably controlled by a software timer and all data is saved in a data file.

4. A method according to claim 3, wherein the time interval between incubation termination and the first reading in the photometer (phot 2) is as short as possible and is limited by the time required for transport of the microplate from incubator module to the photometer (phot 2), said time interval preferably being less than 45 seconds, suitably being 25-35 seconds.

5. A method according to claim 3 or 4 wherein the time interval between absorbance measurements in the photometer (phot 2) is determined by the length of time required for the measurement of the absorbance at the two wavelengths λl and λ2 and the length of time required to save the data in a file, said time interval in general being longer than 25 seconds, preferably being less than 45 seconds and suitably being 30 seconds.

6. A method according to any of the preceding claims, wherein the temperature determined on the basis of absorbance data from the photometer (phot 2) is corrected for the cooling down that occurs between transport from incubator module to said photometer using calculating means on the basis of a mathematical function fitting the pattern of cooling down such that the regression coefficient R² is equal to or larger than 0.98.

7. A method according to any of the preceding claims, wherein calculating means are employed enabling a function for absorbance against time to be fitted in the least squares sense on the absorbance measurement data obtained in the photometer (phot 2) enabling calculation of actual absorbance in the incubator module at the moment of taking the temperature dependent solution from the incubator module i.e. at time zero thereby enabling subsequent calculation of the actual temperature in the incubator module at time zero.

8. A method according to claim 7 wherein the function is an n^th order, suitably a second order polynomial for the absorbance to time relation like A(t)= a + b . t + c . t², with a, b and c being constants and a being the absorbance at t=0, i.e. at the moment the temperature dependent solution is removed from the incubator module to be transported to the photometer (phot 2).

9. A method according to any of the preceding claims, wherein calculating means are also used for filtering out any high frequency components in the temperature distribution due to imprecision ofthe photometer (phot 2).

10. A method according to claim 9, wherein the filtering out occurs using a two dimensional low pass filter.

11. A method according to any of the preceding claims, wherein the effect of interaction between the multiple samples of e.g. a microtitre plate is taken into account in the calculation of the temperature in the incubator module by fitting a two dimensional linear model in the sense of the least squares to the temperature data of a part of a matrix of samples i.e. two dimensional linear regression e.g. in a square block of nine wells for a microtitre plate.

12. A method according to claim 11 wherein the calculation of the temperature in the incubator module by fitting a linear model in the sense of the least squares to the temperature data of a square block of nine wells i.e. two dimensional linear regression e.g. for a microtitre plate involves expressing the temperature after filtering as e.g.

^τfιlt_D6 = fi . T_C5 + f₂ . T_C6 + f₃ . T_C7 + ft . T_D5 + f₅ . T_D6 +

+ f₆ . T_D7 + f₇ . T_E5 + f₈. T_E6 + f₉. T_E7 (8)

wherein CD and E represent the row denomination of the square block and 5, 6 and 7 represent the column denomination of the square block, with fl through f9 being constants which can be calculated.

13. A method according to claim 12, wherein use is made in the calculation of

Model: T_f,ι_t = a.rownumber + b.columnnumber + c (9)

(a, b and c are constants) and minimising in the sense ofthe least squares results in function F:

F = Σ (T_fllt(welli) - T(welli))² (i = 1 .. 9) (10)

with the following being valid:

- for a well not at the edge ofthe microplate f=l/9

- for a well at the edge of the microplate and not at a corner f =f₅=f₆=2/18, f₇=f₈=f<f 1/18 with

'filtAβ = fi . T_A5 + f₂ . T_A6 + f₃ . T_A7 + . TBS + fs . T_B6 +

+ fj . TB? + f7. Tc5 + f₈ • Tcβ + f^*9 ■ c7 (1 1)

- for a well at the corner

^Tfllt_Al = fl.T_Al + f₂-TA2 + S.Tsi + £,.TA3 + f^*5.T_B2 +

+ f₆.Tcι + f₇.T_B3 + fs.Tc2 + f₉.T_C3 (12)

14. A method according to any of the preceding claims, wherein the buffer system of the temperature dependent solution has a large δpH7δT preferably at least an absolute value of 0,02 pH units/°C.

15. A method according to any of the preceding claims wherein the temperature dependent solution exhibits large δA/δpH values, with large being sufficient to result in at least a precision of 0,1 °C and an accuracy of at least 0,5 °C, preferably 0,3 °C.

16. A method according to any of the preceding claims, wherein the solubility of the indicator in the temperature dependent solution and the temperature independent calibration solution is sufficient to ensure measurable and reliable absorbance values, preferably absorbance values between 0,5-1,5 AU at the defined temperature and wavelengths λl and λ2.

17. A method according to any ofthe preceding claims, wherein the concentration of acid-base indicator in the temperature dependent solution and the temperature independent calibration solutions to be used in photometer (phot 1) and photometer (phot 2) is such that the ratio ofthe length of the lightpaths through the samples in the two photometers is identical to the ratio ofthe concentrations.

18. A kit suitable for carrying out a method according to any of the preceding claims comprising at least a container comprising a temperature dependent solution exhibiting at least an absorbance higher than 0,5 AU at wavelengths λl and λ2 at the defined temperature, said temperature dependent solution comprising a buffer system with a temperature dependent pH and an acid-base indicator with an absorbance spectrum linked to pH and a container comprising at least one temperature independent calibration solution exhibiting an absorbance spectrum identical to that of the temperature dependent solution at the defined temperature, said temperature independent calibration solution preferably comprising a composition as close to that of the temperature dependent solution as possible.

19. A kit according to claim 18, wherein the pH ofthe solutions is a) in the working range of the buffer system of the temperature dependent solution at the defined temperature at which measurement is to take place, said defined temperature being within a desired temperature range, preferably being within the range 20-60 °C, b) in the indication range of the acid-base indicator at the defined temperature at which measurement is to take place, c) such that the absorbances at the two wavelengths λl and λ2 are as close to eachother as possible at a temperature in the middle ofthe specified range. d) preferably as low as possible to prevent C02 absorption.

20. A kit according to claim 19 comprising multiple temperature independent calibration solutions for a number of defined temperatures, wherein the pH of each temperature independent calibration solution is selected such that the absorbance spectrum of the temperature independent solution is equal to the absorbance spectrum of the temperature dependent solution at the defined temperature.

21. A kit according to any of claims 18-20, wherein the temperature dependent buffer comprises an aqueous solution of citrate, tartrate, phtalate, phosphate, Tris(hydroxymethyl)aminomethane, Borax or sodium bicarbonate, preferably of Tris(hydroxymethyl)aminomethane.

22. A kit according to any of claims 18-21, wherein the acid base indicator is Cresol Red and is present in soluble form.

23. A method according to any of claims 18-22, wherein the temperature independent calibration solution comprises HEPES, phosphate and a Cresol Red solution and is preferably further identical to the temperature dependent solution which comprises Tris(hydroxymethyl)aminomethane as buffer and a Cresol Red solution as acid-base indicator.

24. A kit according to any ofthe claims 18-23, wherein said containers further comprise agents against microbial degradation such as azide and preferably cinnamaldehyde and gentamicin sulphate.

25. Use of a method or a kit according to any of the preceding claims for determining the temperature dispersion of the incubator module at a number of different defined temperatures, said method or kit requiring a number of temperature independent calibration solutions equivalent to the number of different temperatures.

26. Use of a method or a kit according to any of claims 1-24 for determining the temperature dispersion of the incubator module in an automated multi-sample system whilst the system is in normal operation.