WO1994011725A1 - A procedure and apparatus for correcting intensity decrease in fluorometric quantitative analysis - Google Patents

Abstract

The invention relates to a procedure serving to correct for intensity decrease, due to optical attenuation, of fluorescence from a liquid sample in connection with fluorometric quantitative analysis, in said procedure the liquid sample being placed in a cuvette, an excitation light beam being directed into said cuvette and the intensity of the fluorescent light produced by the excitation light beam in the liquid sample owing to the substance therein contained being determined and the concentration of the substance contained in the liquid sample being determined on the basis of this intensity, wherein the intensity distribution of the image of the fluorescent liquid sample created in the liquid sample by said excitation light beam is determined and the optical attenuation is determined from the intensity distribution that has been measured.

Description

A PROCEDURE AND APPARATUS FOR CORRECTING INTENSITY DECREASE I_N FLUOROMETRIC QUANTITATIVE ANALYSIS

When a substance is irradiated by optical radiant energy (for simplicity and from now on the author will use the word 'light' as well), a portion of light is reflected, another portion is transmitted and some of the radiant energy will be absorbed by the sample. Reflected light is composed of regularly and diffusely reflected fluxes. Regular reflection takes place, for example from optical components used with the system.

Transmitted light is also a mixture of regu- larly and diffusely transmitted portions. The former means that incident energy enters through the sample without chianging the direction. The latter represents the scattered portion of the light flux.

Light absorption plays the most important role in spectrophotometry as well as in spectrofluorometry. The absorbed amount of the energy of radiation is that portion of light which is not reflected scattered or transmitted. Material absorptance α(λ)⁶ is thus repre¬ sented as follows:

α(λ) = 1.0 - [r(λ) + β(λ)],

where τ(λ) and β(λ) represent the total transmittance and the total reflectance of the sample, respectively. The photometric properties of material are not specific. Many independent factors, such as spectral composition of the radiant energy, the state of polari¬ sation, incident and viewing geometry, the thickness of the sample, temperature etc., exist which affect the measured value of a substance.

Fluorometric properties, conversely are more specific. Only a fraction of substances absorbing radi- ation, fluoresce.

Fluorescence is a subclass of photolumine- scence, which is a further subclass of luminescence, meaning any emission of optical radiation resulting from nonthermal excitation of the energy levels of atoms, molecules, polymers and crystals.

Luminescence is classified according to the mode of excitation. The most typical being biolumines- cence due to a biological process or chemiluminescence caused by a chemical reaction or photoluminescence, which may result after light absorption.

Photoluminescence includes fluorescence as well as phosphorescence, depending on the radiative transition between singlet and triplet states. The former is due to the spin-allowed transition between two states with equal multiplicity (singlet-singlet or triplet-triplet) . The latter is caused by the spin-for¬ bidden transition between two states with different multiplicities (triplet-singlet) or, which is more probable, by intersystem crossing. In most cases mole¬ cules are raised from the ground state S₀ to the singlet state S by absorption of energy. The most probable deexcitation is the one which has the shortest life¬ time. In general, fluorescence lifetimes are much shorter than those of phosphorescence, which, on the other hand, means that fluorescence intensities are higher than those of phosphorescence.

Most of the molecules drop back immediately to the lowest energy levels, by nonradiative transition, so that the commonly observed radiative transition is from S and T. states to the ground state. Because part of the excited energy is dissipated thermally, the emission wavelength represents lower energy that exci¬ tation, meaning that the emission wavelength is longer than the excitation wavelength.

Fluorescence is usually emitted isotropically, enabling it to be detected from different directions. Because the intensity compared with the intensity of the scattered light is usually lower, the favoured viewing direction is the one with the least scattered excitation, unless other considerations are important and scattered light can be blocked with optical fil¬ ters. The standard measurement geometries are right- angle or perpendicular geometry, where the sample is irradiated and viewed in mutually perpendicular direc¬ tions, in-line geometry, where the fluorescence emis- sion is collected over the entire width of the cuvette, and front-surface geometry, where irradiation and view¬ ing occur through the same cuvette face.

The advantage of right-angle geometry is obvi¬ ous. The scattering toward 90° is small, and the effect of the cuvette can be minimised, using well-focused excitation and emission optics. The 90° instruments are used when high sensitivity is needed. A disadvantage of right-angle viewing is that the linearity is reduced at high concentrations, because the sample absorbs either an appreciable amount of excitation or emitting light or both before it reaches the detector. This so-called 'inner filter' effect can be reduced by using front- surface geometry.

In 1937¹³ Bowen and Sawtell recognised the value of front-surface geometry for studying opaque and highly absorbing samples.

Bryant et al¹⁴ described an angular separation of excitation and emission beams, using frontal geome¬ try with standard microcell. During the last decades computers have been used as a standard part of an instrument. Ritter et al. describe a microcomputer-based fluorometer system, designed for spectral correction and quantum yield determinations.¹⁵ Puyn and Park report on an inexpensive subnanosecond fluorometer using a microcomputer.¹

Fluorescence quenching refers to a process which decreases the fluorescence intensity of a given substance, when environmental changes take place. Be¬ cause of the reduced energy transfer by a reduction in the number of molecular collisions, fluorescence can be increased by increasing the viscosity of the medium. The opposite happens when the temperature is increased; the intensity is decreased because of the increased molecular collisions. Oxygen is also a well-known strong 'excited state quencher'. The reduction of fluo¬ rescence by a competing deactivating process resulting from the specific interaction between a fluorophore and another substance present in the system, represents

'collisional and complex formation quenching'. 20'21 High concentration of a fluorophore can cause quenching. High optical densities or turbidity may result in de- creased fluorescence intensities. This type of quench¬ ing is referred to as concentration quenching or rather optical quenching.

Absorption errors in molecular fluorescence spectrometry have been discussed in several books and review articles.^20"25 Terms such as "inner filter effect", "pre-filter effect", "post-filter effect", "self-ab¬ sorption" and "reabsorption" have been used to describe these errors, which can be classified as absorption of the exciting or primary radiation or absorption of the fluorescence or secondary radiation. To be more specif¬ ic, the pre-filter effect is due to the absorption of exciting radiation and the post-filter effect is corre¬ spondingly due to the absorption of emitted radiation. "Reemission" can take place if the emission and absorp- tion wavelengths have some overlap, so that the emitted fluorescence radiation undergoes absorption in the solution, producing fluorescence emission.

Inner filter effects produce not only spectral distortions but also a spatial distortion of the exter- nal fluorescence distribution for concentrated samples. The resulting external fluorescence anisotropy is one of the key factors in choosing the proper viewing geom- etry for a given sample.

Because absorption of radiant energy is the basis for photoluminescence, the inner-filter effect has considerable influence in photoluminescence spec- trophotometry. In routine work investigators using fluorescence methods are usually quite aware (or un¬ aware) of this absorption problem but, in general, have consistently either deemed it negligible, assumed it to be constant from sample to sample, or neglected it entirely. The basic rule is to keep the fluorophore concentration low enough, so that the absorption of exciting radiation is low as well; however this method is not really a universal solution in overcoming the inner-filter problem. Consider the case where the fluorophore is initially present at a very low concentration level, but the solution absorbance is high because of the presence of other absorbing species. Continued dilution of such a system would not increase the accuracy of a determination.

In addition, dilution may cause changes in conformation, bonding, solvation, and the degree of association as well as other chemical events that may alter the absorption-fluorescence processes and there- by, introduce large unknown errors into the measure¬ ments. On the other hand, failure to dilute such a solution will introduce serious error into a fluoromet- ric determination if the fluorescence measurements are not corrected for the primary absorption processes. Mode and Sisson²⁷ formulated and analytical expression for correcting sample-cell-related losses due to the differences in the optical densities for

28 front-surface geometry. Holland et al. have derived a correction of right-angle fluorescence measurements for the absorption of excitation radiation. The correspond¬ ing correction for the fluorescence radiation has been presented by Christmann et al..²⁹ Leese and Wehry have derived a correction for inner-filter effects in fluo¬ rescence quenching measurements via both right-angle and front-surface illumination.³⁰

The basic rule for disregarding the inner- filter effect is "for very dilute solutions in which not over 2% of the total excitation energy is ab¬ sorbed..." Thus, if the substance absorption for ei¬ ther exciting or emitting radiation or both is high, inner-filter corrections must be applied. This is most important especially with perpendicular and in-line detection geometries, because high concentration of an analyte may lead to a low signal, misleading the inves¬ tigator.

The common method for inner-filter correction is based on the spectrophotometric measurement of the optical density of a sample, both exciting and emitting wavelengths. However, this procedure suffers from the following problems: many fluorescents samples undergo spectral changes during the delay between the fluores- cence and absorbance measurements; analysis time is increased because of the need for additional measure¬ ments; care must be taken that the spectral bandwidths of excitation and emission are closely duplicated by the spectrophotometer. In order to remove the substan- tial errors that can arise from the use of two differ¬ ent instrument systems for the absorption and the fluo¬ rescence measurements, some instruments for simulta¬ neous absorption and fluorescence measurements have been developed. Another way, presented by Novak, can be used to carry out inner-filter correction with one unit as well.³² In addition to this, direct measurement of the sample absorbance is not required; a spectrofluorometer is the only measurement device needed. The method has been named 'the cell-shift method' because it involves shifting the sample cell in a spectrofluorometer to change the effective pathlength through which the ex-

SUBSTITUTE SHEET citing and fluorescence radiation must travel.

The disadvantage of the cell-shift method is the requirement to move the cell. To be able to correct both pre- and post-filter effects, the cell must be moved in two directions, thus limiting the practical applicability of the method. It can only be used with 'one sample instruments' and not with MTP-systems.

The author presents a new method for perform¬ ing inner-filter correction. The method is based on processing the image of the fluorescing object. The processing dependends on the geometry and component used with the system. When a convolution between a proper slit and the projected image of the fluorescein object is performed, one is able to carry out total correction for optical quenching.

To refer to these different inner-filter cor¬ rection methods, some abbreviations will be used in this thesis. The common method, based on measurement of optical densities with a spectrophotometer, will be called a 'Photometric Inner-Filter Correction' and abbreviated further on by PIFC. With the Cell Shift (CS) and the image Convolution (IC), correction is carried out by methods based on the direct measurement of fluorescence as a function of position along the fluorescing object; these methods are classified as the 'Fluorometric Inner-Filter Corrections' (FIFC) . There is a major difference between the CS and the IC meth¬ ods; in the former, the whole sample cell must be moved to calculate the correction. In the latter, only the image is processed. If an array-type detector is used, no moving parts are required. In this study an inexpen¬ sive solution has been used. A moving slit with a pho- tomultiplier tube performs image convolution, from which the correction can be made. This method is called an 'Image Convolution Inner-Filter Correction' (ICIFC). The relationship between the linear spectral absorption coefficient a. , the Naperian molar spectral

T SHEET absorption coefficient e^ and the molar spectral ab¬ sorption coefficient e is as follows

a_λ = eⁿ _χ c= 2.30 e_χc

The transmitted optical density or simply absorbance is defined as

A_λ = log(l/τ_λ) = ε^ch

The scattering associated to nephelometry or turbidimetry involves no net loss in radiat power (in contrast to Raman spectroscopy), only the direction of propagation is affected. The scattered intensity dis¬ tribution at any angle is a complex relationship be¬ tween several parameters such as number of particles, size, shape, relative refractive index and wavelength of the radiation. An exact theoretical analysis can be made if all of the affecting parameters are known. Theoretical treatment is feasible but seldom applied to specific analytical problems because of its complexity. In practical work most nephelometric and turbidimetric procedures are highly empirical. The quantitative relationship between fluo¬ rescence intensity and concentration may be derived from Beer's law. The flux of light transmitted through the sample in a cylindrical cuvette, the contents of which is excited from beneath using a collimated exci- tation beam, forms the geometrical basics for vertical photometry and fluorometry.

Linearity encountered with fluorescence versus concentration is usually two or more decades; limiting by the blank fluorescence and quenching. Especially the minimum detectable quantity of an analyte is generally limited by the magnitude of the blank, which is further dependent on the instrumental as well as the sample related factors . The high concentration end of the dynamic range, from physical stand point of view, is limited by the fluorometer geometry used.

It is assumed that the internal flux is con- sidered quashiparallel in the neighbourhood of the beam waist so formed and the figure of the slit is focused inside the sample. It is also assumed that the emitted fluorescence is isotropic; polarisation effects are negligible. Finally we assume that the substance con- centration in a cuvette is uniform, so that c(z) is constant.

We can see that the internal flux of the exit¬ ing radiation, denoted as φ (z)Δλ_χ, at a distance z from the bottom of the cuvette is

Φ_lx(z)Δλ_χ = Φ. Δλ τ exp(-eⁿ. c(z)dz)

where Δλ_χ is the excitation bandwidth e"_λ the Naperian attenuation coefficient and τ_χ is the spectral transmit- tance of the bottom of the cuvette for the exciting radiation.

Thus the total external fluorescence intensity

L_m(λχ)= C_m(n,a)f² dL_ima_x,z)e-£ ^{b J}*ι where

C_m(n,α) = τ_ιncosα/[n(λ_m)(n²(λ_ιιι)-sinα)]

27

The bispectral photon convertance B_m(λ_χ) (=4πL_m(λ_χ)/ φ. (z)Δλ ) of the sample is

J'^*~l, where

TE SHEET B mx ( ) ' = r x C n ( ^v n , α ) ' eⁿ x y^J m

When comparing the in-line and front-surface geometries it was assumed that the boundary corrections are equal. This is really not the case with MTP cuvet¬ tes, because the emitted flux passes through the bottom in frontal geometry, but through the free liquid-air surface in in-line geometry. The boundary correction factor, of arbitrary viewing angles of γ(0 < γ < 90°), will vary along γ.

To formulate the bispectral convertance for the front surface detection, we assume that the detec¬ tion system is focused inside the sample. And the pro¬ jection Δh of the viewed part of the excited volume is detected using angle γ. Because the self absorption path length -and the distance of the viewed emitted source varies along the width of the beam, we integrate over the z and x directions, as follows

The inner integration is performed along the z-axis where z is a function of x as follows

φ(x) = -tan(90-α)x+C

where 0 < α < 90°. The constant C is the intersection between the line φ(x) and the z-axis, leading to values C^d-Δh/2 and C₂=d+Δh/2 when x=0, as can be seen from the following figure 2.

The path length b(z) of the emitted light is z/cosα. (B-4.14)

BSTITUTE SHEET where K = eⁿx-eⁿre/cosα.

The outer integration is done along the z-axis from -u_χ/2 to u_χ/2, where u_χ is the width of the exciting beam inside the sample, following to The total emitted luminescence is

Here it is assumed that the exciting beam is coherent and the width of the beam inside the sample to be viewed is independent of z. In reality this is not quite true. Aberration and dispersion of the optics may disturb this idealistic assumption, more or less.

The unit presented in this thesis is able to carry out photometric inner filter correction (PIFC) , The absorbance and the fluorescence can be measured at the same time, using the same excitation filter. Post filter correction can be made, but a filter, similar than the emission filter used, must be used to measure the optical density for the emitting wavelength.

The correction factor C is defined by which an observed fluorescence signal F at a specified wave¬ length can be corrected to the fluorescence signal F₀ expected in the absence of the inner filter effect. It is assumed that F_Q is a linear function of concentration of the fluorophore. The correction factor, as expressed in mathematical terms.

C=F₀/F

is dependent on the geometry and the method used.

When sample cell is excited through the bottom of the cuvette, the intensity at distance z from the bottom is

If the emitted fluorescence is taken to vary linearly with L_λ(z) and F. is similarly related to L..(0), then F₀ can be calculated

The correction factor C can now be calculated to be

sinh(KΔh/2) sinh(Ku/2)

where

K =Ktan(90°-α) .

Because dC/dd=C/k is proportional to e^Kd, the relative error of the correction factor increases the deeper the correction is calculated. The higher the optical density, the bigger is the relative error of correction.

Even this method is basically inert to sedi- mentation or evaporation, author want to point out that the dispensing inaccuracies in practical work as well as the concentration dependent nonflat free surface of the liquid may cause serious absorption errors. Large amounts of proteins or detergents may increase the surface tension of the liquid and cause the edges of the sample to rise on the sidewalls.

When a slit is scanned across the image of a fluorescing object, the intensity as a function of slit position represents convolution between the image func- tion and slit function. Convolution of two real valued functions φ( ) and Ψ(C), for which we use the shorthand notation φ( ) ♦ Ψ(C),

SUBSTITUTE SHEET Ω ( C ) =Φ ( t ) ♦ 7 ( C )

The function φ(C) represents the image of the fluorescein object and the function 7(C) describes the emission slit.

Because the fluorescein object exists within certain limits, the lower limit being the air-cuvette bottom interface and the upper limit being the liquid- air interface of the sample, we use the rect function to define the fluorescing image function F ( ζ ) :

F(C)=rect[(C-∑/2)/Σ]φ(C)

The parameter Σ represents the projected height of the fluorescing object and actually defines the existence range for φ(C)

∑=(u +h tanα)cosγ

where h is the height of the fluorescein object and u is the width of the exciting beam.

Depending on the type of slit used, the cor¬ responding function Ψ(C) varies. The slit being used with our prototype is a rectangular aperture located at the image plane. The corresponding slit function is thus defined by rect function:

Ψ(C)=rect[( -ζ₀)/w ]

where w. is the 'height' or the size vi. of the slit in the C-direction. _{By US}j._ng these relationships, we will write the convolution intergal as follows:

Φ(ζ ♦ Ψ(ζ)= j F(ω)rect [(ζ-ω)/w_ζ]dω

SUBSTITUTE SHEET When the size w. of the slit is made small enough, the rect function approaches another common function, known as delta function. We will employ the widely used symbol δ(C) to denote this impulse func- tion. The convolution of a delta function with any other function merely reproduces the other function. Thus, if Ψ( )=δ(C), then

φ(C) ♦ δ(C) = Φ(C)

This is an important result; if the size w. can be made small enough, the convolution approaches func¬ tion F( ) and the inner-filter correction can easily be extrapolated from that. Based on aforesaid the external fluorescence distribution φ(C) is of the form F_Qexp(-kζ), where F₀ is

In actual practice, it is reasonable to study the case, ' when the size wζ.<<∑ and uX is made small. The corresponding result of such a convolution is presented in the following Figure 5. The maximum value of the convolution occurs at w./2. Between w./2<C<∑^_w./2, the attenuation constant k. of the convoluted image can be estimated from the con¬ volution integral, for example by plotting the convolu¬ tion in a semi-logarithmic scale and finding the slope of the line.

In practice, the convolution integral is pro¬ duced by moving the slit in steps of ΔC, and recording the corresponding signal. The value of F(£) within the exponential region, and from ζ_ to C.+Δw, can be calcu- lated from

F_Δ:=exp [ k_c ( C_{i+ 1} - ζ ) ] which is independent of Δw. This means that the ratio between readings taken from successive slit positions is independent of the size w. of the slit, when the slit increment is constant. If the step size Δζ is made equal to size w. of the slit, we can see that the logarithm of the F., F_Δ=F_Λl(ΔC=w.) , is directly proportional to the attenua¬ tion constant k and slit size w..

LN(F_Δ)=K_cw_c

The maximum value F₀ of the image function can be calculated from the equation

(B-6.15)

The maximum value of convolution Ω(C) can be found from experimental data by extrapolating detected fluorescence to w./2. F₀ can then be calculated from the following equation

_Fa _ k_e MAX(Φ(Q * Ψ(ζ)) l._e-k.w

The accuracy of the extrapolation is indepen¬ dent of the size w. of the slit, but if the slit is made narrow enough, then F_Q approaches MAX(φ(C) ♦ 7(C)), and the extrapolated value is F₀. this can be explained mathematically as follows. When 7(ζ)=δ(C), convolution Ω(ζ)*φ(ζ). The MAX(φ( ) ♦ 7(C)) is the maximum value of φ(C), e.g. MAX(φ(C)). Now, because φ(C) = ₀ exp(-k_cC), the maximum of φ(C) is found at C=0. φ(C) is reduced to a constant F_Q. The MAX(φ(C) ♦ 7(C)) is thus F_Q.

The relationship between the image attenuation

B TITUTE SHEET coefficient K. (=k_c) and the Naperian attenuation coef¬ ficients, eⁿ for the exciting radiation and eⁿ for the emitting radiation is as follows

kς. = (^veⁿx+eⁿ ~/cosα) /tanα cosy '

This relationship is of great importance. Since the attenuation constant k. include both pre- and post-filter components, it is a measure of total opti- cal attenuation of radiation interacting with the sam¬ ple.

The correction factor

for ICIFC is of the form

C = exp(kh)

However, there is no reason for using correc¬ tion factor C with ICIFC, because the process used with the method directly gives an estimate for inner-filter free signal.

In practice, one is not able to change the size of the excitation beam, but changing the size of the emission slit have practical value. The relation¬ ship between the convoluted fluorescence, measured with different slits is as follows:

(B-620)

. , ,κΔh. -_ F(Δh.) ^Slnh —T^')

The image function φ(C) will be projected towards the angle of viewing, as illustrated in Figure 6. The fluorescence of region IV is caused by the sample, alone. It is also an area, in which the change in emitted intensity dF/dx is constant as a function of x or ζ , if the inner-filter effect can be neglected. The last region is also due to the sample fluorescence, but the intensity decreases when x is increased.

The height Σ of the fluorescein sample image at focus plane, in the case of unit magnification (equation B-6.5) depends on the height h of the sample (=volume of the sample) and size u_χ of the exciting beam. Sample fluorescence exists in regions II to V, but only in regions IV and V is it the only source of photoemis- sion. Further, it is clear that in the absence of opti¬ cal quenching, the fluorescence intensity is constant along x, only within region IV. This portion of the sample fluorescence will be referred to as the 'Cons¬ tant Emission Region' , abbreviated as CER. The corr- esponding measured constant emission region is always smaller than CER., because of the finite size w. of the emission slit. CER. of the convolution function Ω(C) is abbreviated as CCER.. The relationship between CER. and CCER. is (B-7.3)

CER. = CCER. +w.

The CCER. is decreased, the wider the excita¬ tion beam size u and the emission slit size w. as shown in the following Figure 7.

From a practical standpoint, the thickness of the bottom of the cuvette should be as small as possi¬ ble, to maximise the relative amount of sample fluo¬ rescence toward the direction of viewing. On the other hand, when the size u_χ of the exciting beam is made smaller, the ratio R between CER- and the total sample fluorescence is increased as seen from the following equation:

R=(htanα-u_χ)/(htanα-u_χ)

If the beam is made very narrow, some dis-

SHEET advantages may exist. Because many fluorochromes, when subjected to a strong excitation light, are decomposed, focusing the light power of the lamp to a very narrow beam should be avoided. If the intensity is kept con- stant, the fluorescence signal will be decreased, lead¬ ing to a lower S/N. Finally, the integration behaviour of the exciting beam is also reduced.

Understandably, the image of the object, even in the most ideal case, is a rather complicated func- tion φ(C)> The complexity is further increased when the image transformation as well as the reflection from the cuvette sidewalls is taken into account.

The dimensions of the cuvette play a rather important role with this type of geometrical arrange- ment, especially from the reflecting point of view.

The reflection overlap will limit the exciting beam, focusing very close to the reflecting sidewall of the cuvette. The small inner diameter of the cuvette will limit the whole fluorescing sample object being seen by the detector, if excitation is focused through the centre of the cuvette.

The value of the sidewall shift will determine the reflection overlap. The smaller Δy is, the larger is Δx. The value of Δy can be increased, by making the sidewall of the cuvette thicker. If the outer sidewall is made to widen downward, Δy is also increases, de¬ creasing the RO. The most effective way of eliminating emission reflection toward the direction of viewing, is to make the outer sidewall nonreflecting, for example by painting it black.

In practice, the most preferred focusing area for the exciting beam is the center of the cuvette. This will allow absorbance and fluorescence measure¬ ments at the same time, without repositioning the exci- tation beam between fluorescence and absorbance mea¬ surements .

One of the limiting factors is the edge of the cuvette sidewall, which cuts the upper end of the pro¬ jected object.

The measurement system has been design to be a part of a complete instrument, which in addition to the measurement unit will include dispensing, shaking and incubation functions. Also an instrument, based on the measurement unit alone, will offer a potential alterna¬ tive. The modular design as well as a compact size sup¬ port both of these goals. The complete instrument has been designed to be controlled by a PC-type computer via fast serial channel (with a special card placed into the computer) or using the RS-line, which offers portability between different commercial computers. The computer will communicate with a master processor, which is further controlling different sub- processors. The computer can also communicate directly with the controllers as drawn in the following figure 8. The controller is responsible of the operation related to the measurement unit. This includes control¬ ling of the lamp, detector electronics and excitation and emission filter motors. The emission slit was oper¬ ated manually with the prototype, but should be con- trolled by the microprocessor as well.

The multifunction features as well as the ICIFC is enabled by the design of the optical kernel. It is the mechano-optical unit with lamp, filtermono- chromators and detectors. The design of this unit is in a key role. To be able to understand the reasons let to the geometrical construction used, some explanations are needed.

Because samples to be measured may contain impurities, quenching substrates and even scattering particles and because the inline detection nonlinearity is known to be a problem with samples of high concen¬ tration, we have decided to use the frontal viewing

SUBSTITUTE SHEET geometry .

Light can be directed through the free surface of the sample (above) or through the bottom of the sample cuvette (below) . From ICIFC point of view viewing through the free sample surface will not work. From that respect we state that the incident light must be directed below the sample, through the bottom of the cuvette. And the emitted light must be collected below the sample. To be able to measure absorbances with this geometry, it is clear that the incident light must be directed normally through the bottom of the cuvette.

The emitted light can be collected just nor¬ mally below the sample using lens system or fibre bun- die. Both of these techniques can be used with this instrument. When estimating the compulsory of the op¬ tics needed for both of these solutions, it seems that the traditional lens system optics will be simpler and cheaper. But, the most important is that the ICIFC will require high quality image, which can't be produced using a fibre bundle. Thus the selected solution is based on using discrete optical components; lenses, mirrors and slits.

To be able to perform ICIFC, the projected image of the fluorescing sample must be made. This will require a small angle between the exciting light path and viewing direction. To be more exact, we finally state that "the incident light beam will be directed below the sample normally through the bottom of the cuvette. The emitted light will be collected below the sample, viewed by a small angle using the frontal geom¬ etry and the transmitted flux will be detected above the sample".

The unit is constructed so, that the ease of manufacturing as well as servicing will be considered. The modular structure of the kernel will support test¬ ing of the different parts of the unit. The optical

SUBSTITUTE SHEET arrangement of the kernel is shown in figure 8 princi¬ pally.

The Xenon flash lamp is located left in the figure. Lamp slit (LS) is followed filter monochoroma- tor with the excitation lens, lens aperture and excita¬ tion filter (XF) . Beam splitter (BS) will direct about 4% of the incident light onto the reference detector (RD). Most of the light is directed upward, vertically through the bottom of the cuvette (C) . The light will pass through the sample, and will be detected using the photodetector (PD); lens in front of the photodiode.

The isotropically emitted light is viewed, through the bottom of the cuvette. The emission mirror (M) will reflect the radiation horizontally, through the emission filter (MF) . Emission lens focuses the image, of the fluorescein sample object, on the emis¬ sion slit (ES), followed by the photomultiplier tube (PMT) .

In general, there are some critical things related to the order and relative distances between different components.

Usually the interference filters are located between two plane-convex lenses with convex surfaces facing, to quarantee coherent flux through the filter, both transmittance and reflectance spectra shift to shorter wavelengths as filters are tilted from normal to oblique incidence.

The unit will use three different detectors. The reference and photometer channels uses semiconduc- tor photodiodes, but the fluorescence channel is con¬ structed with a photomultiplier tube (PMT) .

Because the peak intensity of the Xenon flash lamp is high, the sensitivity of a semiconductor diode is sufficient to be used as a photometer detector. The advantage of a photo diode is its wide dynamic range, small size and low prize, compared to a photomultiplier tube. The prototype was equipped with a high quality

BSTITUTE SHEET photo diode.

The reference channel was also equipped using a photo diode. The quality of the reference detector need not be as high as with the photometer detector, but linearity must be good. The present unit was assem¬ bled with an identical photodiode also in reference channel.

Because of the low emission-collection effi¬ ciency of the used fluorometer geometry, the only rea- sonable component to be used as a photosensitive ele¬ ment, is the photomultiplier tube (PMT) . The PMT should have low dark current, wide dynamic range and moderate gain. Because the electronics was designed to support also photon counting (PC) at very low light levels, PMT should be suitable for photon counting as well. 1P21 PMT was used with the prototype, because of its rather low prize, good DC characteristics and suitability for photon counting ^{" 4}.

The output of photoelectrons from the photo- cathode varies depending upon, where incident light enters. The uniformity of photoelectron output is an important factor affecting the quality of the ICIFC. The uniformity of the side-on type PMT's are rather bad, and must be taken into account. To be able to minimise the uniformity error, the image, and especially the C~direction of it, should be adjusted within the most constant sensitivity area of the photocathode. The image formed, must be made as small as possible, relative to the size of the active area of the PMT photocathode.

The virtual image behind the emission slit can be compressed with additional optics. A convex lens, cylinder lens or any lens or mirror system can be used. A fibre optic homogeniser, with proper optics, will give dispersionless spot on the PMT. The disadvantage of any system is the increased length or size, complex¬ ity and additional cost of the emission channel. The light intensity is also reduced, except with a proper mirror based solution.

With the present prototype, the PMT was placed as close to the slit as possible. No additional compo- nents were used.

Because there exists may factors affecting the quality and position of the image, such as wavelength of light, emission slit size and the very PMT used, uniformity must be corrected as a function of f. A typical example of the uniformity calibration curve is presented in the following figure 10.

The data in figure 10 was a result of follow¬ ing experiment. After adjusting the image to the center of the detector, the detector spatial sensitivity was tested. A low concentration fluorophore was used (fluo¬ rescein 5^*10^"7 mg/ml) to eliminate the attenuation of the exciting or emitting radiation. The sample cell was a class plate bottom polystyrene-cylinder, the diameter of which was 10 mm. The thickness of the bottom was 1.0 mm. The test tube was coated with a black ink to elimi¬ nate the reflected light from passing into the direc¬ tion of viewing. The test cylinder filled with fluores¬ cein was positioned so that the exciting beam passed the bottom of the cylinder 4 mm from the 'reflecting wall' of the cylinder. The test tube was adjusted 2 mm lower than the bottom of the cuvette in the MTP, to ensure that the slit scanning area covers only the sample fluorescence. The experimental convolution was performed scanning the slit between positions 0.5-2.5 mm. in steps of 0.5 mm. The focus of the exciting beam was also varied, to ensure that this will not affect the fluorescein image formed on the emission slit.

The PIFC correction can be performed using a separate photometer to measure the absorbance values for the exciting and emitting wavelengths, if similar filters are available for the separate fluorometer and photometer used. It is clear that the advantage of PIFC

SUBSTITUTE SHEET done with the present concept, is that the bandwidth errors are eliminated, because similar filters can be used for measuring excitation and emission absorbances. If several samples should be corrected, the manual work to measure all the samples with different instruments is just too big. The advantage of the present system is obvious.

There are some error sources of the PIFC done with the present concept, which must be considered. When absorbances are measured, especially for the exci- tating light, a strongly fluorescing sample will emit light, which will decrease the- recorded absorbances. Also a scattering of light may lead to lowered absor¬ bance values. One practical problem with any measure- ment system based on the principles of vertical photom¬ etry, is the error due to the dispensing inaccuracy, which decreases the system accuracy substantially. Evaporation and concentration dependent surface defor¬ mation, especially with bad optical design will in- crease the experimental error of the test.

These problems can be avoided using the ICIFC method. The sample volume and free liquid surface re¬ lated problems of the vertical principles can be avoid¬ ed, by masking the 'close to surface' values out from the calculations. There is also a clear system through¬ put related advantage of ICIFC compared to PIFC.

Absorbances, needed to calculate PIFC should be measured both for the exciting as well as for the emitting light. And to avoid off-axis refraction from the concave free surface of the liquid, repositioning for the plate should be made between the fluorescence and absorbance measurements, if special cuvettes are not used or viewing depth limited.

The practical measurement cycle, to be able to carry out PIFC, with the present system, might be as follows: 1 Select excitation and emission filter pair

SUBSTITUTE SHEET 2 Adjust plate offset for fluorometry

3 Measure fluorescence values of the samples

4 Adjust plate offset for photometry

5 Measure absorbances of the samples for the exciting light

6 Select emission filter for excitation

7 Measure absorbances of the samples for the emitting light.

The corresponding measurement cycle for ICIFC is as simple as this:

1 Select excitation and emission filter pair

2 Adjust plate offset for fluorometry 3 Measure fluorescence values of the wells, with dif¬ ferent emission slit positions.

The slit scanning can be made so fast that the measurement cycle is much shorter that with the PIFC. If reduced dynamic range is satisfactory, linear CCD- matrix as a detector can be used instead of the mechan¬ ical slit, decreasing the measurement time to minimum. In the following the invention is described in more detail by way of reference to the following pic- tures, in which

Figure 1 illustrates coordinates in a cylindrical cu¬ vette, excited through the bottom, with a rectangular beam, focused inside the sample. Viewed through the bottom at an angle of γ. Figure 2 illustrates coordinates in a cylindrical cu¬ vette excited through the bottom, with a rectangular beam.

Figure 3a illustrates inner filter correction factor C as a function of z [cm] with A_χ=A_m={0.05, 0.25, 0.5, 0.75, 1.0} , α=22°, γ=30°, Δh=0.14 in a semilogarithmic scale. Figure 3b illustrates relationship between h and Σ. Figure 4 illustrates convolution functions in ICIFC. Figure 5 illustrates normalised convolution exp(k£)re- ct[(C-∑/2)/Σ]^»rect[(C-C₀)/w] with w=0.25, 0.5, 1 and 1; k=l. Figure 6 illustrates projection of fluorescein object to the direction of viewing.

Figure 7 illustrates CCERS as a function of beam width u with emission slit heights w=w of 0,5 mm, 1 mm and 1.5 mm. Figure 8 illustrates layout of the measurement system. The optical kernel is framed with dashed line; exclud¬ ing the photometer detector, which is located above the MTP.

Figure 9 illustrates the relative arrangement of the components in the optical kernel.

Figure 10 illustrates normalised sensitivity of the PMT photocathode with different emission slit positions; 0.5 mm to 2.5 mm, in steps of 0.5 mm. Focus of the excitation beam was varied, by changing the lamplens distance 10 mm; from position 2 mm to 12 mm.

Figure 11 illustrates QS linearity with emission slit position 1.5 mm; arbitrary fluorescence reading as a function of sample concentration [μg/ml]. Figure 12 illustrates fluorescence of QS as a function of sample concentration [μg/ml], with different emis¬ sion slit positions from 2.5 mm to 0.5 mm (bottom to top) . The ICIFC extrapolated fluorescence values are the most top dots in the graph.

Figure 13 illustrates fluorescence as a function of logarithm of concentration [μmol/1], using emission slit size 0.25 mm. The top dots represent ICIFC cor¬ rected result, calculated based on the readings for different slit positions 0.5-2.5 mm (top to bottom). Figure 14 illustrates fluorescence as a function of logarithm of concentration [μmol/1], using emission slit size 0.5 mm. The top dots represent ICIFC correct¬ ed result, calculated based on the readings for differ- ent slit positions 0.5-2.5 mm (top to bottom). Figure 15 illustrates the absorbance of fluorescein for the exciting wavelength 492 nm, as a function of con¬ centration. Figure 16 illustrates the correction factor C of PIFC as a function of fluorescein concentration [mmol/1], with different viewing depths; 0.17 cm, 0.29 cm, 0.43 cm, 0.57 and 0.72 cm, from bottom to top. Figure 17 illustrates the ^'correction factor C for ICIFC^' as a function of fluorescein concentration [mmol/1], with different viewing depths; 0.17 cm, 0.29 cm, 0.43 cm, 0.57 cm and 0.72 cm from bottom to top. Figure 18 illustrates the ratio between the correction factors of PIFC and ICIFC. Parameters are sample con- centration [μmol/1] and viewing depth d from the bottom of the cuvette [cm].

Figure 19 illustrates fluorescence of NADH as a func¬ tion of NADH concentration [μmol/1], with four differ¬ ent bilirubin concentrations (0, 28/25, 99/25, 305/25 μmol/1 from top to bottom) . The emission slit position was 2.5 mm (=7mm viewing depth) in figure A and 0.5 mm (=2 mm viewing depth) in figure B. The figure C repre¬ sents ICIFC extrapolated results. Figure 20 illustrates absorbance scanned across the bottom of the cuvette for bilirubin samples concentra¬ tions of 0 (lower) and 305/25 μg/ml. The dimension of abscissa is[mm].

Figure 21 illustrates log-log plot of the recorded absorbances as a function of PNP solution, distilled in water.

EXAMPLE 1

The limit of detection (LOD) or the minimum detectable concentration (MDC) or weight (MDW) and the linear range was measured. The LOD was found to be 20 ppb. The lower limit of linearity is limited by the LOD and the upper limit by the pre filter effect of the

SUBSTITUTE SHEET Quinine Sulphate (QS). When ICIFC is applied, the high¬ er end of linearity can be as high as 1000 μg/ml.

Sensitivity The lamp intensity or rather the excitation intensity, the emission collection aperture and emis¬ sion optics, the detector sensitivity and electronic noise turned out to be the most important factors af¬ fecting the sensitivity of the instrument. Because yet being a noncommercial system, author didn't try to maximise the sensitivity, but rather wants to get to know the level of sensitivity for this type of an in¬ strument. The tests were performed using such instru¬ mental setup which have been used also for the other experiments presented in this thesis.

The height u_χ of the lamp slit was 0.5 mm, and the width u 2 mm, leading to a low intensity (but high quality) excitation. The emission slit size w. was 0.5 mm and w 5 mm. In other terms, only a one sixth of the whole fluorescein image was viewed. The total integra¬ tion time of one measurements was 1 s according to the ASTM standard used E578-83 and E579-83 (approved at 1988)⁴⁵. The emission slit position was fixed at 1.5 mm, corresponding 0.4 cm viewing depth. The peak wavelength of the excitation filter used is 325 nm. The bandwidth of that filter is about 40 nm. The peak wavelength of the emission filter is 470 nm, with a FWHM of 20 nm. The sensitivity test was performed according to the ASTM standard E579-83, were the same cuvette was re- filled ten times with blank (H₂S0₄) and sample (0.1 μg/ml of QS in H₂S0₄) . The rms-noise was calculated from the sample data and the LOD was calculated from the formula

LOD=0.1^*(S-B)^"1*3^*rms

where S and B are the average of sample and blank read-

^S ings correspondingly.

The LOD was found to be 20 ppb or 20 ng/ml of MDC. Because the system will detect a part of the sam¬ ple which is limited by the emission slit size and position; MDW is more or less dependent on the slit position. F.ex, if sample volume is 100 μl, slit will be adjusted at position 0.5 mm, leading to 2 ng of MDW, if LOD 20 ng/ml is used.

When comparing these results to other instru- ments, which have the sensitivity of QS in their speci- fications 3'49, it is seen that the sensitivity of the system with used setup is not high, about two orders of magnitude lower that the commercial grating, 90° sys¬ tems, as expected. It must be noted that the exact definition of sensitivity used for the other fluorome- ters compared is not necessarily the same; the MDC is often given as the concentration of the sample giving a signal-to-noise ratio of one, which in fact leads to higher sensitivities than the LOD used here.

Linearity

The linearity was tested according to the ASTM standard E 578-83 (Reapproved 1988)⁴⁴, which covers a procedure for evaluating the limits of linearity of fluorescence-measuring system under operating condi¬ tions. The test is based on using the quinine sulphate dihydrate (QS) in sulphuric acid (H₂S0₄) as standard solution. 0.1N sulphuric acid was used to prepare step¬ wise dilution; solutions with concentrations of 10 , 10², 10¹, 10°, 10^"1 and 10^"2 μg/ml: The concentration of 10^"3 μg/ml was not prepared because of the LOD of the testing setup. A Biohit MTP strip cuvettes were used as the sample cells. One cuvette was reserved for blank solution. The sample volume used was 300 μl. The excitation filter was 320 nm, with FWHM of

40 nm and the emission filter 470 nm with FWHM of 20 nm. The maximum transmittance being 40% and 70% corre- spondingly. The lamp slit was 0.5 mm^* 2 mm and the emission slit 0.5 mm^* 5 mm. To be able to calculate inner filter correction (ICIFC), emission slit was scanned using five different positions from 0.5 mm to 2.5 mm, in steps of 0.5 mm.

The upper limit of linearity is the point (concentration) at which the upper end of the curve deviates more than 5 % of the signal from the straight line defined by the center region of the curve. The upper limit of linearity is dependent on the position of the emission slit; the deeper the viewing is fo¬ cused, the lower the upper limit.

The figure 11 represents arbitrary fluores¬ cence as a function of sample concentration for emis- sion slit position 1.5 mm. The upper limit of detection is reduced because of pre filter effect, and is about 20 μg/ml for that slit position.

To be able to estimate the true upper limit of the present system, ICIFC correction must be made. Figure 12 represents a result of an experiment, where the ICIFC was applied to the high concentration end of the sample.

When ICIFC is applied to the data, the inner filter corrected upper limit of linearity will be in- creased substantially. Author have a good reason to believe that when ICIFC is used, the upper limit of the linearity for QS can be as high as ~ 1000 μg/ml.

The lower limit of linearity is taken as the point (concentration) at which the lower end of the curve deviates from the straight line defined by the central portion of the curve by more than twice the average percent deviation of the points that determine the straight line.

The lower limit of linearity is limited by the detection limit of the setup. Lower limit is indepen¬ dent of the slit position and was estimated to be < 0.10 μg/ml.

SUBSTITUTE SHEET One must point out that the sensitivity as well as linearity range can be increased by changing the setup. The sensitivity can be increased increasing lamp current and lamp slit, emission slit and the mono- chromator apertures. The most important is to replace the low quality electronics used with the prototype f.ex. with the one described before in Chapter C. If PMT will be run as a pulse counter (PC) instead of DC mode, the sensitivity can be further increased. But, this will require changing of the light source.

The upper limit of linearity, on the other hand will be increased along the advanced electronics. A precise linear emission slit scan with a spatial corrected detector will have a significant effect on the precision of the ICIFC. Sophisticated ICIFC system will probably offer the widens range of linearity of any MTP based fluorometer on the marketplace.

EXAMPLE II The inner filter effect of fluorescein was studied using both 0.5 mm and 0.25 mm emission slit size w.. The absorbances for the samples were measured to be able to calculate PIFC and compare it to ICIFC.

The ratio of the absolute fluorescence level of the data collected, using the 0.25 mm and 0.5 mm emission slits, was calculated to be 0.5. And is inde¬ pendent of the viewing depth. These results are in good agreement with the theoretical calculations presented in Chapter B. Both correction methods PIFC and ICIFC can be used, but the PIFC correction turned out to overesti¬ mate at very high concentration levels of fluorescein; ICIFC is preferred.

Fluorescein was used as a fluorophore. Seven different concentrations were made δ^'lO¹, 3.33^'lD¹, 1.67^*10', 5^*10^"°, 3.33^*10^"°, 1.67^*10^"° and S^'lO^"1 mol/1. 0.IN NaOH was used as the diluent. 300 μl of each concentra- tion was pipetted in to the cuvettes of Biohit MTP strip.

The fluorescence of fluorescein was measured using 492 nm excitation filter and 510 nm emission filter. The bandwidth for both filters are within 6-9 nm. The transmission >20% and absolute blocking is 10^"7 for these filters.

The intensities was reduced using 3 mm excita¬ tion and 5 mm emission lens apertures. The lamp slit was 0.5 mm. ICIFC measurements were made with 0.25 mm and 0.5 mm emission slits.

The different fluorescein sample concentra¬ tions were measured using 0.25 mm and 0.5 mm emission slit apertures w.. The corresponding results are pre- sented graphically in figures 13 and 14. The optical quenching is clearly indicated. The fluorescence read¬ ings are dramatically decreased when emission slit position is increased from 0.5 mm to 2.5 mm ICIFC ex¬ trapolated values are drawn in both of the graphs. As it can be seen, the upper limit of linearity is sub¬ stantially increased.

To be able to calculate PIFC one must measure the absorbances of the different samples. A result of this is shown in the following figure 15. Based on the measured absorbances, the PIFC was calculated for different depths 0.17 cm, 0.29 cm, 0.43 cm, 0.57 cm and 0.72 cm, corresponding the emis¬ sion slit positions 0.5-2.5 mm. The correction factor C is presented in figure 16. The corresponding correction factor, calculated from the ICIFC results, is presented in the figure 17. Comparison between these results shows that the PIFC will overcorrect the measured val¬ ues at high concentration levels of fluorescein. This is clearly seen in the last figure of this chapter,

SUBSTITUTE SHEET figure 18 where the ratio between PIFC correction to ICIFC is drawn in 3D-plot.

The overcorrection of photometric correction methods have been noticed also by other researchers, who have been using similar type of absorption correc¬ tion models¹⁵'⁶⁴.

EXAMPLE III

An artificial experiment was made to test the inner filter correction applied to NADH-bilirubin sys¬ tem. The test shows that the fluorescence of NADH is dramatically decreased by post filter quenching of bilirubin. But it came out that the total quenching of the system is not entirely due to the optical quench- ing, but also by some specific interaction between NADH and another substance present in the bilirubin samples.

The optical quenching was corrected using the

ICIFC method. The PIFC was not applied, because of the strongly concave free liquid surface of the bilirubin samples, which made the accurate absorbance readings difficult.

Several different substances are measured spectroscopically from human blood. Usually the tests are done with the plasma separated from the whole blood. A common fluorophore used with fluorometric tests is NAD (Nicotinamide Adenine Dinucleotide) or rather its fluorescing reduced form NADH. An advantage of using NAD is the huge Stokes shift of NADH; the maximum absorbance is at 340 nm and high fluorescence at 460 nm.

The common problem due to NADH-bilirubin sys¬ tem is caused by the absorption spectra of bilirubin; the maximum absorbance is around 470 nm, bandwidth being about 80 nm. The emission spectra of NADH is almost completely overlapped by the absorption spectra of bilirubin. Thus the post filter quenching is of concern. A set of bilirubin samples were made to repre¬ sent the concentration range of a human plasma (0, 28/25, 99/25, 305/25 μmol/1). These were mixed with six different concentrations of NADH (0, 10, 20, 30, 40, 50 μmol/1). 300 μl of each concentration combination was measured using standard Biohit MTP strips.

The setup of the instrument was as follows. The beam width used was 0.5 mm and the emission slit size w. was 0.5 mm as well. Because the sensitivity of the prototype instrument with these settings is quite low, we used broadband excitation and emission inter¬ ference filters. The peak wavelength of the excitation filter was 325 nm. The bandwidth of that filter is about 40 nm. The peak wavelength of the emission filter is 470 nm, with a FWHM of 20 nm. The absolute offband- transmission is 10^"4, for both filters.

The results are collected in figure 19, where the recorded fluorescence values are represented as a function of NADH concentration, up to 50 μmol/1. When the detector is viewing deep into the sample (figure D- 5.1 A) the quenching of bilirubin is clearly seen. If this is compared to Figure D-5.1 B, where the viewing is focused near-the-bottom of the cuvette, the relative difference of readings of the tested bilirubin concen- trations are decreased. This indicates optical quench¬ ing. In the last figure D-5.1 C, the ICIFC was applied to correct the post filter effect caused by the NADH fluorescence emission blocking by bilirubin absorption. After inner filter correction, there still exists clear decrease in fluorescence as bilirubin concentration is increased. This indicates specific interaction between NADH and another substance present in the different samples.

The absorbance readings measured from differ- ent samples varied substantially. The reason for varia¬ tion is caused by the non-flat free liquid surface of the sample. Even a small concentration of bilirubin

TITUTE SHEET made the liquid to rise on the wall of the cuvette; sample acting like a plano-concave lens.

Because the exciting light beam was rather narrow, focused 2 mm from the reflecting sidewall of the cuvette, even a small position inaccuracy between different wells, lowered the accuracy of the recorded optical densities.

Figure 20 shows, how the concave free liquid surface increases the measured absorbances, in presence of position inaccuracy. With bilirubin the 'peaking' is so strong that the position inaccuracy must be less than +/- 0.3 mm.

In addition to this, the optical path length is reduced in the centre of the cuvette, resulting smaller values of optical densities, if not corrected. A simple experiment was made to test the line¬ arity of the photometer. Test was done using non fluo¬ rescent and non scattering solution of paranitrophenol (PNP) in distilled water. A dilution series of four decades was measured using 405 nm interference filter. The linearity covered the whole range of the prepared solution.

EXAMPLE IV Non fluorescent and non scattering solution of paranitrophenol (PNP) in distilled water (H₂0) was prepared to cover the concentration range from 5 μmol/1 to 50 mmol/1. The standard Biohit MTP cuvette strips were used. The lamp slit size was 0.5 mm times 2 mm. The

405 nm interference filter with 6-9 nm bandwidth was used. The absolute blocking of this filter is 10^" , and the transmission >20%.

Absorbances as a function of PNP concentration is represented in the figure 21. The linearity covers the whole range of prepared concentration.

A new method to correct optical quenching has

SUBST_ITUTE SHEET been invented and presented. The theoretical basis have been formulated. The method has been tested with a special designed multifunction, fluorometer based, system. The experimental results are in good agreement with the theoretical calculations.

The multifunction system is an instrument pro¬ totype, designed to be a part of a clinical chemistry analyser or a self working MTP supporting unit. There are two main functions of the system; fluorometer and photometer. In addition to this, the designed geometry will, in principle, offer a possibility to measure luminescence and phosphorescence. The photometer is able to measure turbidity and the emission channel can be used to detect scattered radiation to 150°. Both of these have been experimentally tested with the present unit.

When the presented viewing geometry is com¬ pared to the solutions commonly used with MTP based systems, some clear advantages can be picked up. The effect of the dispensing error is eliminated, because the method measures concentration of the sample, rather than the amount of fluorescing substance in a cuvette, as commercial MTP-instruments. Secondly the selffluore¬ scence of the cuvette bottom can be eliminated, because the viewing can be focused into the sample; fluores¬ cence of the cuvette bottom will not reach the detec¬ tion cone.

Because the system has been designed to sup¬ port both the photometric inner filter correction (PIFC) method and the image convolution inner filter correction (ICIFC) method, author had a possibility to verify the applicability of these in practice. From instrumental standpoint of view the ICIFC turned out to be superior over PIFC. The speed and ease of use are real advantages.

In addition to the well known photometric correction method, some authors have developed so called cell shift correction (CSC) method to carry out correction for optical quenching. The disadvantage of this method lies in its practical inapplicability. CSC method can be used only with 'one sample' instruments, not for batch processing as with MTP systems. The CSC method requires changing of excitation and emission filters and shifting the cell in two dimensions to be able to carry out inner-filter correction. All this is compulsory in respect to the ICIFC method. Using ICIFC, one is able to perform total optical quenching 'at one run'. There is no need to measure separately the pre- and post filter effects.

The shift of the cell is neither needed. In addition to this ICIFC can be applied to any system including the MTP based instruments.

The applicability of ICIFC was found to be useful when scattering of light was measured. Optical quenching was clearly observed and corrected. The au¬ thor is not aware of any report, where scattering quen- ching has been measured before.

The primary task, to develop an MTP-fluorome- ter/photometer system to increase the linearity of the instrumental fluorometry has turned out to be an inter¬ esting new multifunction concept, supporting sophisti- cated method to carry out correction for optical quen¬ ching.

SUBSTITUTE SHEET G REFERENCES

1. Suovaniemi 0, French Patents 7442147, 7800847

2. Suovaniemi 0, US Patent 4144030

3. Suovaniemi 0, US Patent 4290997

4. Suovaniemi O, Jarnefelt J, Am.Lab. 14 (6), 106 (1982)

5. Suovaniemi 0. Int. Biotechnol. Lab. 5/6 (1984)

6. Grum F, Becherer Rj , Optical Radiation Measure- ment, vol. 1, Academic Press, Orlando (1982)

7. Herschel J, Philos. Trans. Roy. Soc. London, 147, 143 (1845)

8. Herschel J, Philos. Trans. Roy. Soc. London, 147, 147 (1845)

9. Stokes G G, Philos. Trans. Roy. Soc. London, 142, 463 (1852)

10. Eisinger J, Flores J, Anal. Biochem. 94, 15 (1979)

11. Bowman R L, Caufield P A, Udenfriend S, Science 122, 276 (1981)

12. Turner, G K, Science 146, 183, (1964)

13. Bowen E J, Sawtell J W, Trans. Faraday Soc, 33, 1425 (1937)

14. Bryant M F, O'Keefe K, Malmstadt H V, Anal. Chem. 47, 2324 (1975)

E SHEET 15. Ritter A W, Tway P C, Cline Love L J, Ashworth H A, Anal.Chem. 53, 280 (1981)

16. Pyun C-H, Park A-M, Anal. Instr., 13, 159 (1984)

17. Harjunmaa H, "Theory, Design and Immunological Applications of a Vertically Measuring Fluorome¬ ter", Labsystems Oy, Helsinki (1986)

18. Labsystems, "fluoroskan II, Operating Instruct¬ ions", Helsinki (1989)

19. Labsystems, "Titertek Fluoroskan Operating In¬ structions", Helsinki

20. Lakowicz J R. , "Principles of Fluorescence Spec¬ troscopy", Plenum Press, New York (1984)

21. Guilbault G G, "Practical Fluorescence", Marcel Dekker, Inc., New York (1973)

22. Skoog D A, West D M, "Principles of Instrumental Analysis", Holt, Rinehart and Winston, Inc., USA (1971)

23. Willard H H, et al. , "Instrumental Methods of Analysis", Wadsworth Publishing Company, USA (1981)

24. Mielenz K D (ed), "Optical Radiation Measurements, Vol. 3, Academic Press, Orlando (1982)

25. Schulman S G, "Fluorescence and Phosphorescence Spectroscopy: Physicochemical Principles and Prac- tise", Pergamon Press, Oxford, U.K. (1977)

TE SHEET 26. Forster T, "Fluoreszenz Organischer Verbindungen", Vandenhoeck & Ruprecht, Gδttingen (1951)

27. Mode V A, Sisson D H, Anal Chem., 46, 200 (1974)

28. Holland J F, Teets R E, Kelly P M, Timnick A, Anal. Cham., 49, 706 (1977)

29. Christmann D R, et al. , Anal. Chem., 52, 291 (1980)

30. Leese R A, Wehry E.L., Anal. Chem., 50, 1193 (1987)

31. Holland J F, Teets R E, Timnick A, Anal. Chem., 45, 145 (1973)

32. Novak A, Collect. Czech. Chem. Commun., 43, 2869 (1978)

33. Christman D R, Crouch S R and Timnick A, Anal. Chem. 53, 276 (1981)

34. Christman D R, Crouch S R and Timnick A, Anal. Chem. 53, 2040 (1981)

35. Lord Rayleigh, Phil. Mag., 81, 5 (1881)

36. Mie G, Ann. Phys. , Lpz., 25, 37 (1908)

37. Schoenberg E, et al., Astr. Nachr., 253, 261 (1934)

38. Walstra P, Brit. J. Appl. Phys., 15, 1545 (1964)

39. Heller W, J. Chem. Phys., 26, 920 (1957)

SUBSTITUTE SHEET 40. Gans R, Ann.Phys. Lpz. , 76, 29 (1925)

41. Van de Hulst, "Light Scattering by Small Parti¬ cles", Dower Publications Inc., N.Y., USA (1981)

42. Jablonski A, Acta Phys. Polon., 16, 471 (1957)

43. Cehelnik E D, Mielenz K D, Velapoldi R A, J. Res. NBS 79A, 1 (1975)

44. Almgren M, Photochem. and Photobiol. 8, 231 (1968)

45. Melles Griot, "Optics Guide 5", Irvine, USA (1991)

46. EG&G Electro Optics, "Flashlamp Applications Manu¬ al", Salem, USA (1983)

47. Heimann, "Xenon flashtubes", Wiesbaden, Germany (1984)

48. Hamamatsu, "Opto-semiconductors", Japan (1990)

49. Hamamatsu, "Accessories for Photomultiplier Tubes", Japan (1990)

50. Hamamatsu, "Photomultiplier Tubes", Japan (1990)

51. Hamamatsu, "How to perform Photon Counting Using Photomultiplier Tubes", Japan (1992)

52. Texas Instruments, "Linear Circuits, Data Acquisi¬ tion and Conversion, Vol. 2", USA (1989)

53. Burr-Brown, "Integrated circuits databook supple- ment, vol. 33b", Tuscon, USA (1990)

54. Analog Devices, "1990/91 Linear Products Data-

SUBSTITUTE SHEET book", Norwood, USA (1990)

55. Linear Technology, "1990 Linear Databook", Milpi- tas, USA (1989)

56. Malmstadt H V; Franklin M L; Horlick G, Anal. Chem., 44, 63A (1972)

57. Franklin M L, Horlick G; Malmstadt H V, Anal. Chem., 41, 2 (1969)

58. Ingle J D Jr, Crouch S R, Anal. Chem., 44, 777 (1972)

59. Ingle J D Jr, Crouch S R, Anal. Chem., 44, 785 (1972)

60. Hayes J M, Schoeller D A, Anal. Chem., 49, 306 (1977)

61. Nakamaru J K, Schwartz S E, Appl. Opt., 7, 1073 (1968)

62. Borders R A, Birks J V, Anal. Chem., 52, 1273 (1980)

63. Borders R A, Birks J W, Anal. Chem., 52, 1273 (1980)

64. Nau V J, Nieman T A, Anal. Chem., 53, 350 (1981)

65. Intel, "8-bit Embedded Controllers", USA (1990)

66. Standard Test Method for Limit of Detection of Fluorescence of Quinine Sulfate, ASTM E 579-84

SUBSTITUTE SHEET 67. Standard Test Method for Linearity of Fluorescence Measuring Systems, ASTM E 578-83 (Reapproved 1988)

68. Rohatgi K K, Singhal G S, Anal. Chem. 34, 1702 (1962)

TITUTE SHEET

Claims

1. A procedure serving to correct for inten¬ sity decrease, due to optical attenuation, of fluores- cence from a liquid sample in connection with fluoro¬ metric quantitative analysis, in said procedure the liquid sample being placed in a cuvette, an excitation light beam being directed into said cuvette and the intensity of the fluorescent light produced by the excitation light beam in the liquid sample owing to the substance therein contained being determined and the concentration of the substance contained in the liquid sample being determined on the basis of this intensity, characterized in that the intensity distribution of the image of the fluorescent liquid sample created in the liquid sample by said excitation light beam is deter¬ mined and the optical attenuation is determined from the intensity distribution that has been measured.

2. Procedure according to claim 1, charac- terized in that the optical attenuation is measured from the oblique projection of the liquid sample's image on an image plane.

3. Procedure according to claim 1 or 2, characterized in that the excitation light beam is directed perpendicularly through the cuvette wall, and that the central axis of the fluorescence measuring direction and the central axis of the excitation light beam enclose an angle of 10 to 40°.

4. Procedure according to any one of claims 1-3, characterized in that the excitation light beam is directed into the cuvette at right angles through the bottom thereof upward from below, and that the fluores¬ cence is measured through the cuvette bottom upward from below.

5. Procedure according to any one of claims

1-4, characterized in that the excitation light beam is coherent.

6. Procedure according to any one of claims 1-5, characterized in that the excitation light beam has rectangular cross section, its sides on the order of (0.5 to 15) x (1 to 4) mm, advantageously about 4 mm.

7. Apparatus for implementing a procedure according to any one of claims 1-6, comprising a meas¬ uring cuvette to be filled with a liquid sample, an excitation light source, optics for directing into the measuring cuvette the light beam formed by said exci¬ tation light source, a detector for measuring the in¬ tensity of the fluorescent light produced by the liquid sample in the cuvette, and a calculator means for de¬ termining the concentration of the substance producing the fluorescent light, characterized in that the appa¬ ratus comprises a measuring instrument for determining the intensity distribution of the fluorescent light.

8. Apparatus according to claim 7, charac¬ terized in that the excitation light beam is directed perpendicularly through a wall, advantageously the bottom, of the cuvette, and that the measuring instru¬ ment is aimed to measure the image of the fluorescent liquid sample under an angle of 10 to 40° relative to the excitation light beam, advantageously from below through the cuvette bottom.

9. Apparatus according to claim 7 or 8, characterized in that the measuring instrument com¬ prises a slit and a fluorescent light detector, said slit being located substantially in the image plane, advantageously at the distance of the image from the sample, and the slit is movable in the image plane.

10. Apparatus according to claim 9, charac¬ terized in that the slit and the fluorescent light detector both are movable in the image plane.