WO2004017056A1

WO2004017056A1 - Determination of mechanical properties of polymer products using raman spectroscopy

Info

Publication number: WO2004017056A1
Application number: PCT/EP2003/008904
Authority: WO
Inventors: Dieter Lilge; Claus Gabriel
Original assignee: Basell Polyolefine Gmbh
Priority date: 2002-08-12
Filing date: 2003-08-11
Publication date: 2004-02-26
Also published as: AU2003255419A1

Abstract

A process for determining mechanical properties of polymer products. A Raman spectrum of the at least one polymer product is recorded and the at least one mechanical property of the polymer product is calculated from the Raman spectrum of the at least one polymer product. This provides a rapid, noninvasive and reproducible method which does not require sample preparation and is suitable for mechanical characterization of polymer products and for use in high throughput screening.

Description

DETERMINATION OF MECHANICAL PROPERTIES OP POLYMER PRODUCTS USING RAMAN SPECTROSCOPY

The invention relates to a process for determining mechanical properties of polymer products with the aid of Raman spectroscopy. 5

At the present time, far-reaching efforts are generally being made to discover novel polymers with the aid of High Throughput Screening (HTS) and to optimize the properties of known polymers.

This involves combinatorial reactions which generate an array of polymer products while varying the reactants, the catalysts used or the reaction conditions under which the polymerization takes 10 place over the array. The properties of the resulting polymer products are then investigated. In this manner, the influences of the variations on the product can be realized over a wide spectrum with relatively low time and material demands.

While molecular properties of polymers can be predicted quite reliably, this is virtually impossible 15 for the physical, in particular for the mechanical, properties, so that rapid and simple characterization of the physical properties for the purposes of HTS will be particularly advantageous.

The classical methods of determining mechanical properties of polymers, for example by tensile

20 or bending tests, have the disadvantage that they involve considerable substance and time demands and are accordingly unsuitable for HTS. Although mini-injection molding machines are available for very low sample sizes of up to 2 g to prepare very small test specimens, the mechanical data obtained with the aid of these specimens is only of limited comparability with the values obtained with the aid of standard test specimens.

25

Efforts were accordingly made to determine the mechanical properties with the aid of spectroscopic methods. The European Polymer Journal 38 (2002) 745 discloses the determination of mechanical properties such as Melt Flow Index (MFI), impact strength or flexural modulus of propylene copolymers with the aid of IR spectroscopy. For the determination, the IR

30 spectra of polypropylene copolymer films are measured and the spectra are calibrated with the aid of a chemometric algorithm against polymers having known mechanical properties in the selected wavelength range of 1270 - 1240 cm^"1. However, disadvantages of IR spectroscopy are the necessary sample preparation and the relatively high cost and inconvenience of injecting the beam path with the aid of IR-transparent optical elements.

35

The use of Raman spectroscopy for the determination of molecular properties such as density and crystallinity of polymer products is known. Applied Spectroscopy Vol. 53 (1999), 55 presents U I \ a calibration model for forecasting the density of LLDPE from Raman spectra by carrying out a

Partial Least Squares Regression for the range from 1600 to 600 cm^"1 of the Raman spectrum. In

40 Journal of Polymer Science, Polymer Physics Edition Vol 16 (1999), 1181 , the proportion of crystalline phases in various polyethylenes is determined by assigning the intensity of a sharp Raman band at 1416 cm^"1 to the crystalline phase which is not present in the amorphous phase.

Furthermore, Journal of Polymer Science Part B: Polymer Physics Vol 28, (1990) 167-185 describes investigations which show that many polyethylene copolymers show a correlation between molecular properties, in this case crystallinity, and macroscopic properties such as the modulus of elasticity or yield stress. The crystallinity was determined by the abovementioned methods from Raman spectra.

It is an object of the present invention to provide a process which allows rapid, simple and noninvasive determination of mechanical properties of polymer products which, however, is also reproducible and does not require sample preparation.

We have found that this object is achieved by a process having the features of claim 1. Claims 2 to 18 contain preferred embodiments of the process according to the invention.

The invention is based on the fact that the mechanical properties of polymers may be determined with the aid of Raman spectroscopy. To this end, a Raman spectrum of the at least one polymer product is recorded and the at least one mechanical property of the polymer product is determined from it.

In this context, polymer products include any natural or synthetic polymer or polymer mixture. These include in particular all polymers which contain C-C units and are derived from monomers having C=C double bonds, irrespective of the type of polymerization. Preference is given to applying the process to C₂ to C₂_ 1-alkene or vinylaromatic homo- or copolymers, more preferably to a polyethylene or polypropylene homo- or copolymer. Useful comonomers are again C₂-C₂₀ 1- alkenes, in particular ethene, propene, 1-butene or 1-hexene, but the process is also applicable to other copolymers such as EPDM, EVA, etc. These may be random or else block or graft copolymers. Blends and compounds of different polymers or polymers provided with colorants and fillers can also be mechanically characterized in a rapid and simple fashion by the method described. In particular, the process is applicable to HDPE, LDPE and LLDPE, but also to polyamides and other semicrystalline polymers, copolymers thereof and blends thereof. The applicability of the process substantially depends only on the mechanical properties in the polymer being reflected in some manner in the Raman-active molecular vibrations, so that a correlation results between the change in the spectrum and the mechanical property.

The Raman spectrum of the polymer products may in principle be recorded in liquid or solid form. The sample form is variable within a wide range so that working the samples up before recording the Raman spectra is normally unnecessary. As well as films, pellets and other shaped polymer products, preference is accordingly also given to granules, lumps, meal or powder.

Mechanical properties of the polymer product may be any property of the polymer product which characterizes its macroscopic-mechanical behavior and can be determined with the aid of mechanical measuring methods. These include in particular the elastic and inelastic deformability and also the rheological properties of the polymer product. Preferred mechanical properties are the modulus of elasticity, the tensile strength, the yield stress, the flexural modulus, the flexural strength and the impact strength, although the process would not be restricted to the properties mentioned. Rather, any mechanical property of the polymer product which is known to those skilled in the art and can be determined from the Raman spectrum can be used. Particularly preferred mechanical properties are the modulus of elasticity and the yield stress.

However, the process according to the invention also makes it possible to realize in a relatively simple manner the online determination of the mechanical properties of a polymer product during an operating polymerization process, whether it be on the laboratory, pilot plant or industrial scale. This involves continuously recording Raman spectra of the polymer product in the particular reactor at predefined intervals and calculating the desired mechanical property from them. This can be technically realized by injecting the exciting laser beam via a glass fiber cable into the reactor and capturing the resulting scattered Raman radiation with the aid of a detector.

Alternatively, samples may be continuously withdrawn from the reactor and their mechanical properties may be investigated with the aid of Raman spectroscopy. Changes in the mechanical properties of the polymer product during the production process may thus be quickly detected and appropriate corrective measures may be taken.

The process according to the invention is particularly effective when the mechanical property is determined for an array of a multiplicity of polymer products. In this manner, many Raman spectra of different polymer products may be recorded in direct succession which allows the combination with a high throughput reactor system which delivers a large number of polymers having varying properties. This is true in particular when the multiplicity of polymer products differs in the desired and determined mechanical properties. The polymer products may be arrayed as a series or preferably as a two-dimensional field. However, any other array is conceivable in principle, as long as only the polymers to be investigated can be physically assigned to the array. In the simplest case, a microtiter plate may be used. It is normally possible to determine more than one different mechanical property of the polymer product from one spectrum. This shows the particular strength of the process, since the calculation is carried out very quickly and a very large amount of useful information on the polymer product may be obtained with only one measurement. For instance, all relevant mechanical elongation, flexural and deformability values may be determined at the same time which allows virtually complete mechanical characterization of the polymer products. In addition to the mechanical properties, it is also possible to determine nonmechanical properties of the polymer such as crystallinity, density and other molecular properties such as the degree of short-chain branching of the polymer product.

Preference is given to determining the at least one mechanical property by comparing the Raman spectrum with a calibration spectrum. The calibration spectrum is determined by analyzing polymers of known mechanical properties. The known statistical methods may be used here. Calibration makes it possible to forecast the mechanical product properties which have been determined by standard testing methods.

In the calibration, the polymer samples of known composition and known mechanical properties which have been previously determined with the aid of conventional methods are placed in the Raman spectrometer under the conditions under which the polymer products to be characterized are subsequently analyzed and at least one Raman spectrum of each is recorded. In the calibration, the mechanical properties to be determined within the polymers are varied within a very wide range, at least within the range to be expected for the analysis of unknown samples, it is also important that the magnitude ratios of the mechanical properties relative to each other are varied to a very large extent in order to avoid redundancies and facilitate very accurate forecasts. The wavelength range used for the calibration is in the range from 560 to 1980 cm^"1, a preferred range for polyethylene is from 1000 to 1500 cm^"1, and a preferred range for polypropylene is from 750 to 1550 cm^"1.

Preference is given to generating the calibration spectra by interpolating the crude spectra recorded in wavelength steps of 2 cm^"1, and the resolution of the analytical instrument is 1.4 cm^"1.

The calibration spectra may be compared with the sample spectra by assuming a linear relationship between scattering intensity and the mechanical property in a similar manner to the Beer-Lambert law for characteristic Raman bands, as described, for example, in Journal of Polymer Science, Polymer Physics Edition Vol. 16 (1978), 1181 for determining the crystallinity. However, it is necessary for this purpose to determine isolated bands which react to the change in the mechanical property to be determined with as little influence as possible from other parameters.

Since assignment of peaks or bands to vibrations which correlate directly with the property to be determined is normally possible only with great effort owing to the complexity of the spectra, preference is given to applying chemometric methods to the determination. Frequently used methods include the Inverse Least Squares (ILS) method and the Classical Least Squares (CLS) method. A preferred chemometric method is Principal Component Regression (PCR) which is generally known to those skilled in the art and is described, for example, in Matthias Otto: Chemometrics, Wiley-VCH, 1999. This combines the advantages of ILS and CLS and delivers optimal results for the application according to the invention. Principal Component Regression involves calibrating selected regions of the Raman spectrum with the aid of polymer products of known mechanical properties and then carrying out a regression of the selected spectral region on the mechanical property using this calibration. Alternatively, preference is given to using two further partial least squares methods PLS1 and PLS2 for the chemometric analysis which are provided by the program used.

The regression may be carried out with the aid of commercial programs. In the present case, the PCR was carried out using the program Spectrum Quant÷ (Version 4-1) from Perkin Elmer.

The calibration is carried out with or without cross-validation, preferably with cross-validation. Criteria taken into account for the quality of the calibration include the number of principal components, the variance and the standard forecast error and also the application of the regression to the predefined and estimated values.

The process according to the invention can be particularly advantageously applied to optimize the preparative process of a polymer product with reference to at least one mechanical property by converting one or more reactants to a polymer product in a multiplicity of reactors under polymerization conditions, then analyzing at least one mechanical property of the polymer products resulting from each reactor according to the above-described processes with the aid of Raman spectroscopy and finally selecting the polymer products from the individual reactors with reference to the at least one mechanical property. Variation of the reactant concentrations or the reaction parameters allows a polymer product to be tailored exactly to a predefined mechanical property or property combination, or a combinatorial database to be constructed with which it is possible to systematically select the reactants, catalysts and reaction conditions.

In this context, a reactant is any chemical material which can be converted with or without a catalyst to one or more polymer products. These also include auxiliaries and additives. In particular, the process is applicable irrespective of whether the reactant is a gas, a liquid or a solid. The reactant only has to be present in fluid form, i.e. it has to be possible to supply it to the reactor. It may be necessary to dissolve, suspend or disperse the reactant in a fluid carrier stream in order to be able to supply it to the reactor. A polymer product is any polymer resulting from the reaction, irrespective of its chemical or physical properties. Preference is also given here to using C₂ to C₂o alkenes and mixtures thereof as reactants. Particular preference is given to polymerizing ethylene, propylene and 1-butene. to prepare graft copolymers, one of the reactants may itself also be a polymer which is reactively grafted with the a further low molecular weight component, for example maleic anhydride, or an oligomeric or polymeric component. The process according to the invention may further be applied to selecting from a multiplicity of catalysts for preparing a polymer product in a high throughput screening and thus optimizing the reactants, additives, preparative processes and processing methods used. In such a screening, a multiplicity of catalysts for preparing a polymer product is first introduced into an array of reactors and one or more reactants is or are added to each individual reactor and contacted with the catalysts under predefined reaction conditions. The mechanical property of each of the polymer products formed is then determined with the aid of Raman spectroscopy, as described above. These mechanical properties of the polymer products formed then serve in the selection of suitable catalysts for the preparative process. Combination with the abovementioned preferred embodiments of polymer analysis in any variation and combination is possible and contemplated by the invention. In this case, it has to be possible to contact the reactant with the catalyst.

In principle, the catalyst may also be in any form and any aggregate state, as long as it can be secured in the reactor. However, preference is given to working with solid or solid-supported catalysts. In this context, catalysts also include catalyst mixtures or catalysts activated by cocatalysts. In particular, the process according to the invention is suitable for selecting metallocene catalysts, chromium catalysts or Ziegler-Natta catalysts for polymerizing polyolefins. Such catalysts are often used in polymer production and generally known to those skilled in the art.

It is advantageous when the reactants or catalysts used in the individual reactors differ in at least one chemical and/or physical property. It is also possible to vary reactant properties and catalyst properties in combination. The chemical property may consist in any molecular property such as constitution, configuration or conformation. This also include enantiomeric forms of catalysts. Examples of physical properties include the particle shape and size of the catalyst or catalyst support chosen, its surface, the average pore volume and the pore volume distribution. However, in particular for supported catalysts systems, it is also possible to vary the type and pretreatment of the support used, the quantity of catalyst applied to the support and the possible use of cocatalysts, among other variables.

Independently thereof or in addition thereto, it is also possible to vary the polymerization conditions of the individual reactors in at least one physical parameter, which also makes it possible to optimize the process to predefined mechanical properties. In this context, physical parameters include any physical quantity which has an influence on the course of polymerization to form the polymer product. These include in particular temperature, pressure, reactant or product concentrations and distributions, introduction and withdrawal of reactants and products, fixed or fluidized bed methods, reaction time or residence time, etc., without the process according to the invention being restricted to the parameters mentioned. Preference is given to a temperature range of from -80 to 200°C, rηore preferably from 0 to 100°C, and a pressure range of from 10^"4 to 10² MPa, more preferably from 10^"2 to 6 MPa.

The process according to the invention may exhibit its particular strength when the reaction conditions in each individual reactor during the polymerization process differ in exactly one reaction parameter. The optimum variation of the reaction conditions may be planned using the generally known principles of the combinatorial approach, in particular of combinatorial chemistry.

When carrying out the high throughput screening, a preferred embodiment of the process according to the invention involves recording the Raman spectrum of the at least one polymer product immediately within each individual reactor which had previously been used to carry out the polymerization. The advantage of Raman spectroscopy of working with visible light and obtaining the information from the vibrations via the shifting ofthe wavelength of the scattered light has the great advantage that simple optical elements of glass or quartz may be used to inject the analytical beam into the reactor and that analysis without transferring the samples to separate analytical holders is realized in a simple manner. IR-transparent windows with which each of the reactors would have to have been equipped are superfluous. When reactors of glass or quartz are used, injection windows may even be omitted entirely.

Furthermore, this variant even opens the possibility of following the mechanical properties of the polymer product on-line during the progress of the polymerization reaction when Raman spectra of the reaction mixture are recorded continuously during it and then converted into the desired mechanical properties.

In a further advantageous embodiment, the reactors are operated continuously under the polymerization conditions, in particular when the catalysts are supported and may be used in a fluidized bed. In the case of fluid reactants, this may be achieved by the reactant or reactants themselves generating a sufficient flow through the reactor to perform the reaction in a suitable manner. Alternatively, an inert carrier fluid may be admixed to the reactant stream. However, batchwise operation is likewise suitable.

A number of reactors used in the array is chosen depending on the use. For a rapid screening to preselect suitable catalysts, a relatively large number of relatively small parallel reactors is used to very quickly limit the number of suitable catalysts. In this case, the polymerization conditions are varied only to a limited extent. In this case, preference is given to using 8, 16, 24, 32, 48, 64, 96 and 112 reactors which have a volume from 1 to 50 cm³, preferably from 5 to 20 cm³, more preferably from 10 to 15 cm³. Preference is likewise given to combining the reactors into blocks which more preferably each consist of 4 or 16 reactors. The reaction parameters can be set separately for each block. The size, the type, the material and other reactor properties are selected according to the reaction to be investigated and in such a manner that comparability with processes on the pilot plant or industrial scale is as easy as possible.

After such a preselection, a more exact selection of the catalysts may take place by more widely varying the activation and reaction conditions for a reduced number of catalysts in a reduced number of reactors of increased volume. Finally, the array of reactors described also allows only one catalyst to be used and an exclusive optimization of the activation and reaction conditions to be carried out with increased analytical precision. In this case, preference is given to using 4, 8 or 16 reactors arranged in parallel which in one embodiment customarily have a larger volume than in the rapid screening, preferably from 30 to 200 cm³, more preferably from 60 to 90 cm³.

The processes according to the invention are illustrated with the aid of the examples hereinbelow in a high throughput screening of polymers, without restricting the invention to the embodiments described.

Example 1

The mechanical properties of unknown polyethylene samples were determined by polymerizing them in a high throughput reactor system from Chemspeed Ltd, Switzerland, which comprises from 4 to 112 parallel reactors. The reactors are combined in blocks of 4 or 16 reactors which can be heated independently and to which pressure can be applied. The reactors of a 16-block have a volume of 13 cm³, those of a 4-block 75 cm³.

After withdrawing the polymer samples from the reactors, they were dried. Alternatively, it is possible to introduce the samples directly to the analysis. The samples, which were present as a powder, meal, lumps or granules, or as a liquid and required no further workup, were introduced into a microtiter plate for 24, 48 or 96 samples. The filling was carried out by hand, although automatic withdrawal is also contemplated. The filled microtiter plate was then introduced into the Raman spectrometer (Multiwell, Jobin Yvon Horiba, Bensheim, 532 nm, 150 mW output, spectral range from 560 to 1980 cm^"1, resolution 1.4 cm^"1, CCD detector with single grating, scattered light suppression by filter).

The number of measurements per polymer sample and the measuring duration were input, and the number of measurements was customarily from 1 to 10, preferably from 3 to 5, and the measuring duration was in the range from 1 s to 60 s, preferably 5-15 s. In this case, the measurements were carried out at room temperature. Using a heating table, it is possible to analyze up to about 200°C. After the type of microtiter plate was input, the Raman spectra of each sample were recorded in succession. To this end, the sample table on which the microtiter plate is disposed is automatically movable in the horizontal x and y directions. The laser beam of the Raman spectrometer, in this case the second harmonic of an Nd:YAG laser (532 nm), was initially prefocused manually on the polymer surface by moving the objective in the vertical direction and observing the "light spot" resulting from the focus on a monitor. The focus was preadjusted using a microtiter plate well which has the average filling height of the polymer samples. Focusing during the measurement was then effected automatically by the Raman spectrometer within a tolerance range of 2.5 mm. The measurements themselves were made by moving to the center of the appropriate wells of the microtiter plate having the samples. This is effected for each well until the entire microtiter plate has been covered and at least one Raman spectrum of each sample has been recorded. In the present example, one spectrum was recorded per well.

The measurements were then stored automatically on an electronic medium and transferred to a computer for chemometric evaluation.

Before the actual regression, the spectra may be processed. Although it is possible to scale and smooth the spectra, this was not done in the present case. If, however, this had been done, it would have been effected with the aid of a reference band at 1300 cm^"1. Normalization and baseline correction were likewise not carried out. If, however, normalization or baseline correction had been carried out, they would have been effected with the aid of the 1st derivative and smoothing over 5 points.

For the principal component regression, the number of PCA factors used was initially defined, which may be varied from one to the number of spectra used. Preference is given to using a maximum of 5 PCA factors. The mechanical property against which the regression is to be effected is then chosen. Preference is given to regression against the modulus of elasticity, the yield stress, the impact strength or the flexural strength, and particular preference to carrying out regression against more than one mechanical property with the aid of one spectrum. Finally, the spectra used for the regression are selected while discarding by hand or automatically those having noticeable errors, significant noise, etc., and not using them for the regression.

The regression, like the calibration, was carried out using the program Spectrum Quant+ (Version 4.1) from Perkin Elmer. The results of the evaluation are output in a file which may either be read directly to a database or printed out. With the aid of a computer program to visualize the data, simple and rapid evaluation of the results is possible. Example 2

It will be shown hereinbelow that it is possible with the aid of a calibration to forecast mechanical properties of polyethylenes from Raman spectra by the method described.

The characteristic mechanical values modulus of elasticity E and yield stress σ_y were taken from the polyethylene product portfolio of Basell. Polyethylenes of highly differing density were used for the investigations (metallocene and Ziegler-Natta LLDPE, LDPE, MDPE, HOPE)

Table 1

For the calibration of the yield stress, a further additional eight test products were used (HDPE). Their yield stresses are substantially in the range of the values of commercial products.

The characteristic mechanical values were used to calibrate the Raman spectra measured on the individual polyethylenes with the aid of Principal Component Regression (PCR) as the chemometric method. Figures 1 and 2 show that Raman spectra allow good values to be determined both for the modulus of elasticity and for the yield stress. In Figure 1 , the moduli of elasticity determined from the Raman spectra are plotted against the moduli of elasticity determined by conventional tensile tests. This results in a linear dependence with a variance of 98.3% and a standard forecast error of 117 MPa. The quality of the correlations is sufficient for the forecasting of mechanical properties of polyethylene from high throughput screening.

In Figure 2, the yield stresses determined from the Raman spectra are plotted against the yield stresses determined with the aid of conventional tensile tests (DIN 53457). This results in a linear dependence with a variance of 99.7% and a standard forecast error of 1.6 MPa. The quality of the correlation is sufficient for the forecasting of mechanical properties of polyethylene from high throughput screening.

Example 3

It will be shown hereinbelow that it is possible with the aid of a calibration to forecast mechanical properties of polypropylen homo- and copolymers from Raman spectra by the method described.

The characteristic mechanical value modulus of elasticity E were taken from propylene (co)polymers produced by the Novolen gas phase process. Either metallocene or Ziegler-Natta catalysts were used. The content of ethylene in the propylene homo- and copolymers were in the range from 0 to about 20 % by weight. The melt flow rates MFR (230 °C / 2,16 kg) according to ISO 1133 of the products were in the range from about 0.5 to 80 g/10 min.

The characteristic mechanical values were used to calibrate the Raman spectra measured on the individual polyethylenes with the aid of Principal Component Regression (PCR) as the chemometric method. Figure 3 shows that Raman spectra allow good values to be determined for the modulus of elasticity.

In Figure 3, the moduli of elasticity determined from the Raman spectra are plotted against the moduli of elasticity determined by conventional tensile tests. This results in a linear dependence with a variance of 89.5% and a standard forecast error of 117 MPa. The quality of the correlations is sufficient for the forecasting of mechanical properties of polyethylene from high throughput screening.

Table 2 shows that the deviations between predicted and measured moduli of eleasticity are in the range of ± 10%. Only in excepted cases deviations up to 20 % were found. A prediction is possible for a wide range of copolymer content.

Table 2

10

15

20

25

30

35

40

Claims

We claim:

1. A process for determining mechanical properties of polymer products, which comprises

a) recording a Raman spectrum of the at least one polymer product and

b) calculating the least one mechanical property of the polymer product from the Raman spectrum of the at least one polymer product.

2. A process as claimed in claim 1 , wherein the mechanical property is determined for an array of a multiplicity of polymer products.

3. A process as claimed in claim 2, wherein the multiplicity of polymer products differ in the at least one mechanical property.

4. A process as claimed in any of the preceding claims, wherein more than one mechanical property is determined at the same time from a Raman spectrum.

5. A process as claimed in any of the preceding claims, wherein the mechanical property of the polymer product is determined continuously during a preparative process of the polymer product.

6. A process as claimed in any of the preceding claims, wherein the mechanical property is selected from the group consisting of modulus of elasticity, the yield stress, the elongation, the stress at break, the elongation at break, the flexural strength and impact strength.

7. A process as claimed in any of the preceding claims, wherein the at least one mechanical property is calculated by comparing the Raman spectrum with calibration spectra which have been measured using polymers of known mechanical properties.

8. A process as claimed in claim 7, wherein the at least one mechanical property of the at least one polymer product is calculated by Principal Component Regression of a selected region of the Raman spectrum compared to the at least one mechanical property of the polymer which has been determined by calibration using polymers of known mechanical properties.

9. A process as claimed in any of the preceding claims, wherein

- at least one reactant is converted under polymerization conditions in a multiplicity of reactors to a polymer product,

- the mechanical property of each polymer product is determined and then

- the polymer products of the individual reactors are selected with reference to the at least one mechanical property.

10. A process as claimed in claim 9, wherein the at least one reactant in each individual reactor differs or differ in at least one chemical and/or physical property.

11. A process as claimed in any of claims 1 to 8, wherein

- a multiplicity of catalysts is introduced for preparing a polymer product in an arrangement of reactors,

- at least one reactant is introduced into the individual reactors and is contacted with the catalysts under predefined reaction conditions to form at least one product,

- the mechanical property of each polymer product is determined and then

- the catalysts of the individual reactors are selected with reference to the at least one mechanical property of the polymer products.

12. A process as claimed in claim 11 , wherein the at least one reactant and/or the catalysts in each individual reactor differs or differ in at least one chemical and/or physical property.

13. A process as claimed in claim 11 or 12, wherein the polymerization catalyst is selected from the group consisting of a metallocene catalyst, chromium catalyst and Ziegler-Natta catalyst.

14. A process as claimed in any of claims 9 to 13, wherein the polymerization conditions of the individual reactors differ in at least one physical parameter.

15. A process as claimed in any of claims 9 to 14, wherein the Raman spectrum of the at least one polymer product is recorded immediately within each individual reactor.

16. A process as claimed in any of claims 9 to 15, wherein each individual reactor is operated continuously under the polymerization conditions.

17. A process as claimed in any of the preceding claims, wherein the at least one polymer product is a C₂ to C₂o-aIkene homo- or copolymer.

18. A process as claimed in claim 17, wherein the at least one polymer product is a polyethylene or polypropylene homo- or copolymer.