DE102007044606A1 - Correlation model developing method for e.g. food industry, involves determining chemical characteristics of cellulose or paper sample and selecting chemical-analytical process, which enables quantitative determination of values - Google Patents

Abstract

The method involves chemical-analytical determining chemical characteristics of cellulose or paper reference sample to obtain reference values e.g. carbonyl group content and carboxyl group content. A reference parameter is generated for each reference sample by mathematical evaluation of reference data. The obtained reference parameter is correlated with reference values by performing multivariate calibration for creating a correlation model. A chemical-analytical process is selected, where the process enables quantitative determination of the reference values.

Description

I. Field of application

The The present invention relates to a nondestructive process Determination of the chemical properties and thus of the condition of pulp and paper, in particular a method of determining the Carbonyl and Carboxylgruppengehaltes, the average molecular weight and the surface pH of pulp and paper.

• a) Korrelation von Referenz-Kerngrößen aus IR-Spektren mit diesen chemischen Eigenschaften unter Verwendung von multivariater Datenanalyse und
• b) darauf basierend die Entwicklung von Modellen, um
• i) den aktuellen Zustand des Papiers zu bestimmen und
• ii) das zukünftige chemische und physikalische Verhalten von Cellulose oder daraus hergestellten Produkten vorhersagen zu können.

The present invention relates in particular to a method for

• a) Correlation of reference core sizes from IR spectra with these chemical properties using multivariate data analysis and
• b) based on this the development of models
• i) to determine the current state of the paper and
• ii) be able to predict the future chemical and physical behavior of cellulose or products made from it.

II. Technical background

In It is very important in various areas of pulp analysis Interest, insight especially in the oxidation state of cellulose to get.

detailed Information on the content of carbonyl and carboxyl groups and the middle Molecular weight, surface pH and other properties Kings can be used, on the one hand the current Status and on the other hand the future chemical and physical Behavior of the cellulose and / or products made from it, especially paper, to be able to predict or quality judgments to substantiate.

In particular, the carbonyl group content allows statements about the oxidative stress, the pulp or paper in the past z. B. has been suspended during processing ( Lewin, M., Macromol. Symp. 1997 (118): 715-724 ).

carbonyl react more strongly to light and temperature, the paper So is against light and climatic fluctuations sensitized. Consequently, increased photosensitivity can result of the pulp / paper examined.

The same applies to carboxyl groups ( Chirat, C., De la Chapelle, V., Journal of Pulp and Paper Science 1999 (25): 201-205 ).

There due to optical properties not on the Carbonylgruppengehalt can be closed - an oxidatively bleached paper can z. B. appear very bright and at the same time a high carbonyl group content - it is necessary to have sound knowledge and have meaningful analytical data.

Carboxyl groups are the last oxidation state, they are formed from carbonyl groups and are therefore also an indicator of oxidative stress. Furthermore, the H ⁺ ion of the carboxyl group may dissociate to give an internal acid. At the resulting COO group transition metal ions can bind to the cellulose, which adversely affect the stability of the molecule.

The average molecular mass, however, gives information about the average Length of the cellulose molecules that are in one Pulp / paper sample are located. Thus, these are conclusions allowed on the mechanical stability.

The pH gives an indication of the acids present in the paper. It is acknowledged that acid hydrolysis is a major factor in paper degradation, and a low pH therefore indicates that degradation is already occurring and degraded durability expectations. But also a higher pH value can lead to the oxidative destruction of the cellulose ( Knill, CJ, Kennedy, JF Carbohyd. Pole. 2003 (51): 281-300 ).

Other parameters, such as paper thickness or whiteness, support the description and assessment of paper quality. Under paper quality here is a term for the description of the chemical and mechanical stability of papers understood.

Of the Whiteness and paper thickness are taken individually rather ambivalent parameters. They do not give a clear name alone Information about the state of paper and can even be contradictory.

The trend However, it is to be expected that a thick paper of rather light color better durability properties than a thin, already strongly discolored paper will have.

Naturally paper thickness and whiteness can also be determined by means of known, conventional methods are determined non-destructively, However, their determination can also with the invention Procedures are performed, for. B. supplementary to other chemical properties.

All the parameters mentioned in common, carbonyl and carboxyl group content as well as average molecular weight, surface pH, but also whiteness and paper thickness, provide information on the current state and the aging tendency of paper ( Whitmore, PM, Bogaard, J., Restorer 1995 (16): 10-30 ). They allow statements about the future behavior of the paper under climatic and mechanical stress caused by transport and storage.

Farther allow planning of appropriate stabilization measures and the consequent necessary treatment.

In Recently, a combination of IR spectroscopy has been successful and multivariate calibration models used to prepare chemical and physical properties of pulp, cellulose, etc. in paper and pulp production.

there are values that are usually complicated with wet-chemical analysis methods must be obtained, with reference parameters from infrared spectra in both the near infrared (NIR) and the medium infrared (MIR), by means of a suitable multivariate Calibration linked in a model, which then the wet chemical Can replace analytics.

This Procedure has already been in the food industry, pharmaceutical industry as well as wood, pulp and paper analysis. Data that otherwise only with considerable effort by wet-chemical analysis methods can be accessed this way quickly and without sampling and thus determined non-destructive become.

In In pulp analysis, this combination became predictive from Z. B. Strength properties of pulp used.

The U.S. Patents 5,680,321 and 5,638,284 For example, methods for determining physical properties of pulp and / or paper, such as wet strength, tear resistance, hydrophilicity and the like, describe.

The U.S. Patent 6,476,915 describes methods for determining paper thickness, basis weight, paper moisture, optical properties, modulus of elasticity, tear propagation resistance, bursting strength, breaking length of paper and / or cardboard on a moving web.

• (I) die Entwicklung eines Kalibrationsmodells durch
• (Ia) Aufnahme von Absorptions-, Reflektions- oder Emissionsspektren von Referenzproben an Papier mit bekannten Eigenschaften,
• (Ib) Bearbeiten der Rohdaten aus den Spektren und
• (Ic) Analysieren der Daten mittels chemometrischer Methoden,
• (II) Bestimmen der unbekannten Eigenschaften von Papierbrei und/oder Papier durch
• (IIa) Aufnahme von Absorptions-, Reflektions- oder Emissionsspektren von Papierproben mit unbekannten Eigenschaften,
• (IIb) Bearbeiten der Rohdaten und
• (IIc) Anwendung des entwickelten Kalibrationsmodells aus (I).

These methods include

• (I) the development of a calibration model
• (Ia) recording absorption, reflection or emission spectra of reference samples on paper with known properties,
• (Ib) Processing the raw data from the spectra and
• (Ic) Analyzing the Data Using Chemometric Methods
• (II) determining the unknown properties of pulp and / or paper
• (IIa) recording absorption, reflection or emission spectra of paper samples with unknown properties,
• (IIb) edit the raw data and
• (IIc) Application of the developed calibration model from (I).

Thus, no spectra of the wet-chemically examined reference samples are recorded there men.

Furthermore, NIR spectra are also used to determine the yield, the kappa number, the content of lignin, glucose, xylose and uronic acids in pulps ( Olsson, RJO, Tomani, P., Karlsson, M., Joseffson, T., Sjöberg, K., Björklund, C., Tappi Journal 1995 (78): 158-166 ).

Other parameters that have been determined are intrinsic viscosity, degree of polymerization, copper number and whiteness of various pulps and papers ( Fardim, P., Ferreira, MMC, Durán, N., Journal of Wood Chemistry and Technology 2002 (22): 67-81 ).

MIR spectra in connection with multivariate calibration have been so far for the prediction of extractives, lignin, total carbohydrate content and density, but not yet for the determination of chemical properties, which correlate with the oxidative stress of the paper, in particular not used in the content of carbonyl and carboxyl groups.

The in the prior art described wet chemical processes for Determination of properties of pulp or strength properties of papers, such as intrinsic viscosity and copper number, as frequently predicted chemical properties, or the physical methods for determining, for example, the tensile strength, zero tensile strength and bursting strength have the disadvantage that for the determination of these parameters great Sample volumes and multiple repetitions are needed.

One further disadvantage of the determination of mechanical paper properties is that they cause oxidative damage to the cellulose reflect only at high levels of damage, so high Carbonyl and carboxyl groups or strongly reduced middle Molecular weights.

III. Presentation of the invention

a) Technical task

It It is therefore an object of the invention to provide a method for a non-destructive as accurate as possible Statements about the chemical properties of cellulose and to make paper and this to another already in a very early stage of the injury of cellulose and Paper in which the mechanical properties of the paper barely are affected.

b) Solution of the task

These The object is achieved by the features of independent claims 1, 11 and 12 solved, with advantageous developments in the dependent Claims are given.

The The invention provides a method for nondestructive determination the carbonyl and carboxyl group content, the average molecular weight and the surface pH. Other parameters, such as As the whiteness of pulps and papers, and The thickness of papers can also be determined by this method become.

1. inventive aspect

• (a) chemisch-analytische Bestimmung der chemischen Eigenschaften von Zellstoff- oder Papierreferenzproben, um Referenzwerte zu erhalten;
• (b) Messen der Absorption oder Reflexion derselben Zellstoff- oder Papierreferenzproben aus Stufe (a) im NIR- und/oder MIR-Bereich, um Referenzdaten zu erhalten;
• (c) gegebenenfalls mathematische Vorbehandlung der Referenzdaten;
• (d) Erzeugung von Referenz-Kenngrößen für jede Referenzprobe durch mathematische Auswertung der aus Stufe (b) oder (c) erhaltenen Referenzdaten;
• (e) Korrelation der aus Stufe (d) erhaltenen Referenz-Kenngrößen mit den aus Stufe (a) erhaltenen Referenzwerten mittels multivariater Kalibration zur Erstellung von Korrelationsmodellen, dadurch gekennzeichnet, dass für Schritt (a) eine solche chemisch-analytische Methode ausgewählt wird, die eine quantitative Bestimmung der Referenzwerte ermöglicht.

The first aspect of the invention relates to a method for the development of correlation models for the nondestructive determination of chemical properties of pulp or paper by means of IR spectra, which comprises:

• (a) chemical analytical determination of the chemical properties of pulp or paper reference samples to obtain reference values;
• (b) measuring the absorption or reflection of the same pulp or paper reference samples from step (a) in the NIR and / or MIR range to obtain reference data;
• (c) optionally mathematical pretreatment of the reference data;
• (d) generating reference characteristics for each reference sample by mathematically evaluating the reference data obtained from step (b) or (c);
• (e) correlation of the reference parameters obtained from step (d) with the reference values obtained from step (a) by means of multivariate calibration to produce correlation models, characterized in that for step (a) such a chemical-analytical method is selected which a quantitative best the reference values.

The chemical properties of the reference samples are wet-chemical Methods determined. Under "chemical properties" are herein preferred is the carbonyl and carboxyl group content and the mean molecular weight, carbonyl and carboxyl group content are particularly preferred.

According to the first aspect of the invention is a method to develop a correlation model with which it is possible to very accurately determine the content of carbonyl and Carboxyl groups, the average molecular weight and the surface pH of pulp and paper only from IR spectra.

in the The following are the stages of the method according to the first aspect the invention explained in more detail.

Stage (a)

First the chemical-analytical determination of the sought-after chemical Property of pulp and paper reference samples, by means of Reference methods that provide a very accurate determination of this property, z. For example, the content of carbonyl or cellulose contained in the pulp or paper. and carboxyl groups and the average molecular weight.

Of the pH of the paper surface is determined by TAPPI (T 529 om-88) as the reference method.

• – die CCOA-Methode, die beispielsweise in Biomacromolecules 2002 (3): 959–968 ; Biomacromolecules 2002 (3): 969–975 ; und Biomacromolecules 2003 (4): 743–747 beschrieben ist, und
• – die FDAM-Methode, die in Biomacromolecules 2006, 7 (6), 1743–1750 , beschrieben ist, zur Anwendung.

As a reference method for the determination of the content of carbonyl and carboxyl groups and the determination of the average molecular weight come

• - the CCOA method, for example, in Biomacromolecules 2002 (3): 959-968 ; Biomacromolecules 2002 (3): 969-975 ; and Biomacromolecules 2003 (4): 743-747 is described, and
• - the FDAM method used in Biomacromolecules 2006, 7 (6), 1743-1750 , is described, for use.

Likewise, other methods, such. B. by means of hydrazine ( Analytica Chimica Acta 1962 (27): 434-440 ) or the copper number (TAPPI T 430 om-94) can be used to determine the carbonyl group content, which results in significantly less accurate values and thus inaccurate correlation models.

Prefers are thus the CCOA and the FDAM method, since thereby the content on carbonyl and carboxyl groups present in pulp and paper and the average molecular weight very precisely in the form of reference values can be determined.

Of Furthermore, these methods have the advantage that they already deliver very accurate values with very low content of these groups.

In order to can be oxidative damage already in an early stage Stage can be detected, which is not possible with other methods is, for. B. from a content of 3 .mu.mol / g of carbonyl groups and a content of 10 μmol / g of carboxyl groups.

in the Compared to previous publications and patents that have the characteristics of pulp or the strength properties of papers, are the errors of the reference methods used here, in particular concerning carbonyl and carboxyl groups, average lower, so that the resulting correlation models are much more accurate are.

So show the determination methods chosen here even if the wet chemical provisions over a period of collected at least one year, at a selected Pulp has a standard deviation of still less than 5%, what a very good reproducibility (within a Laboratories) of the results obtained by these methods.

The Intrinsic viscosity and copper number, as common certain chemical properties are also average higher errors (intrinsic viscosity according to TAPPT T 230 om-94, repeatability within a laboratory 4%, reproducibility between two laboratories 19%, copper number according to TAPPT T 430 om-94, repeatability within a laboratory 10%).

Common methods for determining the strength, on the other hand, are the tensile strength (repetition within one laboratory and reproducibility between two laboratories according to TAPPT 404 om-92 4-13%), the zero tensile strength (reproducibility within a laboratory and reproducibility between two laboratories according to TAPPT 231 cm-96 3-10%) and the bursting strength ( Repeatability within a laboratory and reproducibility between two laboratories according to TAPPT 403 om-97 22-28%), which, however, still provide inconspicuous values for a small cause of damage.

Stage (b)

In this stage of the process becomes the absorption or reflection of the same Pulp and / or paper reference samples in the NIR and / or MIR range from which in step (a) the sought chemical properties determined by chemical analysis. The wavelength range is to be determined so that the sought chemical property reflected in the spectral data.

For this purpose, spectra of these reference samples are recorded in the range of the near (NIR) and / or middle (MIR) infrared, thereby obtaining reference data. The NIR covers the wavelength range between 12,800 and 4,000 cm ^-1 and the MIR covers the wavelength range between 4,000 and 400 cm ^-1 .

It is recommended, inter alia, for the determination of meaningful reference data, that the measurement of the spectra at the reference samples the entire relevant wavelength range, ie at NIR of 12,800-4,000 cm ^-1 and MIR from 4,000 to 400 cm ^-1 , or both together, includes.

Further It should be clarified that for each chemical property, for example, the content of a particular chemical-functional group, separate reference data determined for which, however, the same spectra, possibly only differentiated according to the wavelength range, can be used.

Out determined from the individual samples of a set of reference samples Spectra may also have an average range before further processing be generated, especially if several samples have a very similar Reference value.

Stage (c):

Preferably the reference data of the recorded spectra are pretreated, irrelevant, accidental or systematic sources of variation to reduce or completely remove.

• – verschiedene Standardvorbehandlungen, wie Normierungen, Minimum-/Maximum-Normierung und/oder besonders Vektornormierung (Standard Normal Variate Transformation, SNV) ( Zeaiter, M., Roger, J.- M., Bellon-Maurel, V., TrAC – Trends in Analytical Chemistry, 24 (5), 2005: 437–445, Robustness of models developed by multivariate calibration. Part II: The influence of pre-processing methods ) und/oder Subtraktion einer Geraden (straight line substraction), und/oder mathematische Ableitungen (1. und 2.), und/oder
• – multiplikative Streuungskorrektur (Multiplicative Scatter Correction/MSC), beschrieben in Isaksson, T., Naes, T., Effect of multivariate scatter correction (MCS) and lineartity improvement in NIR spectroscopy, Applied Spectroscopy 42 (7) 1988: 1273–1284 , und/oder
• – erweiterte Multiplikative Streuungskorrektur (Extended Multiplicative Scatter Correction/EMSC,) beschrieben in Martens, H., Stark, E., Extended multiplicative signal correction and spectral interference subtraction: New preprocessing methods for near infrared spectroscopy, Journal of Pharmaceutical and Biomedical Analysis, 9 (8) 1991: 625–635 und/oder
• – orthogonale Signalkorrektur (Orthogonal Signal Correction/OSC, US-Patent 6,754,543 ), beschrieben in Wold, S., Antti, H., Lindgren, F., Öhman, J., Orthogonal signal correction of near-infrared spectra, Chemometrics and Intelligent Laborstory Systems, 44 (1–2) 1998: 175–185 und in Fearn, T., On orthogonal signal correction, Chemometrics and Intelligent Laborstory Systems, 50 (1) 2000: 47–52 und/oder
• – multivariate Curve Resolution (MCR), beschrieben in Schoonover, J. R., Marx, R., Zhang S. L., Multivariate curve resolution in the analysis of vibrational spectroscopy data files, Applied Spectroscopy, 57 (5) 2003: 154A–170A
• – bzw. Kombinationen der genannten von Datenvorbehandlungsmethoden.

Various methods exist to overcome these problems. From these known methods, the following are selected according to the invention:

• Various standard pretreatments, such as standardization, minimum / maximum normalization and / or particularly standard vector transformation (SNV) ( Zeaiter, M., Roger, J.-M., Bellon-Maurel, V., TrAC Trends in Analytical Chemistry, 24 (5), 2005: 437-445, Robustness of models developed by multivariate calibration. Part II: The influence of pre-processing methods ) and / or subtracting a straight line (substraction), and / or mathematical derivatives (1 and 2), and / or
• Multiplicative Scatter Correction (MSC), described in Isaksson, T., Naes, T., Effect of Multivariate Scatter Correction (MCS) and linearity improvement in NIR spectroscopy, Applied Spectroscopy 42 (7) 1988: 1273-1284 , and or
• - Extended Multiplicative Scatter Correction (EMSC,) described in Martens, H., Stark, E., Extended multiplicative signal correction and spectral interference subtraction: New preprocessing methods for near infrared spectroscopy, Journal of Pharmaceutical and Biomedical Analysis, 9 (8) 1991: 625-635 and or
• Orthogonal signal correction (Orthogonal Signal Correction / OSC, U.S. Patent 6,754,543 ), described in Wold, S., Antti, H., Lindgren, F., Öhman, J., Orthogonal signal correction of near-infrared spectra, Chemometrics and Intelligent Laboratory Systems, 44 (1-2) 1998: 175-185 and in Fearn, T., On Orthogonal Signal Correction, Chemometrics and Intelligent Laboratory Systems, 50 (1) 2000: 47-52 and or
• Multivariate Curve Resolution (MCR), described in Schoonover, JR, Marx, R., Zhang SL, Multivariate curve resolution in the analysis of vibrational spectroscopy data files, Applied Spectroscopy, 57 (5) 2003: 154A-170A
• - or combinations of said data pre-treatment methods.

at The mathematical pretreatment will be the reference data for each chemical property to be determined separately, in particular treated with different methods.

Stage (d):

In This level will be the one for each reference sample in step b) or c) obtained reference data reference characteristics determined for each chemical property to be determined separately. The goal is to recognize of course occurring groupings or classes or possible Outliers, d. H. untypical reference data or reference characteristics. Because only in this way is it possible for the next step for Each chemical property to be determined has its own correlation model to provide.

Especially becomes in this step from the reference data of the spectra as a reference characteristic determines a sum parameter that matches the content of a particular chemically functional group or another chemical property correlated.

• – Principle Component Analysis (PCA, Hauptkomponentenanalyse)

The raw or pretreated measurement data from step b) or c) of the spectra are loaded into a corresponding computer program. There, multivariate methods, such as computational methods for orientation data analysis or structure recognition, are used to get an insight into the data structure:

• - Principle Component Analysis (PCA, Principal Component Analysis)

at Principal Component Analysis (PCA) is about getting a record to study natural group formation without that a certain class affiliation is assumed. It is a mathematical transformation of a data matrix with the aim of reducing the variance contained in the data by a smaller one Number of variables to represent. These variables are called "factors" or "main components". A new coordinate system is calculated, in which the samples newly defined axes instead of the original measurement coordinates be assigned.

In order to can be based on the reference parameters generated in this stage Reference samples determined from the total number of reference samples be selected later for the Correlation in stage (e) should be used as it is a specific Paper type or have other common parameters.

Stage (s):

In this step is correlated between those obtained from step a) wet-chemically determined reference values and those from stages b) to d) using the spectra determined reference parameters produced by multivariate calibration. The advantage of a multivariate data calibration against the univariate Data calibration is that not like a univariate Data calibration is a measure of a sample is assigned to create a calibration function, but multiple measures (typically spectra with a plurality of individual data points) with a property or several properties of a sample can. This will be for each chemical to be determined Property achieved its own correlation model.

• – Soft Independent Modeling of Class Analogies (SIMCA).

First, the unknown sample to be determined must be classified, ie assigned to a specific class (pulp, rag paper, etc.). One possible method for this is:

• - Soft Independent Modeling of Class Analogies (SIMCA).

If used a known class affiliation is to develop a model for future samples SIMCA is used. For SIMCA becomes a multidimensional box for each classifier Affiliation created by PCA. The classification future samples is to place the samples in the on best fitting box to classify.

The data analysis performed in step (d) can be used to classify the reference samples using SIMCA. This classification is necessary in that a clear assignment of different classes of paper (ie pulp, rag paper, groundwood paper, hybrids, etc.) to a model specific to this class of paper is necessary in order to establish reliable property determinations of the model.

• – Multiple Lineare Regression (MLR), und/oder
• – Partial Least Squares Regression (PLS-R), und/oder
• – Hauptkomponentenregression (Principle Component Regression, PCR).

By mathematical methods, namely multivariate methods for data analysis, applied, are obtained from the measured data of the spectra reference characteristics, such. By:

• - Multiple Linear Regression (MLR), and / or
• - Partial Least Squares Regression (PLS-R), and / or
• - Principal Component Regression (PCR).

The multiple linear regression (MLR) has the task of describing the relationship between more than two variables X ₁ , X ₂ ... and thus to arrive at prognostic statements for a variable Y considered to be dependent on which it is assumed depends not only on one but several variables X.

The Partial Least Squares Regression (PLS-R) is a statistical procedure for the estimation of causal models. The PLS-R is divided in PLS-1 and PLS-2. While in PLS-1 a variable (Paper property) for the correlation with the spectra PLS-2 can handle multiple variables at the same time. The be obtained from the spectra characteristics used to model the desired quality parameters.

The principle component regression (PCR) is different from the PLS in how from collinear spectra an independent Matrix is calculated. Although the mathematical algorithms are different, the two methods are common considered comparable. Your application can too related to the available computer program.

One such created calibration model must be validated, for example through cross-validation or test-set validation, on the one hand to test the quality of the created calibration model and on the other hand also, for each chemical to be determined Property a specific, most appropriate method for select the multivariate data analysis in step (e).

For example a cross validation is performed, wherein for the calibration according to e) and the following Validation using the same sample set and record becomes.

at the cross validation is in each case - separately for every chemical property to be determined - a sample, to construct the correlation model according to e) was used, left out and a new calibration model without she created. This new calibration model is used to the chemical property of the sample omitted only by reference the reference parameter, ie based on the recorded Spectra to determine.

the one multivariate reduction method, which in this way a Reference characteristic showed that the best wet-chemical the reference value of this omitted sample, is thus considered to be the specific chemical property most suitable multivariate reduction method selected.

At the Example of over the concentration of a specific functional chemical group applied spectral sum parameter and the calibration line generated from these values, this means that one the value pairs is initially omitted and from the rest Value pairs a new calibration line is formed and as the best reduction method the one is selected, in which the deviation of the new calibration line from the original calibration line, which consists of all value pairs was created is the lowest.

The Correlation with cross validation is thus a summary the correlations of each sample through the model.

to further optimization of this correlation model can now for more reference samples that are not in the build of the correlation model, according to the Steps b) to d) determines reference parameters and with the other wet-chemically determined reference values These new reference samples are compared with the validation now consists solely of the reference parameters and the matching model the desired values without any to determine wet-chemical analysis.

2. Inventive aspect

After this a correlation model between wet chemical reference values and by reference to IR spectra determined reference parameters created and, if necessary, through validation and further optimization can be improved by applying this correlation model to an unknown sample a chemical property of interest, For example, the content of carboxyl groups alone by the application the correlation model according to step (e) and determining one according to steps (b) to at least (c), better (d) determined spectral characteristic be determined.

For this Advantageously, NIR or MIR spectra will be directly on the intact paper or pulp without special sample preparation, preferably in reflection mode, measured. This may be due to of the correlation model are dispensed with sampling the method extensively for the investigation of cellulosic materials and very sensitive papers can be used as no sampling more is needed.

By the nondestructive recording of the IR spectrum from the From this determined characteristic of the unknown sample becomes immediately the desired one from the correlation model chemical property, in particular the content of a particular chemical-functional group, obtained without being a wet-chemical Treatment and thus a partial destruction of the paper sample becomes necessary according to step (a).

This will ultimately also by using the above listed, optimal reduction methods for individual properties (CCOA and FDAM) allows, through the correlation models can be developed with which it is possible is already at a very early stage the state in particular the oxidative damage of cellulosic material or paper.

3. Inventive aspect

• – unter Ermittlung einer Kenngröße für diese chemische Eigenschaft mittels Aufnahme von Infrarot-Spektren und daraus gewonnenen Daten; und
• – die Auswahl eines von mehreren zur Verfügung stehenden Korrelationsmodellen für diese chemische Eigenschaft gemäß dem ersten und zweiten Aspekt der Erfindung.

Another aspect of the invention is to determine an unknown chemical property of a new pulp or paper sample according to the invention

• - Determining a characteristic for this chemical property by recording infrared spectra and derived data; and
• The selection of one of several available correlation models for this chemical property according to the first and second aspects of the invention.

• (i) Messen der Absorption oder Reflexion des Zellstoffs oder Papiers im NIR- und/oder MIR-Bereich;
• (ii) gegebenenfalls mathematische Vorbehandlung der in Stufe (i) erhaltenen Messdaten;
• (iii) Erzeugung von Kenngrößen durch mathematische Auswertung der aus Stufe (ii) erhaltenen Referenzdaten; dadurch gekennzeichnet, dass
• (iv) Auswahl eines der gemäß den Ansprüchen 1 bis 20 erstellten Korrelationsmodells,
• (v) die Bestimmung der unbekannten chemischen Eigenschaften durch Anwendung des Korrelationsmodells der Stufe (iv) alleine anhand der in Stufe (iii) erhaltenen Kenngrößen ohne Probennahme direkt an der intakten Zellstoff- oder Papierprobe erfolgt.

Such a method comprises the following steps:

• (i) measuring the absorption or reflection of the pulp or paper in the NIR and / or MIR region;
• (ii) optionally mathematical pretreatment of the measurement data obtained in step (i);
• (iii) generating characteristics by mathematically evaluating the reference data obtained from stage (ii); characterized in that
• (iv) selecting a correlation model created according to claims 1 to 20,
• (v) determining the unknown chemical properties by applying the correlation model of step (iv) solely on the basis of the parameters obtained in step (iii) without sampling directly on the intact pulp or paper sample.

there Steps (i), (ii) and (iii) correspond to steps (b), (c) and (d) according to the first two aspects of the invention.

The in steps (v) and (iv) to be selected and applied Correlation model is one of the in step (e) of the first Aspect of the invention resulting correlation models.

there are made from the same reference data as in step (b) or (c) by using different multivariate reduction methods obtained different correlation models.

Farther can or must be based on the parameters generated in step iii) also a classification of the one or more unknown samples take place, for example, according to the basic paper (cellulose paper or rag paper) or other sample-determining parameters, z. B. by means of SIMCA, since partially only then the selection of the calibration model can be done.

If this happens, the selection of the appropriate correlation model for the determination of the ge sought chemical property to be selected.

• – Restaurierung/Konservierung von Kunst- und Kulturgut (historische Papiermaterialien, besonders Hadernpapiere)
• – Zellstoffproduktion/Zellstoffverarbeitung/Papierherstellung (verringerte Analysezeiten, Vorhersagen von Vergilbungsneigung, Weißgradstabilität und Lagerfähigkeit).

The methods for determining the carbonyl and carboxyl group content and the average molecular weight, the surface pH, the paper thickness and the whiteness are particularly suitable for the non-destructive examination of cellulosic materials in two areas:

• - Restoration / conservation of artistic and cultural property (historical paper materials, especially rag paper)
• Pulp production / pulp processing / papermaking (reduced analysis times, prediction of yellowing tendency, whiteness stability and shelf life).

The advantage of the method according to the invention is clear from the following applications:
By non-destructive determination of parameters such as carbonyl and carboxyl group content and average molecular weight, recommendations can be made for the further treatment of pulps and papers. The following table lists three chemical properties to be determined, namely carbonyl group content, carboxyl group content and average molecular weight.

The influence of these parameters on pulp and paper is shown. These are chemo-physical properties that relate to interactions with environmental conditions and other treatment steps. These factors have a corresponding effect on further measures that are recommended to be carried out or even avoided with the pulp or paper concerned. It deals in particular with practical aspects of processing, use and storage. parameter influence effects carbonyl group • Light stability • Alkaline stability • Resistance to external influences • Interactions with carboxyl groups • Long-term stability • storage, transport • exhibition, loan • deacidification treatment • bleaching treatments • further processing carboxyl • Light stability • Water absorption capacity • Ion binding capacity • Resistance to external influences • Storage • Exhibition loan • Aqueous treatments • Deacidification treatments • Value processing Mean molecular weight • Mechanical strength • Long-term stability • usability • exhibition, loan • all treatment, especially watery • further processing

Some These relationships are explained below by way of example.

The light stability, the alkaline stability, the mechanical strength, the long-term stability, the stability to environmental influences and the water absorption capacity decrease as the number of carbonyl groups in the pulp or paper increases. In transport, loan, storage, further processing, restorative treatment, usability, this must be taken into account by minimizing contact with light, acid and water, as well as mechanical stresses.

takes the number of carboxyl groups increases, the acid stability decreases and the stability of the pulp or paper Environmental influences.

In contrast, the ion binding capacity and the water absorption capacity increase with increasing Number of carboxyl groups too.

Increasing the carboxyl group content leads to a deterioration of some pulp or paper properties while others increase. It therefore depends on the further course of action whether this increase is rated as positive or negative.

The decrease in the average molecular weight affects the mechanical strength, the long-term stability, the acid stability and the stability against environmental influences in such a way that these factors decrease in each case. The decrease in the average molecular weight thus leads to a deterioration of the pulp or paper.

from that let statements be made, like the materials in the future further processed, stored and transported, which conditions are required for the exhibition of exhibits are, so that the exhibits take as little harm as possible, or how to restore pieces to be treated. Also, statements can be made about which conditions must be met when lending.

In summary, the effects of these three chemical properties are:

The using the MIR and NIR spectra resulting analytical Method for the investigation of cellulose in the form of pulp and Paper sheets are also called predictive models.

The Applicability of the predictive models can be for one new pulps (papers, hand sheets), as used in the paper industry find wide application, and for historical Rag paper, which is relevant for conservatory purposes are to be shown.

There the historical rag paper both in their color as well in their composition significantly from modern pulp sheets Different, different models for pulp paper on the one hand and rag paper on the other hand.

c) embodiments

A Embodiment according to the invention in the following explained by way of example with reference to the figures. Show it:

1 : a variety of recorded spectra,

2 : Scoreplot of spectral data,

3 : Correlation of wet-chemically determined values with the spectral characteristics,

4 : Validation of the correlation model 3a .

5 : tabular representation of different correlation models, and

6 : Graphical representations of some of the correlation models 5 ,

The Figures show the procedure for creating a correlation model, with its help later on new additional Sample chemical properties - such as the carbonyl group content - non-destructive exclusively by means of spectral measurements and application of the correlation model should be determined. The correlation model is created using available reference samples, from which also removed material shares and wet-chemically processed be allowed to.

1a shows the NIR spectra generated by these reference samples, in this case 100 spectra from 100 different pulp reference samples.

Since the frequency range of 12,000 cm ^-1 - at least 9,000 cm ^{-1 contains} no characteristic rashes, it is not included in the modeling. In 1b the selected location is displayed (step b) or i)).

Subsequently be - according to step c) or ii) - this Subjected spectral data to a mathematical pretreatment and furthermore from these spectral data for each individual Sample a reference characteristic determined according to step d) or iii).

To Fixing on a particular mathematical pretreatment is with The data pretreated in this way is the subject of a principal component analysis. This supplies several main components.

2 FIG. 12 shows the exemplary graphical representation of samples in which the second versus the fourth major component is plotted and re-sorted for the largest occurring variance. FIG.

It In such a representation often also shows a Classify the samples by adding value pairs in two or more individual areas are highly concentrated, indicating the presence a common strong characteristic, z. B. one Main component, in each case the samples suggesting the a point in the diagram are concentrated.

After selecting the most suitable mathematical reduction method, determining the principal components, and determining the obvious outliers among the reference samples and omitting to proceed, the remaining reference samples are determined according to 3a the correlation between the reference characteristics of the chemical property sought for each sample from the spectral data and the reference value of the same property previously determined wet-chemically from the same sample.

in the In the present case, the correlation with the module PLS-1 was the Software Opus Quant created from the existing value pairs also immediately marks a resulting calibration line. The investigated chemical property here is the content of carbonyl groups.

3b shows an analogous representation for the same samples for the content of carboxyl groups.

Such a selected correlation model requires further optimization, first by validation, as in 4a for the carbonyl group content and in 4b shown for the carboxyl group content:
There is by means of a test set, ie a group of new reference samples, in accordance with the creation of the correlation model 3a was not used again, a comparison of the wet chemical values with the spectrally determined parameters, which are obtained with the same pretreatments and reduction methods and their location to that of the correlation model of 3a drawn Eichstraaden.

Provided the deviations of the test set from the calibration line on average not larger than the one at the time of creation the correlation model with the first scope of reference samples, the correlation model can be classified as successful.

is If this is not the case, it is necessary to search for a new correlation model become.

at the determination of one and the same chemical property on the same Group of reference samples are used alone by using different mathematical pretreatment methods and mathematical reduction methods different reference parameters from the reference samples can be obtained, and thus in the - at the investigation of unknown samples - several new ones spectral characteristics. The juxtaposition these different parameters with each the same wet-chemically determined values in a correlation model thus results in different correlation models for the same chemical property.

5 shows in tabular representation for a series of samples, the wet-chemically determined "measured values" shown in the first column of values and compared the spectral parameters KG1 to KG7 in the last four columns determined by different pretreatment and reduction methods.

The correlation of the column KG1 with the "measured value" is in 6a graphically represented, that of the column KG3 in 6b ,

The comparison of these two 6a and 6b and the correlation coefficients given there on the top right each, clearly shows that the correlation model according to 6a better correlated than that of 6b , which is already visible graphically from the position of the pairs of values for the correlation line (calibration line) resulting from the individual value pairs, but also from the 6a closer to the ideal value of 1.0 correlation coefficients.

• – für die Bestimmung der mittleren Molmasse: die erste Ableitung (bevorzugter Spektralbereich: 4250–7500 cm^–1),
• – bei der Bestimmung des Gehalts an Carbonylgruppen: zweite Ableitung (bevorzugter Spektralbereich: 4250–7500 cm^–1),
• – bei der Bestimmung des Gehalts an Carboxylgruppen: und Multiplicative Scattering Correction (bevorzugter Spektralbereich: 4600–6100 cm^–1), Bei Hadernpapieren:
• – für die Bestimmung der mittleren Molmasse: Straight Line Substraction (bevorzugter Spektralbereich: 6100–7500 cm^–1,
• – bei der Bestimmung des Gehalts an Carbonylgruppen: erste Ableitung, Multiplicative Scattering Correction (bevorzugter Spektralbereich: 3800–4500 cm^–1 sowie 5000–5300 cm^–1 sowie 6000–7100 cm^–1),
• – bei der Bestimmung des Gehalts an Carboxylgruppen: erste Ableitung, Vektornormierung (bevorzugter Spektralbereich: 3800–7500 cm^–1),
• – Oberflächen-pH-Wert: Multiplicative Scattering Correction (bevorzugter Spektralbereich: 3800–5400 cm^–1 sowie 5800–7500 cm^–1).

In summary, it can thus be said that the following mathematical pretreatments have proved to be advantageous:
For pulp papers:

• For the determination of the average molecular mass: the first derivative (preferred spectral range: 4250-7500 cm ^-1 ),
• In the determination of the content of carbonyl groups: second derivative (preferred spectral range: 4250-7500 cm ^-1 ),
• In the determination of the content of carboxyl groups: and Multiplicative Scattering Correction (preferred spectral range: 4600-6100 cm ^-1 ), In rag paper:
• For the determination of the average molecular mass: straight line subtraction (preferred spectral range: 6100-7500 cm ^-1 ,
• In the determination of the content of carbonyl groups: first derivative, multiplicative scattering correction (preferred spectral range: 3800-4500 cm ^-1 and 5000-5300 cm ^-1 and 6000-7100 cm ^-1 ),
• In the determination of the content of carboxyl groups: first derivative, vector normalization (preferred spectral range: 3800-7500 cm ^-1 ),
• Surface pH: Multiplicative Scattering Correction (preferred spectral range: 3800-5400 cm ^-1 and 5800-7500 cm ^-1 ).

QUOTES INCLUDE IN THE DESCRIPTION

This list The documents listed by the applicant have been automated generated and is solely for better information recorded by the reader. The list is not part of the German Patent or utility model application. The DPMA takes over no liability for any errors or omissions.

Cited patent literature

• US 5680321 [0022]
• US 5638284 [0022]
• - US 6476915 [0023]
• - US 6754543 [0055]

Cited non-patent literature

• Lewin, M., Macromol. Symp. 1997 (118): 715-724 [0005]
• - Chirat, C., De la Chapelle, V., Journal of Pulp and Paper Science 1999 (25): 201-205 [0007]
• - Knill, CJ, Kennedy, JF Carbohyd. Pole. 2003 (51): 281-300 [0011]
• - Whitmore, PM, Bogaard, J., Restaurator 1995 (16): 10-30 [0016]
• Olsson, RJO, Tomani, P., Karlsson, M., Joseffson, T., Sjöberg, K., Björklund, C., Tappi Journal 1995 (78): 158-166 [0026]
• Fardim, P., Ferreira, MMC, Durán, N., Journal of Wood Chemistry and Technology 2002 (22): 67-81 [0027]
• - Biomacromolecules 2002 (3): 959-968 [0040]
• - Biomacromolecules 2002 (3): 969-975 [0040]
• Biomacromolecules 2003 (4): 743-747 [0040]
• - Biomacromolecules 2006, 7 (6), 1743-1750 [0040]
• Analytica Chimica Acta 1962 (27): 434-440 [0041]
• - Zeaiter, M., Roger, J.-M., Bellon-Maurel, V., TrAC - Trends in Analytical Chemistry, 24 (5), 2005: 437-445, Robustness of models developed by multivariate calibration. Part II: The influence of pre-processing methods [0055]
• Isaksson, T., Naes, T., Effect of Multivariate Scatter Correction (MCS) and linearity improvement in NIR spectroscopy, Applied Spectroscopy 42 (7) 1988: 1273-1284 [0055]
• Martens, H., Stark, E. Extended multiplicative signal correction and spectral interference subtraction: New preprocessing methods for near infrared spectroscopy, Journal of Pharmaceutical and Biomedical Analysis, 9 (8) 1991: 625-635 [0055]
• Wold, S., Antti, H., Lindgren, F., Öhman, J., Orthogonal signal correction of near-infrared spectra, Chemometrics and Intelligent Laboratory Systems, 44 (1-2) 1998: 175-185 [0055]
• Fearn, T., On Orthogonal Signal Correction, Chemometrics and Intelligent Laboratory Systems, 50 (1) 2000: 47-52 [0055]
• - Schoonover, JR, Marx, R., Zhang SL, Multivariate curve resolution in the analysis of vibrational spectroscopy data files, Applied Spectroscopy, 57 (5) 2003: 154A-170A [0055]

Claims (15)

A method of developing at least one correlation model for nondestructive determination of a chemical property of pulp or paper by IR spectra, comprising: (a) chemical analytical determination of the chemical property of the pulp or paper reference sample to obtain reference values; (b) measuring the absorption or reflection of the same pulp or paper reference sample from step (a) in the NIR and / or MIR range to obtain reference data; (c) optionally mathematical pretreatment of the reference data; (d) generating a reference characteristic for each reference sample by mathematically evaluating the reference data obtained from step (b) or (c); (e) correlation of the reference parameters obtained from step (d) with the reference values obtained from step (a) by means of multivariate calibration to produce a correlation model, characterized in that for step (a) such a chemical analytical method is selected which allows a quantitative determination of the reference value. Method according to claim 1, characterized in that the reference value is chosen from carbonyl group content, carboxyl group content, average molecular weight and surface pH of the pulp or paper is selected. Method according to claim 2, characterized in that the reference value from step (a) is the carbonyl or carboxyl group content is. Method according to claim 2 or 3, characterized that the reference value is determined by the CCOA method or FDAM method becomes. Method according to claim 2, characterized in that that the reference value from step (a) is the pH of the pulp or Paper surface is. Method according to claim 5, characterized in that that the reference value for the pH is determined according to TAPPT becomes. Method according to Claim 6, characterized in that the measurement of the spectra in step (b) covers the entire relevant wavelength range, in particular for measurements in the mid-infrared range (MIR) of 4,000 to 6,000 cm ^-1 and in the near infrared range (NIR) of 12,800 to 4,000 cm ^{-1 is} performed. Method according to one of claims 1 to 7, characterized in that generated by the in step (d) Reference parameters a classification of the reference samples is made. Method according to one of claims 1 to 7, characterized in that based on the in step (d) generated Reference characteristics specific reference samples be selected for the correlation in stage (e) are used. Method according to one of the preceding claims, characterized in that a validation of the obtained in step (e) Correlation models take place to one for each chemical property of the pulp or paper sample Method for multivariate data analysis in step e) select. Method according to claim 10, characterized in that that validation by means of cross validation or test set validation is carried out. Use of the correlation model according to one of Claims 1 to 11, for the non-destructive Determination of an unknown chemical property of pulp or paper by using steps (b) to (d) of the process according to one of claims 1 to 10 to an unknown sample, characterized in that solely on the basis of the thus determined characteristic the sample with the unknown property and with the help of the correlation model according to feature e) of claims 1 to 10 unknown chemical property of pulp or paper is determined without application of step (a). Method for nondestructive determination of an unknown chemical property of a cell fabric or paper sample by IR spectra, which comprises: (i) measuring the absorption or reflection of the pulp or paper sample in the NIR and / or MIR region; (ii) optionally mathematical pretreatment of the measurement data obtained in step (i); (iii) generating characteristics by mathematically evaluating the reference data obtained from stage (ii); characterized in that (iv) selecting one of the correlation models established in accordance with claims 1 to 11, and (v) determining the unknown chemical property solely by applying the correlation model of step (iv) to the characteristic obtained in step (iii) without Sampling directly on the intact pulp or paper sample. Method according to claim 13, characterized in that that based on the parameters generated in step (iii) a classification of the unknown pulp and paper samples he follows. Method according to claim 14, characterized in that that based on the classification, a suitable correlation model selected for determining the respective chemical property becomes.