Prediction of physical properties of superabsorbent polymers

Abstract

The present disclosure relates to a method for predicting physical characteristics, particularly performance parameters, of superabsorbent polymers.

Description

[0001] The present disclosure relates to a method for predicting physical characteristics, particularly performance parameters, of superabsorbent polymers. Background Art

[0002] Superabsorbent polymers (SAPs) are well-known materials that are commonly used in personal care products, such as diapers. These polymers are known to absorb several times their weight in fluids such as water, saline solutions, urine, blood and serum.

[0003] The preparation of water-absorbing polymer particles is described in the monograph “Modern Superabsorbent Polymer Technology”, FL Buchholz and AT Graham, Wiley-VCH, 1998, pp. 71-103.

[0004] An important requirement for superabsorbent polymers is the ability of the hydrogel to conduct liquid (permeability) and distribute liquid. The permeability of a superabsorbent is recorded in the form of saline flow conductivity (SFC). Other important performance parameters include the centrifuge retention capacity (CRC) of the superabsorbent particles, the absorption capacity under load, and the absorption rate, which is recorded as the amount of liquid absorbed per gram of superabsorbent per second, i.e., the free swelling rate (FSR); or the time required for the superabsorbent to absorb 20 g of liquid / g of superabsorbent (T20). Typically, water-absorbent polymer particles do not have a uniform particle size, but rather a particle size distribution (PSD). The particle size distribution also affects the performance of superabsorbent polymers.

[0005] The polymer chains of the water-absorbing polymer particles are crosslinked to one another. One effect of this is that the polymer particles are water-insoluble. The properties of the water-absorbing polymer particles can be adjusted via the amount of crosslinking agent used. As the amount of crosslinking agent increases, the centrifuge retention capacity (CRC) decreases and the absorption under pressure (AAP) passes through a maximum value.

[0006] In order to improve the use properties, such as permeability and absorption under load, the water-absorbing polymer particles are usually surface crosslinked. This increases the level of crosslinking at the particle surface and in this way can at least partially eliminate the mutual influence between absorption under load and retention capacity. Crosslinking agents suitable for surface crosslinking are compounds which can form covalent bonds with at least two carboxylate groups of the water-absorbing polymer particles.

[0007] The preparation of water-absorbing polymer particles is described, for example, in WO 01 / 038402 A1, WO 03 / 022896 A1, WO 03 / 051415 A1, WO 2006 / 034806 A1, WO 2006 / 034853 A1 and WO 2009 / 115472 A1.

[0008] WO 2005 / 097313 A1 discloses a process for producing water-absorbing polymer particles, in which an aqueous polymer gel is comminuted and pressed through a die plate by means of an extruder, dried and surface crosslinked to give polymer particles having high CRC, SFC and FSR.

[0009] Measuring the performance properties of superabsorbent polymers requires extensive analytical effort, especially if an industrial production process of superabsorbent polymers has to be monitored to ensure a constant quality of the produced product. It would be desirable to have a fast, simple and reliable method for determining the physical properties of superabsorbent polymers involving limited measurement effort.

[0010] One object of the present disclosure is to provide a reliable tool for predicting the physical properties, in particular the performance parameters, of superabsorbent polymers.

[0011] US 2004 / 133364 A1 discloses a method for determining and controlling a property of a polymer product (i.e., polyolefin particles) online in a slurry reactor system (such as a stirred slurry reactor or a slurry loop reactor). The method includes obtaining a regression model for determining a property of the polymer product, the regression model including a principal component input and a principal component score, obtaining a Raman spectrum of the polymer product in the slurry reactor system, calculating a new principal component score from at least a portion of the Raman spectrum and the principal component input, and calculating the polymer product property by applying the new principal component score to the regression model. The property can be controlled by adjusting at least one polymerization parameter based on the calculated polymer product property.

[0012] Adar, F. et al., "Raman Spectroscopy for Process / Quality Control", Applied Spectroscopy Reviews 32 (1997) 45-101 discloses the use of Raman spectroscopy in process control. The cross-linking of polyacrylamide gels was monitored to determine the percentage of reacted resin; and the chemical characteristics of the gel were determined. Summary of the invention

[0013] A computer-implemented method for predicting physical properties, particularly performance parameters, of superabsorbent polymers is provided. In the chemometric model of the present disclosure, spectral data (typically Raman scattering spectra) are associated with physical properties, particularly performance parameters, of superabsorbent polymers. A mathematical model based on a regression algorithm (e.g., a partial least squares regression algorithm) associates spectral data of superabsorbent polymers with their physical properties, particularly performance parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1An example of a set of 10 Raman spectra from a superabsorbent polymer is shown with subtraction of a linear fit as baseline correction and a SNV (standard normal variate) normalization step as data preprocessing;

[0015] Figure 2 Predicted online data and measured data for SFC of superabsorbent polymers are shown;

[0016] Figure 3 The correlation of predicted and measured CRC for a range of superabsorbent polymers is shown;

[0017] Figure 4 is a schematic diagram of a method for preparing a superabsorbent polymer;

[0018] Figure 5 The correlation of predicted and measured Absorption Under Pressure (AAP) for a range of superabsorbent polymers is shown;

[0019] Figure 6 The correlation of predicted and measured centrifuge retention capacity (CRC) for a range of superabsorbent polymers is shown;

[0020] Figure 7 The correlation of predicted and measured effective capacities (EFFC) for a range of superabsorbent polymers is shown;

[0021] Figure 8 The correlation of predicted and measured fixed height absorbency (FHA) for a range of superabsorbent polymers is shown;

[0022] Fig. 9 The correlation of predicted and measured permeability-dependent absorption under pressure (PDAUP) for a range of superabsorbent polymers is shown;

[0023] Fig.10 The correlation of predicted and measured saline flow conductivity (SFC-1.5) for a range of superabsorbent polymers is shown;

[0024] Fig.11 The correlation of predicted and measured absorption time under pressure (T20) for a range of superabsorbent polymers is shown;

[0025] Fig.12 The correlation of predicted and measured absorption times (VORTEX) for a range of superabsorbent polymers is shown;

[0026] Fig.13 The correlation of predicted and measured bulk density for a range of superabsorbent polymers is shown;

[0027] Fig.14The predicted and measured correlations for a range of superabsorbent polymers are shown;

[0028] Fig.15 Shown is the correlation of predicted and measured residual monomer content (ReMo) for a range of superabsorbent polymers. DETAILED DESCRIPTION

[0029] The present disclosure provides a method for predicting at least one physical property, particularly a performance parameter, of a super absorbent polymer (SAP), the method comprising collecting a Raman spectrum of the super absorbent polymer and using the Raman spectrum as an input to a model that determines the value of the corresponding performance parameter from the spectral data. In one embodiment, the model is based on a partial least squares (PLS) regression algorithm. Other statistical modeling techniques, such as ridge regression or elastic network regression, may also be used. These techniques produce results similar to PLS modeling. In the present disclosure, PLS modeling is used as an example to describe modeling.

[0030] Fundamentally, there may be a correlation between the spectral data and the performance data of superabsorbent polymers because the chemical composition and morphology of superabsorbent polymers directly affect the performance parameters.

[0031] Additionally, spectroscopic analysis methods, such as Raman scattering spectroscopy, probe the chemical composition and to some extent the morphology of polymers.

[0032] It has been discovered that spectroscopic methods can be used to predict physical characteristics, such as performance parameters, of superabsorbent polymers.

[0033] Raman spectroscopy can be used to characterize superabsorbent polymers in both on-line and off-line applications. In one embodiment, the Raman spectra at a fundamental wavelength of 785 nm are recorded over a time range of 10 seconds to 2 minutes, depending on the application.

[0034] Algorithms, statistical factor models are used to convert Raman spectra into SAP physical properties, such as performance parameters. In one embodiment, partial least squares or latent structure projection (PLS) models are used. Typically, each physical property uses a separate PLS model (PLS1). In alternative embodiments, one PLS model is used for multiple properties (PLS2).

[0035] Partial least squares regression (PLS) modeling is used to relate the spectral data to the performance data (calibration model). This iterative algorithm determines which parts of the spectral data (e.g., Raman scattering spectra) can explain the observed changes in the measured performance parameters. In the simplest model, one wavelength, wave number, or peak in the spectral data completely determines the change in the performance parameter. However, due to the amount of different chemical groups present in the polymer, chain length, morphology, etc., their weighted contribution to the performance parameter, and other boundary conditions (such as temperature or operation mode of the experiment), the obtained model is usually complex and uses a large amount of available data to predict the results.

[0036] The quality of any PLS calibration model is directly related to the quality of the data to be correlated. Spectral data, such as Raman scattering spectra, are generally of high quality. In the case of superabsorbent polymers, the limiting factor is the accuracy of the measured physical properties, especially the performance parameters. Therefore, it is important to obtain high quality data from laboratory reference measurements by reducing or eliminating systematic errors as much as possible. Examples are reducing the bias of laboratory technicians by selecting multiple people, or reducing the standard deviation of measurements by increasing the number of repetitions of the same experiment on the same polymer sample.

[0037] It is also important to note that calibrated PLS models are only valid within the range in which they are calibrated. For example, if the CRC in the reference data is in the range of 20-25 g / g, then the model is not suitable for accurately predicting CRC values ​​outside of this range.

[0038] In one embodiment, the model has been trained using Raman spectra of a variety of superabsorbent polymers covering a wide range of physical property values ​​and measured values ​​of at least one physical property, particularly a performance parameter.

[0039] The first step to obtain a model is a training phase, in which the (Raman scattering) spectra of several reference superabsorbent samples are correlated with the measured performance parameters to obtain a "calibrated" model that can predict the performance parameters of the samples with a certain accuracy based on the Raman spectra. The choice of samples (according to the statistical distribution of the parameters, quantity, etc.) and other settings during calibration (number of iterations, complexity of the model, data preprocessing, etc.) result in different calibration models with different characteristics in terms of cross-validation robustness, accuracy and error.

[0040] The choice of training samples for this first step is critical. In the case of superabsorbent polymers, two options can be considered. The first option is to train the model based on process samples, and the second option is to train the model using synthetic samples.

[0041] For the first option, samples with as large a process variation as possible should be selected. This will extend the calibration range as wide as possible and reduce the impact of measurement errors on a single sample on the entire training sample set (e.g., if the calibration range is 20-25 g / g, it will be more difficult to model a combined measurement error with a 0.9 g / g CRC than if the calibration range is 15-30 g / g).

[0042] The second option is to use a synthetic sample, by which physical properties such as performance parameters are manipulated in a laboratory or pilot plant. These manipulations can be completed based on polymers from a preparation process, polymers from a pilot plant, or polymers synthesized in a laboratory. A typical example of manipulation is to mix a variety of particle sizes from a production sample (via a sieve cut) to obtain a new "artificial" particle size distribution (PSD), which produces a wide range of performance characteristics. The second example of manipulation is to use a base polymer from a production plant and convert these base polymers into surface post-crosslinked superabsorbent polymers in a laboratory under a variety of conditions (such as different surface crosslinking agents, a wide temperature range, changes in reaction time, etc.). The third example of manipulation is to use the same reagents as in the production process, but to synthesize superabsorbent polymers in a laboratory environment. The fourth example of obtaining a series of synthetic samples is to change the process parameters in the pilot plant to extreme situations that rarely occur in an actual production environment. Other methods of producing superabsorbent polymers with greatly varying performance characteristics can also be used to synthesize "synthetic" samples as training samples for chemometric models. This second option is only associated with an offline model.

[0043] In the second step of the modeling approach, the calibration model obtained in the first step is validated using independent test samples (batches not used in the training phase). The calibration model that produced the best results in terms of prediction error, reduced complexity, robustness, etc. during validation is then selected to predict the performance parameters for other samples.

[0044] In order to ensure the robust performance of the prediction model, continuous validation is used in one embodiment of the method. This is done on the computer by determining the fit of any newly measured spectrum to the spectrum used to train the model, and in the laboratory by comparing the predicted results with the actual measured data (with the same accuracy level). A validation plan can be designed based on the model performance (as evaluated in step 2 of the modeling method) and the required specifications of the polymer, the validation plan including frequency and sample points. For example, a model that predicts parameters with a prediction error of 1 within a specification range of 20-25 will require more frequent validation than a model with an error of 0.1. If it is detected that the performance of the model does not meet the requirements (for example, if the parameter is out of the calibration range), additional reference samples can be included in the model, or the model can be retrained.

[0045] When a formulation change occurs in the production of superabsorbent polymers, the validity of the existing model for the new formulation will largely depend on the changes made to the formulation. If the changes are limited to small deviations in the proportions of the reagents used or small changes to physical parameters (such as temperature during processing), and the resulting performance parameters are within the calibration range of the model, then the probability of successful transfer to the new formulation using the existing model is high. The probability of successful transfer is lower when changing the chemical nature of the product (such as new reagents) or imposing larger deviations in process parameters. In both cases, the validity of the existing model needs to be verified. Two possible options are pre-startup validation using synthetic samples, or an intensive validation plan at startup.

[0046] In one embodiment of the training step, multiple Raman spectra have been collected for each superabsorbent polymer and the spectral data have been preprocessed by baseline correction, optionally smoothing or derivative generation, and subsequent normalization of each Raman spectrum.

[0047] In one embodiment, the spectra are preprocessed on both the x-axis and the y-axis. On the x-axis, this is accomplished by three actions: (1) resampling with a reference spectrum to ensure that the Raman shift of the model corresponds to the Raman shift of the measured spectrum; (2) defining the global range of the spectrum considered by the model, typically 200-1800 cm -1 ; and (3) define “excluded” ranges, which are ranges ignored by the model, such as 920-1130 cm -1 , the latter being an optional optimization step, depending on the product being characterized. Preprocessing on the y-axis can be accomplished by three actions: (1) baseline correction; (2) smoothing and / or differentiation; and (3) normalization. Depending on the product being characterized, a combination of these preprocessing steps can be optimized.

[0048] In an exemplary embodiment, a baseline correction is first performed by subtracting a straight line or linear fit from the Raman spectrum. Alternatively, a simpler baseline correction is to use offset subtraction or no correction at all. A more complex correction can be rubber band subtraction.

[0049] The second y-axis preprocessing is to smooth the spectrum or use the derivative (first or second order derivative) of the spectrum. Considering that this may lead to information loss, this preprocessing step is used to optimize the signal-to-noise ratio. Therefore, it is preferred to improve the quality of the spectrum before using this preprocessing, for example by recording over a longer time frame or taking an average of more recordings of the same sample.

[0050] The final pre-processing is to normalize the spectrum. In an exemplary embodiment, the spectrum is normalized based on area or by standard normal variable (SNV) normalization. The former scales the signal area under the spectrum to 1. The latter scales the signal intensity to meet the standard normal distribution. Therefore, this is a weighted normalization, in which not all points contribute equally to normalization. An alternative normalization step can be a minimum-maximum normalization, in which the minimum value of the signal intensity is equal to 0 and the maximum value is equal to 1, thereby producing a relative intensity ranging from 0 to 1. Another alternative is peak normalization, in which the signal intensity is scaled relative to a reference peak (usually a peak not affected by performance parameters).

[0051] The combination and choice of preprocessing depends on the model and the product being studied, but typically involves a combination of subtracting a straight line or a linear fit with area or SNV normalization without the use of smoothing filters or derivatives.

[0052] During the modeling / training step of the PLS model, several model performance parameters or model KPIs are optimized: root mean square error of cross validation (RMSECV), root mean square error of prediction (RMSEP), R 2 , spectral residuals and Mahalanobis distances. Typically, several models are made / trained and tested / validated using process samples. The best model is selected based on the reliability, accuracy and precision of the predicted / calculated values ​​of the product performance parameters. The validation samples are usually selected from two independent production campaigns of the same product. Once a set of model parameters and preprocessing are selected, the model can be evaluated by continuously estimating (trends of) Mahalanobis distances and spectral residuals and regular control samples (usually one sample per week). Additional training samples can be added to the model to improve the predictive power over time, such as when the difference between the model prediction and the reference value exceeds a user-defined limit (such as a confidence interval). Feature selection in Raman spectroscopy can be used to optimize model performance.

[0053] In one embodiment of the method of the present disclosure, the Raman spectrum of the superabsorbent polymer is collected online during the production process of the superabsorbent polymer. In another embodiment, the Raman spectrum of the superabsorbent polymer is collected offline on a sample of the superabsorbent polymer.

[0054] In one embodiment of the method of the present disclosure, the physical property predicted is a performance parameter of the superabsorbent polymer.

[0055] The performance parameters of superabsorbent polymers include centrifuge retention capacity (CRC), absorbency under load (AUL), effective capacity (EFFC), saline flow conductivity (SFC), T20, free swelling ratio (FSR), fixed height absorbency (FHA), permeability-dependent absorbency under pressure (PDAUP), and vortex. Other important physical properties of superabsorbent polymers include flow rate, bulk density, and residual monomer content in the core structure.

[0056] In a specific embodiment, the performance parameter is the centrifuge retention capacity (CRC) of the superabsorbent polymer. CRC measures the liquid absorbed by superabsorbent polymer particles when they swell in excess liquid. CRC is measured according to EDANA test method NWSP 241.0.R2 (15) (ISO 17190-6:2001) "Polyacrylate Superabsorbent Powders—Determination of the Fluid Retention Capacity in Saline Solution by Gravimetric Measurement Following Centrifugation", wherein for higher centrifuge retention capacity values, larger tea bags must be used.

[0057] In another embodiment, the performance parameter is the absorbency under load (AUL) or absorbency against pressure (AAP) of the superabsorbent polymer. The AUL of superabsorbent polymer particles is determined according to EDANA Test Method No. NWSP 242.0.R2 (15) (ISO 17190-7:2001) "Polyacrylate Superabsorbent Powders—Gravimetric Determination of Absorption Against Pressure".

[0058] In another embodiment, the performance parameter is the effective capacity of the superabsorbent polymer (EFFC). The effective capacity represents the average of the centrifuge retention capacity (CRC) value and the absorption under pressure (AAP) value of the superabsorbent polymer particles. The effective capacity is calculated as EFFC=(CRC+AAP) / 2.

[0059] In a further embodiment, the performance parameter is the saline flow conductivity (SFC) of the superabsorbent polymer. The saline flow conductivity represents the gel layer permeability of the swollen gel layer of the water-absorbent polymer particles. It is determined as described in EP 0 640 330 A1 and the evaluation of the measurement is carried out as described in EP 0 640 330 A1. The flow rate is recorded automatically. The data of pages 19 and 20 of EP 0 640 330 A1 are reproduced in full. Figure 8 The apparatus described in was modified so that the frit (40) was no longer used, the plunger (39) consisted of the same polymer material as the barrel (37) and now included 21 holes of 9.65 mm diameter evenly distributed over the entire contact surface. The saline flow conductivity (SFC) was calculated as follows:

[0060] SFC[cm 3 s / g]=(F g (t=0)x L 0 ) / (dxAxWP),

[0061] Among them, F g (t = 0) is the flow rate of NaCl solution in g / s, which is measured by the flow rate F g (t) Data were subjected to linear regression analysis by extrapolating to t = 0, L 0 is the thickness of the gel layer, in cm, and d is the density of the NaCl solution, in g / cm 3 A is the surface area of ​​the gel layer, in cm 2 WP is the hydrostatic pressure on the gel layer, in dyn / cm 2 count.

[0062] In a variant of this method, the saline flow conductivity (SFC 1.5) of the superabsorbent polymer is measured according to the so-called UPM method described in EP 2 535 027 A1, pages 19-22.

[0063] In another specific embodiment, the performance parameter is the time (T20) for the superabsorbent polymer to reach a liquid absorption of 20 g liquid / g superabsorbent polymer. The time to reach a liquid absorption of 20 g / g is measured by the method "K(t) test method (Dynamic Effective Permeability and UptakeKinetics Measurement Test Method)" disclosed in EP 2 535 027 A1, pages 13 to 18.

[0064] In another specific embodiment, the performance parameter is the free swelling rate (FSR) of the superabsorbent polymer. The free swelling rate indicates the rate at which a superabsorbent polymer can absorb a known amount of saline without stirring or applying external pressure. FSR is measured according to the following process: 1.00 g (= W1) of dry water-absorbent polymer particles are weighed, placed in a 25 ml glass beaker, and evenly distributed at the bottom of the glass beaker. Then 20 ml of a 0.9 wt % sodium chloride solution is dispensed into a second glass beaker. The contents of the beaker are quickly added to the first beaker, and a stopwatch is started. Once the last drop of saline solution is absorbed (determined by the disappearance of reflection on the surface of the liquid), the stopwatch is stopped. By weighing the second beaker again (= W2), the exact amount of liquid poured from the second beaker and absorbed by the polymer in the first beaker is accurately determined. The time required for absorption, measured with a stopwatch, is denoted as t. The free swelling rate (FSR) is calculated as FSR [g / gs] = W2 / (W1xt). When the moisture content of the hydrogel-forming polymer is greater than 3% by weight, the weight W1 must be corrected for this moisture content.

[0065] In another specific embodiment, the performance parameter is the moisture content (MC) of the superabsorbent polymer. The MC of the superabsorbent polymer particles is determined according to EDANA Test Method No. NWSP 230.0.R2 (15) (ISO 17190-4:2001)) "Polyacrylate Superabsorbent Powders—Estimation of the Moisture Content as Weight Loss Upon Heating".

[0066] In yet another specific embodiment, the performance parameter is the fixed height absorption (FHA) of the superabsorbent polymer. FHA represents the absorption capacity of a superabsorbent polymer for saline solution under a fixed hydrostatic pressure applied to the superabsorbent. FHA can be measured using the method described in EP 1 493 453 A1, pages 9 / 10.

[0067] In another specific embodiment, the performance parameter is the permeability-dependent absorption under pressure (PDAUP) of the superabsorbent polymer. PDAUP represents the absorptive capacity of the superabsorbent under constant pressure from bottom to top. The permeability determines the total amount of saline absorbed in the superabsorbent. PDAUP can be measured according to EDANA test method NWSP 243.0.R2 (15) "Polyacrylate Superabsorbent Powders—Determination of the PermeabilityDependent Absorption Under Pressure of Saline Solution by Gravimetric Measurement".

[0068] In another specific embodiment, the physical property is the particle size distribution (PSD) of the superabsorbent polymer. The particle size distribution of the superabsorbent polymer particles is determined according to EDANA Test Method No. NWSP 220.0.R2 (15) (ISO 17190-3:2001) "Polyacrylate Superabsorbent Powders—Determination of the Particle Size Distribution by Sieve Fractionation".

[0069] In yet another specific embodiment, the physical property is the bulk density of the superabsorbent polymer. Bulk density is measured according to EDANA method NWSP 251.0.R2 (15) "Polyacrylate Superabsorbent Powders—Gravimetric Determination of Flow Rate and Bulk Density".

[0070] In yet another specific embodiment, the performance parameter is the flow rate of the superabsorbent polymer. The flow rate of the superabsorbent polymer particles is determined according to EDANA Test Method No. NWSP 251.0.R2 (15) "Polyacrylate Superabsorbent Powders—Gravimetric Determination of Flow Rate and Bulk Density".

[0071] In yet another specific embodiment, the physical property is the color value of the superabsorbent polymer.The color value is measured according to the tristimulus method according to DIN 5033-6.

[0072] In another specific embodiment, the performance parameter is the surface tension of the water extract of the superabsorbent polymer. The surface tension of the water extract is measured according to the following method: 0.50 g of superabsorbent polymer particles are weighed into a small beaker and 40 ml of a 0.9 wt % sodium chloride solution is added. The contents in the beaker are stirred at 500 rpm for 3 minutes with a magnetic stirring bar and then allowed to settle for 2 minutes. Finally, the surface tension (ST) of the supernatant aqueous phase is measured with a K10-ST digital tensiometer or a similar instrument with a platinum plate (Krüss GmbH, Hamburg, Germany). The measurement is carried out at a temperature of 23 ° C.

[0073] In yet another specific embodiment, the performance parameter is the vortex of the superabsorbent polymer. The vortex is measured according to the following method: 50.0 ± 1.0 ml of 0.9% NaCl solution is added to a 100 ml beaker. A cylindrical stirring bar (30x6 mm) is added and the saline solution is stirred at 60 rpm on a stirring plate. 2.000 ± 0.010 g of superabsorbent polymer particles are added to the beaker as quickly as possible, and a stopwatch is started at the beginning of the addition. When the surface of the mixture becomes "stationary", the stopwatch is stopped. The "stationary" means that there is no turbulence on the surface, and although the mixture may still rotate, the entire surface of the particles rotates as a whole. The display time of the stopwatch is recorded as the vortex time.

[0074] In one embodiment of the method of the present disclosure, the superabsorbent polymer comprises crosslinked polymer chains of at least partially neutralized acrylic acid monomer units. In one embodiment, the crosslinking agent is N-(2-hydroxyethyl)-2-oxazolidinone (HEONON).

[0075] Exemplary methods for preparing superabsorbent polymer particles are described in detail below.

[0076] The superabsorbent polymer particles can be prepared by polymerizing a monomer solution or suspension comprising

[0077] a) at least one ethylenically unsaturated monomer which carries acid groups and may be at least partially neutralized,

[0078] b) at least one cross-linking agent,

[0079] c) at least one initiator,

[0080] The resulting polymer gel is dried and the dried polymer gel is ground.Superabsorbent polymers are generally insoluble in water.

[0081] Suitable monomers a) are, for example, ethylenically unsaturated carboxylic acids such as acrylic acid, methacrylic acid and itaconic acid. Particularly preferred monomers are acrylic acid and methacrylic acid. Very particular preference is given to acrylic acid.

[0082] Acrylic acid generally contains a polymerization inhibitor, preferably hydroquinone monoether, as a storage stabilizer.

[0083] The monomer solution preferably comprises up to 250 ppm by weight, preferably up to 150 ppm by weight, more preferably up to 100 ppm by weight and preferably at least 10 ppm by weight, more preferably at least 30 ppm by weight and in particular about 50 ppm by weight of hydroquinone monoether, based in each case on the unneutralized monomers a). For example, the monomer solution can be prepared by using ethylenically unsaturated monomers carrying acid groups with an appropriate content of hydroquinone monoether.

[0084] Suitable crosslinkers b) are compounds having at least two groups suitable for crosslinking. Such groups are, for example, ethylenically unsaturated groups which can be free-radically polymerized into the polymer chain, and functional groups which can form covalent bonds with the acid groups of the monomers a). In addition, polyvalent metal salts which can form coordinate bonds with at least two acid groups of the monomers a) are also suitable as crosslinkers b).

[0085] Crosslinkers b) are preferably compounds having at least two polymerizable groups which can be polymerized free-radically into the polymer network. Suitable crosslinkers b) are, for example, ethylene glycol dimethacrylate, diethylene glycol diacrylate, polyethylene glycol diacrylate, allyl methacrylate, trimethylolpropane triacrylate, triallylamine, tetraallyl ammonium chloride, tetraallyloxyethane, as described in EP 0 530 438 A1; diacrylates and triacrylates, as described in EP 0 547 847 A1, EP 0 559 476 A1, EP 0 632 068 A1, WO 93 / 21237 A1, WO 2003 / 104299 A1, WO 2003 / 104300 A1, WO 2003 / 104301 A1 and DE 103 31 450 A1; mixed acrylates which, in addition to the acrylate groups, contain further ethylenically unsaturated groups, as described in DE 103 31 456 A1. A1 and DE 103 55 401 A1; or crosslinker mixtures, as described, for example, in DE 195 43 368 A1, DE 196 46 484 A1, WO 90 / 15830 A1 and WO 2002 / 032962 A2.

[0086] The amount of crosslinker b) is preferably from 0.05 to 1.5% by weight, more preferably from 0.1 to 1% by weight and most preferably from 0.3 to 0.6% by weight, in each case based on monomer a). With increasing crosslinker content, the centrifuge retention capacity (CRC) decreases, while 21.0 g / cm 2 The absorption under pressure passes through a maximum.

[0087] The initiators c) used may be all compounds which generate free radicals under the polymerization conditions, for example thermal initiators, redox initiators, photoinitiators. Suitable redox initiators are sodium peroxodisulphate / ascorbic acid, hydrogen peroxide / ascorbic acid, sodium peroxodisulphate / sodium bisulfite and hydrogen peroxide / sodium bisulfite. Preference is given to using mixtures of thermal initiators and redox initiators, such as sodium peroxodisulphate / hydrogen peroxide / ascorbic acid. However, the reducing component used is preferably disodium 2-hydroxy-2-sulfonatoacetate or a mixture of disodium 2-hydroxy-2-sulfinatoacetate, disodium 2-hydroxy-2-sulfonatoacetate and sodium bisulfite. Such mixtures may be used in combination with the initiators c) of the polymerization reaction. FF6 and FF7 (Brüggemann Chemicals; Heilbronn; Germany).

[0088] Typically, aqueous monomer solutions are used. The water content of the monomer solution is preferably 40 to 75% by weight, more preferably 45 to 70% by weight and most preferably 50 to 65% by weight. It is also possible to use monomer suspensions, i.e. monomer solutions with an excess of monomer a), such as sodium acrylate. With increasing water content, the energy requirement in the subsequent drying increases, while with decreasing water content, the heat of polymerization can only be removed inadequately.

[0089] For optimal action, preferred polymerization inhibitors require dissolved oxygen. Therefore, dissolved oxygen can be removed from the monomer solution by inertization, i.e., by passing an inert gas (preferably nitrogen or carbon dioxide) through it, before polymerization. The oxygen content of the monomer solution is preferably reduced to less than 1 ppm by weight, more preferably less than 0.5 ppm by weight, and most preferably less than 0.1 ppm by weight before polymerization.

[0090] For better control of the polymerization reaction, all known chelating agents can optionally be added to the monomer solution or suspension or to the starting material thereof. Suitable chelating agents are, for example, phosphoric acid, pyrophosphoric acid, triphosphoric acid, polyphosphoric acid, citric acid, tartaric acid or their salts.

[0091] The monomer solution or suspension is polymerized. Suitable reactors are, for example, kneading reactors or belt reactors. In a kneader, the polymer gel formed in the polymerization of the aqueous monomer solution or suspension is comminuted continuously, for example by counter-rotating stirrer shafts, as described in WO 2001 / 038402 A1. Polymerization on a belt is described, for example, in DE 38 25 366 A1 and US 6,241,928 A. Polymerization in a belt reactor forms a polymer gel which has to be comminuted in a further process step, for example in an extruder or kneader.

[0092] In order to improve the drying properties, the comminuted polymer gel obtained by means of a kneader can additionally be extruded.

[0093] The acid groups of the resulting polymer gel are usually partially neutralized. Neutralization is preferably carried out at the monomer stage. This is usually achieved by mixing in a neutralizing agent as a solid or preferably in the form of an aqueous solution. The degree of neutralization is preferably 50 to 90 mol %, more preferably 60 to 85 mol % and most preferably 65 to 80 mol % and particularly preferably 65 to 75 mol %, for which conventional neutralizing agents can be used, preferably alkali metal hydroxides, alkali metal oxides, alkali metal carbonates or alkali metal hydrogencarbonates and mixtures thereof. Instead of alkali metal salts, ammonium salts can also be used. Particularly preferred alkali metals are sodium and potassium, but very particularly preferred are sodium hydroxide, sodium carbonate or sodium hydrogencarbonate and mixtures thereof.

[0094] The resulting polymer gel is dried. The dryer is not subject to any restrictions. However, the drying of the polymer gel is preferably carried out with a belt dryer until the residual moisture content is preferably 0.5 to 10% by weight, more preferably 1 to 7% by weight and most preferably 2 to 5% by weight, the residual moisture content being determined by the EDANA test method No. WSP 230.2-05 "Mass Loss Upon Heating". In the case of excessively high residual moisture contents, the dried polymer gel has a too low glass transition temperature T g , and further processing is difficult. In the case of too low a residual moisture content, the dried polymer gel is too brittle and, in the subsequent grinding step, an undesirable large amount of polymer particles with too low a particle size ("fines") is obtained. The solids content of the gel before drying is preferably 25 to 90% by weight, more preferably 35 to 70% by weight and most preferably 40 to 60% by weight. However, a fluidized bed dryer or a paddle dryer can also optionally be used for drying purposes.

[0095] Subsequently, the dried polymer gel is ground and classified. The average particle size of the polymer particles removed as the product fraction is preferably at least 200 μm, more preferably 250 to 600 μm, very particularly 300 to 500 μm. The average particle size of the product fraction can be determined by EDANA test method No. WSP 220.2-05 "Particle Size Distribution", in which the mass proportions of the sieve fractions are plotted in cumulative form and the average particle size is determined from the graph. In this context, the average particle size is the value of the sieve opening size that produces a cumulative 50% by weight.

[0096] To improve the properties, the polymer particles can subsequently be subjected to thermal surface postcrosslinking. Suitable surface postcrosslinkers are compounds which contain groups which can form covalent bonds with at least two acid groups of the polymer particles. Suitable compounds are, for example, polyfunctional amines, polyfunctional amidoamines, polyfunctional epoxides, as described in EP 0 083 022 A2, EP 0 543 303 A1 and EP 0 937 736 A2; difunctional alcohols or polyfunctional alcohols, as described in DE 33 14 019 A1, DE 35 23 617 A1 and EP 0 450 922 A2; or β-hydroxyalkylamides, as described in DE 102 04 938 A1 and US 6,239,230 A.

[0097] Preferred surface post-crosslinking agents are ethylene carbonate, ethylene glycol diglycidyl ether ( NagaseChemteX Corp, Tokyo, Japan), the reaction product of polyamide with epichlorohydrin, and a mixture of propylene glycol and 1,4-butanediol.

[0098] The amount of surface postcrosslinkers is preferably 0.001 to 2% by weight, more preferably 0.02 to 1% by weight and most preferably 0.05 to 0.2% by weight, based in each case on the polymer particles.

[0099] In a preferred embodiment of the invention, polyvalent cations are applied to the particle surface in addition to the surface postcrosslinkers before, during or after the surface postcrosslinking.

[0100] Polyvalent cations that can be used in the process according to the invention are, for example: divalent cations, such as cations of zinc, magnesium, calcium, iron and strontium; trivalent cations, such as cations of aluminum, iron, chromium, rare earth elements and manganese; tetravalent cations, such as cations of titanium and zirconium. Possible counterions are chloride; bromide; sulfate; hydrogensulfate; carbonate; hydrogencarbonate; nitrate; phosphate; hydrogenphosphate; dihydrogenphosphate; and carboxylates, such as acetate and lactate. Preference is given to aluminum sulfate and aluminum lactate. In addition to metal salts, polyamines can also be used as polyvalent cations.

[0101] The polyvalent cations are used in amounts of, for example, 0.001 to 1.5% by weight, preferably 0.005 to 1% by weight, more preferably 0.02 to 0.8% by weight, based in each case on the polymer particles.

[0102] The surface post-crosslinking is usually carried out by spraying a solution of the surface post-crosslinking agent onto the dried polymer particles. After spraying, the polymer particles coated with the surface post-crosslinking agent are thermally dried, and the surface post-crosslinking reaction can be carried out before or during drying.

[0103] The spraying of the surface postcrosslinker solution is preferably carried out in mixers with moving mixing tools, such as screw mixers, disk mixers and paddle mixers. Particular preference is given to horizontal mixers, such as paddle mixers, and very particular preference to vertical mixers. The difference between horizontal mixers and vertical mixers lies in the position of the mixing shaft, i.e. horizontal mixers have a horizontally mounted mixing shaft, while vertical mixers have a vertically mounted mixing shaft. Suitable mixers are, for example, horizontal Ploughshare mixer (Gebr. Maschinenbau GmbH; Paderborn; Germany), Vrieco-Nauta continuous mixer (Hosokawa Micron BV; Doetinchem; the Netherlands), Processall Mixmill mixer (Processall Incorporated; Cincinnati; USA), and Schugi (Hosokawa Micron BV; Doetinchem; the Netherlands). However, it is also possible to spray on the surface postcrosslinker solution in a fluidized bed.

[0104] The surface postcrosslinkers are generally used in the form of an aqueous solution. The penetration depth of the surface postcrosslinkers into the polymer particles can be adjusted via the content of nonaqueous solvent and the total amount of solvent.

[0105] The thermal surface postcrosslinking is preferably carried out in a contact dryer, more preferably a paddle dryer, most preferably a tray dryer. Suitable dryers are, for example, Hosokawa Horizontal paddle dryer (Hosokawa Micron GmbH; Leingarten; Germany), Hosokawa Tray dryers (Hosokawa Micron GmbH; Leingarten; Germany) and Nara paddle dryers (NARA Machinery Europe; Frechen; Germany). In addition, fluidized bed dryers can also be used.

[0106] The thermal surface post-crosslinking can be carried out in the mixer itself by heating the jacket or blowing in warm air. Also suitable are downstream dryers, such as cabinet dryers, rotary tube furnaces or heatable screws. It is particularly advantageous to carry out mixing and drying in a fluidized bed dryer.

[0107] The preferred surface post-crosslinking temperature range is from 100 to 250° C., preferably from 110 to 230° C., more preferably from 120 to 210° C., most preferably from 150 to 190° C. The preferred residence time at this temperature in the reaction mixer or dryer is preferably at least 10 minutes, more preferably at least 20 minutes, most preferably at least 30 minutes, and typically at most 60 minutes.

[0108] Subsequently, the surface postcrosslinked polymer particles can be classified again, and too small and / or too large polymer particles can be removed and recirculated into the process.

[0109] To further improve the properties, the surface-post-crosslinked polymer particles can be coated or rewetted.

[0110] Remoistening is preferably carried out at 30 to 80° C., more preferably at 35 to 70° C., most preferably at 40 to 60° C. At too low a temperature, the superabsorbent polymer particles tend to form lumps, while at higher temperatures, the water has evaporated to a significant extent. The amount of water used for remoistening is preferably 1 to 10% by weight, more preferably 2 to 8% by weight and most preferably 3 to 5% by weight. Remoistening increases the mechanical stability of the polymer particles and reduces the tendency of the polymer particles to be electrostatically charged.

[0111] Suitable coatings for increasing the free swelling rate and the saline flow conductivity (SFC) are, for example, inorganic inert substances, such as water-insoluble metal salts, organic polymers, cationic polymers and divalent metal cations or polyvalent metal cations. Suitable coatings for powder bonding are, for example, polyols. Suitable coatings for counteracting the undesirable agglomeration tendency of the polymer particles are, for example, fumed silica, such as 200 (Evonik Industries AG, Essen, Germany), or precipitated silica, such as D17 (Evonik Industries AG, Essen, Germany); and surfactants such as 20 (Merck KGaA, Darmstadt, Germany).

[0112] The methods of the present disclosure can also be used to obtain other information about the chemical composition of the superabsorbent polymer, for example, the concentration of residual acrylic acid or HEONON; solvents such as 1,2-propylene glycol or 1,3-propylene glycol; or extractables.

[0113] The residual monomer (ReMo) content in superabsorbent polymer particles is determined according to EDANA test method No. WSP 210.2 (04) "Determination of the Amount of Residual Monomers in Superabsorbent Materials".

[0114] Another aspect of the present disclosure is the use of the predicted values ​​of the physical properties of the superabsorbent polymer obtained by the method of the present disclosure for guiding the production process of the superabsorbent polymer. The predicted values ​​of the physical properties of the superabsorbent polymer obtained by the method of the present disclosure, such as the performance parameters, can be used to guide the production process, i.e., to control and optimize the process parameters or process steps, respectively, for example, by changing the concentration of the crosslinking agent in the polymerization step or by changing the concentration of the surface crosslinking agent or by optimizing the process guidance parameters. A feedback loop is established, which produces optimized product properties and ensures consistent product quality, i.e., less variation in product properties.

[0115] The present disclosure also provides a software product configured to perform the method of the present disclosure. In one embodiment, the software product is a computer program implemented in an apparatus for producing superabsorbent polymers, for example in a computing unit (computer) integrated therein and / or connected thereto.

[0116] According to another embodiment, the software product is a computer program product which, when loaded into a memory of a computing device and executed by at least one processor of the computing device, performs the steps of the above-described computer-implemented method.

[0117] The software product, i.e., computer program product, may be used with or incorporated into a computer system, which may be a stand-alone unit or include one or more remote terminals or devices that communicate with a central computer via a network (e.g., the Internet or an intranet). Thus, the computer or processor and related components described herein may be part of a local computer system or a remote computer or online system or a combination thereof. Any database and software product, i.e., computer program product described herein may be stored in a computer internal memory or in a non-transitory computer readable medium.

[0118] The present disclosure also provides a computer system for predicting at least one physical property of a superabsorbent polymer, the computer system comprising at least:

[0119] - an interface component configured to access and read the Raman spectrum of the superabsorbent polymer;

[0120] - a processor unit implementing a model and configured to use the Raman spectrum provided via the interface component as input to the model, the model determining the value of the physical property from the spectral data, the model being based on a regression algorithm.

[0121] In one embodiment, the regression algorithm is a partial least squares (PLS) regression algorithm. In another embodiment, the regression algorithm is based on ridge regression. In yet another embodiment, the regression algorithm is an elastic net regression algorithm.

[0122] According to one embodiment, the system is configured to be coupled to a Raman spectrometer via a wired and / or wireless communication connection, and to access and read out the Raman spectrum from the Raman spectrometer at least partially automatically via the interface component. The system can be connected to a device for producing superabsorbent polymers, or integrated in a corresponding device.

[0123] According to another embodiment, the system is configured to determine the value of the physical property from the spectral data on-line during the production process of the superabsorbent polymer.

[0124] The system may also include a database that at least temporarily stores a plurality of measured Raman spectra and measured values ​​of at least one physical property of a plurality of superabsorbent polymers covering a wide range of physical property values, and the system is further configured to train the model using the stored measured Raman spectra and measured values ​​of the at least one physical property.

[0125] The system may also be configured to update the model again and again by iteratively training the model with new measured Raman spectra of superabsorbent polymers and new measured values ​​of the at least one physical property.

[0126] The system is particularly configured to perform the above-described method for predicting at least one physical property of a superabsorbent polymer.

[0127] The subject matter of the present disclosure is further described and illustrated in the following working examples.

[0128] Example

[0129] Raman methods for SAP characterization

[0130] Raman spectroscopy is used to characterize superabsorbent polymers in both online and offline applications. Offline applications use the benchtop RamanRXN2 TM A hybrid spectrometer (wavelength 785 nm) (Kaiser Optical Systems, Inc., Ann Arbor, MI48103, USA) was built. The online application also used a RamanRXN2 cooled in a dust-free cabinet. TM A hybrid spectrometer was built and the laser and detector were connected via optical fiber to an insertion probe with a PhAT probe (Kaiser Optical Systems, Inc., Ann Arbor, MI 48103, USA). The insertion probe was located at a critical location in the production process where the product flows freely. Raman spectra at a fundamental wavelength of 785 nm were recorded in a time range between 10 seconds and 2 minutes, depending on the application.

[0131] The detection window range is 150cm -1 Up to 1900cm -1 . Use 200cm -1 Up to 1800cm -1 The window is used for modeling.

[0132] The modeling software used is (S·PACT GmbH, 52064 Aachen, Germany). Other modeling software can also be used, such as R, python, Mathlab, etc. For semi-automatic data collection in the laboratory and autonomous measurements in the process, software solutions specially developed by S·PACT are used. The models are supplemented by software control of the Raman spectrometer. The Raman-predicted properties of superabsorbent polymers reduce the spread of the property data, because the models are trained based on more accurate data.

[0133] Due to the online capabilities of the technology, process guidance is faster and more efficient. Using online Raman equipment, the number of samples in process is reduced, thereby reducing the work that must be performed in the laboratory. Product discharge can be correlated with the evaluation of online Raman spectroscopy, further reducing the required laboratory work.

[0134] Furthermore, data on the chemical properties of the superabsorbent polymers, such as data on the chemical composition of the superabsorbent polymers, can be extracted from the spectra, which can be used for product optimization. Automated process guidance can be achieved by integrating Raman predictions into the decision-making step to optimize process settings.

[0135] Figure 1 An example of a set of 10 Raman spectra obtained from a sample of superabsorbent polymer is shown after subtracting a linear fit as baseline correction and a SNV normalization step as data pre-processing. Each spectrum was obtained in the laboratory by measuring the accumulation of 6 values, each for 10 seconds, for a total of 60 seconds.

[0136] Figure 2 Predicted online data and smoothed curve calculated from individual data points (a) and measured data and smoothed curve calculated from individual data points (b) for the saline flow conductivity (SFC) of superabsorbent polymers are shown. Measured data with error bars, predicted data without error bars are shown. The data were collected over a period of one week. Raman spectra were measured online, with each spectrum recorded as 6 accumulated values, each 20 seconds, so each spectrum totaled 2 minutes.

[0137] Figure 3 The correlation of the predicted CRC to the measured CRC for a range of superabsorbent polymers is shown. A total of 169 samples were included in the evaluation. A range of 34 to 38 g / g was modeled. It is evident from the figure and the parameters given that the predicted values ​​are in good agreement with the experimentally measured values.

[0138] Figure 4 Schematic diagram of a method for preparing a superabsorbent polymer. The method comprises the following steps: (1) polymerization, (2) drying of the polymer, (3) grinding of the polymer, (4) surface post-crosslinking (SXL) of the polymer; and finally, (5) packaging of the finished product. The predicted values ​​of the physical properties (e.g., performance parameters) of the superabsorbent polymer obtained by the method of the present disclosure from the spectral data (6) collected by the online Raman probe during the process steps SXL and packaging can be used in any process step to guide (7) the process, i.e., control and optimize the process parameters.

[0139] Figure 5 The predicted online data of the absorption under pressure (AAP) of a series of superabsorbent polymers and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. The Raman spectra were measured online, where each spectrum was recorded as 6 accumulated values, each for 20 seconds, so each spectrum totaled 2 minutes. The range of 21.5 to 25.2g / g was modeled.

[0140] Figure 6 The predicted online data of the centrifuge retention capacity (CRC) of a series of superabsorbent polymers and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. The Raman spectra were measured online, where each spectrum was recorded as 6 cumulative values, each for 20 seconds, so each spectrum totaled 2 minutes. The range of 24.5 to 29g / g was modeled for 44 samples.

[0141] Figure 7 The predicted online data of the effective capacity (EFFC) of a series of superabsorbent polymers and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra are measured online, where each spectrum is recorded as 6 accumulated values, each for 20 seconds, so each spectrum totals 2 minutes. The range of 24.5 to 29g / g is modeled for 44 samples.

[0142] Figure 8 The predicted online data of fixed height absorption (FHA) of a series of superabsorbent polymers at a height of 20 cm and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra are measured online, where each spectrum is recorded as 6 accumulated values, each for 20 seconds, so each spectrum totals 2 minutes. The range of 20.5 to 25g / g is modeled for 44 samples.

[0143] Fig. 9 The predicted online data of the permeability-dependent absorption (PDAUP) of a series of superabsorbent polymers under pressure and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra are measured online, where each spectrum is recorded as 6 accumulated values, each for 20 seconds, so each spectrum totals 2 minutes. The range of 12 to 19g / g is modeled for 44 samples.

[0144] Fig.10 The predicted online data of the saline flow conductivity (SFC-1.5) of a series of superabsorbent polymers are shown together with the actual measured data measured offline in the laboratory. The line depicts the correlation function between the predicted and measured data. It is obvious from the figure that the predicted values ​​are in good agreement with the experimentally measured values. The Raman spectra were measured online, where each spectrum was recorded as 6 accumulated values, each for 20 seconds, so each spectrum totaled 2 minutes. 44 samples were modeled from 25.5 to 80 10 -7 cm 3· The range of s / g.

[0145] Fig.11 The predicted online data of the absorption time (T20) under pressure of a series of superabsorbent polymers and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra are measured online, where each spectrum is recorded as 6 accumulated values, each 20 seconds, so each spectrum totals 2 minutes. The range of 128 to 175s is modeled for 43 samples.

[0146] Fig.12 The predicted online data of the absorption time (VORTEX) of a series of superabsorbent polymers and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra are measured online, where each spectrum is recorded as 6 accumulated values, each 20 seconds, so each spectrum totals 2 minutes. The range of 34 to 46s is modeled for 43 samples.

[0147] Fig.13 The predicted online data of the bulk density of a series of superabsorbent polymers and the actual measured data measured offline in the laboratory are shown. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra were measured online, where each spectrum was recorded as 6 accumulated values, each for 20 seconds, so each spectrum totaled 2 minutes. The range of 0.55 to 0.63 g / ml was modeled for 44 samples.

[0148] Fig.14 The predicted online data of the flow rate of a series of superabsorbent polymers are shown with the actual measured data measured offline in the laboratory. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra were measured online, where each spectrum was recorded as 6 accumulated values, each for 20 seconds, so each spectrum totaled 2 minutes. The range of 7.8 to 9.4 g / s was modeled for 44 samples.

[0149] Fig.15 The predicted online data of the residual monomer content (ReMo) of a series of superabsorbent polymers are shown with the actual measured data measured offline by HPLC in the laboratory. The line depicts the correlation function between the predicted data and the measured data. It can be clearly seen from the figure that the predicted values ​​are very consistent with the experimentally measured values. Raman spectra are measured online, where each spectrum is recorded as 6 accumulated values, each for 20 seconds, so each spectrum totals 2 minutes. The range of 350 to 430ppm is modeled for 44 samples.

Claims

1. A method for predicting at least one physical property of a superabsorbent polymer, the method comprising collecting a Raman spectrum of the superabsorbent polymer and using the Raman spectrum as input to a model that determines the value of the physical property from the spectral data, the model being based on a regression algorithm, characterized in that The superabsorbent polymer comprises cross-linked polymer chains of at least partially neutralized acrylic acid monomer units, the physical property is at least one of the following performance parameters: centrifuge retention capacity (CRC), absorption under pressure (AAP), fixed height absorption (FHA), effective capacity (EFFC), permeability-dependent absorption under pressure (PDAUP), saline flow conductivity, time for the superabsorbent polymer to reach a liquid absorption capacity of 20 g / g (T20) and vortex, particle size distribution (PSD) of the superabsorbent polymer or bulk density of the superabsorbent polymer, wherein the Raman spectrum of the superabsorbent polymer is collected online during the production process of the superabsorbent polymer, wherein the production process comprises the following steps: polymerization, drying of the polymer, grinding of the polymer, surface post-crosslinking of the polymer and packaging of the finished product, and the Raman spectrum is collected during the process steps of surface post-crosslinking and / or packaging.

2. The method of claim 1, wherein the model has been trained using Raman spectroscopy and measured values ​​of the at least one physical property of a plurality of superabsorbent polymers covering a wide range of values ​​of the physical property.

3. The method of claim 2, wherein a plurality of Raman spectra have been collected for each superabsorbent polymer and the spectral data have been preprocessed by baseline correction of each Raman spectrum, optionally smoothing or derivative generation of each Raman spectrum, and subsequent normalization of the Raman spectra.

4. The method according to any one of claims 1 to 3, wherein the Raman spectrum of the super absorbent polymer is collected online during the production process of the super absorbent polymer.

5. The method of any one of claims 1 to 3, wherein the physical property is a performance parameter of the superabsorbent polymer.

6. The method of any one of claims 1 to 3, wherein the physical property is the particle size distribution (PSD) or bulk density of the superabsorbent polymer.

7. Use of the predicted values ​​of the physical properties of a super absorbent polymer obtained by the method according to any one of claims 1 to 6 for controlling a process for preparing the super absorbent polymer.

8. A computer readable medium having stored thereon a computer program, wherein when loaded into a memory of a computing device and executed by at least one processor unit of the computing device, the computer program performs the steps of the computer-implemented method according to any one of claims 1 to 6.

9. A computer system for predicting at least one physical property of a superabsorbent polymer, the computer system comprising at least: - an interface component configured to access and read the Raman spectrum of the superabsorbent polymer; a processor unit implementing a model and configured to use the Raman spectrum provided via the interface component as input to a model, the model determining the value of the physical property from the spectral data, the model being based on a regression algorithm, wherein the superabsorbent polymer comprises cross-linked polymer chains of at least partially neutralized acrylic acid monomer units, the physical property is at least one of the following performance parameters: centrifuge retention capacity (CRC), absorption under pressure (AAP), fixed height absorption (FHA), effective capacity (EFFC), permeability-dependent absorption under pressure (PDAUP), saline flow conductivity, time for the superabsorbent polymer to reach a liquid absorption capacity of 20 g / g (T20) and vortex, particle size distribution (PSD) of the superabsorbent polymer or bulk density of the superabsorbent polymer, wherein the Raman spectrum of the superabsorbent polymer is collected online during the production process of the superabsorbent polymer, wherein the production process comprises the following steps: polymerization, drying of the polymer, grinding of the polymer, surface post-crosslinking of the polymer and packaging of the finished product, and the Raman spectrum is collected during the process steps of surface post-crosslinking and / or packaging.

10. The system of claim 9, configured to couple with a Raman spectrometer via a wired and / or wireless communication connection, and to at least partially automatically access and read out the Raman spectrum from the Raman spectrometer via the interface component.

11. The system of claim 9 or 10, the system being configured to determine the value of the physical property from the spectral data on-line during the production process of the superabsorbent polymer.

12. The system of claim 9 or 10, further comprising a database at least temporarily storing a plurality of measured Raman spectra and measured values ​​of at least one physical property of a plurality of superabsorbent polymers covering a wide range of values ​​of the physical property, the system further configured to train the model using the stored measured Raman spectra and measured values ​​of the at least one physical property.

13. The system of claim 12, the system being configured to continuously update the model over and over again by iteratively training the model with new measured Raman spectra of superabsorbent polymers and new measured values ​​of the at least one physical property.

Images