Information
On-line measurements in the pulp and paper industry are among the most difficult challenges of process analytical chemistry. Pulp and paper production streams are typically very high in both dissolved and suspended solid material. The samples are often optically opaque or nearly so. Sample conditioning systems are prone to plugging and corrosion. On-line titrators tend to be very troublesome due to the problems with handling the streams, dispensing measured amounts of sample, reagent consumption and disposal of the titrated wastes. Here we report on the application of the InfraSpec NR800 FT-NIR analyzer to the determination of alkali salts in various Kraft process streams.
Kraft Process
Although there are a number of different processes for making pulp and paper from wood, the Kraft or "sulfate" process of chemical pulping is the most common. Wood chips are fed into a solution of Sodium Hydroxide and Sodium Sulfide at high temperature in a digester which breaks the lignin chemical bonds which bind the cellulose fibers together. After cooking, the fiber or "pulp" is then separated by screening and/or filtering from the undigested wood chips and washed. Additional steps to remove residual lignin are often employed in modern plants. The pulp is then bleached to whiten the fibers, formed into a sheet, dried and rolled.
Figure 1: Simplifies schematic of pulp and paper mill
Chemical reprocessing in the Kraft process is illustrated in Fig. 2
Figure 2: Chemical reprocessing block flow diagram
Measurement of Alkali
The spectra of white liquor and green liquor are shown in Fig.3. Changes in the absorbance spectra are related to changes in the hydrogen bonding of water caused by the concentration of the various alkali salts dissolved in the water.
Figure 3: Spectrum of Green Liquor
PLS calibration models were created for NaOH, Na2S and Na2CO3 concentrations and are shown in Figures 4-6. Correlations for all 3 calibrations are 0.97 or greater and the Standard Error of the Prediction (SEP) is approximately 1 g/L for all 3 calibrations.
A key point to be noted is that these spectra and calibrations were made at a fixed temperature. One issue that must be addressed in this application of FTNIR is the change in the spectrum of water caused by temperature. Figure 7 shows the spectrum of water at various temperatures. The change in the spectrum due to temperature change is great compared to the change in spectrum caused by the alkali concentration.
Figure 4: Predicted vs. Measured Plot for NaOH (g/L)
There are basically three ways to deal with the effects of temperature on the spectra such that it does not affect the measurement. The first, and most obvious is to control the sample temperature using a sample conditioning system. Figure 8 shows such a system for white liquor, green liquor and black liquor. Note the provisions for flushing the system with water and with Hydrochloric Acid to remove deposits. This approach has the advantage of being the most precise, but is also usually the most expensive.
Figure 5: Predicted vs. Measured Plot for Na2CO3 (g/L)
Figure 6: Predicted vs. Measured Plot for Na2S (g/L)
Figure 7: Spectrum of Water at Various Temperatures
Figure 8: System for Pulp and Paper Liquors
Another approach is to create the calibration at a constant and convenient temperature, then to measure the effect of temperature on the output of the calibration model. Temperature changes in the sample often apply a linear bias to the output that can be compensated by a temperature measurement and a factor applied to the analyzer output.
The third approach is to include the variations in temperature into the calibration model. One of the advantages of Partial Least Squares (PLS) calibration models is that they separate effects that are not correlated with the measurement. To do this, the set of samples used to make the calibration model (typically 30-50 samples) must have spectra taken over the entire range of temperatures likely to be seen. This need not be done systematically, that is, by varying the temperature of each sample by set amounts. The spectral scans can be taken at random temperatures, provided the range of temperatures covers the range likely to be observed in the process. This approach is referred to as making the model "robust" to temperature effects.
Figure 9: Temperature compensated FT-NIR measurement
Figure 10: Robust Calibration Model
Figure 11 shows an on-line trend graph of FT-NIR measurements on green liquor. Figure 12 shows the measurement of effective alkali in black liquor.
Figure 11: On-Line Measurement of Alkali in Green Liquor
Figure 12: On-Line Measurement of Effective Alkali in Black Liquor
Other Applications of FT-NIR and Paper Process
Measurements of kappa number in black liquor have been reportedi. Near infrared spectral measurements have been used to predict pulp yield ii. In-situ ATR probes have been used on spent liquorsiii. Andersson and Wilsoniv report excellent measurements of effective alkali (EA), lignin, Xylan, total organic material, Cellulose and Glucomannan among other parameters in black liquor by NIR.
Clearly, we have only begun to scratch the surface of applications for NIR in the pulp and paper industry. Given the capital-intensive nature of the industry, and the size of the facilities, the economic justifications will be found for new and creative applications of NIR technology over the next several years.
- R. E. Hodges, "Applications of near infrared spectroscopy in the pulp and paper industry," PhD thesis, Auburn University, Auburn Alabama, US, 1997.
- T. Lindgren and U. Edlun, "Prediction of lignin content and pulp yield," Nor. Pulp Pap. Res. J., 13(1):78-80, 1998.
- V.M. Saucedo and G.A. Krishnagopalan, "Applications of in-situ near-infrared analysis for the measurement of cooking liquor components during kraft pulping," J. pulp Paper Sci., 26(1):25-30, 2000.
- N. Andersson and D. I. Wilson, http://www.ee.kau.se/forskning/ModSim/nir_tappi.pdf