5.1.1. ICP-AES

The ICP-AES method has low detection limits and permits several metals to be determined simultaneously. It is a typical liquid analysis method and so the pulp sample has to be wet ashed or metals have to be extracted into the aqueous phase. Since the wet ashing would be difficult to perform on-line, this work focused on developing a simple but reliable method for the extraction [VI]. Target metals were Al, Ba, Ca, Cr, Cu, Fe, Mg, Mn, Ni, Ti, Zn and the performance of four different chelating agents (EDTA, DTPA, nitrilotriacetic acid NTA and diethylenetriamine pentakis (methylenephosphonic acid) DTPMP) was compared. In addition, acid hydrolysis without any chelating agent was performed.

As an experimental procedure, 20 g of pulp (dry matter about 10%) was added to 150 ml of extraction solution in which the dosage of chelating agent was 1.83 mmol. This amount was approximately ten times higher than the molar concentration of the metals. Nitrogen bubbles carried out mixing. After the extraction, 20 ml liquid was bled with a filtrated syringe, to which 1 ml concentrated nitric acid had been added, to adjust acid concentration corresponding that of ICP standards. To determine the total contents of metals in the pulp, the samples were dry-ashed and the recovery of the extraction was calculated by comparing the amount of metals in the extraction solution to the total content.

Regarding the experiments with the chelating agents, optimal recoveries were obtained using DTPMP at 85 °C and pH 3. A contact time of 15 minutes was sufficient. In general, recoveries increased as a function of stability constants of the chelating agents and temperatures. When the temperature approached 100 °C, mixing was complicated due to swelling of fibres. Overall, the best extraction method was acid hydrolysis [86, 87] by 0.1 mol/l HCl at 95 °C, but this kind of solution may cause corrosion damages, if metal constructions are used in on-line equipment.

Over 95% recoveries were obtained for Mn and also for Ba, Ca, Mg, Zn. Recoveries for iron and aluminum were matrix dependent, 40-80% and 15-40%, respectively. Samples contained very little Cr, Cu Ni and Ti and so conclusions for these metals were difficult to draw. Mn, Ca, Mg and Zn showed no differences between pulp types. However, Al and Fe were easier to extract from softwood pulp than from hardwood pulp. The method was linear in the range of investigations. The reproducibilities, repeatabilities and determination limits varied between 1.7 and 7.0 %, 0.3 and 1.7 % and between 0.05 (Mn) and 2.1 (Al) mg/kg, respectively.

5.1.2. X-ray fluorescence

Besides being rapid, the XRF-technique requires little or no sample pretreatment, so that process control can be achieved with a minimal delay. This study concentrated on the three technically most important metals, Mn, Fe and Cu [VII]. As a complex matrix, pulp makes demands on the XRF, too. The solid content of pulp taken from bleaching lines is typically 10%, the distribution of metals between fibres and the aqueous phase may differ and precipitation may occur [10, 88]. Differences in distribution of fibres on the sample film might also influence on the detected fluorescence.

To avoid matrix effects, a standard addition method was investigated. Preparation of addition standards and samples was commenced by weighing 10.00 g hardwood pulp into a polypropylene decanter. Further, to avoid the distribution problems, concentrated nitric acid was added to all addition standards and samples to enhance solubility and equalize the distribution. The target metals were added to the standards as 15-300 mg/l solutions. The total solution addition to the 10.00 g of pulp was always 2 ml. The solution added to the samples consisted of pure nitric acid. Five standard addition calibration lines were prepared for Mn, Fe and Cu in their realistic concentration ranges; 0-15 mg/kg for Mn and 0-5 mg/kg for Fe and Cu. The pulp to which the additions were made was taken from a final wash of an ECF line and contained very low amounts of metals initially.

The averages of squares of the linear correlation coefficients between measured intensity and added metal concentrations were 0.994, 0.950 and 0.932, for Mn, Fe and Cu, respectively. The reproducibilities and repeatabilities turned out to be concentration dependent and varied between 3 and 19 % and between 1 and 17%, respectively. The quantification limits were 2-3 mg/kg.

To find out whether the analysis could be carried out without the addition of acid, which would simplify on-line construction, four different process samples were analysed. The results are given in Table 3 in which ”Result without pretreatment” refers to the situation where no nitric acid was added. In that case, the pulp was simply pressed lightly onto the Mylar film. The difference between the two measurements is not considered to be significant and evidently manganese is distributed uniformly in the samples.

Using air instead of helium as a chamber gas would also simplify the on-line equipment construction. To study this, acidic metal solutions of 4 mg/l were measured as before, the only difference was the chamber gas. The measured intensities with air were 60% of the intensities with helium, indicating that air does attenuate the fluorescence of the elements. This effect could be minimized if the path length of the X-rays before and after the sample could be shortened.

On-line XRF spectrometry has been successfully applied in the mining industry for a long time. In these applications slurry courses are driven continuously in a cuvette containing a sample window with changeable film. With the wet pulp matrix this is problematic, because the aquatic phase easily separates from the fibres. A screw pump might the choice with the most potential. Discrete pulp dosages could be possible to move on to an XRF sample window. There could be also another way to analyse metals on-line in pulp. Present pulp sampling technology allows preparation of a dry briquette in five minutes. Only simple robotics would be needed to move the briquette to the sample window of an XRF-spectrometer.

5.1.3. Ion selective electrodes

As a cost effective technique, potentiometry is a noteworthy alternative for on-line determination of K+, Ca2+, Na+ and Cl- ions. In this study, a series of ISE measurements of calcium and sodium from pulp samples taken from a TCF bleaching line was performed [VIII, IX]. The purpose of these measurements was to compare ISE measurents carried out in different ways and to create information about the distribution of these metals. The total concentrations were measured by AAS. The results for calcium are presented in Table 4. The ”ISE in line” column refers to a measurement without any sample preparation. This type of measurement can be thought to represent a situation in which an ion selective electrode is placed directly in a process stream or in a separated side stream. The ion strength of the calibration standards was 0.1, which is typical of bleaching lines, and the error resulting from differences in activity coefficients can thus be assumed to be insignificant. Considering the results in the ”ISE on line” column, an ionic strength adjustor was used. These measurements simulate a situation in which a sample has been taken from a process stream into a separated vessel in which potassium chloride solution has been added to adjust the total ionic strength to unity, which was now the ionic strength of the calibration standards, too.

The P-phases were alkaline in which only a negligible amount of calcium can exist as free ions in solution according to thermodynamics [89]. Perhaps due to kinetic hindering by the DOC, however, [13, 80] a certain amount of calcium could be detected in a solution, as Table 4 reveals. The pH in the Z/Q phases was adequately low to allow measurements in the atmosphere (solubility of the free Ca2+ ion is about 80 mg/l at pH 8 [89]), and it can be seen that a considerable part of calcium indeed exists as free Ca2+ ions. A similar observation has been made earlier under paper machine conditions [90]. It deserves to be mentioned that in the absence of the DOC, precipitation of calcium carbonate increases very rapidly as a function of pH [VIII].

As an interesting observation, the results obtained without using the ionic strength adjustor were lower than those obtained with the adjustor. The reason might be that a part of the ionic strength adjustor adsorbted to the fibres. Because of this, activity coefficients in samples were higher than those in the calibration standards and the measured concentrations were thus higher.

Since glass electrodes bearing high temperatures have been developed, the ISE technique allows good facilities also for monitoring sodium ion levels. A series of sodium ion selective measurements similar to that of calcium was performed in this work. The results are presented in Table 5. Considering on-line measurements, the same systematic error as in the case of calcium was observed. The ion strength was adjusted with an NH₄Cl+NH₄OH, solution which also ensured pH values high enough to avoid interference of the hydrogen ions. Table 5 reveals that both the in-line and the on-line measurements are reliable corresponding well to the total concentrations analysed by AAS. Thus, the main part of the sodium existed as a free or weakly adsorbed Na+ ion.

5.1.4. Comparison of metal analysis methods

The main points of the methods discussed above are summarized in Table 6 to which have been included some practical expert opinions dealing with on-line feasibility.