|[Home] [APPW 2004] [Journal papers]|
ESTIMATION OF PROPORTIONS OF DIFFERENT PULP TYPES IN RECYCLED PAPER AND DEINKED PULP BY ACID METHANOLYSIS
Anna Sundberg, Stefan Emet, Patrik Rehn, Lari Vähäsalo and Bjarne Holmbom
Presented at the 59th APPITA Annual Conference, Auckland, New Zealand
We have developed and evaluated a method to estimate the proportion of bleached kraft pulp from hardwoods (HP), bleached kraft pulp from softwoods (SP) and mechanical pulp (MP) in deinked pulp (DIP) or recycled paper (RP) by analysis of the sugar units in hemicelluloses and pectins.
Different pulps, i.e., HP, SP and MP, were obtained from Nordic pulp mills. The pulps were mixed in different proportions and dried. The content of arabinose, xylose, mannose, galactose, rhamnose, galacturonic acid and 4-O-methylglucuronic acid in hemicelluloses and pectins in the different pulps and in the mixtures were determined by acid methanolysis and gas chromatography (GC).
The contents of the sugar units were significantly different in the three pulp types, but quite similar in the same pulp type, irrespectively of bleaching sequence.
A linear model for estimating the relative proportions of HP, SP and MP in DIP or RP were formulated. Hence, given a predefined set of errors, εi , and data of HP, SP and MP, the relative proportions can be estimated by solving a linear least-squares problem.
It was concluded that the proportion of hardwood kraft pulp, softwood kraft pulp and mechanical pulp in deinked pulp or recycled paper can be estimated using a mathematical model based on the contents of the sugar units in hemicelluloses and pectins. The difference between the actual mixing ratio and the estimated values was on average less than ± 5%.
KEYWORDS: Carbohydrates, Hemicelluloses, Pectins, Recycled Paper, Deinked Pulp, Gas Chromatography, Parameter Estimation, Constrained Least-Squares Optimization.
The proportion of different pulp types, i.e., hardwood kraft pulp (HP), softwood kraft pulp (SP) and mechanical pulp (MP) will vary much in deinked pulp (DIP) and in recycled paper (RP) depending on the raw material. The proportion of kraft pulp will influence the strength properties of the recycled paper, and hence how much additional kraft pulp are needed as reinforcement pulp. In order to use as small additions of expensive kraft pulp as possible, the proportions should be known.
One possibility of estimating the proportion of the different pulp types is to calculate the fibres originating from different types of pulps by microscopy. This, however, requires a skilled operator and the reproducibility of the determination is low. The mechanical pulp furthermore contains much fines, which are difficult to count and will therefore not be included in the estimation.
During analysis, the polysaccharides are cleaved into sugar units. Acid methanolysis is a convenient method that only cleaves the hemicelluloses and pectins and not cellulose, as acid hydrolysis (Sundberg et al. 1996). The neutral and the more sensitive acidic sugar units can be analysed with the same method. Only small sample amounts are needed, about 10 mg. The results are reproducible.
The drawbacks are that the method is time consuming and that resorbed mannans can be difficult to analyse.
Galactoglucomannans are the dominating hemicelluloses in softwoods, comprising about 20% of the wood, while the amount in hardwoods is much lower, about 1-2% (Sjöström 1993). The ratio of Gal:Glc:Man in spruce is about 0.5:1:4. Birch consists of about 15-30% of 4-O-methylglucuronoxylans, with a ratio of MeGlcA:Xyl about 1:10. In softwoods, xylans also contain arabinose and the ratio of Ara:MeGlcA:Xyl is about 1.3:2:10.
Galacturonic acid and rhamnose are constituents of pectins. A ratio of GalA:Rha about 6:1 has been found in spruce TMP (Sundberg et al. 2003). Arabinogalactans are water-soluble hemicelluloses. They consist of arabinose, galactose and glucuronic acid in the ratio Ara:Gal:GlcA about 1:4:0.5 (Willför et al. 2002). Note that some sugar units, e.g., arabinose and galactose, are found in two hemicelluloses.
The composition of hemicelluloses is thus very different in hardwoods and in softwoods. During kraft pulping much of the hemicelluloses and pectins are degraded and dissolved into the cooking liquor (Sjöström 1993), the amounts of hemicelluloses and pectins are therefore much lower in kraft pulps. The composition of certain sugar units, constituents of hemicelluloses and pectins, could therefore be used as a base for a mathematical model.
The amounts of the sugar units in hemicelluloses and pectins were determined for several hardwood kraft pulps, softwood kraft pulps and mechanical pulps. The amounts were used to construct a linear mathematical model. Based on the sugar units in deinked pulps or recycled papers the model was used to estimate the proportions of the three types of pulps.
Four fully bleached hardwood kraft pulps (HP) were obtained in dry lap form from pulp mills in Finland (Table 1). One of the pulps was bleached with a TCF sequence; the other pulps were bleached with ECF sequences. The pulps were stored at -24oC and used without further purification.
Four fully bleached softwood kraft pulps (SP) were obtained in dry lap form from pulp mills in Finland (Table 1). All pulps were bleached with ECF sequences. The pulps were stored in a freezer at -24oC until used.
The raw material for all mechanical pulps (MP) was Norway spruce (Picea abies). The pulps were: an unbleached groundwood pulp, a dithionitebleached groundwood pulp, a dithionite -bleached thermomechanical pulp (TMP) and a peroxidebleached TMP (Table 1). The dissolved and colloidal substances were removed by washing. The pulps were suspended in distilled water, stirred at 60oC for 30 min and filtered on a paper machine wire. The filtrate was passed through the same fibre pad one more time in order retain all fines. This washing procedure was repeated until the total organic carbon (TOC) value of the filtrate was lower than 10 mg/l. The dry content was determined and the pulps were stored in a freezer until needed.
Two deinked pulps (DIP) were obtained from a Swedish paper mill. One recycled paper, consisting of 2.3% chemical pulp, 40% DIP and the rest TMP, was obtained from the same Swedish paper mill. A Finnish newsprint paper, composition unknown, was also tested.
The different pulps (SP, HP and MP) were mixed in different proportions (Table 2), diluted to 1% consistency and stirred with a magnetic stirrer at about 900 rpm for 30 min at about 60oC. The suspension was then disintegrated with a mixer for 2 min. The suspension was dewatered on a paper machine wire and the filtrate was passed through the same fibre pad one more time, in order to retain all fines in the pulp. The pulp was homogenised and part of the pulp was freeze-dried while the rest was stored in a freezer.
Hemicelluloses and pectins were determined by acid methanolysis and GC (Sundberg et al. 1996). All pulps or mixtures of pulps were freeze-dried. About 10 mg was submitted to acid methanolysis with 2 mol/L HCl in anhydrous methanol. The methanolysis degrades non-cellulosic polysaccharides into their monomeric sugar units. After 5 h at 100oC, the samples were neutralised with pyridine, and sorbitol was added as internal standard. Part of the clear sample was transferred to a new flask and dried, silylated and analysed by GC.
The sugar units quantified were arabinose (Ara), xylose (Xyl), galactose (Gal), glucose (Glc), mannose (Man), rhamnose (Rha), glucuronic acid (GlcA), 4-O-methylglucuronic acid (MeGlcA) and galacturonic acid (GalA).
RESULTS AND DISCUSSION
Composition of hemicelluloses and pectins in pulp used
The contents of sugar units in the different pulp are presented in Table 3. The relative standard deviation of the analytical method has been shown to be below 5%, both for neutral and acidic sugars (Bertaud et al. 2002).
Table 3: The compositions of sugar units in hemicelluloses and pectins in the original pulps
The amount of xylose was high in HP (Table 3) due to the high content of xylans in hardwood (Sjöström 1993). The amount of especially mannose but also galactose was high in MP, due to the presence of high amounts of mannans in mechanical pulp from spruce. The amount of mannans was much smaller in SP than in MP, since a large part of the mannans are degraded and dissolved in the cooking liquor during kraft cooking and bleaching.
Significant amounts of galacturonic acid, found in the pectins, were found only in MP. The amount of glucose was rather similar in all types of pulps (Table 3), which suggests that glucose is not useful in the model. Furthermore, amorphous cellulose present in the kraft pulps can partly be degraded to glucose during acid methanolysis (Sundberg et al. 1996). The recycled paper can also contain starch, which is degraded to glucose during acid methanolysis. Glucose was therefore not used in the model.
The amounts of glucuronic acids were small in all types of pulps. The determination of glucuronic acid can be difficult, especially if the column is not in a very good condition. This sugar unit was also excluded from the model.
A linear model for estimating the relative fractions of HP, SP and MP within DIP (or RP) can be formulated as follows:
where α, β and γ are parameters that denote the relative fractions, respectively. These should also satisfy the following constraints:
α + β + γ = 1, 0 ≤ α ≤ 1, 0 ≤ β ≤ 1 and 0 ≤ γ ≤ 1 .
Note, that HP, SP, MP and DIP are n-dimensional vectors that consists of the determined sugar units. The following sugar units were considered: arabinose (Ara), xylose (Xyl), galactose (Gal), mannose (Man), rhamnose (Rha), 4-O-methylglucuronic acid (MeGlcA) and galacturonic acid (GalA). The problem can be formulated as a constrained linear least-squares problem:
The problem (2) can thus be solved using, for example, the Matlab lsqlin-function (The Mathworks Inc. 2000). Above it is assumed that the given data; HP, SP and MP, and the observations of DIP are exact ones. It should, however, be noted that the absolute measure errors in obtained sugar units was approximately less than 5%. An alternative model where these errors are considered can be formulated as:
where the errors, εi , are modelled as uniformly distributed random variables such that
Hence, given a predefined set of errors, εi, the problem (4) can be solved similarly as problem (2) using, e.g., Matlab. The degree of confidence of the estimated parameters in (4) can be approximated using Monte Carlo analysis. In this study, the problem (4) was solved for 10000 different sets of random errors, and, hence, the meanvalues and the variances of the estimated parameters were obtained.
ESTIMATION OF PULP PROPORTIONS IN MIXTURES
The sugar units can be combined in different groups according to the composition in the hemicelluloses and pectins (Table 4). Note that in combination I, arabinogalactans and mannans were combined in the same group. In combination III, methylglucuronic acid was excluded from the xylans and from the mathematical model. In combination IV, pectins were also excluded.
Table 4. Different combinations of sugar units
In tables 5-8 the actual mixing proportion was compared with the estimated proportion, calculated from the determined amounts of the sugar units in the mixed pulps using the mathematical model. The differences between the actual and the estimated value were also included in the tables; a negative value for the difference means that the estimated value were lower than the actual mixing ratio.
In most mixtures, the determined amount of mannose was too low (not shown), calculated from the content in HP, SP and MP with the content in the mixtures. This was probably due to the sorption of mannans especially from MP onto kraft pulp. The sorbed mannans are more difficult to cleave during acid methanolysis and are therefore not analysed completely (Sundberg et al. 1996). This was seen as too low estimated values of MP in all combinations (Table 5-8).
Table 5. Actual mixing proportions, estimated proportions and differences between the two, using combination I. Negative difference: estimated value too low
Table 6. Actual mixing proportions, estimated proportions and differences between the two, using combination II.
Table 7. Actual mixing proportions, estimated proportions and differences between the two, using combination III
Table 8. Actual mixing proportions, estimated proportions and differences between the two, using combination IV
The estimated values of HP were very close to the actual mixing proportion for all combinations. The difference was larger for SP and MP, which both originates from softwoods. The combination that gave the smallest difference between the actual mixing proportion and the estimated proportion was combination I. The average difference was below 5%.
ESTIMATED PROPORTIONS IN DIP AND RP
The estimation of the three pulp types on basis of sugar unit composition was tested on two deinked pulps, one unprinted recycled paper and one printed newsprint.
The DIP contained most MP, about 65-68%. DIP 1 contained more SP than HP, 24 and 11% respectively, while DIP 2 contained about equal amounts of HP and SP.
The RP contained very small amounts of chemical pulp and the newsprint contained about 94% MP, 6% SP and no HP. The proportions were reasonable; the RP was known to contain at least 2% chemical pulp and about 60% MP but the DIP of course contained all types of pulps.
Table 9. Estimated proportions in DIP and RP.
This paper presents a new method to estimate the proportions of hardwood kraft pulp, softwood kraft pulp and mechanical pulp in deinked pulp or recycled paper.
The pulp or paper is dried and submitted to acid methanolysis, which cleaves the hemicelluloses and pectins into their sugar units. The amount of the sugar units are determined by gas chromatography.
The sugar units are then combined according to the composition of the hemicelluloses and pectins. A linear mathematic model was used to estimate the proportion of the different pulps.
The method can be used to estimate the proportions in the raw material used in the deinking plant or to classify the recycled paper. However, the determination of the sugar units is still time consuming.
This work is part of the activities at the Ǻbo Akademi Process Chemistry Centre within the Finnish Centre of Excellence Programme (2000- 2005) by the Academy of Finland.
The Mathworks Inc. (2000), Optimization Toolbox User's Guide. Natick, MA
Bertaud, F., Sundberg, A., and Holmbom, B. (2002), Carbohydr. Polym. 48:319-324
Sjöström, E. (1993), Wood Chemistry, Fundamentals and Applications, Academic Press, San Diego, USA
Sundberg, A., Sundberg, K., Lillandt, C. and Holmbom, B. (1996), Nord. Pulp Pap. Res. J. 11(4): 216-219, 226
Sundberg, A., Pranovich, A.V. and Holmbom, B. (2003), J. Pulp Pap. Sci.29(5):173-178
Willför, S., Sjholm, R., Laine, C., Holmbom, B., (2002), Wood Sci. Technol. 36:101-110
Willför, S., Sundberg, A., Hemming, J. and Holmbom, B. (2004), Wood Sci. Technol. in press