Effect of wood stiffness on radiata pine kraft pulp quality

Effect of wood stiffness on radiata pine kraft pulp quality ROBERT EVANS*, R. PAUL KIBBLEWHITE** AND MARK J. C. RIDDELL** * Ensis / CSIRO, Private Bag 10, Clayton South, Victoria 3169, Australia ** Ensis / SCION, Private Bag 3020, Rotorua, New Zealand ABSTRACT Longitudinal modulus of elasticity (MOE) of solid wood from eleven radiata pine clones grown in New Zealand was previously found to strongly affect the properties of handsheets made from those clones. This work has now been extended to include a wider range of samples, including seedling trees. In this work, the pulping conditions were kept constant (in this case, all pulps were produced at a nominal kappa number of 30), and the degree of PFI refining was used as a model variable. Wood stiffness, as estimated by diffractometry and densitometry on SilviScan-2, is a composite property. It includes information on density, and (indirectly) microfibril angle, and the proportion of microfibrillar reinforcement in the cell wall. Density itself is a composite property, including fibre wall thickness and fibre diameter. High density implies a large wall thickness/ diameter ratio, which imparts high collapse resistance to the fibre, reducing sheet density. Low microfibril angle and a high proportion of microfibrillar reinforcement impart high intrinsic longitudinal stiffness to the fibre, also reducing sheet density. The MOE of the wood conveniently encapsulates these fundamental properties, allowing simple statistical models to be produced for the prediction of several sheet properties, such as density, tear index, and the product of tear index and tensile strength index (used as an indicator of the quality of reinforcement pulps). The models indicate that increasing wood MOE decreases handsheet density and tensile strength, and increases tearing strength. High MOE has a generally positive influence on the product of tear index and tensile index. INTRODUCTION Optimisation of the value of a wood resource starts with its characterisation and with the development of relationships between wood properties and product properties. In an increasingly competitive commercial environment, wood quality in plantation trees is becoming a major issue. Although every physical and chemical property of wood has the potential to affect the properties of products, it is not practical to have structure/property models that include large numbers of variables. In this work, we examine a few key properties that are relatively easy to measure and account for a large proportion of variability in paper handsheet density, tearing strength and tensile strength. Pulping and papermaking conditions have been kept constant, so that the only processing condition that needs to be considered is the degree of pulp refining in the PFI mill. The wood properties chosen are wood density, microfibril angle (MFA) and MOE. Our previous work (1) involved the analysis of 16-year-old radiata pine clones. In the current study we will examine three other sets of samples using relationships generated from the earlier data. As the MOE of wood has not been explored as a contributor to paper quality, we will concentrate here on this property at the expense of many other properties such as fibre length, fibre cross-sectional dimensions and chemistry EXPERIMENTAL Sample selection and preparation Three sets of trees previously characterised by SilviScan and by pulp evaluation were chosen for this analysis: a) Tree-set16: Eleven 16-year-old radiata pine clones, selected to cover the extremes of basic density and tracheid length, as determined from breast height increment cores. Two clones were selected for each density/tracheid length grouping, and each clone was represented by two trees. b) Tree-set13/15: Twenty-five half-sib progeny of 25 parents selected from a pool of 200 progeny at age 13, and a further 25 selected at age 15 on the basis of extremes in tracheid length and perimeter. c) Tree-set 27T: Toplogs from twenty 27- year-old clones, 2 trees from each clone. d) Tree-set 27S: Slabwood from the same 27- year-old clones, 2 trees from each clone. Discs were cut from the trees at approximately 5-8 heights. SilviScan strips of tangential width ~15 mm were cut from the discs, solvent-exchanged in ethanol (two soakings each of one week duration), and airdried. Thin pith-to-bark strips 2mm wide (tangential direction) x 7mm high (longitudinal direction) were cut with twin-blade saws from each dried ~15mm strip, continuously extracted overnight with hot acetone to remove resins then conditioned to 20 C and 40% relative humidity. Wood property measurement The average density of each strip was measured gravimetrically and used to normalise the corresponding density profile obtained by x-ray scanning densitometry on SilviScan-2. MFA was estimated from the cellulose-i Appita 2006-267

002 azimuthal x-ray diffraction peaks obtained with the SilviScan-2 scanning x-ray diffractometer system (2). Wood stiffness (MOE) was obtained by the method of Evans (3) using densitometric and diffractometric data from SilviScan-2. The MOE model employed here was calibrated using the sonic resonance method on short clears (4). It is important to note that the MOE values reported here are comparable only with other values measured by the sonic resonance method. Static stiffness methods, for example, give lower values. The only set of samples for which wood properties were not directly measured is Tree-set27T. These toplogs were above the highest sampling point for wood property analysis. However, it was assumed that there would be a reasonable correlation between the properties of the toplog and the those of the rest of the tree. Kraft pulping and handsheet properties Individual trees, logs and slabwood samples were chipped in a commercial chipper, and the chips retained between 40mm and 10 mm hole-screens. A ~5kg sample of chips from each tree was pulped to Kappa number 30±2 using 200g o.d. of chips with a 15% effective alkali charge at 4:1 liquor to wood ratio and H- factor levels in the range 1500 2000. Handsheets were prepared and pulp physical evaluations carried out according to APPITA standards. PFI refining to 4 levels (500, 0, 2000 and 3000 or 4000 revs) was carried out using a 3.4 N/mm load and 10% stock concentration. Handsheet physical evaluation data are reported on an o.d. basis. RESULTS AND DISCUSSION Handsheet density Paper properties are strongly governed by the degree of consolidation of the sheet. The greater the bonding, the higher the tensile strength and (in the case of softwoods) the lower the tearing strength, after an initial increase. Sheet consolidation is reflected in its density. Wood fibre properties that control sheet density would be expected to influence sheet strengths. Wood density is one property widely known to correlate strongly with paper density. Wood density is a composite property that depends on both cell wall thickness and cell diameter, or perimeter. High density wood has thick fibre walls relative to fibre crosssectional perimeter. Such fibres resist collapse and therefore produce paper with a more open structure (Figure 1). In addition, such fibres do not bend as easily as those that collapse into ribbons. High bending stiffness therefore prevents efficient contact where fibre intersect (Figure 1). The longitudinal stiffness, or modulus of elasticity (MOE) of fibres is also controlled by microfibril angle (MFA). Low MFA is associated with high MOE. SilviScan is routinely used to estimate variations in density, MFA and MOE of wood. The method used by SilviScan to estimate MOE takes into account density and microfibril orientation (3). MOE was expected, therefore, to be an efficient predictor of paper density and strengths. Figure 1. Schematic cross-section of fibre overlap area in a low density paper sheet. Intra- and interfibre void spaces contributing to the low sheet density are shown in black. Tear index Tear index is a measure of the energy expended as the sheet is torn out of plane. It is governed to a large extent by fibre strength and the degree of bonding in the sheet. An increase in sheet density results in increased fibrefibre bonding and a greater likelihood that fibre will break, expending less energy than they would if they pulled out of the sheet. Greater tearing strength can be obtained by increasing the strength of the fibres, allowing higher sheet densities before fibre breakage. The increase in bonding results in greater energy consumption when the sheet is torn. Higher wood MOE should result in stronger fibres and a more open sheet, which should result in higher tearing strength. Tensile index Tensile strength is also controlled by interfibre bonding and fibre strength. Both need to be high if high tensile strength is required. However, high fibre MOE should have a negative influence on bonding and a positive influence on fibre strength. The influence of MOE on handsheet tear index is therefore not as easy to predict. Tear index*tensile index Softwood pulps used for reinforcement are required to exhibit high tearing strength at a given tensile strength. As paper density increases, tear index decreases and tensile index increases. The product of tear index and tensile index is therefore a convenient indicator of the quality of such pulps. The data from Tree-set16 was previously found (1) to produce strong relationships between handsheet properties and wood properties (especially MOE and density). In the following analysis we will use models generated from Tree-set16 to examine the other three sets of data. Table 1 lists the selected wood properties and handsheet properties for Tree-set16. The data for each clone is the average for 2 trees. Appita 2006-268

Table 1. Selected properties of 16-year-old clones (average of two trees per clone, and 4 trees for clone 7), together with handsheet properties used for statistical modelling Clone Wood density kg/m³ MFA deg MOE GPa PFI revs Sheet density kg/m³ Tear index mn.m²/g Tensile index N.m/g 1 355 21.3 7.2 500 730 9.1 97.0 0 749 8.5 101.5 2000 767 7.9 109.0 4000 783 7.6 114.0 2 361 17.6 9.3 500 703 10.2 91.0 0 719 10.0 99.0 2000 736 9.2 105.0 4000 757 8.4 111.0 3 397 19.1 9.0 500 708 10.5 85.5 0 710 9.7 91.5 2000 736 8.7 98.0 4000 750 8.4 107.5 4 399 16.5 10.4 500 673 12.6 79.5 0 695 11.3 91.0 2000 715 10.3.5 4000 738 9.5 106.5 5 414 22.3 8.0 500 726 10.3 86.0 0 736 9.6 93.5 2000 761 8.5 103.0 4000 778 8.2 108.5 6 410 18.1 10.1 500 682 12.0 87.0 0 704 11.2 92.5 2000 730 10.0 99.5 4000 739 9.1 110.5 7 421 18.8 10.3 500 683 12.5 81.0 0 701 11.4 90.0 2000 726 10.6 96.5 4000 744 8.9 107.0 8 424 17.5 10.9 500 673 12.6 84.5 0 699 11.9 91.0 2000 717 10.6 99.5 4000 738 9.2 108.5 9 444 17.8 11.3 500 672 14.0 84.0 0 693 12.3 92.5 2000 717 10.8.0 4000 733 9.7 111.0 10 440 17.5 12.0 500 668 14.3 82.5 0 683 13.2 90.5 2000 716 11.5 98.5 4000 734 10.4 105.5 11 452 17.1 11.5 500 673 13.5 75.0 0 684 12.3 86.0 2000 705 11.3 95.0 4000 726 10.3 106.5 Multiple linear regression models As the aim of this exercise was to identify very simple relationships between wood properties and paper properties, we generated multiple linear regression models containing only one wood property together with the degree of refining (PFI revs). The parameters of these models are listed in Table 2. We see from the model equations that MOE has a positive influence on tear index, a negative influence on tensile index, and a positive influence on their product. Figure 5 (Tree-set27S) indicates strong correlations between wood properties and paper properties even though the predicted values drift away from the predictions from Tree-set16. As mentioned above, the properties of the more mature Tree-set27S generally lie beyond those of Tree-set16. By including Tree-set27S in the calibration, we should improve the predictions overall. Appita 2006-269

0 Sheet density (revs, MOE) R 2 = 0.96 Sheet density (revs, wood density) R 2 =0.79 Sheet density (revs, MFA) R 2 =0.90 750 650 650 750 16 Tear index (revs, MOE) 14 R 2 =0.97 650 750 Tear index (revs, wood density) R 2 =0.81 650 750 0 Tear index (revs, MFA) R 2 =0.72 12 10 8 6 6 8 10 12 14 Tensile index (revs, MOE) 110 R 2 =0.91 90 70 70 90 110 1400 Tear * Tensile (MOE) R 2 =0.85 0 6 8 10 12 14 Tensile index (revs, wood density) R 2 =0.93 70 90 110 Tear * Tensile (wood density) R 2 =0.51 Tensile index (revs, MFA) R 2 =0.87 70 90 110 Tear * Tensile (MFA) R 2 =0.50 0 0 0 0 0 0 0 0 0 0 0 1400 Figure 2. Illustration of the strengths of simple statistical models (for selected handsheet properties) generated using Tree-set16. The variables contributing to each model are shown in parentheses. Measured values are on the vertical axis and predicted values on the horizontal axis Appita 2006-270

Table 2. Paramaters of multiple linear regression models based on data in Table 1. Offset log(pfi revs) MOE MOE*log(PFI revs) MOE^3.15 wood density MFA R 2 dependence on MOE Sheet density 645.4 64.6-13.1 0.96 Tear index 0.676 2.09-0.397 0.97 Tensile index 33.5 26.5-2.02 0.91 Tear index * tensile index 786 0.144 0.85 dependence on wood density Sheet density 706.4 64.6-0.467 0.79 Tear index 6.55-3.25 0.0345 0.81 Tensile index 56.2 26.5-0.104 0.93 Tear index * tensile index 90 2.22 0.51 dependence on MFA Sheet density 330.6 64.6 9.97 0.90 Tear index 30.5-3.25 0.528 0.72 Tensile index -8.6 26.5 1.19 0.87 Tear index * tensile index 1710-38.4 0.50 For the models involving MOE, the results of combining Tree-set16 and Tree-set27S are illustrated in Figure 6. The model parameters are in Table 3. Note that the greater range of properties has significantly improved the strength of the tear*tensile model. For the combined sets, log(tear index) was more efficiently modelled than tear index itself because the greater range of values introduced more curvature into the relationship. Note that the sensitivity of tear index * tensile index to MOE required the use of an exponent (3.15) on MOE to account for curvature. Figure 2 indicates the strength of the Tree-set16 models. The R 2 values presented here relate to the proportion of variance between clones that is explained by the model. MOE is clearly a strong predictor of sheet density, tear index, tensile index and the product of tear index and tensile index. Density is a weaker predictor of sheet density, tear index and tear*tensile but is a strong predictor of tensile index. MFA has a strong influence on handsheet density, but is weaker in its prediction of tear index and tensile index. Wood density nor MFA were relatively poor predictors of tear*tensile. Note that the product of tear index and tensile index does not depend on the degree of refining for these models. The effects of applying the models for Tree-set16 to Tree-set13/15, tree-set27t and Tree-set27S are illustrated in Figures 3-5. In several cases, the measured values tend to drift away from the predictions. The prediction of sheet density using MOE is good for Tree-set13/15 and Tree-set27T, although scatter (precision) is lower than for the calibration set (Treeset16). The slabwood set (Tree-set27S) exhibits wood and paper properties that fall well outside the range of Tree-set16, so it is not surprising that the predictions drift further from the measurements in this case. Tear*tensile prediction is poor for Tree-set13/15 (Figure 3), although the average of the whole set was well predicted by MOE and by wood density. In spite of the fact that the wood properties in the toplogs were not measured (they were estimated from the values obtained from the rest of the stem), the quality of the predictions for tree-set27t (Figure 4) was similar to that for Treeset13/15. This indicates that the properties of the wood in the toplogs are strongly correlated with the properties in the rest of the stem. Appita 2006-271

0 Sheet density (revs, MOE) Sheet density (revs, wood density) Sheet density (revs, MFA) 750 650 650 750 18 Tear index (revs, MOE) 16 14 12 10 8 650 750 Tear index (revs, wood density) 650 750 0 Tear index (revs, MFA) 6 Tensile index (revs, MOE) Tensile index (revs, wood density) 18 Tensile index (revs, MFA) 1400 Tear * Tensile (MOE) Tear * Tensile (wood density) Tear * Tensile (MFA) 0 0 0 0 0 0 0 0 0 0 0 0 1400 Figure 3. Application of Tree-set16 models to seedling trees (Tree-set13/15). No changes were made to the model constants. Lines of exact correspondence are shown. Measured values are on the vertical axis and predicted values on the horizontal axis. Appita 2006-272

0 750 Sheet density (revs, MOE) Sheet density (revs, wood density) Sheet density (revs, MFA) 650 650 750 18 Tear index (revs, MOE) 16 14 12 10 8 6 Tensile index (revs, MOE) 650 750 Tear index (revs, wood density) Tensile index (revs, wood density) 650 750 0 Tear index (revs, MFA) 18 Tensile index (revs, MFA) 1400 Tear * Tensile (MOE) Tear * Tensile (wood density) Tear * Tensile (MFA) 0 0 0 0 0 0 0 0 0 0 0 0 1400 Figure 4. Application of Tree-set16 models to toplogs from 10 clones (Tree-set27T). No changes were made to the model constants. Lines of exact correspondence are shown. Measured values are on the vertical axis and predicted values on the horizontal axis. Appita 2006-273

0 Sheet density (revs, MOE) Sheet density (revs, wood density) Sheet density (revs, MFA) 500 500 30 Tear index (revs, MOE) 25 500 Tear index (revs, wood density) 500 0 Tear index (revs, MFA) 20 15 10 5 5 10 15 20 25 Tensile index (revs, MOE) 5 10 15 20 25 Tensile index (revs, wood density) 5 10 15 20 25 30 Tensile index (revs, MFA) 40 40 2000 Tear * Tensile (MOE) 40 Tear * Tensile (wood density) 40 Tear * Tensile (MFA) 1500 0 500 500 0 1500 500 0 1500 500 0 1500 2000 Figure 5. Application of Tree-set16 models to slabwood from 10 clones (Tree-set27S). No changes were made to the model constants. Lines of exact correspondence are shown. Note the increased scale required for tear index and tear * tensile. Measured values are on the vertical axis and predicted values on the horizontal axis. Appita 2006-274

measured handsheet density kg m -3 0 Tree-set16 + Tree-set27S 750 R 2 = 0.94 650 550 500 500 550 650 750 0 predicted handsheet density kg m -3 measured tensile index N m/g Tree-set16 + Tree-set27S R 2 = 0.91 40 40 predicted tensile index N m g -1 measured tear * tensile 1 Tree-set16 + Tree-set27S R 2 = 0.83 1400 0 0 measured log(tear index) 1.5 1.4 1.3 1.2 1.1 1.0 Tree-set16 + Tree-set27S R 2 = 0.92 0 0 0 0 1400 1 predicted tear * tensile 0.9 0.9 1.0 1.1 1.2 1.3 1.4 1.5 predicted log(tear index) Figure 6. Strengths of calibration models using the combined Trees-set16 and Tree-set27S. Table 3. Parameters of multiple linear regression models from combined Tree-set16 and Tree-set27S data sets (see also Figure 6). Offset log( PFI revs) MOE MOE*log(PFI revs) MOE^3.15 R 2 dependence on MOE Sheet density 868 2.46-36.5 6.4 0.94 Log(Tear index) 0.4186 0.0541 0.1087-0.0205 0.92 Tensile index 108.1 8.19-9.84 1.88 0.91 Tear index * Tensile index 1040-88.6 2.47 0.83 CONCLUSIONS The longitudinal modulus of elasticity of wood appears to be a strong predictor of the physical properties of handsheets made from that wood by the kraft process. The strong relationships between MOE and paper properties can be explained qualitatively by the effect of MOE on paper structure. Although (eg Ref. 5) other fibre properties such as length, wall thickness and perimeter affect paper properties, it appears that MOE is of considerable importance. An obvious extension to this work is the application of acoustic velocity and resonance tools to the large assessment and segregation of pulpwood (trees and logs) for potential papermaking quality. In addition, for sawlogs, appropriate acoustic models could allow the value of the wood to combine sawn timber quality with residual woodchip quality. REFERENCES 1. Evans, R., Kibblewhite, R.P. and Riddell, M.J.C. - 58 th Appita Annual General Conference, Canberra (2004). 2. Evans, R. - Appita J. 52(4): 283-289, 294 (1999). 3. Evans, R..- Proceedings of the workshop Characterisation of the cellulosic cell wall, 25-27 August, Grand Lake, Colorado, USA. Southern Research Station, University of Iowa and the Society of Wood Science and Technology (2003) 4. Evans, R. and Ilic, J. -. Forest Products Journal 51(3), 53-57 (2001) 5. Kibblewhite, R.P., Evans, R. and Riddell, M.J.C. - th Appita Annual General Conference, Melbourne (2006). Appita 2006-275