DOCTORA L T H E S I S

Department of Engineering Sciences and MathematicsDivision of Materials Science

Extrusion Processing of

Wood-Based Biocomposites

Maiju Hietala

ISSN: 1402-1544 ISBN 978-91-7439-541-9

Luleå University of Technology 2012

Maiju H

ietala Extrusion Processing of W

ood-Based B

iocomposites

ISSN: 1402-1544 ISBN 978-91-7439-XXX-X Se i listan och fyll i siffror där kryssen är

Extrusion Processing of Wood-Based Biocomposites

DOCTORAL THESIS

Maiju Hietala

Luleå University of TechnologyDepartment of Engineering Sciences and Mathematics

Division of Wood and BionanocompositesSE-97187 Luleå, Sweden

University of OuluDepartment of Process and Environmental Engineering

Laboratory of Fibre and Particle EngineeringFI-90014 Oulun yliopisto, Finland

January 2013

Printed by Universitetstryckeriet, Luleå 2012

ISSN: 1402-1544 ISBN 978-91-7439-541-9

Luleå 2012

www.ltu.se

i

Abstract

Interest in biocomposite materials and their use in various applications has been growing steadily over the past 10-15 years. Increasing environmental awareness and lower material costs are the main driving forces for using renewable materials, such as wood and cellulose fibers, as reinforcement in polymer composites. In addition to the materials used in the composite preparation, the processing has a large impact on the final properties of the composite. Therefore, the main thrust of this work has been on processing of wood-based biocomposite materials using twin-screw extrusion.

In the first part of the work (Papers I, II and III), wood-plastic composites weremanufactured using twin-screw extrusion. Currently, wood flour consisting of small wood particles with low aspect ratio is used as the main wood raw material in commercial wood-plastic composites. A better reinforcement is achieved by using wood fibers with a higher aspect ratio, but individual fibers are seldom used in composite manufacturing due to processing problems and higher cost. Therefore, the objective of the first part of this work was to study the use of wood chips as raw material in wood-plastic composites and the possibility to separate individual fibers from the wood chips during the composite manufacturing process. The effect of extrusion parameters and raw materials on the aspect ratio of the wood particles/fibers and on the mechanical properties of the composites was evaluated. The studyshowed that wood chips can be used as raw material in a one-step process for manufacturing wood-plastic composites, and that it is possible to separate individual fibers with a higher aspect ratio than wood flour from the wood chips under suitable processing conditions.

The second part of the work (Papers IV and V) focused on the twin-screw extrusion of cellulose nanocomposites. The use of nanosized cellulose fibers to reinforce polymer matrices has many benefits over the macrosized fibers, such as the high surface area and large aspect ratio. However, the preparation of cellulose nanocomposites is more complicated due to the high hydrophilicity and aggregation tendency of nanocellulose, meaning that drying of the nanofibers is not recommended when good dispersion of nanofibers is needed. Therefore, the aim was to study the processing of green cellulose nanocomposites with twin-screw extrusion using thermoplastic starch as the matrix polymer and cellulose nanofibers with high water content as the reinforcement (Paper IV). In addition, the effect of twin-screw extrusion on separating micro/nanoscale fibers from cellulose fibers during the compounding of biocomposites was studied (Paper V). The fibrillation of nanocellulose is a highly energy intensive process; therefore, it would be very beneficial if it could be done at the same step as the compounding of the composites. The preparation of thermoplastic starch and composite compounding was performed in one step, and the effects of extrusion compounding on the dispersion of the cellulose nanofibers, on the micro/nanofibrillation of cellulose fibers, and on the composites’ mechanical, optical and moisture absorption properties were studied. The results showed that some aggregation of cellulose nanofibers occurred during the extrusion process, but that the addition of cellulose nanofibers had a positive effect on the properties of the prepared bionanocomposites. Nanofibrillation of cellulose was not accomplished using theselected processing conditions; however, dispersion of the fibers was enhanced.

ii

Table of contents

Abstract ............................................................................................................................... i

Table of contents................................................................................................................ii

List of papers ....................................................................................................................iii

List of conference contributions...................................................................................... iv

1 Introduction .................................................................................................................. 1

1.1 Background.............................................................................................................. 1 1.2 Wood-plastic composites......................................................................................... 2

1.2.1 Raw materials ................................................................................................... 3 1.2.2 Processing......................................................................................................... 4 1.2.3 Properties.......................................................................................................... 5

1.3 Cellulose nanocomposites ....................................................................................... 6 1.3.1 Raw materials ................................................................................................... 6 1.3.2 Manufacturing .................................................................................................. 7 1.3.3 Properties.......................................................................................................... 8

1.4 Wood as raw material .............................................................................................. 8 1.4.1 Structure of wood cell wall .............................................................................. 9 1.4.2 Chemical composition.................................................................................... 10 1.4.3 Isolation processes of fibers and nanofibers................................................... 11

1.5 Objectives .............................................................................................................. 14 2 Experimental procedure ............................................................................................ 15

2.1 Materials ................................................................................................................ 15 2.1.1 Raw material pre-treatments .......................................................................... 15

2.2 Extrusion................................................................................................................ 16 2.3 Characterization..................................................................................................... 17

2.3.1 Fractionation................................................................................................... 17 2.3.2 Image analysis ................................................................................................ 18 2.3.3 Microscopy..................................................................................................... 18 2.3.4 Mechanical testing.......................................................................................... 19 2.3.5 Density............................................................................................................ 20 2.3.6 Transparency .................................................................................................. 20 2.3.7 Moisture absorption........................................................................................ 20

3 Summary of appended papers................................................................................... 21

4 Conclusions ................................................................................................................. 24

5 Future work ................................................................................................................ 24

6 Acknowledgements ..................................................................................................... 26

7 References.................................................................................................................... 27

Appendices

Papers I-V

iii

List of papers

Paper I

Hietala, M., Niinimäki, J., Oksman, K. (2011). The use of twin-screw extrusion in processing of wood: The effect of processing parameters and pretreatment. Bioresources 6(4), 4615-4625.

Paper II

Hietala, M., Niinimäki, J., Oksman, K. (2011). Processing of Wood Chip-Plastic Composites: Effect on Wood Particle Size, Microstructure and Mechanical Properties. Plastics, Rubber and Composites 40(2), 49-56.

Paper III

Hietala, M., Samuelsson, E., Niinimäki, J., Oksman, K. (2011). The effect of pre-softened wood chips on wood fibre aspect ratio and mechanical properties of wood-polymer composites. Composites: Part A 42(12), 2110-2116.

Paper IV

Hietala, M., Mathew, A.P., Oksman, K. (2012). Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured using twin-screw extrusion, European Polymer Journal, doi: http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016.

Paper V

Hietala, M., Rollo, P., Kekäläinen, K., Oksman, K. One-step extrusion processing of green biocomposites: compounding, fibrillation and dispersion. To be submitted.

iv

List of conference contributions

Hietala, M., Grubbström, G., Oksman, K. Use of wood chips in extrusion of wood plastic composites. 10th International Conference on Wood & Biofiber Plastic Composites & Cellulose Nanocomposites Symposium, May 11-13, 2009, Madison, Wisconsin, USA. Poster presentation.

Hietala, M., Niinimäki, J., Oksman, K. Wood Chips as Raw Material in Wood Plastic Composites. Proceedings of the 11th International Conference on Biocomposites: Transition to Green Materials. May 2-4, 2010, Toronto, Ontario, Canada. Oral presentation.

Hietala, M., Oksman, K. Effect of extrusion process on wood chip particle size in manufacturing of wood-plastic composites. The 21st Annual SICOMP Conference on Manufacturing and Design of Composites, June 3-4, 2010, Piteå, Sweden. Oral presentation.

Hietala, M., Mathew, A. P., Oksman, K. Extrusion of thermoplastic starch reinforced with cellulose nanofibers. 12th International Conference on Biocomposites: Transition to Green Materials. May 6-8, 2012, Niagara Falls, Ontario, Canada. Oral presentation.

Hietala, M. Extrusion processing of wood and wood based biocomposites. The 23rd Annual International SICOMP Conference, June 5-6, 2012, Piteå, Sweden. Oral presentation.

1

1 Introduction

1.1 Background

Interest in biocomposite materials and their use in various applications has been growing steadily over the past 10-15 years, especially in the case of natural fiber reinforced polymer composites. Increasing environmental awareness and lower material costs are the main driving forces for the growing interest in using natural fibers as reinforcement in polymer composites. Natural fibers, such as wood and cellulose fibers, are considered environmentally sustainable materials due to their renewability and biodegradability. A very appealing characteristic of natural fibers is also the fact that they can be considered as carbon dioxide neutral materials, i.e. they do not release excess carbon dioxide into the atmosphere when composted or combusted [1]. Other main advantages of natural fibers over the synthetic and inorganic fibers traditionally used as reinforcement in polymeric composites are low specific weight, high specific strength and stiffness, safer handling and working conditions and non-abrasiveness to the processing equipment [1]. When the used reinforcement is in nanoscale, such as nanocellulose, even further improvements in the composite properties can be obtained.

As the needs to replace the less environmentally friendly materials with more sustainable ones are growing and new applications for biocomposites are being developed, demands on the performance of composites are also increasing. So far, research in the field of biocomposites has focused strongly on the materials used in the composite preparation, and not so much on the processing on the materials, even though the latter has a large impact on the final properties of the material. One of the most commonly used methods in the industrial processing of polymer composites is the extrusion process; therefore, the focus on this work is on processing of wood-based biocomposites using the twin-screw extrusion method, and especially on development of new processing methods in order to be able to use cheaper raw materials and to improve the properties of the prepared composites.

In the case of wood-plastic composites (WPC), the main wood raw material used nowadays in the composites is wood flour consisting of small wood particles with rather low aspect ratio. Earlier studies have shown that individual wood fibers with higher aspect ratio improve the strength properties of the composites [2, 3], but due to difficult feeding and processing fibersare seldom used in the WPC manufacturing. In twin-screw extrusion mechanical stresses, such as shear and compression, are subjected to the material as it goes into the clearances between the two rotating screws or between the screw and the barrel [4]. Shear forces could also be used to reduce the particle size and to separate fibers with a higher aspect ratio from the wood particles. If the defibration of wood chips could be done in the same extrusion step as the composite compounding, it would be possible to use cheaper wood residues in composite manufacturing and simultaneously improve the mechanical properties of the composite.

2

The typical methods used currently in manufacturing of cellulose nanocomposites are laboratory-scale methods such as solution casting and resin impregnation. However, using more efficient processing methods in the nanocomposite preparation would be important in order to obtain more commercially viable products. Twin-screw extrusion is a suitable method for large-scale processing, but extrusion compounding of cellulose nanocomposites is studied only to some extent, mainly due to the difficult feeding and poor dispersion of cellulose nanoparticles in melt compounding [5]. Especially, if non-polar polymers are used as matrices, dispersion of cellulose nanoparticles with polar surfaces is difficult [6]. Problems also arise from the fact that when nanocellulose is dried prior to compounding, it forms aggregates which are extremely difficult to break in order to achieve good dispersion in the polymer matrix [5]. Chemical pre-treatments can be used to produce dried, non-aggregated cellulose nanoparticles; however, feeding and handling of dried nanoparticles can be a safety issue. Earlier studies have shown that liquid feeding of cellulose nanoparticles in melt compounding of cellulose nanocomposites is possible, but problems have arisen due to fast evaporation of water and degradation of the matrix polymer [6, 7]. However, if the nanofibers could be used in a suspension with high water content in the composite manufacturing using a polymer with similar surface characteristics, better dispersion could be possible. Even more cost-effective bionanocomposites could be manufactured if the shear forces in a twin-screw extruder could be used for nano/microfibrillation of cellulose in the same step as the compounding of the biocomposite, as the isolation of nanocellulose is a highly energy-intensive process.

This chapter gives an overview of wood-plastic composites and cellulose nanocomposites. Also, the separation processes typically used to isolate wood fibers and cellulose nanofibers from wood raw material, as well as the structure of wood, are briefly discussed.

1.2 Wood-plastic composites

Wood-plastic composites (WPCs) are materials which combine the properties of wood and matrix polymer. In general, wood-plastic composite materials contain varying contents of wood, plastics and additives and their processing is done using the same techniques as processing of neat thermoplastic polymers. The use of wood-plastic composites in different applications has been steadily growing since the 1990s. So far, the major markets for wood-plastic composites have been North America and Asia, although the use of natural fiber and wood reinforced polymer composites is also on the increase in Europe. Currently, the main use of WPCs is in building and construction to replace impregnated wood in outdoor applications such as decking, railings, and window and door frames. Other uses of WPCs are, e.g. automotive interior panels and interior materials such as flooring and furniture. Examples of some WPC products are shown in Fig. 1.

3

Fig. 1 Examples of wood-plastic composite products [8].

The use of natural cellulose materials such as wood fibers and wood flour as reinforcement in thermoplastics has many advantages over the inorganic and synthetic fillers traditionally used in polymer composites. Natural fibers are light and inexpensive, they are abundant and they have high specific stiffness and strength. The environmental aspects are also important; natural fibers are renewable, recyclable and biodegradable materials. However, there are also problems related to the use of natural fibers in thermoplastic composites. Major drawbacks are the poor interfacial adhesion between the hydrophobic matrix and the hydrophilic fibersand the difficult dispersion of fibers in the matrix. Because of the hydrophilic nature of natural fibers, they need to be dried before the compounding step. In addition, the processing temperatures are limited due to the thermal degradation of the wood fibers in temperatures above 200°C, hence limiting the range of suitable matrix polymers. [9]

1.2.1 Raw materials

The polymer matrix forms a continuous phase which surrounds the wood component in WPCs. Because of the low thermal stability of wood, only polymers with processing temperatures lower than 200°C are typically used in WPCs. The polymers used are mostly low-cost commodity thermoplastics which flow easily when melted, and the most common polymers used in WPCs are polyethylene (PP), polypropylene (PP) and polyvinylchloride (PVC). [10]

The wood raw material used in wood-plastic composites is either in the form of wood flour or wood fibers. The term fiber refers to a single spindle-shaped wood cell, even though the term is sometimes used inaccurately to describe wood particles. Individual fibers can be separated

4

from wood using various chemical or mechanical methods familiar from papermaking processes. Fibers have typically high strength and quite high aspect ratios (length-to-diameter ratio, l/d), resulting in good reinforcing potential. However, due to the higher cost and difficulties in processing, particularly in feeding and metering, wood fibers are not used to the same extent as wood flour in commercial manufacture of wood-plastic composites. Wood flour is the most commonly used material in manufacturing of wood-plastic composites; it is less expensive and feeding and metering wood flour is easier compared to wood fibers. Wood flour consists of small wood particles composed of fiber bundles and the aspect ratio of the particles is low, typically between 1 and 5. Therefore, the reinforcing capability of wood flour is not as good as that of fibers. Wood flour is provided in various particle size ranges; however, typically the particle size of wood flour is within 180-840 m. Wood flour can be obtained directly from the forest product companies or from companies specialized in producing wood flour [11]. Wood flour is mainly produced from the residues of various wood processors, and the typical processing steps are: 1) size reduction using different types ofmills, 2) size classification by screening and 3) drying. [10]

In addition to wood and matrix polymer, wood-plastic composites contain small amounts of additives. These additives affect the processing and performance of the material. The most common additives are coupling agents, light stabilizers, pigments, lubricants, fungicides and foaming agents. Coupling agents (or compatibilizers) are especially important in wood-plastic composites. They are used to improve the interfacial adhesion between the hydrophobic matrix and the hydrophilic wood. Maleic anhydride grafted polypropylene, MAPP, is one of the most commonly used coupling agents in WPCs. MAPP is believed to enhance the interfacial adhesion via two mechanisms: 1) anhydride forms an ester bond with a wood cell wall hydroxyl group, and 2) PP segment attached to the anhydride entangles with the PP or PE network in the melt. [11]

1.2.2 Processing

Processing of wood-plastic composites typically consists of three steps: 1) wood raw material processing, 2) compounding and 3) molding of the products by injection molding, profile extrusion or compression molding. The purpose of the wood processing step is to increase the value and quality of the wood raw material by separating it into different sizes and species and drying it before the compounding step. There are also wood fiber companies which provide pre-processed wood fibers for many applications. In the compounding step the actual composite is manufactured by blending the reinforcement with the matrix polymer. Typically, a twin-screw extruder is used in the compounding. The compounded material can be immediately pressed or shaped into an end product or formed into pellets for further processing. The products can be manufactured using sheet or profile extrusion, injection molding, calendering, thermoforming or compression molding. Most WPCs are made by profile extrusion, in which the molten composite is forced through a die of specific shape to produce a continuous profile. These kinds of products are mainly used in building materials,

5

such as decking boards. When more complex shapes are wanted, other methods such as injection or compression molding can be used. [9, 11]

A variety of extruder types and processing methods are used in the production of WPC. For example, some processors use single-screw extruders to produce the end product from compounded pellets, whereas some compound and shape the final products in one step using twin-screw extruders. This kind of processing is also called direct extrusion. In direct extrusion either profiles or sheet materials for compression molding can be manufactured. A schematic diagram of a direct extrusion line for profile manufacture is shown in Fig. 2. [9, 11]

Fig. 2 In-line compounding and profile extrusion of WPCs [12].

1.2.3 Properties

Because of the wide variety of wood-plastic composites, the properties and performance of the composites vary considerably. The individual properties of wood and matrix polymer determine the properties of a composite. The stiffness is especially influenced by the form, size, dispersion and content of the wood in the composite. The interfacial adhesion between the wood and matrix determines the strength, moisture uptake and long-term properties ofwood-plastic composites. [9]

In general, it can be said that addition of a wood component enhances the stiffness of the polymer matrix. However, at the same time, the material becomes more brittle. When compared to solid wood, WPCs are typically less stiff. When fibers are used instead of wood flour as reinforcing agent, the mechanical strength, elongation and the unnotched impact strength properties of the composites usually increase. [13]

6

The fungal resistance and dimensional stability of wood-plastic composites is better than that of solid wood because they absorb moisture more slowly due to the thermoplastic matrix. However, the mechanical properties such as creep resistance, stiffness and strength are typically lower than those of solid wood. Therefore, wood-plastic composites have been used mainly as non-structural members in building applications. However, there is a need for WPCs suitable for more demanding structural applications, and a lot of research is being done to improve the mechanical performance, durability and service life of wood-plastic composites. For example, WPCs for structural applications such as the sill plate for housing,bridges and waterfront applications are under development. [13, 14]

1.3 Cellulose nanocomposites

In recent years there has been a remarkable growth in interest in the use of nanocellulose as polymer reinforcement in order to create high-performance biomaterials. The main reason for the appeal of nanosized cellulose is that by reducing the size of the cellulose fiber, material with higher uniformity and less defects and, thereby, improved mechanical properties, can be obtained [15]. Nanocellulose also has the advantage of being a sustainable material due to its biodegradability and renewability. When compared to conventional composite materials, nanocomposites are expected to have improved mechanical, thermal and barrier properties [16, 17]. There are a wide variety of possible applications for cellulose nanocomposites,varying from replacement of synthetic materials with more environmentally friendly materialsto creating completely new types of biomaterials. Currently, cellulose nanocomposites are being considered for use in, e.g. packaging, medical, automotive, electronics, construction and water treatment applications [5].

1.3.1 Raw materials

In cellulose nanocomposites the reinforcement can be either in the form of cellulose nanofibers or cellulose nanocrystals. Cellulose nanofibers (also known as nanofibrillated cellulose, microfibrillated cellulose, cellulose nanofibrils, etc.) are prepared by mechanical and/or chemical or enzymatic treatments, and they have a fibrous structure containing the amorphous and crystalline structure of cellulose. The thickness/width of the cellulose nanofibers is typically in the range of 20-40 nm, as they consist of cellulose microfibrilaggregates [16]. The length of cellulose nanofibers is estimated to be several micrometersand, therefore, they have high aspect ratio (l/d) [17, 18]. The rod-like cellulose nanocrystals(a.k.a. cellulose nanowhiskers, nanocrystalline cellulose, cellulose nanorods, etc.) consist of crystalline cellulose, and they are prepared by acid hydrolysis, whereby the amorphous partsof cellulose are removed. The width of cellulose nanocrystals is usually between 5 to 70 nm, and the length is from 100 nm to 250 nm when plant cellulose is used as the source material[18].

7

A wide range of polymers have been used as the matrix in cellulose nanocomposites,including both biodegradable and non-biodegradable thermoset and thermoplastic resins.Particularly, use of nanocellulose as biopolymer reinforcement has gained considerable interest as efforts have been made to create more environmentally sustainable high performance materials. The inherent properties of cellulose, biodegradability and hydrophilicity, explain why, in many studies, matrices with similar properties have been used, such as starch [19], chitosan [20], polyvinyl alcohol (PVOH/PVA) [21, 22] and polyethylene oxide (PEO) [23]. Also, non-biodegradable thermoset resins such as phenol formaldehyde[24-26], melamine formaldehyde [27], epoxy [26, 28] and acrylic resins [26] have been used to prepare cellulose nanocomposites using the resin impregnation method. Hydrophobic thermoplastic polymers, such as polypropylene (PP) [29], polyethylene (PE) [29], polylactid acid (PLA) [30], cellulose acetate butyrate (CAB) [31, 32] and polycaprolactone (PCL) [33],have been used as matrices in cellulose nanocomposites using, e.g. solvent exchange or chemical modification to ensure better dispersion of the hydrophilic nanocellulose in the hydrophobic matrix.

1.3.2 Manufacturing

Despite the great prospects of cellulose nanocomposites, there are still challenges related to the preparation of these materials limiting their use in commercial applications. Development of suitable processing methods is one of the major challenges with respect to the energy-efficient separation of nanocellulose and the compatibilization of reinforcements with the matrix polymer [5]. The main concern in the processing of cellulose nanocomposites is to obtain a homogeneous dispersion of nanoparticles in the polymer matrix. Due to the hydrophilic nature of cellulose, drying of cellulose leads to formation of irreversible aggregates and aggregation in most non-polar matrices as a result of hydrogen bond formationbetween cellulose nanoparticles [6, 34, 35]. For this reason, the highest level of dispersion of nanocellulose is expected to be achieved when water is used as the processing medium and,therefore, polymers which can be processed in aqueous solutions are often preferred. Solvent exchange methods have been used to prepare cellulose nanocomposites with hydrophobic polymers to change the water from nanocellulose suspension to a less polar solvent. Also surfactants or chemical modifications of the nanoparticle surfaces have been studied to reduce the surface energy of the nanoparticles and thus improve the dispersion and compatibility with a hydrophobic material. [15]

So far, the most common and successful methods used in the preparation of cellulose nanocomposites have been solution casting of a dilute suspension of the nanocellulose and matrix [22, 36-38], and resin impregnation of dried nanocellulose network [24-26,39]. In the solvent casting method the matrix polymer and the nanocellulose must be dispersed in the same solvent, and water soluble polymers are favored when using this method. In the resin impregnation method the used polymer must be in dissolved state or have a low viscosity to ensure good impregnation. The advantage of both solution casting and resin impregnation

8

method is that no specific equipment is needed for the composite preparation, but for industrial-scale manufacturing these methods are rather inefficient, and the shape of the prepared composites is limited to sheet-like structures. Extrusion compounding would be asuitable method for larger scale processing; however, extrusion processing of cellulose nanocomposites has been studied by only a few authors, especially by Oksman and co-workers [6, 7, 40, 41]. Obtaining good dispersion of the nanoparticles in the polymer matrix without degrading the matrix or the reinforcement has been one of the key problems in melt compounding of cellulose nanocomposites. [5, 6]

1.3.3 Properties

The enhancement of the properties of nanocomposites arises from the nanosize of the reinforcement (at least one dimension <100 nm), resulting in a large interfacial area between the reinforcement and matrix if good dispersion is achieved. In general, the large interface is expected to affect the molecular mobility, relaxation behavior and the consequent thermal and mechanical properties of the nanocomposite. [42]

Especially in the case of cellulose nanocomposites, the mechanical properties are expected to be higher than for composites with micro- or macrosized reinforcements [16]. One reason for this is the theoretical value of axial Young’s modulus of crystalline cellulose, 167 GPa [43], which is higher than steel and similar to Kevlar, although experimentally determined values for Young’s modulus of cellulose crystals have been somewhat lower, between 130 and 143 GPa [44-46]. Several researchers have also reported improvements in the strength and modulus of the polymer with addition of nanocellulose [19, 21, 24, 25, 27]. Other improved properties of cellulose nanocomposites with respect to the matrix polymer include better thermal properties [33, 38, 47-49], improved coefficient of thermal expansion (CTE) [26, 39, 50], decreased moisture sensitivity [51-53] and improved barrier properties [54, 55]. Due to the small size of the reinforcement, highly transparent composites can also be prepared when well-dispersed nanocellulose is used as polymer reinforcement [26, 39, 50].

1.4 Wood as raw material

Wood has had a great importance as a building material for homes, furniture, tools and vehicles throughout the history of humankind. Today, wood is used as raw material for a great variety of products including construction timber, paper products and sports equipment. Wood can be considered nature’s composite material, since it consists of a cellular structure of cellulose in a matrix of lignin, hemicelluloses and extractives. There are at least 60,000 different wood species in the world [56] and, depending on the species and growth conditions, variability can be found in the characteristics and quantity of wood constituents and cell structure. Wood species are divided into two main classes; namely, softwoods and

9

hardwoods. In botanical terms, softwoods are gymnosperms and mostly conifers, such as pine and spruce. Hardwoods are angiosperms, usually broadleaf, deciduous trees such as birch and maple. The cell structures of hardwoods and softwoods differ somewhat. In general, hardwoods have more complex structure and a larger number of different kinds of cells than softwoods. The vertical structure of softwoods consists mainly of long and tapering cells called tracheids, which are considered fibers in softwood species. The average length of Scandinavian softwood tracheids is 2-4 mm and the average width in the tangential directionis 0.02-0.04 mm. [57-60]

1.4.1 Structure of wood cell wall

The structure of a wood cell wall is very complex, and it is composed of several layers. The typical cell wall structure of a softwood tracheid is shown in Fig. 3. The cell wall consists of middle lamella (ML), primary wall (P), and secondary wall (S) with three layers and lumen (L) (Fig. 3). These layers have different structures and chemical compositions, but the main difference is in their microfibril alignments. Microfibrils are cellulose molecules bundled together. [60, 61]

Fig. 3 Cell wall structure of softwood tracheid [62].

The middle lamella (ML) does not actually belong to the cell wall because it separates two adjacent fibers/cells from each other. The middle lamella has very high lignin content and it is amorphous. Primary wall (P) is the outermost layer of the cell wall of wood fiber. It is a thin layer consisting mainly of lignin and pectins. The secondary wall (S) has three layers: S1, S2

and S3. The layers differ from each other on different microfibril alignments and compositions. S1 is a thin and lignin-rich outer layer, S2 is a thick middle layer and S3 is a thin inner layer. The S2 layer of softwood tracheid is technically the most valuable because of its high cellulose content. The inner layer of the secondary wall, S3, is also called the tertiary wall (T), and it consists mainly of hemicelluloses. In the middle of softwood tracheid there is a canal called lumen (L). [57, 60, 61]

10

The microfibrils in the wood cell wall consist of even smaller components, sometimes called elementary fibrils in the terms of plant physiology. Elementary fibrils are bundles of cellulose molecules, and their diameter is approximately 3.5 nm. Cellulose molecules can be arranged or non-arranged forming crystalline and non-crystalline amorphous regions within the fibrils. The crystalline regions are stronger and it is more difficult for solvents or reagents to penetrate into them. A schematic description of the structure of a wood cell wall is shown in Fig. 4. When the microfibrils or elementary fibrils are separated from the wood cell wall theyare called cellulose nanofibers. Basically, an elementary fibril can be considered a single cellulose nanofiber. [60, 61, 63, 64]

Fig. 4 Schematic description of the structure of a wood cell wall.

1.4.2 Chemical composition

The main components of wood are cellulose, hemicelluloses, lignin and extractives. The amounts of each of these components in softwoods and hardwoods are presented in Table 1. Cellulose is the most important substance in wood because it determines the character of the fiber. The chemical formula of cellulose is (C6H10O5)n, in which n is the number of repeating sugar units or the degree of polymerization, DP. The repeating unit in cellulose consists of two glucose molecules, also called the cellobiose unit. The DP varies depending on the cellulose source and the treatments it has received. For native cellulose the DP is around 3500, and for most papermaking fibers the DP is between 600 and 1500. The properties of cellulose-containing materials depend strongly on the DP of cellulose molecules. [60]Hemicelluloses are branched heterogeneous polysaccharides consisting of five different sugars: glucose, mannose, galactose, xylose and arabinose. Hemicelluloses have lower molecular weight than cellulose, and they have lower DP [57]. Hemicelluloses degrade and dissolve more easily than cellulose, and their amount in pulp is always less than in the

11

original wood. Besides cellulose and hemicelluloses, wood contains lignin and extractives. Lignin is an amorphous and highly polymerized substance with a very complex structure. The primary purpose of lignin is to form the middle lamella and glue the fibers together. In addition, different kinds of substances, e.g. resin acids, fatty acids, turpenoid compounds and alcohols, are present in wood. These substances are typically called extractives, contributing to wood properties such as color, odor, taste, decay resistance, density, hygroscopicity and flammability. The name “extractives” comes from the fact that these substances can be removed from the wood by extraction with solvents. [60, 65]

Table 1 Chemical components of wood [60].Component Softwood Hardwood

Cellulose 42 ± 2% 45 ± 2 %

Hemicelluloses 27 ± 2 % 30 ± 5 %

Lignin 28 ± 2 % 20 ± 4 %

Extractives 3± 2 % 5± 3 %

1.4.3 Isolation processes of fibers and nanofibers

1.4.3.1 Wood fibers

In the papermaking industry, pulping methods are used to separate individual fibers from wood. The pulping processes can be mechanical, chemical or combinations of chemical and mechanical treatments. The principle in mechanical pulping is to bring wood raw material into a cyclic oscillating stress field to break down the raw material and to separate individual fibers. The main mechanical pulping processes are the stone groundwood process (GW), thepressure groundwood process (PGW), thermo-mechanical pulping (TMP) and chemi-thermo-mechanical pulping (CTMP). In the groundwood process pulp is produced by pressing wood blocks against a rotating stone surface. In the TMP and CTMP process the defibration occurs in a refiner consisting of two parallel discs separated by a narrow gap. The surfaces of the discs are grooved, and either one or both of the discs rotate (opposite directions) during the refining. The wood mass is gradually broken down into smaller particles as it moves from middle of the refiner disc to towards the edges. [60, 66]

When producing fibers mechanically from wood it is very important that the polymeric wood material is softened before mechanical treatment. Because wood is a polymer composite with viscoelastic behavior its response to mechanical treatment depends on the temperature, moisture and time under load. Thus, the glass transition temperatures (Tg) of wood polymers (cellulose, hemicelluloses and lignin) are very important from a fiber separation point of view. Lignin, especially, has a major influence on the wood behavior in mechanical fiber separation.As Fig. 5 shows, moisture content has an effect on the softening temperatures of wood

12

polymers. In dry conditions the softening temperature (Tg) of lignin and hemicelluloses is between 180 and 220°C. In water-saturated conditions the softening temperature (Tg) of hemicelluloses and amorphous cellulose is close to 20°C, and for water-saturated isolated lignin the Tg is between 80°C and 90°C. The softening temperature of lignin as a part of woodmatrix is typically from 100°C to 130°C. Therefore, lignin softening is the critical factor in mechanical separation of wood fibers. If lignin remains stiff, the fracture between the fibers occurs in an uncontrollable way, and more broken fibers are formed. For practicality, the wood for mechanical fiber separation processes should have sufficient moisture content, and the temperature should be above the lignin glass transition temperature. Chemical treatments can also be used to soften the wood and to reduce the energy needed in fiber separation. Due to the viscoelastic nature of wood, increasing the frequency of the mechanical action also increases the softening temperature of wood polymers. [66]

Fig. 5 Influence of moisture content on the softening temperature of wood [67].

When undamaged fibers are wanted, chemical pulping processes are used. The aim in chemical pulping is to remove lignin and maintain cellulose and hemicelluloses in the fibers. However, in practice, some of the hemicelluloses and celluloses are also dissolved, and thus the yield of chemical pulping is lower when compared to mechanical pulping. The yield is usually between 40% and 50% of the original raw material. The main methods used in chemical pulping are the kraft (alkaline) process and the sulphite (acidic) process, the kraft process being more popular due to the recovery of chemicals and better pulp strength properties. [60]

13

1.4.3.2 Cellulose nanofibers

The separation process of cellulose nanofibers from wood raw material is usually a combination of mechanical and chemical (or enzymatic) treatments. The isolation of cellulose nanofibers from the original cellulose fiber, also known as the nanofibrillation step, is usually done using mechanical treatment. Prior to mechanical treatment, chemical treatments are used to remove lignin and other substances besides cellulose from the fibers. If the raw material is already in the form of purified cellulose (e.g. kraft pulp), further chemical pre-treatments are not necessary.

The methods most commonly used in the nanofibrillation of cellulose are ultra-fine grindingand high-pressure homogenization. In the ultra-fine grinding, refiner-type equipment is used in which the dilute fiber suspension is forced through a narrow gap between a rotary and a stator disk. The grinder disks are made of ceramic materials, and they have a special surface pattern which makes the fibers disintegrate due to repeated cyclic stresses. The disks can be produced with different grit classes and groove configurations to produce very fine particles, and by adjusting the clearance between the disks, ultra-fine grinding conditions can be obtained. High-pressure homogenization was the method which was used in the first publications about separation of cellulose nanofibers from wood pulp in 1983 [68, 69]. In this method, the raw material is defibrillated into nanofibers using rapid pressure drops and high shear and impact forces against a valve and an impact ring [15]. In another type of homogenization method, microfluidization, the raw material suspension is exposed to high shearing stresses in the interaction chambers of the equipment, which can have either y- or z-type geometry [70]. Due to high velocity created by pressure, the fibers are disintegrated by shear forces and impact against the chamber walls and colliding streams [16, 71].

Often, several passes and a combination of different methods are needed to obtain cellulose nanofibers with uniform size distribution. Therefore, the major problem in the mechanical disintegration of cellulose nanofibers is the high energy consumption. For this reason, many researchers are currently studying ways to reduce the amount of energy used in the nanofibrillation of cellulose. For example, combinations of enzymatic and/or chemical treatments with mechanical treatments have been shown to be useful ways to reduce energy use in the nanofibrillation process [72, 73]. However, it should be noted that the use ofchemicals and/or enzymes also increases the cost of cellulose nanofibers, and the treatment used may have a negative effect on the properties of nanocellulose.

14

1.5 Objectives

The objective of this work has been to study defibration/fibrillation of wood-based raw materials and processing of wood-based biocomposites using a twin-screw extruder.

In the first part of this work (Papers I, II and III) the aim was to study the use of wood chips as raw material in wood-plastic composites and the separation of individual fibers with higher aspect ratios from the wood chips in order to enhance the strength properties of the composites during the composite compounding process in a twin-screw extruder.

The aim of the second part of this work (Papers IV and V) was to study the extrusion processing of green cellulose nanocomposites using thermoplastic starch as the matrix and cellulose nanofibers with high water content as the reinforcement. In order to reduce the cost of produced nanocomposites, the use of twin-screw extrusion to isolate micro/nanofibers fromof cellulose fibers during the extrusion compounding of TPS/CNF composites was alsostudied.

15

2 Experimental procedure

2.1 Materials

In the first part of the study, the used wood raw materials were spruce pin chips and pre-crushed spruce chips (Picea abies). Polypropylene (PP), in a powder form, was used as matrixpolymer and was supplied by Borealis Polyolefins, Austria. The melt flow index (MFI) of the used PP was measured to 16 g/10 min (200°C/2.16 kg). Maleic anhydride grafted polypropylene Epolene E-43 (MAPP) supplied by Eastman Chemical Company (USA) was used as coupling agent to improve the interaction between the wood and PP. A lubricant (TPW 113, supplied by Struktol, USA) was used as processing aid and an antioxidant agentreceived from Borealis Polyolefins (Austria) was used as stabilizer. Chemical pre-treatment of the wood chips was done using sodium sulfite (Na2SO3, Merck KGaA, Germany).

In the second part of the study, softwood wood flour supplied by Scandinavian Wood Fiber AB, (Orsa, Sweden) with a particle size range of 200-400 µm was used as a starting material for preparation of cellulose nanofibers and bleached cellulose fibers. Never-dried bleached hardwood kraft pulp (Betula pendula) was used to prepare the TEMPO-oxidized cellulose fibers. The matrix polymer was potato starch, ELIANETM 100 supplied by AVEBE, Netherlands, with high amylopectin content (>99%). D-Sorbitol (Merck, Germany) was used as plasticizer for starch, and stearic acid (VWR, Leuven, Belgium) was used as a lubricant.

2.1.1 Raw material pre-treatments

Sulfonation pre-treatment

To soften the wood raw materials (spruce chips and pre-crushed spruce chips) chemically, sodium sulfite (Na2SO3) was used in the chemical pretreatment. The wood raw material was diluted to 10% consistency and 5% sodium sulphite (based on the total mass of the mixture) was used in the treatment. The pre-treatment was done at 90°C and at pH 9 for 210 minutes.After the pre-treatment, wood chips were washed with water.

Preparation of cellulose nanofibers

The softwood flour was delignified using acidified sodium chlorite solution using the Jayme-Wise method following the procedures of Abe et al. 2007 [74], Iwamoto et al. 2008 [75] and Gong et al. 2011 [76]. 250 g of wood flour (dry mass) was treated in a flask containing 3500 mL deionized water with 5 mL acetic acid and 33.5 g of sodium chlorite at 70-75°C. Additions of acetic acid and sodium chlorite were continued at 2-hour intervals until the wood became white (11 additions). The wood was washed twice during the treatment and at the end

16

with a minimum of 60L of deionized water. Before washing, the wood was left in the acidified chlorite solution for 12h at 70-75°C. The total treatment time of the wood flour in acidified chlorite solution was 58h. To obtain cellulose nanofibers (CNF), a 3 wt% suspension was prepared from the delignified wood flour and deionized water, dispersed homogenously using a laboratory shear mixer (Silverson L4RT, Silverson Machine Ltd., England), and passed twice through an ultrafine grinder MKCA6-3 (Masuko Sangyo Co, Ltd., Japan), at a rotational speed of 1440 rpm. The total grinding time was 16 minutes. After grinding, the dry matter content of the CNF suspension was increased by centrifuging.

Preparation of the TEMPO-oxidized cellulose

The TEMPO-mediated oxidation was performed under alkaline conditions according to the procedure described by Saito et al. 2006 [72] and 2007 [77]. A 50 g cellulose sample was diluted to 1% consistency and 0.1 mmol/g of TEMPO and 1 mmol/g of NaBr were added to the suspension. The reaction was started by adding 10% NaClO solution, and the pH was adjusted to 10-10.5 and maintained at this level with 0.5 M NaOH during the oxidation. The total amount of NaClO used was 10 mmol/g. The reaction was finished after 5.5 hours and the oxidized sample was washed with deionized water until the conductance of the filtrate reached a value under 10 µC/cm. The carboxyl content of the oxidized samples was determined using conductometric titration in duplicate [78, 79].

2.2 Extrusion

A co-rotating twin-screw extruder with twin-screw side feeder, Coperion W&P ZSK-18 MEGALab (Stuttgart, Germany) equipped with K-Tron gravimetric feeders (Niederlenz,Switzerland), two atmospheric vents and a vacuum vent was used in the experiments (Fig. 6).The screw shafts (Fig. 7) in the Coperion W&P ZSK-18 MEGALab twin-screw extruder have an elemental structure and the screw configurations can therefore be varied.

17

Fig. 6 Co-rotating twin-screw extruder, Coperion W&P ZSK-18 MEGALab [80].

Fig. 7 The screw shafts of Coperion W&P ZSK-18 MEGALab co-rotating twin-screw extruder.

2.3 Characterization

2.3.1 Fractionation

The fractionation method was used in papers I, II and III to divide the wood particle samples into different size categories and to obtain images of the samples for the image analysis. The fractionation was done either by using a tube flow fractionator (Metso Automation, Kajaani, Finland) or a Bauer McNett classifier. Tube flow fractionation is a method in which the sample (in water suspension) is injected into a continuous plug flow of water, which then enters a long plastic tube. As the sample flows in the tube it is fractionated by size. The fractionation device is equipped with a CCD camera, which records approximately 600 images of the sample during the analysis. The basic principle of the device is presented in Fig. 8. Tube flow fractionation is typically used in the analysis of wood pulp samples for papermaking purposes.

18

Fig. 8 Schematic description of the tube flow fractionator [81].

2.3.2 Image analysis

The length and width of the wood particles were measured using an optical fiber analyzer FiberLab (Metso Automation, Kajaani, Finland) and image analysis software kajaaniIMG (Metso Automation, Kajaani, Finland). The average length (length weighted) and averagewidth measurements were then used to calculate the aspect ratio of the wood particles and fibers. The images captured by the tube flow fractionator were used in the kajaaniIMG image analysis. In brief, the kajaaniIMG software uses several image analysis algorithms to measure the properties of the objects (particles, fines and fibers) in the captured images. A projection area is determined for each object and, based on the shape of the object, parameters such as length and width are decided.

2.3.3 Microscopy

Optical microscopy, scanning electron microscopy (SEM, FESEM) and transmission electron(TEM) microscopy were used to study the morphology and microstructure of the wood particles, fibers and nanofibers as well as the fracture surfaces of the composites. Optical microscopy was also used to visually examine the dispersion and size of cellulose fibers in the thermoplastic starch films (paper V). The fracture surfaces were created by first freezing the specimen under liquid nitrogen and then breaking it. A SEM image of a fracture surface is shown in Fig. 9. The wood particle samples were freeze-dried before microscopy. The samples for scanning electron microscopy were sputter-coated with platinum or gold before observation.

19

Fig. 9 Fracture surface of a wood chip-polypropylene composite.

2.3.4 Mechanical testing

Mechanical properties of the composites were measured using flexural (papers II-III), impact(paper III) and tensile testing (papers IV and V). Flexural testing was performed according to ASTM D790 standard and with a Shimadzu Autograph AG-X universal testing machine(Shimadzu Corp, Kyoto, Japan). A falling weight impact test was done with a Dynatup Minitower (Instron, UK) falling weight impact testing machine and according to ASTM D3763 standard. Tensile testing was done using an Instron 4411 tensile testing machine. In Fig. 10 a typical stress-strain curve obtained from tensile testing is shown.

Fig. 10 Stress-strain curve for thermoplastic starch reinforced with 10 wt% of cellulose nanofibers.

20

2.3.5 Density

The density of the wood-plastic composites was measured using either gas displacement technique (AccuPyc 1330, Micromeritics Instrument Corp, USA, paper II) or the water displacement method according to ASTM D792 standard (test method A, paper III).

2.3.6 Transparency

The light transmittance (T %) of the composite films (papers IV and V) was measured using aUV/Vis spectrometer Lambda 2S (Perkin Elmer, Überlingen, Germany) in a light wavelength area from 300 to 1000 nm. A typical light transmittance curve obtained from UV/Vis spectroscopy is shown in Fig. 11.

Fig. 11 Light transmittance of thermoplastic starch composite film reinforced with 10wt% of cellulose nanofibers.

2.3.7 Moisture absorption

The moisture absorption of the composite films (Papers IV and V) was studied according to the ASTM D 5229 standard by exposing the samples to 98% relative humidity and determining the water uptake as a function of time until the moisture equilibrium content was reached.

21

3 Summary of appended papers

Paper I: The use of twin-screw extrusion in processing of wood: The effect of processing parameters and pretreatment

In Paper I the effect of extrusion parameters on different types of wood raw materials was studied. The aim was to produce wood particles with as high aspect ratio as possible. The study consisted of two parts: 1) the breaking and separation of individual fibers from woodchips during the extrusion process, and 2) the effect of chemical pre-treatment and reversed screw elements (RSE) on pre-crushed wood raw material. This study showed that a twin-screw extruder can be used to separate individual fibers from wood chips, and the separated fibers have higher aspect ratios than the wood flour particles typically used in wood-plasticcomposites. When more fibrous and chemically softened wood raw material was used, fiberswith even higher aspect ratios were obtained. According to the statistical analysis, the screw speed was the main factor affecting wood fiber aspect ratio in twin-screw extrusion of wood chips. It was also observed that the screw configuration can affect the fiber properties and the reverse screw elements reduced the shive content in the samples.

Paper II: Processing of Wood Chip-Plastic Composites: Effect on Wood Particle Size, Microstructure and Mechanical Properties

In Paper II wood chip-polypropylene composites were manufactured using twin-screw extrusion. The aim was to study the effect of the extrusion process on wood chip aspect ratio and to measure the mechanical properties of the prepared composite materials. Two different compounding methods were used in manufacturing the composites, and dried and undried wood chips were used as wood raw materials. The results showed that both the processing method and the wood chips’ moisture content had an effect on the wood particle size and aspect ratio. Individual fibers and wood particles with high aspect ratio could be separated from the wood particles when wet wood chips were used as raw material. When dried wood chips were used, the particle size was significantly reduced, but wood remained in particulate form. The composites manufactured with dry wood chips had slightly better flexural properties compared to composites manufactured with wet wood chips, despite the fact that the aspect ratio of wood particles in wet wood chip composites was higher. In general, this study showed that it is possible to use undried wood chips as raw material for WPCs, and thus reduce the cost of wood raw material.

22

Paper III: The effect of pre-softened wood chips on wood fibre aspect ratio and mechanical properties of wood-polymer composites

In Paper III chemically pre-treated wood chips were used as raw material for wood-polypropylene composites. The aim was to study the effect of the pre-treatment and moisture content of wood chips on the wood particle aspect ratio after extrusion and the mechanical properties of the composites. The results showed that chemical pre-treatment with sodium sulfite improved the composites’ mechanical properties and that the wood particle aspect ratio was increased in the extrusion process when chemically pre-treated wood chips were used.Combination of pre-treated, undried wood chips resulted in the highest wood particle aspect ratio and the highest flexural strength and impact properties of the composites. As in Paper II, the flexural modulus of the composites decreased when undried wood chips were used as raw material. In general, aspect ratios were considerably higher when using undried wood chips as raw material instead of dried wood chips. Overall, the study shows that the defibration of wood chips in a twin-screw extruder can be enhanced by using a simple chemical pre-treatment, and that wood chips can be used as raw material in the production of wood-plastic composites.

Paper IV: Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured using twin-screw extrusion

In Paper IV the processing and properties of green nanocomposites of thermoplastic starch and cellulose nanofibers (CNF) were studied. The aim was to determine if CNF gels with high water content together with starch and plasticizer can be processed into nanocomposites using continuous twin-screw extrusion. Nanocomposites with 0, 5, 10, 15 and 20 wt% cellulosenanofiber content were successfully compounded using extrusion, and the properties of the composites were measured. The microscopy showed that the CNF aggregated during the processing, indicating that the screw configuration should be more distributive and dispersive to obtain a more homogeneous material. Despite the aggregation of CNF, the tensile strength and modulus of the starch matrix were improved, and the moisture sensitivity was reduced with the addition of cellulose nanofibers. Also, the translucency of the films was maintained,even with 20 wt% of CNF content. The tensile modulus increased linearly with the increasing nanofiber content, but the strength properties were highest for the material with 10 wt% of cellulose nanofibers.

23

Paper V: One-step extrusion process of green biocomposites: compounding, fibrillation and dispersion

In Paper V the aim was to study if the twin-screw extrusion process can be used to fibrillate cellulose fibers into micro- or nanosize fibers and at the same time prepare green bionanocomposites of thermoplastic starch with 10 wt% fiber content. The effect of the processing set-up on micro/nanofibrillation and fiber dispersion and distribution in starch was studied. Two different types of cellulose fibers were used in the composite preparation;namely, bleached wood fibers and TEMPO-oxidized cellulose fibers, in order to examine the effect of fiber treatment on the fibrillation efficiency. A composite with cellulose nanofibers was also prepared to examine the nanofiber distribution and dispersion in the composite as well as to compare the properties with the composites containing bleached fibers and TEMPO-oxidized fibers. The study showed that the shear forces in a twin-screw extruder were not enough to nanofibrillate cellulose fibers, though the size of the TEMPO-oxidized fibers was affected by the extrusion process more than the bleached wood fibers. Overall, the dispersion of the fibers in the thermoplastic starch matrix was good. The transparency of the composites with bleached wood fibers and TEMPO-oxidized fibers was lower than that of the composite reinforced with cellulose nanofibers. The mechanical properties of thermoplastic starch were improved by the addition of cellulose fibers, as expected. However, there were no significant differences in the mechanical properties and moisture uptake of the composite materials reinforced with cellulose nanofibers, bleached wood fibers or TEMPO-oxidized fibers.

24

4 Conclusions

This study shows that reduction of wood raw material particle size during the extrusion compounding of biocomposite materials is possible when using suitable processing conditions and, therefore, it is possible to use cheaper raw materials as reinforcement in the production of wood-based biocomposites.

In the first part of the work, wood-plastic composites (WPCs) were prepared using wood chips as wood raw material. The results show that it is possible to obtain wood particles with higher aspect ratios than the wood flour particles typically used in wood-plastic compositesby separating wood particles and fibers from wood chips in the twin-screw extrusion process. When undried wood chips are used as raw material, individual fibers and wood particles with high aspect ratio can be separated, thereby obtaining composites with similar stiffness values as with WPCs made using wood flour and thereby decreasing the raw material cost for the final product. If the bound moisture from the undried wood chips could be removed during the process, it might be possible to improve the composites’ mechanical properties even further.

In the second part of the work processing of green cellulose nanocomposites using twin-screw extrusion was studied. The results show that extrusion processing of cellulose nanocomposites using cellulose nanofibers (CNF) with high water content as the reinforcement is possible, and that the preparation of thermoplastic starch (TPS) and compounding of the TPS/CNF composites can be done in one extrusion step. Study of the effect of twin-screw extrusion on separating even smaller components from wood pulp, micro- or nanofibers, during the composite compounding step showed that the amount of shear forces generated was not enough to obtain micro- or nanofibers from cellulose fibers. Study showed, however, that good dispersion and distribution of the fibers could be obtained in twin-screw extrusion under suitable processing conditions.

5 Future work

As environmental awareness is increasing and more environmentally friendly materials are developed to replace the non-sustainable materials, more focus should be placed on the processing of these materials using energy-efficient methods. In general, it can be said thatmore studies about extrusion processing of biocomposite materials are needed to increase the scientific knowledge concerning these materials and their commercial potential.

Regarding the first part of this work, more efficient removal of moisture from the undried raw material during extrusion would be a significant improvement. As proper softening of the wood raw material is essential for separation of wood particles and fibers with a higher aspect ratio, pre-softening of wood raw material to greater extent would be interesting. In the case of cellulose nanocomposites, more effort should be devoted to the processing of composites

25

using industrially viable ways. Finding a way to obtain good dispersion of nanocellulose in the matrix polymer in extrusion compounding of nanocomposites would be especially important. Using suitable chemical pre-treatments under the appropriate processing conditions, nanofibrillation of cellulose could also be possible, and this would be an extremely interesting subject for future work.

26

6 Acknowledgements

This work was carried out as part of a joint project between Luleå University of Technology, Sweden and the University of Oulu, Finland. Financial support from the PhD Polis (an LTU program to enhance collaboration with the University of Oulu) and the European CommissionWOODY project (contract number NMP2-LA-2008-210037) is gratefully acknowledged.

I wish to thank both of my supervisors, Prof. Kristiina Oksman and Prof. Jouko Niinimäki, formaking this work possible and also for their ideas, guidance and support during this work. Especially, I want to thank Kristiina for her enthusiastic attitude towards my work. Mysincere gratitude also goes to Prof. Aji Mathew for her help and advices. I also wish to thank all of my wonderful colleagues in Luleå and Oulu, who have been a great help and companyduring these years.

In addition, I also wish to thank Mr. Runar Långström from Swerea SICOMP for his help in specimen preparation, Mr. Jani Österlund and Mr. Jarno Karvonen from the University of Oulu for their valuable help with fiber analyses, Mr. Johnny Grahn for helping with SEM andDr. Mehdi Jonoobi for his help with TEM. My thanks go also to great Master students, Erik Samuelsson and Pierre Rollo, for doing a good job in their projects.

To Martha, Natalia, Liva and Newsha - Thank you for all the talks we have had!

Finally, I wish to thank Jani and my parents and sisters for giving me strength, support and courage.

27

7 References

1. John, M. J., & Thomas, S. (2008). Biofibres and biocomposites. Carbohydrate Polymers, 71(3), 343-364.

2. Stark, N. M. (1999). Wood fiber derived from scrap pallets used in polypropylene composites. Forest Products Journal, 49(6), 39-46.

3. Stark, N. M., & Rowlands, R. E. (2003). Effects of wood fiber characteristics on mechanical properties of wood/polypropylene composites. Wood and Fiber Science, 35(2), 167-174.

4. Bierdel, M. (2008) Computational Fluid Dynamics. In: Kohlgrüber, K. & Wiedmann, W. Co-Rotating Twin-Screw Extruders. Fundamentals, Technology, and Applications. Carl Hanser Verlag, Munich, Germany. pp. 138-158.

5. Oksman, K., Mathew, A. P., & Sain, M. (2009). Novel bionanocomposites: Processing, properties and potential applications. Plastics, Rubber and Composites, 38(9/10), 396-404.

6. Oksman, K., Mathew, A. P., Bondeson, D., & Kvien, I. (2006). Manufacturing process of cellulose whiskers/polylactic acid nanocomposites. Composites Science and Technology, 66(15), 2776-2784.

7. Mathew, A. P., Chakraborty, A., Oksman, K., & Sain, M. (2006). The structure and mechanical properties of cellulose nanocomposites prepared by twin screw extrusion. In: Sain, M., & Oksman, K. (eds.), Cellulose nanocomposites: Processing, characterization and properties. Oxford, UK, Oxford University Press. pp. 114-131.

8. Wood-plastic profiles. Greiner Extrusionstechnik GmbH. [image online]. Available at:http://presse.fabrikderzukunft.at/_internal/cimg_ms!0/3u7ylu71mckoa3vy.jpg. [accessed 6th Dec 2012].

9. Oksman, K., & Bengtsson, M. (2007). Wood Fibre Thermoplastic Composites: Processing, Properties and Future Developments. In: Fakirov, S., & Bhattacharyya, C. (ed.), Handbook of engineering biopolymers: homopolymers, blends and composites. Munich, Germany, Carl Hanser Verlag. pp. 655-671.

10. Clemons, C. Raw materials for wood-polymer composites. (2008). In: Oksman Niska, K., & Sain, M. (ed.), Wood-polymer composites. Cambridge, England, Woodhead Publishing Limited. pp. 1-22.

11. Caulfield, D. F, Clemons, C., Jacobson, R. E., & Rowell, R. M. (2005). Wood Thermoplastic Composites. In: Rowell, R. M. (ed.), Handbook of Wood Chemistry and Wood Composites. Boca Raton, USA, CRC Press. pp. 365-378.

28

12. Schwendemann, D. (2008). Manufacturing technologies for wood-polymer composites. In: Oksman Niska, K., & Sain, M. (ed.), Wood-polymer composites. Cambridge, England, Woodhead Publishing Limited. pp. 72-100.

13. Clemons, C. (2002). Wood-Plastic Composites in the United States. Forest Products Journal 52(6), 10-18.

14. Tichy, R. J. (2008). Performance measurement and construction applications of wood-polymer composites. In: Oksman Niska, K., & Sain, M. (ed.), Wood-polymer composites. Cambridge, England, Woodhead Publishing Limited. pp. 257-272.

15. Spence, K., Habibi, Y., & Dufresne, A. (2011). Nanocellulose-based composites. In: Kalia, S., Kaith, B. S. & Kaur, I. (eds.), Cellulose fibers: Bio- and nano-polymer composites. Springer Berlin Heidelberg. pp. 179-213.

16. Siró, I., & Plackett, D. (2010). Microfibrillated cellulose and new nanocomposite materials: A review. Cellulose, 17(3), 459-494.

17. Svagan, A. J., Azizi Samir, M. A. S., & Berglund, L. (2007). Biomimetic polysaccharide nanocomposites of high cellulose content and high toughness. Biomacromolecules. 8(8), 2556-2563.

18. Klemm, D., Kramer, F., Moritz, S., Lindström, T., Ankerfors, M., Gray, D., & Dorris, A. (2011). Nanocelluloses: A new family of nature-based materials. Angewandte Chemie International Edition, 50(24), 5438-5466.

19. Dufresne, A., Dupeyre, D., & Vignon, M. R. (2000). Cellulose microfibrils from potato tuber cells: Processing and characterization of starch-cellulose microfibril composites. Journal of Applied Polymer Science, 76(14), 2080-2092.

20. Li, Q., Zhou, J., & Zhang, L. (2009). Structure and properties of the nanocomposite films of chitosan reinforced with cellulose whiskers. Journal of Polymer Science Part B: Polymer Physics, 47(11), 1069-1077.

21. Bhatnagar, A., & Sain, M. (2005). Processing of cellulose nanofiber-reinforced composites. Journal of Reinforced Plastics and Composites, 24(12), 1259-1268.

22. Kvien, I., & Oksman, K. (2007). Orientation of cellulose nanowhiskers in polyvinyl alcohol. Applied Physics A, 87(4), 641-643.

23. Brown, E. E., & Laborie, M. G. (2007). Bioengineering bacterial Cellulose/Poly(ethylene oxide) nanocomposites. Biomacromolecules, 8(10), 3074-3081.

29

24. Nakagaito, A. N., & Yano, H. (2004). The effect of morphological changes from pulp fiber towards nano-scale fibrillated cellulose on the mechanical properties of high-strength plant fiber based composites. Applied Physics A: Materials Science & Processing, 78(4), 547-552.

25. Nakagaito, A. N., & Yano, H. (2005). Novel high-strength biocomposites based on microfibrillated cellulose having nano-order-unit web-like network structure. Applied Physics A: Materials Science & Processing, 80(1), 155-159.

26. Yano, H., Sugiyama, J., Nakagaito, A. N., Nogi, M., Matsuura, T., Hikita, M., & Handa, K. (2005). Optically transparent composites reinforced with networks of bacterial nanofibers. Advanced Materials, 17(2), 153-155.

27. Henriksson, M., & Berglund, L. A. (2007). Structure and properties of cellulose nanocomposite films containing melamine formaldehyde. Journal of Applied Polymer Science, 106(4), 2817-2824.

28. Ruiz, M. M., Cavaillé, J. Y., Dufresne, A., Graillat, C., & Gérard, J. (2001). New waterborne epoxy coatings based on cellulose nanofillers. Macromolecular Symposia, 169(1), 211-222.

29. Wang, B., & Sain, M. (2007). Isolation of nanofibers from soybean source and their reinforcing capability on synthetic polymers. Composites Science and Technology, 67(11–12), 2521-2527.

30. Kvien, I., Tanem, B. S., & Oksman, K. (2005). Characterization of cellulose whiskers and their nanocomposites by atomic force and electron microscopy. Biomacromolecules, 6(6), 3160-3165.

31. Grunert, M., & Winter, W. (2002). Nanocomposites of cellulose acetate butyrate reinforced with cellulose nanocrystals. Journal of Polymers and the Environment, 10(1-2), 27-30.

32. Petersson, L., Mathew, A. P., & Oksman, K. (2009). Dispersion and properties ofcellulose nanowhiskers and layered silicates in cellulose acetate butyrate nanocomposites. Journal of Applied Polymer Science, 112(4), 2001-2009.

33. Siqueira, G., Bras, J., & Dufresne, A. (2009). Cellulose whiskers versus microfibrils: Influence of the nature of the nanoparticle and its surface functionalization on the thermal and mechanical properties of nanocomposites. Biomacromolecules, 10(2), 425-432.

34. Hult, E., Larsson, P. T., & Iversen, T. (2001). Cellulose fibril aggregation — an inherent property of kraft pulps. Polymer, 42(8), 3309-3314.

30

35. Eyholzer, C., Bordeanu, N., Lopez-Suevos, F., Rentsch, D., Zimmermann, T., & Oksman, K. (2010). Preparation and characterization of water-redispersible nanofibrillated cellulose in powder form. Cellulose, 17(1), 19-30.

36. Dufresne A, Vignon MR. (1998). Improvement of starch film performances using cellulose microfibrils. Macromolecules, 31(8):2693-2696.

37. Malainine ME, Mahrouz M, Dufresne A. (2005). Thermoplastic nanocomposites based on cellulose microfibrils from Opuntia ficus-indica parenchyma cell. Composites Science and Technology, 65(10):1520-1526.

38. Alemdar A, Sain M. (2008). Biocomposites from wheat straw nanofibers: Morphology, thermal and mechanical properties. Composites Science and Technology 68(2):557-565.

39. Iwamoto S, Nakagaito AN, Yano H, Nogi M. (2005). Optically transparent composites reinforced with plant fiber-based nanofibers. Applied Physics A - Materials Science & Processing, 81(6):1109-1112.

40. Bondeson, D., & Oksman, K. (2007). Polylactic acid/cellulose whisker nanocomposites modified by polyvinyl alcohol. Composites Part A: Applied Science and Manufacturing, 38(12), 2486-2492.

41. Jonoobi, M., Harun, J., Mathew, A. P., & Oksman, K. (2010). Mechanical properties of cellulose nanofiber (CNF) reinforced polylactic acid (PLA) prepared by twin screw extrusion. Composites Science and Technology, 70(12), 1742-1747.

42. Sorrentino, A., Gorrasi, G., & Vittoria, V. (2007). Potential perspectives of bio-nanocomposites for food packaging applications. Trends in Food Science & Technology, 18(2), 84-95.

43. Tashiro, K., & Kobayashi, M. (1991). Theoretical evaluation of three-dimensional elastic constants of native and regenerated celluloses: Role of hydrogen bonds. Polymer, 32(8), 1516-1526.

44. Sakurada, I., Nukushina, Y., & Ito, T. (1962). Experimental determination of the elastic modulus of crystalline regions in oriented polymers. Journal of Polymer Science, 57(165), 651-660.

45. Hepworth, D. G., & Bruce, D. M. (2000). A method of calculating the mechanical properties of nanoscopic plant cell wall components from tissue properties. Journal of Materials Science, 35(23), 5861-5865.

46. Šturcová, A., Davies, G. R., & Eichhorn, S. J. (2005). Elastic modulus and stress-transfer properties of tunicate cellulose whiskers. Biomacromolecules, 6(2), 1055-1061.

31

47. Petersson, L., & Oksman, K. (2006). Biopolymer based nanocomposites: Comparing layered silicates and microcrystalline cellulose as nanoreinforcement. Composites Science and Technology, 66(13), 2187-2196.

48. Kvien, I., Sugiyama, J., Votrubec, M., & Oksman, K. (2007). Characterization of starch based nanocomposites. Journal of Materials Science, 42(19), 8163-8171.

49. Azeredo, H. M. C., Mattoso, L. H. C., Wood, D., Williams, T. G., Avena-Bustillos, R. J., & McHugh, T. H. (2009). Nanocomposite edible films from mango puree reinforced with cellulose nanofibers. Journal of Food Science, 74(5), N31-N35.

50. Okahisa, Y., Yoshida, A., Miyaguchi, S., & Yano, H. (2009). Optically transparent wood–cellulose nanocomposite as a base substrate for flexible organic light-emitting diode displays. Composites Science and Technology, 69(11–12), 1958-1961.

51. Anglès, M. N., & Dufresne, A. (2000). Plasticized Starch/Tunicin whiskers nanocomposites. 1. Structural analysis. Macromolecules, 33(22), 8344-8353.

52. Mathew, A. P., & Dufresne, A. (2002). Plasticized waxy maize starch: Effect of polyols and relative humidity on material properties. Biomacromolecules, 3(5), 1101-1108.

53. Svagan, A. J., Hedenqvist, M. S., & Berglund, L. (2009). Reduced water vapour sorption in cellulose nanocomposites with starch matrix. Composites Science and Technology, 69(3–4), 500-506.

54. Sanchez-Garcia, M. D., Gimenez, E., & Lagaron, J. M. (2008). Morphology and barrier properties of solvent cast composites of thermoplastic biopolymers and purified cellulose fibers. Carbohydrate Polymers, 71(2), 235-244.

55. Sanchez-Garcia, M., & Lagaron, J. (2010). On the use of plant cellulose nanowhiskers to enhance the barrier properties of polylactic acid. Cellulose, 17(5), 987-1004.

56. Coulson, J. (2011). Wood in construction - How to avoid costly mistakes. US: John Wiley & Sons Inc. 220 p.

57. Hakkila, P. (1998). Structure and properties of wood and woody biomass. In: Kellomäki, S. (ed.), Papermaking science and technology. Forest resources and sustainable management. Jyväskylä, Finland, Fapet Oy. pp. 117–185.

58. Miller, R. B. (1999). Characteristics and availability of commercially important woods. Wood Handbook - Wood as an engineering material. Madison, WI, Forest products laboratory, U.S. Department of Agriculture. Ch. 1., pp. 1-31.

32

59. Wiedenhoeft, A. C., & Miller, R. B. (2005). Structure and Function of Wood. In: Rowell, R. M. (ed.), Handbook of Wood Chemistry and Wood Composites. Boca Raton, Florida, USA, CRC Press. pp. 9-33.

60. Smook, G. A. (1999). Handbook for pulp and paper technologists (2nd ed.). Vancouver, BC, Canada, Angus Wilde Publications Inc. 419 p.

61. Ilvessalo-Pfäffli, M.-S. (1977). Puun rakenne. In: Jensen, W. (ed.), Puukemia. Turku, Finland, Suomen Paperi-insinöörien Yhdistys. pp. 7-81.

62. Côté, W. A., Jr. (1967). Wood ultrastructure: An atlas of electron micrographs. Seattle, USA, University of Washington Press. 62 p.

63. Meier, H. (1962). Chemical and morphological aspects of the fine structure of wood. Pure and Applied Chemistry, 5, 37-52.

64. Chinga-Carrasco, G., Yu, Y., & Diserud, O. (2011). Quantitative electron microscopy of cellulose nanofibril structures from eucalyptus and pinus radiata kraft pulp fibers. Microscopy and Microanalysis, 17(04), 563-71.

65. Miller, R. B. (1999). Structure of wood. Wood Handbook - Wood as an engineering material. Madison, WI, USA, Forest products laboratory, U.S. Department of Agriculture.Ch. 2, pp. 1-4.

66. Salmén, L., Lucander, M., Härkönen, E., & Sundholm, J. (1999). Fundamentals of mechanical pulping. In: Sundholm, J. (ed.), Papermaking Science and Technology. Book 5: Mechanical pulping. Jyväskylä, Finland, Fapet Oy. pp. 35-61.

67. Back, E. L. & Salmén, L. (1982). Glass transitions of wood components hold implications for molding and pulping processes. Tappi J, 65(7), 107-110.

68. Herrick, F. W., Casebier, R. L., Hamilton, J. K., & Sandberg, K. R. (1983). Microfibrillated cellulose: Morphology and accessibility. Journal of Applied Polymer Science: Applied Polymer Symposium, 37, 797-813.

69. Turbak, A. F., Snyder, F. W., & Sandberg, K. R. (1983). Microfibrillated cellulose, a new cellulose product: Properties, uses, and commercial potential. Journal of Applied Polymer Science: Applied Polymer Symposium , 37, 815-823.

70. Zimmermann, T., Bordeanu, N., & Strub, E. (2010). Properties of nanofibrillated cellulose from different raw materials and its reinforcement potential. Carbohydrate Polymers, 79(4), 1086-1093.

33

71. Spence, K., Venditti, R., Rojas, O., Habibi, Y., & Pawlak, J. (2011). A comparative study of energy consumption and physical properties of microfibrillated cellulose produced by different processing methods. Cellulose, 18(4), 1097-1111.

72. Saito, T., Nishiyama, Y., Putaux, J., Vignon, M., & Isogai, A. (2006). Homogeneous suspensions of individualized microfibrils from TEMPO-catalyzed oxidation of native cellulose. Biomacromolecules, 7(6), 1687-1691.

73. Henriksson, M., Henriksson, G., Berglund, L. A., & Lindström, T. (2007). An environmentally friendly method for enzyme-assisted preparation of microfibrillated cellulose (MFC) nanofibers. European Polymer Journal, 43(8), 3434-3441.

74. Abe, K., Iwamoto, S., & Yano, H. (2007). Obtaining cellulose nanofibers with a uniform width of 15 nm from wood. Biomacromolecules, 8(10), 3276-3278.

75. Iwamoto, S., Abe, K., & Yano, H. (2008). The effect of hemicelluloses on wood pulp nanofibrillation and nanofiber network characteristics. Biomacromolecules, 9(3), 1022-1026.

76. Gong, G., Mathew, A. P., & Oksman, K. (2011). Strong aqueous gels of cellulose nanofibers and nanowhiskers isolated from softwood flour. TAPPI Journal, 10(2), 7-14.

77. Saito, T., Kimura, S., Nishiyama, Y., & Isogai, A. (2007). Cellulose nanofibers prepared by TEMPO-mediated oxidation of native cellulose. Biomacromolecules, 8(8), 2485-2491.

78. Araki, J., Wada, M., & Kuga, S. (2001). Steric Stabilization of a Cellulose MicroCrystal Suspension by Poly(ethylene glycol) Grafting. Langmuir 17(1), 21-27.

79. Katz, S., Beatson, R. P., & Scallan, A. M. (1984) The determination of strong and weak acidic groups in sulfite pulps. Svensk Papperstidning 87(6), 48-53.

80. Coperion. ZSK MEGAlab laboratory extruder. [image online]. Available at:http://www.coperion.com/en/compounding-extrusion/machines-systems/zsk-series/zsl-megalab-laboratory-extruder/. [accessed 7th Dec 2012].

81. Krogerus, B., Fagerholm, K. & Löytynoja, L. (2003). Analytical fractionation of pulps by tube flow. Paperi ja Puu – Paper and Timber 85(4), 209-13

Paper I

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4615

THE USE OF TWIN-SCREW EXTRUSION IN PROCESSING OF WOOD: THE EFFECT OF PROCESSING PARAMETERS AND PRETREATMENT Maiju Hietala,a,b Jouko Niinimäki,b and Kristiina Oksman a,*

In this study the effect of processing parameters on different types of wood raw material in extrusion was examined. The study consisted of two parts: the first part was to break and separate individual fibers from wood chips during the extrusion process; in the second part the effect of chemical pre-treatment and screw elements on wood raw material was evaluated. Statistical analysis was performed to evaluate the most important factors affecting wood particle size in extrusion. The statistical analysis showed that the screw speed is the main factor affecting wood fiber length in twin-screw extrusion of wood chips. This study showed that a twin-screw extruder can be used to separate individual fibers from wood chips, and the separated fibers have higher aspect ratios than the wood flour particles typically used in wood-polymer composites. When more fibrous and chemically softened wood raw material was used, fibers with even higher aspect ratios were obtained.

Keywords: Wood; Twin-screw extrusion; Processing parameters; Statistical analysis

Contact information: a: Department of Engineering Sciences and Mathematics, Luleå University of Technology, SE- 97187, Luleå, Sweden; b: Department of Process and Environmental Engineering, FI-90014 University of Oulu, Finland; * Corresponding author: kristiina.oksman@ltu.se

INTRODUCTION Wood-polymer composites (WPC) are typically described as materials that combine the properties of wood and thermoplastic materials. The biggest market for WPCs at the moment is outdoor building materials, but automotive parts and furniture are also manufactured from WPCs (Clemons 2002). When compared with inorganic and synthetic fillers and reinforcements traditionally used in polymer composites, wood is considered a low-cost and more environmentally friendly raw material. Most of the commercially manufactured WPCs today are made by compounding wood flour consisting of small wood particles with polyolefin polymer, e.g. polyethylene or polypropylene as matrix (Caulfield et al. 2005). Wood flour is produced from the wood residue of various wood processors by size reduction with different kinds of mills and size classification of the pulverized wood by screening (Clemons and Caulfield 2005). The particles in wood flour actually consist of fiber bundles, not individual fibers, and therefore the aspect ratio (length-to-diameter ratio) of wood flour is quite low, typically between 1 and 5 (Clemons 2008). Because of the low aspect ratio the reinforcing potential of wood flour is limited, and wood flour acts mainly as filler material in composites. The aspect ratio of a softwood fiber, for example, can be 100 (Smook 1999) and earlier studies have shown that wood fibers with a higher aspect ratio

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4616

have better reinforcing capability in wood-polymer composites (Stark 1999; Stark and Rowlands 2003). However, fibers are seldom used in manufacturing of commercial WPCs because feeding and metering of low-bulk-density fibers is difficult and they are also more expensive than wood flour (Caulfield et al. 2005; Clemons and Caulfield 2005). Problems with poor dispersion of longer wood fibers in composites have also been reported (Klason et al. 1984; Rozman et al. 1998). WPC research has strongly concentrated on enhancing the interfacial adhesion between the wood and polymer (Dalväg et al. 1985; Maldas et al. 1989; Raj et al. 1989; Felix and Gatenholm 1991; Coutinho et al. 1997; Ichazo et al. 2001), and the effect on wood characteristics is less frequently reported. The effect of wood particle size and aspect ratio on composite properties has been studied to some extent (Stark and Berger 1997; Stark and Rowlands 2003; Bledzki and Faruk 2003; Chen et al. 2006; Migneault et al. 2009; Bouafif et al. 2009), but only a few studies can be found where the effect of extrusion compounding on the dimensional properties of wood fibers and particles is examined (Bledzki et al. 2005; Le Baillif and Oksman 2009; Bouafif et al. 2010; Schirp and Stender 2010), even though twin-screw extrusion is one of the main processing methods used in the manufacture of WPC. In twin-screw extrusion the material is subjected to shear forces, and the amount of shear force depends greatly on the used screw configuration. Shear forces could also be used to reduce the particle size and to separate fibers with a higher aspect ratio from the wood particles if suitable processing conditions can be found. Thus, the wood raw material would not need as much pre-processing as the commonly used wood flour requires before it can be compounded with plastic in an extruder. If larger, undried wood particles can be used as raw material in WPCs, use of cheaper wood residues in composite manufacturing without pre-processing could be possible. In the current study the effect of extrusion parameters on different types of wood raw material was investigated. The study was made as a preliminary study with the goal of using wood chips as raw material in one-step extrusion of WPCs. If wood processing can be done in the same step as the compounding of WPC, then energy-consuming processing steps could be omitted from the composite manufacturing process. The main goal in this work was to produce wood particles/fibers with as high aspect ratio as possible. Basically, the study consisted of two parts. In the first part, the possibility to break down and separate individual fibers from wood chips in a co-rotating twin-screw extruder was examined. Statistical analysis was performed to evaluate the most important factors affecting wood particle size in extrusion. The effects of raw material moisture content, screw configuration, and screw speed on the aspect ratio of wood particles were evaluated using multiple linear regression analysis. In the second part of the study the effect of chemical raw material pre-treatment and reverse screw elements (RSE) on more fibrous wood raw material was evaluated. The dimensional properties of the extruded wood fibers and particles were measured using optical fiber analysis. Scanning electron microscopy was also used to study the morphology of the samples. To avoid laborious matrix removal, all the experiments were carried out without addition of matrix polymer in the extruder. Therefore the extrusion conditions used in the study were more severe for the wood raw material than they would be with addition of polymer.

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4617

EXPERIMENTAL Materials Spruce (Picea abies) wood chips (Fig. 1a) from Finland were used as raw material in the first part of the study. The average wood particle size was of 16.2 mm × 3.3 mm. Wood chip particle size was determined by measuring the length and the width of 100 particles manually. Wood chips were stored frozen until use. Both untreated (defrosted) wood chips and wood chips pre-treated in boiling water for 4 minutes immediately before extrusion were used. The purpose of the pre-treatment was to soften the wood chips and thus avoid crushing them into small particles, thereby facilitating the separation of individual fibers. The initial dry matter content (before freezing) of the wood chips was ca. 51 wt%. After defrosting, the dry matter content of the non-treated wood chips was 57 wt%, and the dry matter content of the pre-treated wood chips was 30 to 35 wt%. In the second part of the study mechanically pre-crushed spruce (Picea abies, from Finland) (Fig. 1b) was used as a raw material. This raw material was more fibrous than wood chips. The pre-crushed spruce was used as such (untreated) and as chemically pre-treated. The pre-treatment, a mild sulphonation was performed to achieve more softened raw material. The sulphonation was done at 90°C and at pH 9 for 210 minutes. The raw material was diluted to 10% consistency and 5% sodium sulphite, Na2SO3, was used in the treatment. The dry matter content of the untreated, pre-crushed spruce (referred as PC-U) was 46.3 wt%, and the dry matter content of sulphonated pre-crushed spruce (referred as PC-S) was 36.3 wt%.

Fig. 1. a) Spruce wood chips used in this study; b) Pre-crushed spruce used in this study

Extrusion A co-rotating twin-screw extruder (Coperion W&P ZSK-18 MEGALab, Stuttgart, Germany) without the extrusion die was used in the experiments. The screw length to diameter ratio (L/D) of the extruder was 40 with a screw diameter of 18 mm. The extruder consisted of seven (7) zones, and wood was fed into the first zone. Feeding of the wood chips was done manually, since the feeding equipment was not suitable for the materials used in the study. The extruder screws were fed as full as possible to maintain the filling degree as consistent as possible in all experiments.

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4618

The extruder screws had an elemental structure, and the screw elements used in this study were conveying elements, two types of kneading elements, and reverse screw elements. Conveying elements are used to take the material into the extruder and transport it downstream through the processing sections. Kneading elements are typically used for melting, dispersion, and mixing processes. Kneading elements with wide discs have a dispersive mixing effect while narrow kneading discs are more distributive. The reverse screw elements are conveying the material in the contrary direction and will resist the material flow effectively (Sämann 2008). Screening of the important factors The aim of the first part of this study was to determine the most important factors affecting wood aspect ratio in a co-rotating twin-screw extruder. MODDE 8 (Umetrics) statistical software and multiple linear regression analysis (MLR) were used to analyze the experimental data. A confidence level of 95% was used in the regression analysis. The experiments were performed according to a non-replicated 23 full-factorial design with three centre points. The total number of experimental runs was 11, and they were performed in randomized order. Experimental design is presented in Table 1. The factors studied were 1) rotational speed of the screws, 2) screw configuration, and 3) raw material treatment. The aspect ratio of wood particles after extrusion was used as a response.

Table 1. Experimental Design Exp. no. Run order Raw material Speed Configuration*

1 11 non-treated 100 rpm A 2 2 non-treated 500 rpm A 3 10 non-treated 100 rpm B 4 3 non-treated 500 rpm B 5 1 pre-treated 100 rpm A 6 6 pre-treated 500 rpm A 7 8 pre-treated 100 rpm B 8 4 pre-treated 500 rpm B 9 5 non-treated 300 rpm A 10 7 non-treated 300 rpm A 11 9 non-treated 300 rpm A

*A: screw configuration with WKE, B: screw configuration with WKE and RSE. The rotational speeds used were 100, 300, and 500 rpm. The speed 300 rpm functioned as a centre point in the design. The rotational speeds were chosen as extreme as possible, but there were some limitations when selecting suitable levels. 100 rpm was chosen as the lowest rotational speed to keep the processing times reasonable, and 500 rpm was chosen as highest rotational speed to avoid exceeding the maximum motor load when using the screw configuration with reverse screw element. The two different screw configurations used are presented in Fig. 2. Configuration A consisted of conveying elements and one wide kneading element (WKE), and configuration B consisted of conveying elements with one wide kneading element (WKE) followed by reverse screw element (RSE). The screw elements were chosen so that configuration A would be gentler than B. Pre-runs made with different screw configurations also showed that using

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4619

of wide kneading element in front of RSE was necessary to avoid motor overload of the extruder when using wood chips as raw material. The raw material used was either non-treated or pre-treated wood chips. To avoid excessive drying or burning of the wood in the extruder, the barrel temperature was set to 80°C in all experimental runs.

Fig. 2. Screw configurations A and B. WKE = wide kneading element; RSE = reverse screw element Effect of raw material and screw configuration Because the pre-treatment of the wood chips in the first part of the study did not increase the wood fiber aspect ratio, it was decided to use a more fibrous raw material with and without chemical pre-treatment in the second part of the study. Due to the change in raw material, reverse screw elements, RSE, could now be used as the first screw element. Three different screw configurations were used in the experiments (Fig. 3). Configuration C had one reverse screw element (RSE), configuration D had two RSEs, and configuration E had two RSEs with a thin kneading element between them (TKE) in addition to conveying elements (Fig. 3). The temperature of the extruder barrel was set to 70°C to avoid unwanted drying of the raw material. Screw speed in all experimental runs was 500 rpm. 500 rpm was chosen based on the results from the first part of the study.

Fig. 3. Screw configurations C, D and E. RSE = reverse screw element, TKE = thin kneading element.

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4620

Characterization The average fiber length (length weighted) and average fiber width used to calculate the aspect ratio (l/d) were measured using a FiberLab optical fiber analyzer (Metso Automation, Kajaani, Finland). Morphology of selected samples was studied using a field emission scanning electron microscope, FESEM (Zeiss ULTRA Plus, Carl Zeiss SMT AG, Jena, Germany) and scanning electron microscope, SEM (Jeol JSM-6400, Jeol Ltd., Tokyo, Japan). The samples for scanning electron microscopy were freeze dried and coated with platinum (FESEM) or gold (SEM) before examination. In the experiments in which wood chips were used as raw material the samples were fractionated into four different size categories before measuring the fiber lengths. The fractionation was done using a tube flow fractionator (Metso Automation, Kajaani, Finland). The purpose of the fractionation was to remove the first fraction containing the largest wood particles (shives) from the sample, thereby preventing blockage of the FiberLab analyzer. The average fiber lengths were then measured by combining the fractions 2, 3 and 4 from the tube flow fractionator into one sample. The tube flow fractionation method is described in more detail in the following publications: Krogerus et al. (2003) and Laitinen et al. (2006). The shive content of the samples in which wood chips were used as raw material were measured according to TAPPI T275 standard (Somerville method, 0.10 slots). In addition, a Bauer McNett classifier was used to divide samples into fractions R14, R28, R48, and R200 before examination with SEM in the second part of the study. The R14 fraction is typically considered as the shive fraction, R28 as the long fiber fraction, R48 as the short fiber fraction and R200 as the fines fraction. RESULTS AND DISCUSSION Screening of the Important Factors The average fiber length (length weighted), fiber width, aspect ratios (l/d) calculated from the length and width measurements, as well as the shive contents of the samples are presented in Table 2. Table 2. Average Fiber Length, Average Width, and Aspect Ratio of SamplesExp. Raw

material Speed,

rpm Screw

configuration Length, L(l), mm

Width, m

Aspect ratio, l/d

Shives, %

1 non-treated 100 A 0.65 31.1 20.9 38.5 2 non-treated 500 A 0.68 31.2 21.8 47.9 3 non-treated 100 B 0.55 30.6 18.0 6.9 4 non-treated 500 B 0.68 30.9 22.0 9.3 5 pre-treated 100 A 0.59 29.9 19.7 33.1 6 pre-treated 500 A 0.67 29.4 22.8 47.6 7 pre-treated 100 B 0.55 28.9 19.0 5.4 8 pre-treated 500 B 0.67 27.8 24.1 13.2 9 non-treated 300 A 0.64 30.7 20.8 43.7 10 non-treated 300 A 0.67 30.3 22.1 44.1 11 non-treated 300 A 0.63 30.5 20.7 46.2

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4621

Fig. 4. Scaled and centered regression coefficients for aspect ratio before model reduction

The statistically significant factors affecting the aspect ratio of wood raw material in twin-screw extrusion were estimated using MODDE 8 statistical software and multiple linear regression analysis (MLR). Figure 4 shows the coefficient plot for the aspect ratio before reducing the insignificant factors from the model. The insignificant factors were removed from the model, and the results from the regression analysis for the reduced model are presented in Tables 3 and 4. Table 3 shows the regression coefficients for the significant ( = 0.05) factors affecting the aspect ratio of wood raw material in a co-rotating twin-screw extruder together with the R2, R2(adj.), and Q2 parameters measuring the goodness of fit of the model. The values of these parameters presented in Table 3 indicate that the model fit the data moderately. According to the MLR analysis, screw speed (Scr) had the largest effect on aspect ratio, and it was also the only significant factor with a 95% confidence level affecting the aspect ratio. The final model for aspect ratio can be written as: (rpm) speedscrewl/dratio,Aspect 0082.06256.18 (1) The analysis of variance, ANOVA, for the fitted model is presented in Table 4 showing the P-value of 0.399 indicating that there was no lack of fit.

Table 3. The Unscaled regression coefficients for the fitted model for aspect ratio Response Constant Scr Con Raw Scr×Con Scr×Raw Con×Raw

Aspect ratio 18.6256 0.0082 - - - - - R2 =0.704, R2(adj.) =0.671, Q2= 0.536 Scr = screw speed, Con = configuration, Raw = raw material treatment. From Table 3 it can be seen that the shive content of the samples made using configuration B were clearly lower in comparison with samples made with configuration A. Therefore it seems that the use of RSE in the screw configuration subjected the material to more shear, thus reducing the shive content of the samples. According to the

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4622

statistical analysis and the results presented in Table 3, the raw material pre-treatment did not have much effect either on the aspect ratio or the shive content of the samples. Therefore it was thought that the pre-treatment was insufficient.

Table 4. Analysis of Variance (ANOVA) for the Fitted Model for Aspect Ratio Response Source DF SS MS F-value P-value

Aspect ratio Regression 1 21.45 21.45 21.44 0.001 Lack of Fit 7 7.79 1.11 1.82 0.399 Residual 9 9.01 1.00 Pure Error 2 1.22 0.61 Total 11 4919.33 447.21 DF = degrees of freedom, SS = sum of squares, MS = mean squares.

FESEM pictures from experiments No. 2, 4, 6 and 8 are presented in Fig. 5. These samples were chosen for the analysis because they had the longest fiber lengths and highest aspect ratios according to the analysis. The screw speed in these four experiments was 500 rpm. From Fig. 5 it can be seen that the samples in which screw configuration A (configuration with wide kneading element, WKE) was used (exp. 2 and 6) contained more coarse material (more shives) in comparison with the samples in which the screw configuration B was used (exp. 4 and 8). However, only minor differences can be seen when comparing the non-treated and pre-treated samples made with same screw configurations (exp. 2 and 6, exp. 4 and 8).

Fig. 5. Scanning electron microscopy images of experiments No. 2, 4, 6 and 8.

Even though the regression analysis showed that the screw configuration did not have a significant effect on aspect ratio and it had only a very small effect on fiber length, it was thought that the real effect of screw configuration was not clearly seen in this experiment. The wide kneading element, WKE, was used in both screw configurations as the first element to avoid motor overload, and it was suspected that the results actually show that the reverse screw element (RSE) had a small fiber shortening effect after the wood had first gone through the wide kneading element. Thus, it is possible that the kneading element was responsible of the extensive fiber shortening.

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4623

Effect of Raw Material Type and Reverse Screw Elements The results from the average fiber length and fiber width measurements together with the calculated aspect ratios are presented in Table 5. In these experiments the pre-crushed, more fibrous wood raw material was used.

Table 5. Fiber Properties of the Pre-Crushed Spruce after Extrusion Exp. Configuration Length, L(l),

mm Width,

mAspect ratio,

l/dPC-U1 C 1.21 27.3 44.3 PC-U2 D 0.82 25.0 32.8 PC-U3 E 0.85 25.5 33.3 PC-S1 C 1.59 30.0 53.0 PC-S2 D 1.20 27.9 43.0 PC-S3 E 1.24 27.8 44.6

PC-U = pre-crushed, untreated spruce. PC-S = pre-crushed, sulphonated spruce. Based on the results obtained from the first part of this study, three different screw configurations were used in the experiments. The highest aspect ratios were achieved using screw configuration C (samples PC-U1 and PC-S1) with one reverse screw element. The screw configurations with two reverse screw elements (configuration D and E) were most likely too harsh; therefore, the aspect ratios were lower in the samples made with these. As expected, the sulphonation pre-treatment affected the aspect ratio of the extruded fibers positively. This indicates that a proper softening of the wood raw material is very important to gain high aspect ratio fibers in extrusion of wood raw material. In Fig. 6 the SEM images of R14, R28, R48 and R200 mass fractions of two of the samples (PC-U1 and PC-S1) are shown. The samples were divided into fractions to achieve a better understanding of the morphology of the fibers. The images in Fig. 6 show that both of the samples contain individual fibers and only a small amount of shives. When comparing the images taken from the different fiber fractions, especially R14 fractions of the samples, it seems that PC-U1 contains coarser material than the PC-S1 sample (Fig. 6).

Fig. 6. Scanning electron microscopy images of R14, R28, R48, and R200 fractions of samples PC-U1 & PC-S1. PC-U = untreated pre-crushed spruce, PC-S = sulphonated pre-crushed spruce

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4624

CONCLUSIONS This study shows that a twin-screw extruder can be used to separate individual fibers from wood chips. The separated fibers have clearly higher aspect ratios than the wood flour particles typically used in wood-polymer composites. When using more fibrous and chemically softened wood raw material, fibers with even higher aspect ratios were obtained. According to the statistical analysis, the screw speed was the main factor affecting wood fiber aspect ratio in twin-screw extrusion of wood chips. It was also observed that the screw configuration can affect the fiber properties and the reverse screw elements reduced the shive content in the samples. REFERENCES CITED Bledzki, A. K., and Faruk, O. (2003). “Wood fibre reinforced polypropylene composites:

Effect of fibre geometry and coupling agent on physico-mechanical properties,” Appl.Compos. Mater. 10, 365-79.

Bledzki, A. K., Letman, M., Viksne, A., and Rence, L. (2005). “A comparison of compounding processes and wood type for wood fibre-PP composites,” Compos. Part A 36(6), 789-97.

Bouafif, H., Koubaa, A., Perré, P., and Cloutier, A. (2009). “Effects of fiber characteristics on the physical and mechanical properties of wood plastic composites,” Compos. Part A 40, 1975-1981.

Bouafif, H., Koubaa, A., Perré, P., and Cloutier, A. (2010). “Effects of composite processing methods on wood particle development and length distribution: Consequences on mechanical properties of wood-thermoplastic composites”, WoodFiber Sci. 42(1), 62-70.

Caulfield, D. F., Clemons, C., Jacobson, R. E., and Rowell, R. M. (2005). “Wood thermoplastic composites,” In: Handbook of Wood Chemistry and Wood Composites, Rowell, R. M., (ed.), CRC Press, Boca Raton, 365-378.

Chen, H. C., Chen, T. Y., and Hsu, C. H. (2006). “Effects of wood particle size and mixing ratios of HDPE on the properties of the composites,” Holz Roh Werkst. 64, 172-77.

Clemons, C. (2002). “Wood-plastic composites in the United States,” For. Prod. J. 52(6), 10-18.

Clemons, C. (2008). “Raw materials for wood-polymer composites,” In: Wood-polymer composites, Oksman, K., and Sain, M. (eds.), Woodhead Publishing Limited, Cambridge, 1-22.

Clemons, C. M., and Caulfield, D. F. (2005). “Wood flour,” In: Functional Fillers for Plastics, Xanthos M (ed.), Wiley-VCH Verlag GmbH, Weinheim, 249-270.

Coutinho, F. M. B., Costa, T. H. S., and Carvalho, D. (1997). “Polypropylene-wood fiber composites: Effect of treatment and mixing conditions on mechanical properties,” J.Appl. Polym. Sci. 65, 1227-35.

Dalväg, H., Klason, C., and Strömvall, H. E. (1985). ”The efficiency of cellulosic fillers in common thermoplastics. Part II. Filling with processing aids and coupling agents,”

PEER-REVIEWED ARTICLE bioresources.com

Hietala et al. (2011). “Extruder effects on wood,” BioResources 6(4), 4615-4625. 4625

Int. J. Polym. Mater. 11(1), 9-38. Felix, J. M., and Gatenholm, P. (1991). “The nature of adhesion in composites of

modified cellulose fibers and polypropylene,” J. Appl. Polym. Sci. 42(3), 609-620. Ichazo, M. N., Albano, C., González, J., Perera, R., and Candal, M. V. (2001). “Polypro-

pylene/wood flour composites: Treatments and properties,” Compos. St. 54, 207-214. Klason, C., Kubát, J., and Strömvall, H. E. (1984). “The efficiency of cellulosic fillers in

common thermoplastics. Part 1. Filling without processing aids or coupling agents,” Int. J. Polym. Mater. 10(3), 159-187.

Krogerus, B., Fagerholm, K., and Löytynoja, L. (2003). “Analytical fractionation of pulps by tube flow,” Pap. Puu–Pap. Tim. 85(4), 209-213.

Laitinen, O., Löytynoja, L., and Niinimäki, J. (2006). “Tube flow fractionator: A simple method for laboratory fractionation,” Pap. Puu–Pap. Tim. 88(6), 1-5.

Le Baillif, M., and Oksman, K. (2009). “The effect of processing on fibre dispersion, fibre length and thermal degradation of bleached sulphite cellulose fibre polypropylene composites,” J. Thermoplast. Compos. Mater. 22(2), 115-133.

Maldas, D., Kokta, B. V., and Daneault, C. (1989). “Influence of coupling agents and treatments on the mechanical properties of cellulose fiber-polystyrene composites,” J.Appl. Polym. Sci. 37(3), 751-75.

Migneault, S., Koubaa, A., Erchiqui, F., Chaala, A., Englund, K., and Wolcott, M. P. (2009). “Effects of processing method and fiber size on the structure and properties of wood–plastic composites,” Compos. Part A 40(1), 80-85.

Raj, R. G., Kokta, B. V., and Daneault, C. (1989). “Use of wood fibers in thermoplastics. VII. The effect of coupling agents in polyethylene-wood fiber composites,” J. Appl. Polym. Sci. 37(4), 1089-1103.

Rozman, H. D., Kon, B. K., Abusamah, A., Kumar, R. N., and Mohd Ishak, Z. A. (1998). “Rubberwood –high density polyethylene composites: Effect of filler size and coupling agents on mechanical properties,” J. Appl. Polym. Sci. 69, 1993-2004.

Schirp, A., and Stender, J. (2010). “Properties of extruded wood-plastic composites based on refiner wood fibres (TMP fibres) and hemp fibers,” Eur. J. Wood Pr. 68, 219-231.

Smook, G. A. (1999) Handbook for Pulp and Paper Technologists, 2nd Edition, Angus Wilde Publications, Vancouver.

Stark, N. M. (1999). “Wood fiber derived from scrap pallets used in polypropylene composites,” For. Prod. J. 49(6), 39-46.

Stark, N. M., and Berger, M. J. (1997). “Effect of particle size on properties of wood-flour reinforced polypropylene composites,” Proceedings of 4th International Conference on Woodfiber- Plastic Composites, Madison, 134-43.

Stark, N. M., and Rowlands, R. E. (2003). “Effects of wood fiber characteristics on mechanical properties of wood/polypropylene composites,” Wood Fiber Sci. 35(2), 167-174.

Sämann, H.-J. (2008). “Screw elements for co-rotating, closely intermeshing, twin-screw extruders,” In: Co-Rotating Twin-screw Extruders. Fundamentals, Technology and Applications, Kohlgrüber, K., and Wiedmann, W. (eds.), Hanser, Munich, 215-236.

Article submitted: July 21, 2011; Peer review completed: August 31, 2011; Revised version received and accepted: September 22, 2011; Published: September 24, 2011.

Paper II

Processing of wood chip–plastic composites:effect on wood particle size, microstructureand mechanical properties

M. Hietala1,2, J. Niinimaki2 and K. Oksman*1

Wood chips were used as raw material in extrusion of wood–plastic composites. Wood–plastic

composites with y50 wt-% wood content were manufactured by using two different compound-

ing methods. Dried and undried wood chips were used to investigate the effect of wood moisture

content on the wood particle size and whether the drying process could be carried out in the

same step. Wood particle properties were measured using optical fibre analysis. Microscopical

methods were used to examine the microstructure of wood particles. Furthermore, the prepared

composites’ mechanical properties were studied. The particle size of wood chips was significantly

reduced during extrusion in both processing methods. The undried wood chips had higher

aspect ratios in comparison with the dried wood chips after extrusion. Despite the higher aspect

ratio, the mechanical properties of composites manufactured with undried wood chips were not

better than the properties of composites with dried wood chips.

Keywords: Wood–plastic composites, Extrusion, Wood particle size, Mechanical properties

This paper is part of a special issue on Manufacturing and Design of Composites

IntroductionWood–plastic composites (WPCs) are materials whichhave a wood-like appearance but can be processed likeplastic materials. When compared to solid wood, WPCshave better durability and lower maintenance require-ments. Wood is also considered a low cost and renew-able material in comparison with inorganic andsynthetic fillers traditionally used in polymer compo-sites. Currently the biggest market for WPCs is outdoorbuilding materials, but automotive parts and furnitureare also made from WPCs.1

Wood–plastic composites are manufactured by com-pounding wood particles with thermoplastic matrix. Themost typical processing methods used to produce WPCsare extrusion compounding followed by profile extru-sion or injection moulding. The matrix polymer isusually a low cost polymer with a processing tempera-ture below the wood degradation temperature. The mostcommon matrix polymers in WPCs are polyethylene,polypropylene (PP) and polyvinyl chloride.2

The most widely used wood derived raw material incommercially manufactured WPCs is wood flour. Woodflour is produced from the scrap materials of variouswood processors by size reduction with different kinds ofmills and size classification of the pulverised wood by

screening.3 The particle size of commercially manufac-tured wood flour is usually smaller than 425 mm, and theaspect ratio (length/diameter ratio) is between 1 and 5.4

The low aspect ratio limits the reinforcing potential ofwood flour,5 and studies have shown that wood fibres witha higher aspect ratio have better reinforcing capability.6,7

Especially when good adhesion between fibre and matrixexists, fibres provide better stress transfer to the fibres.5

However, fibres are seldom used in manufacturing ofcommercial WPCs because feeding and metering of lowbulk density fibres is difficult and they are more expensivethan wood flour.2,5 Problems with poor dispersion ofwood fibres in composites have also been reported.8,9

A great deal of WPC research has concentrated onenhancing the interfacial adhesion between hydrophilicwood and hydrophobic thermoplastic matrix due to theincompatibility problems of these components. Parti-cularly, the use of different coupling agents and fibresurface treatments has been studied extensively.10–15

Research activities focusing on the effect of woodcharacteristics and processing conditions on WPCproperties have been less frequently reported. The effectof wood particle size on composite properties has beenstudied to some extent,7,16–19 but only a few studies canbe found where larger wood particles have been used inthe manufacture of WPCs.17,20 In addition, there havebeen few studies on the effect of extrusion compoundingon wood particle size and fibre length,21,22 even thoughtwin screw extrusion is one of the main processingmethods used in the manufacture of WPCs.

In the study by Bledzki and Faruk17 the physical andmechanical properties of composites made of PP and

1Division of Manufacturing and Design of Wood and Bionanocomposites,Lulea University of Technology, Lulea, Sweden2Laboratory of Fibre and Particle Engineering, University of Oulu, Oulu,Finland

*Corresponding author, email kristiina.oksman@ltu.se

� Institute of Materials, Minerals and Mining 2011Published by Maney on behalf of the InstituteReceived 26 July 2010; accepted 9 August 2010DOI 10.1179/174328911X12988622800855 Plastics, Rubber and Composites 2011 VOL 40 NO 2 49

different types of wood fibres and wood chips werecompared. They concluded that wood chips–PP compo-sites showed better tensile and flexural properties thanthe other wood–PP composites when 5% coupling agentwas added. Balasuriya et al.20 used wood flakes with asize of (1–4) mm6(0?1–2) mm in wood–high densitypolyethylene composites in their study of the effect ofwood flake distribution and wetting on the structure–property relationship of WPCs. They noticed that thelength of wood flakes was reduced after compounding.

Other studies have also shown that wood particle sizeand fibre lengths are reduced during the compounding.21–24

Usually, this reduction is considered a disadvantage, butthe shear forces in extrusion could also be used to reducethe particle size of larger wood particles. It is possible thatindividual fibres with a higher aspect ratio can be separatedfrom wood particles if an optimal screw configuration andprocessing method is found. Thus, the wood raw materialwould not need as much preprocessing as the commonlyused wood flour requires before it can be compounded withplastic in an extruder. If larger, undried wood particles canbe used as rawmaterial in aWPC, the use of cheaper woodresidues in composite manufacturing is possible.

In this study the use of wood chips as a raw materialin a WPC is investigated. The high shear forces in a co-rotating twin screw extruder are utilised to grind woodchips into smaller particles, possibly even into individualfibres, which are then compounded with polymer matrixin the same processing step. Two different extrusionprocessing methods are used, and both dried andundried wood chips are used to study the possibility ofcarrying out the drying process of wood in the sameprocess step. In addition, the effect of wood moisturecontent on wood particle size during extrusion is

examined. Wood particle properties such as aspect ratioand size distributions after extrusion are analysed.Furthermore, the mechanical properties of the preparedcomposites are studied.

Experimental

MaterialsSpruce pin chips were used as wood raw material(Fig. 1). Wood chip particle size was determined bymeasuring the length and the width of 100 particlesmanually. According to manual measurements, averageparticle length was 16?2 mm and average particle widthwas 3?3 mm. Initial moisture content of the wood chipswas 40 wt-% based on total weight. Wood chips wereused as dried and as undried. The dried wood chips werekept at 105uC for 24 h to reach moisture content below1 wt-% (based on dry weight) before processing. Theundried wood chips were used as such.

Polypropylene in powder form (melt flow index532 g/10 min, 230uC/2?16 kg and 16 g/10 min, 200uC/2?16 kg)supplied by Borealis Polyolefins (Austria) was used asmatrix polymer. Maleic anhydride grafted PP (MAPP,Epolene E-43; Eastman Chemical Company, USA),lubricant (TPW 113; Struktol, USA) and antioxidantagent were used as additives.

Processing of WPCsExtrusion compounding

A co-rotating twin screw extruder with twin screw sidefeeder, Coperion W&P ZSK-18 MEGALab (Stuttgart,Germany) equipped with a K-Tron gravimetric feeder(Niederlenz, Switzerland), two atmospheric vents and avacuum vent was used to compound the compositematerials. Before extrusion the PP and the additiveswere premixed. The gravimetric feeder was used to feedPP into the extruder while wood chips were fed into theextruder manually because it was not possible to use thegravimetric feeder system for this kind of material.

Two different processing methods (methods I and II)were used in the manufacturing of the composites. Themethods differed from each other in feeding orders ofthe raw materials, in screw configurations and intemperature profiles. In method I, PP and additiveswere fed first and wood chips were fed using the sidefeeder. In method II, wood chips were fed first and PPand additives were fed using the side feeder. Extrusionset-ups for methods I and II are shown in Figs. 2 and 3.The screw configuration used in method I (PP fed first) ispresented in Fig. 4, and the screw configuration used inmethod II (wood chips fed first) is presented in Fig. 5.

1 Wood chips used in this study

2 Extrusion set-up for processing method I

ONLINECOLOURONLY

Hietala et al. Processing of wood chip–plastic composites

50 Plastics, Rubber and Composites 2011 VOL 40 NO 2

The screw configurations were based on the feedingorder of the materials.

The compounded composites were profiled intorectangular profile with a cross-section of 5620 mm.Altogether four types of WPCs were manufacturedusing two different processing methods and two kinds ofwood chips:

(i) PP–DWC: composite made with method I anddried wood chips

(ii) PP–UWC: composite made with method I andundried wood chips

(iii) DWC–PP: composite made with method II anddried wood chips

(iv) UWC–PP: composite made with method II andundried wood chips.

Owing to the manual feeding of the wood chips, thewood content of composites was calculated according tothe feeding rate of PP (kg h21) and sampling time usedto collect the composite material. The initial goal was toprepare composites with a wood content of 50 wt-%;however, the actual wood contents varied to someextent.

Characterisation of wood particlesExtraction of matrix polymer

To measure wood particle size and shape after thecompounding, PP matrix was extracted from compo-sites. Before extraction, extruded pieces of each compo-site material were pressed into thin sheets with heatedpress (LPC-300; Fontjine Grotnes, Netherlands) by firstpreheating the material for 2 min and then pressing for30 s at 200uC. Five grams of each pressed material waskept in boiling xylene until the matrix was completely

dissolved (15–35 h). The solution was filtered andwashed with hot xylene. After this, the sample waswashed with ethanol to remove xylene and filtered. Atlast, the sample was washed with water to removeethanol and filtered again.

Fractionation

To achieve better understanding of particle size dis-tributions in each sample, the extracted samplesconsisting of wood particles and fibres were dividedinto four different size categories (fractions) by using thetube flow fractionation method (Metso Automation,Kajaani, Finland). After fractionation, mass fractions ofeach fraction were determined by filtering.

Tube flow fractionation is a method in which thesample (in water suspension) is injected into a contin-uous plug flow of water, which then enters a long plastictube. As the sample flows in the tube it is fractionated bysize. The fractionation device is equipped with a CCDcamera, which records approximately 600 images of thesample during the analysis. The images recorded of eachsample were saved and used in the image analysis. Thetube flow fractionation method is described in moredetail in publications by Laitinen et al. and Krogeruset al.25,26

Image analysis

To measure wood particle and fibre dimensions, theimages taken by the tube flow fractionation device wereanalysed using image analysis software developed byMetso Automation for research purposes. During imageanalysis, captured images are segmented to regions thatmay have one or several objects. The objects (particles,

3 Extrusion set-up for processing method II

4 Screw configuration used in method I

5 Screw configuration used in method II

Hietala et al. Processing of wood chip–plastic composites

Plastics, Rubber and Composites 2011 VOL 40 NO 2 51

fines and fibres) in one region are then interpreted basedon geometrical criteria. To measure properties of eachobject, several image analysis algorithms are performed.The basis for these algorithms is the model oninteraction of light and various objects. This enablesthe use of so called subpixel algorithms to improve theresolution. For each of the objects, the projection area isdetermined. Based of the shape of the object, the otherparameters such as length and width are then decided.These parameters were then used to calculate the aspectratio of the wood particles. Settings used in the imageanalysis are presented in Table 1.

Microscopy

To study the morphology of wood particles, microscopicimages were taken from the fractionated samples. ALeica MZ LIII stereomicroscope (Leica MicrosystemsLtd, Switzerland) was used to take images of woodparticle fractions FR1, FR2 and FR3. Owing to thesmall size of wood particles in fraction FR4, images weretaken with field emission SEM (Zeiss ULTRA Plus; CarlZeiss SMT AG, Germany). All the microscopy sampleswere freeze dried and field emission SEM samples weresputter coated with platinum before observation.

Characterisation of compositesCompression moulding

Test specimens for mechanical testing were mouldedwith a conventional compression moulding press(Fjellman Press AB, Mariestad, Sweden). The mouldtemperature was 50uC and the pressure applied on thesamples was y70 MPa. Before the moulding, extrudedprofiles (34 g batch weight) were heated in a hot air ovenat 200–210uC for 30 min. The composite samples inwhich undried wood chips were used as raw materialwere additionally dried at 75uC for 12–34 h beforecompression moulding in order to remove possiblemoisture left in the composites.

Density

Density of the compression moulded composite specimenwasmeasured using a pycnometer (AccuPyc 1330;Micro-meritics Instrument Corp., USA). The measurement is

based on gas displacement technique. The average spec-imen size was 3061065 mm. Three replicates for eachmaterial was tested.

Mechanical properties

Flexural testing of the wood chip–PP composites wasperformed according to ASTM D790 standard (three-point bending). A Shimadzu Autograph AG-X universaltesting machine (Shimadzu Corp., Kyoto, Japan) with aload cell of 1 kN was used for testing. A minimum offive specimens of each material were tested.

Microstructure

Fractured surfaces of composites were studied with aJeol JSM-6460 SEM (Jeol Ltd, Tokyo, Japan). Thespecimen from flexural testing were first frozen usingliquid nitrogen and then broken to create the fracturedsurface. The sample surfaces were sputter coated withgold before observation. Acceleration voltage of 15 kVwas used in the observation.

Results and discussion

Processing of WPCsTable 2 shows the material compositions of the manu-factured composites. Owing to the manual feeding of thewood chips, there was some variation in the woodcontents between the composite materials. The compo-sites with dried wood chips had a higher wood contentcompared to those with undried wood.

The processability of the composites with dried woodchips was better than composites with undried due to thefact that the moisture present in the wood chips wasvaporised during extrusion. During processing theactual processing temperatures increased slightly fromthe set temperatures in the mixing zones where thekneading elements were located, due to the high shearforces. This was observed especially when method II(wood fed first) and dried wood chips were used.

The extruded profiles of the manufactured compositesare shown in Fig. 6. The composites manufactured withprocessing method I had a more rough surface incomparison with those manufactured with method II, ascan be seen in Fig. 6. The composites with dried woodchips (PP–DWC and DWC–PP) had a typical WPCcolour, while those with undried wood chips (PP–UWCand UWC–PP) were lighter (Fig. 6).

Characterisation of wood particlesFractionation

The extracted samples consisting of wood particles andfibres were divided into four different size categories

Table 1 Parameters used in image analysis.

Parameter Value

Minimum fibre width, mm 1Maximum fibre width, mm 500Minimum fibre length, mm 0.01Maximum fibre length, mm 7.6

a PP–DWC, method I; b PP–UWC, method I; c DWC–PP, method II; d UWC–PP, method II

6 Extruded profiles of composites

Table 2 Material compositions of wood chip–PPcomposites*

Wt-%

Material Method PP{ Lubricant MAPP Wood

PP–DWC I 39.3 2.6 1.8 56.3PP–UWC I 42.5 2.8 1.9 52.8DWC–PP II 40.0 2.7 1.8 55.5UWC–PP II 43.8 2.9 2.0 51.3

*PP: polypropylene; DWC: dried wood chips; UWC: undriedwood chips; MAPP: Maleic anhydride grafted polypropylene. I:PP fed first, II: wood fed first.{PP includes 0?18 wt-% of antioxidant agent.

ONLINECOLOURONLY

Hietala et al. Processing of wood chip–plastic composites

52 Plastics, Rubber and Composites 2011 VOL 40 NO 2

(fractions) using a tube flow fractionation device. Afterfractionation, mass percentage of each fraction wasdetermined by filtering. The mass fractions of analysedsamples are shown in Fig. 7.

From Fig. 7, it can be seen that all the samples havedissimilar particle size distributions, which indicates thatboth the processing method and the moisture content ofraw material affected the wood particle size duringextrusion. It can also be seen from Fig. 7 that the shareof shive fraction FR1 is larger in samples that are madeusing processing method I (PP fed first) in comparisonwith those made using processing method II (wood fedfirst). This is reasonable, because in method I wood chipswere fed from the second feeding port, and thus wereexposed to less shear forces caused by the screw elements.The use of undried wood chips as raw material resulted ina larger share of bigger particles (FR1) compared tosamples in which dried wood chips was used. Especiallythe share of fraction FR1 is large when method I andundried wood chips were used (PP–UWC). The amountof fines (fraction FR4) was largest when method II anddried wood chips were used, indicating that this was theharshest processing combination.

Image analysis

Results from image analysis of fractionated samples areshown in Table 3. It can be seen from the results thatwhen undried wood chips have been used as raw mate-rial the samples contain longer particles in comparison

with the samples with dried wood chips. The woodparticles extracted from composite specimens manufac-tured with processing method I and undried wood chips(PP–UWC) had the greatest aspect ratios, as shown inTable 3.

Microscopy

Microscopic images of the whole wood particle samplesas well as fractions FR1, FR2, FR3 and FR4 of eachsample are presented in Fig. 8. The images show that ineach of the four cases wood chips have been significantlybroken down during extrusion compounding. There arealso clear differences in the particle shape and sizebetween samples. Especially the shape of wood particlesextracted from composites made with undried woodchips differs greatly from that of samples extracted fromcomposites in which dried wood chips were used. Withdried wood chips, the particle size is greatly reduced, butwood has remained in particulate form. The particles aremore fibrous and also individual fibres can be seen in thesamples extracted from composites in which undriedwood chips were used as raw material.

Microscopic observations confirm the results obtainedfrom fractionation and image analysis. Processingmethod I (PP fed first) appeared to be gentler on woodchips because, after extrusion, the wood particle lengthsand aspect ratios were higher in comparison withsamples in which method II was used. The use ofundried wood chips as raw material resulted in woodparticles and fibres with highest aspect ratio.

Characterisation of compositesDensity

The densities of the wood–chip composites are presented inFig. 9. There were small differences between all thedensities of the composites, but especially the density ofthe composite manufactured with method I and withundried wood chips (PP–UWC) was lower than others.The difference is possibly due to the lower wood content inthe composites with undried wood chips. However,porosity may also cause the lower density of the composite.

Mechanical properties

The flexural properties of the manufactured compositesare presented in Figs. 10 and 11. The results fromflexural testing were normalised to respond 50 wt-% of

7 Mass fractions of fractionated wood samples

Table 3 Results from image analysis*

Material Fraction Mass-% Average length, mm Average width, mm Aspect ratio (length/diameter)

PP–DWC FR1 15.6 0.59 167.4 3.5FR2 18.1 0.64 74.7 8.5FR3 32.7 0.22 42.3 5.3FR4 33.7 0.08 17.9 4.4

PP–UWC FR1 55.1 1.15 57.6 19.9FR2 11.6 1.12 32.8 34.3FR3 22.0 0.59 21.1 27.9FR4 11.4 0.08 13.7 5.6

DWC–PP FR1 4.4 … … …FR2 11.8 0.48 100.6 4.8FR3 34.7 0.24 49.2 4.9FR4 49.1 0.08 33.2 2.4

UWC–PP FR1 11.9 0.44 71.0 6.2FR2 11.0 0.50 56.3 8.8FR3 42.1 0.35 28.7 12.2FR4 34.9 0.10 20.3 4.8

*PP: polypropylene; DWC: dried wood chips; UWC: undried wood chips.

Hietala et al. Processing of wood chip–plastic composites

Plastics, Rubber and Composites 2011 VOL 40 NO 2 53

wood content, because of the variation in wood contentsof the composites.

The results from the flexural testing show that theaddition of the coupling agent and the moisture contentof wood chips had the greatest influence on thecomposites flexural properties and that the processingmethod did not affect the properties significantly.

Figure 10 shows a trend that the composites withdried wood chips (PP–DWC and DWC–PP) had thehighest flexural modulus being around 4 GPa while thecomposites with undried wood chips (PP–UWC andUWC–PP) had a bending stiffness around 3?5 GPa. Themanufactured composites with dried wood chips hadsimilar bending stiffness, to that reported earlier by

8 Microscopic images of extracted wood particle samples (top row) and fractions FR1, FR2, FR3 and FR4 of each

sample

9 Densities of wood chip–PP composites 10 Normalised flexural modulus of composites

ONLINECOLOURONLY

Hietala et al. Processing of wood chip–plastic composites

54 Plastics, Rubber and Composites 2011 VOL 40 NO 2

Oksman and Sanadi for wood polymer composites with50 wt-% wood flour.27

The composites maximum strength values when thecoupling agent was added were between 35 and 41 MPa,with no significant differences as can be seen in Fig. 11.However, the composites with undried wood chipsmanufactured with method I (PP–UWC) showed atrend with higher strength if compared to the others.The composites with no coupling agent showed lowerproperties and also very high scattering in the data. Thecomposites with wood chips as raw material showedlower flexural strength values when compared to earlierreported values with wood flour when the strength wasreported to be 59 MPa.27

The reason to test the properties without the couplingagent was because it was suspected that the moisture inthe undried wood chips could react with MAPP, andtherefore affect the composites properties negatively(Figs. 10 and 11). However, the results showed that the

addition of MAPP had a positive effect on both flexuralmodulus and strength (PP–UWC and UWC–PP) incomparison with the composites without coupling agent(PP–UWC* and UWC–PP*).

In general, it can be concluded that the mechanicalproperties of composites manufactured with undriedwood chips were lower or similar when compared tocomposites manufactured with dried wood chips, despitethe higher aspect ratio of wood particles. It is possible thatbound moisture was not removed from the wood chipsduring the extrusion process, which then act a plasticiserfor the wood and decreased the stiffness of the composites.

Microstructure

The SEM images from the fractured surfaces of thecomposites are shown in Fig. 12. The wood particles aresignificantly larger in the composites with dried woodchips (Fig. 12a and c) in comparison with undried woodchips (Fig. 12b and d). Wood particles are also clearlyfibrous in the composites made with undried woodchips. Even long individual fibres can be seen in thefractured surfaces of the composite made with method IIand undried wood chips (Fig. 12d).

However, cavities from fibre pull-outs and clean fibresurfaces can be seen in all of the composites indicatingthat the adhesion between the matrix and the woodparticles is not very good in any of these composites.The composites with dried wood chips (PP–DWC andDWC–PP) have large cavities from fibre pull-outs andthe composites with undried wood chips have plenty ofsmaller fibre pull-outs, as the fractured surfaces of thecomposites show (Fig. 12).

The results from flexural tests showed a trend that thecomposites stiffness was lower when undried wood chips

11 Normalised flexural strength of composites

a PP–DWC; b PP–UWC; c DWC–PP; d UWC–PP12 Fractured surfaces of composites

Hietala et al. Processing of wood chip–plastic composites

Plastics, Rubber and Composites 2011 VOL 40 NO 2 55

were used which was unexpected when seeing the micro-graphs of these materials. The undried wood chips withlarger aspect ratios should results in higher bendingstiffness than composites with particulate wood. Onepossible reason for this behaviour could be the remainingmoisture in the wood. The moisture can cause porosity inthe matrix polymer and it can also act as a plasticiser forthe wood. However, when the fractured surfaces of thecomposites were studied, no obvious signs of porosity inthe matrix could be seen (Fig. 12b and d). Therefore, thereason the composites did not reach better stiffness mightbe that the moisture was not efficiently removed duringprocessing and that softened to wood fibres.

ConclusionsWhen wood chips were used as raw material in WPCs,both the processing method and the wood chipsmoisture content affected the wood particle size andthe aspect ratio. In the case of dried wood chips, theparticle size was significantly reduced, but woodremained in particulate form. Individual fibres andwood particles with high aspect ratios were separatedduring the compounding process when undried woodchips were used as raw material.

The used processing methods did not have a greateffect on the mechanical properties of the composites.Composites manufactured with dried wood chips hadslightly better flexural properties compared to thosemanufactured with undried wood chips, despite the factthat the aspect ratio of wood particles in undried woodchip composites was higher.

This work shows that it is possible to use undried woodchips as raw material for WPCs and reach stiffness valuessimilar to the composites where wood flour is used as rawmaterial and by that decrease the rawmaterial cost for thefinal product. Furthermore, if the bound moisture fromwood chips can be removed during the process it might bepossible to improve themechanical properties because thewood is in the form of fibre. Fibres have larger aspectratios than wood particles and should therefore be moreefficient as reinforcement for polymers. It might bepossible to remove the moisture by increasing theresidence time in the extruder by using lower processingspeeds, a larger extruder or an extruder with a higherlength/diameter ratio.

Acknowledgement

The authors wish to thank Borealis Polyolefins, Austria,for supplying polypropylene.

References1. C. Clemons: Forest Prod. J., 2002, 52, (6), 10–18.

2. D. F. Caulfield, C. Clemons, R. E. Jacobson and R. M. Rowell:

‘Wood thermoplastic composites’, in ‘Handbook of wood chem-

istry and wood composites’, (ed. R. M. Rowell), 365–378; 2005,

Boca Raton, FL, CRC Press.

3. L. H. Reineke: ‘Wood flour’, US Forest Service research note FPL-

0113, USDA Forest Service, Forest Products Laboratory, Madison

WI, USA, 1966.

4. C. M. Clemons and D. F. Caulfield: ‘Wood flour’, in ‘Functional

fillers for plastics’, (ed. M. Xanthos), 249–270; 2005, Weinheim,

Wiley-VCH Verlag GmbH & Co KGaA.

5. C. Clemons: ‘Raw materials for wood-polymer composites’, in

‘Wood-polymer composites’, (ed. K. Oksman Niska and M. Sain),

1–22; 2008, Abington, Woodhead Publishing Limited.

6. N. M. Stark: Forest Prod. J., 1999, 49, (6), 39–46.

7. N. M. Stark and R. E. Rowlands: Wood Fiber Sci., 2003, 35, (2),

167–174.

8. C. Klason, J. Kubat and H. E. Stromvall: Int. J. Polym. Mater.,

1984, 10, (3), 159–187.

9. H. D. Rozman, B. K. Kon, A. Abusamah, R. N. Kumar and Z. A.

Mohd Ishak: J. Appl. Polym. Sci., 1998, 69, 1993–2004.

10. H. Dalvag, C. Klason, H. E. Stromvall: Int. J. Polym. Mater., 1985,

11, (1), 9–38.

11. D. Maldas, B. V. Kokta and C. Daneault: J. Appl. Polym. Sci.,

1989, 37, 751–775.

12. R. G. Raj, B. V. Kokta and C. Daneault: J. Appl. Polym. Sci.,

1989, 37, 1089–1103.

13. J. M. Felix and P. Gatenholm: J. Appl. Polym. Sci., 1991, 42, 609–

620.

14. F. M. B. Coutinho, T. H. S. Costa and D. Carvalho: J. Appl.

Polym. Sci., 1997, 65, 1227–1235.

15. M. N. Ichazo, C. Albano, J. Gonzalez, R. Perera and M. V.

Candal: Compos. Struct., 2001, 54, 207–214.

16. N. M. Stark and M. J. Berger: Proc. 4th Int. Conf. on ‘Woodfiber–

plastic composites’, Madison, WI, USA, May 1997, Forest

Products Society, 134–143.

17. A. K. Bledzki and O. Faruk: Appl. Compos. Mater., 2003, 10, 365–

379.

18. H. Bouafif, A. Koubaa, P. Perre P and A. Cloutier: Compos. Part

A, 2009, 40A, 1975–1981.

19. H. C. Chen, T. Y. Chen and C. H. Hsu: Holz. Roh. Werkst., 2006,

64, 172–177.

20. P. W. Balasuriya, L. Ye and Y.-W. Mai: Compos. Part A, 2001;

32A, 619–629.

21. A. K. Bledzki, M. Letman, A. Viksne and L. Rence: Compos. Part

A, 2005; 36A, (6), 789–797.

22. H. Bouafif, A. Koubaa, P. Perre and A. Cloutier: Wood Fiber Sci.,

2010, 42, (1), 62–70.

23. K. Oksman, A. Mathew, R. Langstrom, B. Nystrom and

K. Joseph: Compos. Sci. Technol., 2009, 69, 1847–1853.

24. M. Le Baillif and K. Oksman: J. Thermoplast. Compos. Mater.,

2009, 22, (2), 115–133.

25. O. Laitinen, L. Loytynoja and J. Niinimaki: Pap. Puu–Pap. Tim.,

2006, 88, (6), 1–5.

26. B. Krogerus, K. Fagerholm and L. Loytynoja: Pap. Puu–Pap.

Tim., 2003, 85, (4), 209–213.

27. K. Oksman Niska and A. R. Sanadi: ‘Interaction between wood

and synthetic polymers’, in ‘Wood-polymer composites’, (ed.

K. Oksman Niska and M. Sain), 41–71; 2008, Abington, Wood-

head Publishing Limited.

Hietala et al. Processing of wood chip–plastic composites

56 Plastics, Rubber and Composites 2011 VOL 40 NO 2

Paper III

The effect of pre-softened wood chips on wood fibre aspect ratioand mechanical properties of wood–polymer composites

Maiju Hietala a,b, Erik Samuelsson a, Jouko Niinimäki b, Kristiina Oksman a,⇑aDivision Manufacturing and Design of Wood and Bionanocomposites, Department of Engineering Sciences and Mathematics, Luleå University of Technology,SE-97187 Luleå, Swedenb Fibre and Particle Engineering Laboratory, Department of Process and Environmental Engineering, FI-90014, University of Oulu, Finland

a r t i c l e i n f o

Article history:Received 12 May 2011Received in revised form 13 September2011Accepted 17 September 2011Available online 23 September 2011

Keywords:A. WoodA. Polymer–matrix composites (PMCs)B. Mechanical propertiesE. Extrusion

a b s t r a c t

The objective of this work was to study the effect of chemical pre-treatment and moisture content ofwood chips on the wood particle aspect ratio after compounding in a twin-screw extruder and on themechanical properties of wood–polymer composites (WPCs). Composites with 50 wt.% wood contentwere manufactured using pre-treated and untreated wood chips. The effect of wood moisture contenton composite properties was studied by using dried and undried wood chips. The mechanical propertiesand fracture surfaces of the composites as well as the microstructure and aspect ratio of wood particlesafter compounding were studied. The highest wood particle aspect ratio after extrusion was achieved byusing pre-treated, undried wood chips as raw material. The chemical pre-treatment was found toenhance the defibration of wood chips as well as the mechanical properties of the composites.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Wood–polymer composites (WPCs) are materials which com-bine the properties of wood and thermoplastic polymers. Theyare considered more environmentally friendly options for woodand plastic products in several applications, such as outdoor build-ing materials, furniture and interior panels in automobiles [1].WPCs are manufactured by compounding wood raw material witha molten matrix polymer together with additives such as lubri-cants, antioxidants and coupling agents in various compositions.In commercial manufacturing of wood–polymer composites themain wood rawmaterial used is wood flour. Wood flour is typicallyproduced from the residues of wood processors by milling andscreening the material into a specific particle size range [2]. Usu-ally, the aspect ratio (length-to-diameter ratio) of wood flour par-ticles is low, between 1 and 5 [2]. Earlier studies have shown thatwood and other natural fibres with higher aspect ratio give betterreinforcement than wood particles with low aspect ratio [3,4].Especially when there is a good adhesion between fibres and thematrix, a better stress transfer to the fibres is achieved [2]. How-ever, since the processing of longer, bulky fibres is more difficultand fibres are more expensive than wood flour, wood fibres arenot often used in the manufacture of wood–polymer composites[5].

The focus in wood–polymer composite research has been on theimprovement of the interfacial adhesion between the hydrophilicwood and hydrophobic polymer matrix. Studies have concentratedespecially on various coupling agents and chemical treatments toimprove adhesion between these two components [6–10]. Consid-erably less research has been focused on studying the effects ofwood particle properties on composite properties, and these stud-ies are mainly about how the wood particle size and wood speciesaffect composite properties [11–14]. Only a few studies can befound where the effect of the aspect ratio of wood fillers has beenaddressed [15–17], even though it has been suggested that it is theaspect ratio but not the particle size of wood filler which has themain influence on the mechanical properties of the wood–polymercomposites [4]. The possibility to use larger wood particles as rawmaterial for WPC has been studied by only a few researchers[18,19], though it would be an economic advantage if wood resi-dues could be used directly in the compounding process ofwood–polymer composites without pre-processing steps such asmilling, sieving and drying. In our earlier study it was shown thatwood chips can be used directly as raw material for WPC, and it iseven possible to separate individual fibres from the wood chipsduring the extrusion compounding if undried wood chips are usedas raw material [20].

The objective of this work was to investigate the effect of chem-ical pre-treatment and moisture content of wood chips on thewood particle aspect ratio after compounding in a twin-screw ex-truder and on the mechanical properties of prepared composites.The purpose of the chemical pre-treatment was to soften the wood

1359-835X/$ - see front matter � 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.compositesa.2011.09.021

⇑ Corresponding author. Tel.: +46 920 493371; fax: +46 920 491399.E-mail address: kristiina.oksman@ltu.se (K. Oksman).

Composites: Part A 42 (2011) 2110–2116

Contents lists available at SciVerse ScienceDirect

Composites: Part A

journal homepage: www.elsevier .com/locate /composi tesa

raw material prior to extrusion. In the earlier study by Hietala et al.[20] a decrease in the flexural properties was noticed when un-dried wood chips were used as raw material, even though the as-pect ratio of wood particles was increased. The high moisturecontent (40 wt.%) of the wood raw material feed was believed tobe the reason for this and therefore, in this work the effect of lowermoisture content of the undried wood raw material was alsostudied.

2. Experimental

2.1. Materials

Spruce chips were used as wood raw material (Fig. 1). Woodchips were used as: (1) untreated and undried (UU), (2) untreatedand dried (UD), (3) pre-treated and undried (PU), and (4) pre-trea-ted and dried (PD). The moisture content of the undried woodchips was 33 wt.% based on total weight, and the moisture contentof the dried wood chips was below 1 wt.% based on the dry weight.The wood chips were dried in an oven at 105 �C for 24 h.

The matrix polymer used was polypropylene, PP, in powderform with a Melt Flow Index (MFI) of 32 g/10 min, 230 �C/2.16 kgand 16 g/10 min, 200 �C/2.16 kg supplied by Borealis Polyolefins,Schwechat, Austria. A coupling agent, maleic anhydride graftedpolypropylene, (MAPP, Epolene E-43, Eastman Chemical Company,Longview, Texas, USA), a lubricant (TPW 113, Struktol, Stow, Ohio,USA) and antioxidant agent were used as additives.

2.2. Wood chip pre-treatment

Sodium sulphite, Na2SO3 (Merck KGaA, Darmstadt, Germany),was used in the chemical pre-treatment. The pre-treatment wasdone at 90 �C and at pH 9 for 210 min. The wood raw materialwas diluted to 10% consistency and 5% Na2SO3 (based on the totalmass of the mixture) was used in the treatment. After the pre-treatment, wood chips were washed with water, and then left todry at room temperature until a moisture content of 33 wt.% wasreached.

2.3. Compounding process

A co-rotating twin-screw extruder with twin-screw side feeder,ZSK-18 MEGALab (Coperion W&P, Stuttgart, Germany) equippedwith a K-Tron gravimetric feeder (Niederlenz, Switzerland), twoatmospheric vents and a vacuum vent was used to compoundthe composite materials. Polypropylene and the additives werepre-mixed prior to extrusion. The gravimetric feeder was used tofeed PP into the extruder, while wood chips were fed into the ex-truder manually, because it was not possible to use the gravimetric

feeder system for this kind of material. The wood content of com-posites was calculated according to the following formula:

Wood content;% ¼ mwood

mcomposite� 100%

¼ mcomposite � ðQPP � tÞmcomposite

� 100% ð1Þ

wheremwood andmcomposite are the masses of wood (total mass) andcomposite (kg), QPP is the feeding rate of PP(kg/h), and t is the sam-pling time (h) used to collect the composite. The aim was to preparecomposites with a wood content of 50 wt.%, but the actual woodcontents varied between 45 wt.% and 54 wt.%.

The extrusion set-up and the screw configuration used in thecompounding of the composites are shown in Fig. 2. The extrusionconditions were the same for all prepared composites. The screwconfiguration used was chosen based on the reported results ofan earlier study [20]. The compounded composites were profiledinto rectangular profile with a cross section of 5 mm � 20 mm.Four types of wood chip-PP composites were manufactured:

(a) UD-PP: composite made with untreated and dried woodchips.

(b) UU-PP: composite made with untreated and undried woodchips.

(c) PD-PP: composite made with pre-treated and dried woodchips.

(d) PU-PP: composite made with pre-treated and undried woodchips.

Except for the manual feeding of the wood chips, there were nogreat difficulties in the processing of the composites. Steam wascreated during compounding when undried wood chips were used,but it was ventilated out from the extruder. Blockage of the atmo-spheric vents after a longer compounding period posed a minorproblem. The overall impression was that using undried woodchips was less detrimental to the extruder screws than the use ofdried wood chips.

The appearance of the composite profiles made using undriedwood chips was slightly different from the profiles made withdried wood chips indicating that there could be more voids inthe material due to vaporized moisture. However, after compres-sion moulding the specimens appeared to be identical.

2.4. Characterization

2.4.1. Specimen preparationTest specimens for flexural testingweremouldedwith a conven-

tional compression moulding press (Fjellman Press AB, Mariestad,Sweden). Before moulding, the extruded profiles were broken intoshorter pieces (approximately 60 mm � 5 mm � 20 mm) andweighed intobatchesof�34 gwhichwere thenheatedbetweenalu-minium plates and Teflon sheets in a hot air oven at 200 �C for 20–45 min. After heating, the composites were transferred into thecompression moulding press. The mould temperature was 50 �Cand the pressure applied on the samples was approximately70 MPa. A plate of five flexural specimens could bemoulded at once.After cutting, the size of the flexural specimens was 5 mm �10 mm � 120 mm.

Plate-shaped test specimens for impact testing were madewith an LPC-300 Fontijne Grotnes heat press (Vlaardingen, Neth-erlands) and using a metal frame with an opening size of1.5 mm � 100 mm � 100 mm. Before pressing, the extrudedmaterial was broken into shorter pieces (approximately 90mm � 5 mm � 20 mm, five pieces per plate), placed between alu-minium plates and Teflon sheets with the frame, and preheated inFig. 1. Spruce wood chips.

M. Hietala et al. / Composites: Part A 42 (2011) 2110–2116 2111

the press for 10 min at 200 �C without pressure. The preheatedmaterial was then pressed for 2 min, with a load of 150 kN at200 �C. The dimensions of the test plates were approximately1.6 mm � 100 mm � 100 mm.

2.4.2. Mechanical propertiesFlexural testing of the wood chip-PP composites was performed

according to ASTM D790 standard (three-point bending). A Shima-dzu Autograph AG-X universal testing machine (Shimadzu Corp,Kyoto, Japan) with a load cell of 1 kN was used for testing. A mini-mum of 13 specimens of each material were tested. The supportspanused in the testingwas 84.5 mmand the speed of the crossheadwas 2.254 mm/min. Impact properties of the prepared compositesweremeasuredwitha fallingweight impact testingmachine,Dynat-up Minitower (Instron, Norwood, MA, USA) according to ASTMD3763 standard. The total mass of the impactor was 2.31 kg andspeed of testing was 2.8 m/s. At least three specimens of eachmate-rial were tested.

Because there were differences in the wood contents of thecomposite materials, the results from the flexural testing were nor-malized to correspond to 50 wt.% wood content assuming that lin-ear relationship between wood loading and properties exists. Thenormalization was done using the following formula:

Normalized value ¼ Actual value�wnormalization

wspecimenð2Þ

where wnormalization is the chosen wood content for normalization(50 wt.%) and wspecimen is the actual wood content of the specimen(wt.%). For example, the flexural strength value of UD-PP specimenNo. 8 was 46.11 MPa with a wood content of 54 wt.%. The normal-ized strength value was calculated using Eq. (2) as follows:

Normalized flexural strength ¼ 46:11 MPa� 50 wt:%54 wt:%

¼ 42:69 MPa:

2.4.3. DensityThe density of the compression moulded composite specimen

was measured using the water displacement method accordingto ASTM D792 standard (test method A). The specimen size was65 mm � 10 mm � 5 mm. Four replicates for each material weretested.

2.4.4. Extraction of matrix polymerTo study thewood particle size and shape after compounding, PP

matrix was extracted from the composites. Prior to extraction,piecesof each compositematerialwerepressed into thin sheetswitha heat press (LPC-300, Fontijne Grotnes, Vlaardingen, Netherlands)by first preheating the material for 5 min without pressure at200 �C and then pressing for 2 min at 150 kN at 200 �C. A few gramsof each pressedmaterialwere kept in boiling xylene until thematrixwas completely dissolved (24 h). After extraction, the wood parti-cles were washed with ethanol and water to remove xylene.

2.4.5. Aspect ratio of wood particlesFor the determination of the aspect ratio of wood particles ex-

tracted from the composites, the samples were first analyzed witha tube flow fractionator (Metso Automation, Kajaani, Finland). Thetube flow fractionator is typically used to fractionate wood fibresamples into different size categories. However, in this case, thedevice was used to capture images of the wood particle samplesfor the image analysis, because during analysis approximately600 images of the sample are recorded. The tube flow fractionationmethod is described in more detail in publications by Krogeruset al. and Laitinen et al. [21,22].

After fractionation the images were analyzed using kajaaniIMGimage analysis software supplied by Metso Automation (Kajaani,Finland). During image analysis, captured images are segmentedto regions that may have one or several objects. The objects (parti-cles, fines and fibres) in one region are then interpreted based ongeometrical criteria. Several image analysis algorithms are per-formed to measure the properties of each object. The basis for thesealgorithms is the model on interaction of light and various objects.This enables the use of so-called sub-pixel algorithms to improvethe resolution. For each of the objects, the projection area is deter-mined. Based on the shape of the object, other parameters such aslength andwidth are then decided. The aspect ratio of thewood par-ticles was calculated using the length and width parameters. Thesettings used in the image analysis are presented in Table 1. The set-tings were chosen so that the smallest wood particles would be ex-cluded from the analysis.

2.4.6. Morphology and microstructureA scanning electron microscope (Jeol JSM-6460, Jeol Ltd., Tokyo,

Japan) was used to observe the size and shape of wood particlesafter extraction of the polymer matrix as well as the fracture sur-faces of the manufactured composites. The wood particle samples

Fig. 2. The extrusion set-up and the screw configuration used in the study.

2112 M. Hietala et al. / Composites: Part A 42 (2011) 2110–2116

were freeze-dried before microscopy. The fracture surfaces of thecomposites were created by freezing flexural testing specimensin liquid nitrogen and then breaking the frozen specimen. All thespecimens were sputter-coated with gold before observation. Anacceleration voltage of 10 kV was used in the observation of thefracture surfaces and an acceleration voltage of 5 kV was used inthe observation of the wood particles.

2.4.7. Statistical analysisOne-way analysis of variance (ANOVA) followed by Tukey–

Kramer multiple comparison tests with a 0.05 significance levelwas used to analyze the results from mechanical testing.

3. Results and discussion

3.1. Composites

In Table 2 the compositions of the wood chip-PP composites areshown. The wood contents of the composites varied to some extentdue to the manual feeding of the wood chips in the extruder. Ingeneral, the composites with dried wood chips had higher woodcontent compared to the composites with undried wood.

3.1.1. Mechanical propertiesThe results from the mechanical testing are shown in Table 3.

Because there were differences in the wood contents of the com-posite materials, the results from flexural testing were normalizedto correspond to 50 wt.% wood content.

From Table 3 it can be seen that the chemical pre-treatment af-fected the flexural properties positively: the composites with pre-

treated wood chips had better strength properties than compositeswith untreated wood chips and the same wood chip moisture lev-els. The flexural properties on the composites were also affected bythe moisture content of the wood chips: the use of undried woodchips with a moisture content of 33 wt.% enhanced the flexuralstrength but, simultaneously, the modulus of elasticity was de-creased. The composite manufactured with pre-treated, undriedwood chips (PU-PP) had the highest flexural strength and the com-posite with pre-treated, dried wood chips (PD-PP) had the highestflexural modulus among the prepared composites. The lower mod-ulus of the composites with undried wood chips could be ex-plained by the fact that moisture makes wood more flexible.From Table 3 it can be seen that the strains of the composites withundried wood chips were larger than with the composites withdried wood chips.

The impact properties of the composites were studied using afalling weight impact test, and the results are presented in Table3. Characteristic load–deflection and energy–deflection curvesfrom the impact testing are shown in Fig. 3. The shape of the curveswas similar for all the materials. It can be seen from Table 3 thatthere were small differences between the peak load and total im-pact energy values of the composites; however, the differenceswere not statistically significant.

The results from the mechanical testing showed that thechemical pre-treatment enhanced the flexural properties of thecomposites. The flexural strength of the composites with pre-treated wood chips was higher than that of the composites withuntreated chips. No significant improvement could be seen in theimpact properties due to the chemical pre-treatment. Theenhancement of the flexural strength of the composites withpre-treated wood chips is believed to be caused mainly by thehigher aspect ratio of wood fibres and particles due to lignin soft-ening. This is assumed to be especially pronounced when undriedwood chips are used. It is also possible that pre-treatment withsodium sulphite changed the surface properties of the wood par-ticles [23], and thus improved the mechanical properties of thecomposites by improving the adhesion between wood andpolypropylene.

3.1.2. DensityThe results from the density measurements are shown in Fig. 4.

The results were normalized to correspond to 50 wt.% wood con-tents. The composite with pre-treated, undried chips (PU-PP) hadthe highest density, 1.23 g/cm3, and the composite with untreated,dried chips had the lowest density, 1.03 g/cm3. According to Cle-mons 2008 [2], increased density of the wood–polymer compositesis often related to the collapsing or filling of the hollow wood fi-bres. The reason for the higher density of the PU-PP composite ismost likely that there are more individual, collapsed wood fibresin comparison with the larger wood particles in the other compos-ites due to the softening of the wood caused by moisture and thechemical pre-treatment. The composite with the highest density(PU-PP) had the best flexural properties among the preparedmaterials.

Table 1Settings used in the image analysis.

Parameter Value

Minimum fibre width 10 lmMaximum fibre width 3000 lmMinimum fibre length 0.1 mmMaximum fibre length 7.6 mm

Table 2Material compositions of the wood chip-PP composites.

Material Weight (%)

PPa Lubricant MAPP Wood

UD-PP 41.4 2.8 1.8 54.0UU-PP 44.1 2.9 2.0 51.0PD-PP 45.0 3.0 2.0 50.0PU-PP 49.5 3.3 2.2 45.0

PP = polypropylene, UD = untreated and dried wood chips, UU = untreated andundried wood chips, PD = pre-treated and dried wood chips, PU = pre-treated andundried wood chips.

a Polypropylene includes 0.18 wt.% of antioxidant agent.

Table 3Averages and standard deviations of flexural and impact properties of the composites. The flexural properties were normalized to correspond to 50 wt.% wood content.

Material Normalized flexural properties Impact properties

Strength* (MPa) MOE* (GPa) Strain at max* (%) Peak load* (N) Total energy* (J)

UD-PP 43.6 ± 3.9a 4.5 ± 0.4ac 1.4 ± 0.1a 148.6 ± 10.2a 1.9 ± 0.1a

UU-PP 48.5 ± 4.5b 3.9 ± 0.5b 2.1 ± 0.2b 167.6 ± 3.6a 2.1 ± 0.4a

PD-PP 50.5 ± 2.6b 4.9 ± 0.5a 1.6 ± 0.2a 156.3 ± 13.6a 1.7 ± 0.4a

PU-PP 57.7 ± 5.4c 4.3 ± 0.3bc 2.5 ± 0.2c 169.9 ± 25.8a 2.0 ± 0.3a

* Means marked with the same superscript letter within the same column are not significantly different at 5% significance level based on the ANOVA and Tukey–Kramerpairwise comparison test. PP = polypropylene, UD = untreated and dried wood chips, UU = untreated and undried wood chips, PD = pre-treated and dried wood chips,PU = pre-treated and undried wood chips.

M. Hietala et al. / Composites: Part A 42 (2011) 2110–2116 2113

3.1.3. Fracture surfacesFigs. 5 and 6 show SEM images taken from the fracture surfaces

of the wood chip-PP composites. The images in Fig. 5 are takenwith less magnification to show a general view of the fracture sur-face of the composite. From Fig. 5 it can be seen that the fracturesurface of composites with dried wood chips (Fig. 5a and c) isrougher in comparison with composites with undried wood chips(Fig. 5b and d). Also, large cavities from pull-outs and parts of largewood particles can be seen in the fracture surfaces of the compos-

ites with dried wood chips. The fracture surfaces of the compositeswith undried wood chips are smoother, most likely due to thesmaller wood particles and fibres contained in them.

Fig. 6 shows details from the fracture surfaces of the wood chip-PP composites. It can be seen that wood particles have not beenseparated into individual fibres when dried wood chips have beenused as raw material (Fig. 6a and c), even though the particle sizeappears to be greatly reduced. The presence of individual fibres inthe composites with undried wood chips is obvious (Fig. 6b and d).Also, larger wood particles could be seen in the fracture surfaces ofthe composites with undried wood chips, but the amount was sig-nificantly lower than in the composites with dried wood chips. Noobvious differences between the composites with untreated andpre-treated wood chips could be seen in the SEM images, and theadhesion appeared to be moderately good in all the compositematerials. No large gaps between the matrix and the wood parti-cles could be seen in the images; however, fibre and particlepull-outs were observed in all the composites.

3.2. Wood particles

3.2.1. MicrostructureFig. 7 shows SEM micrographs of the wood particles after ma-

trix removal. Individual fibres are clearly seen in the images whichshow the wood particles extracted from the composites manufac-tured with undried wood chips (Fig. 7b and d). In addition, moreindividual fibres can be seen in Fig. 7d in comparison withFig. 7b, indicating that the wood chip pre-treatment improvedthe fibre separation from the undried wood chips. When driedwood chips have been used as raw material, the wood has re-mained in particulate form, as can be seen from Fig. 7a and c. Thereare no apparent differences between the wood particles extractedfrom the composites made with dried, untreated wood chips(Fig. 7a) and dried, pre-treated wood chips (Fig. 7c).

3.2.2. Aspect ratioThe average aspect ratios (l/d) as well as the average lengths

and widths of the wood particles extracted from the compositesamples are shown in Table 4. The aspect ratios were calculated

Fig. 3. Characteristic load and energy vs. deflection curves from impact testing.

Fig. 4. Densities of the wood chip-PP composites.

Fig. 5. Overviews of the fracture surfaces of the wood chip-PP composites. (a) UD-PP, (b) UU-PP, (c) PD-PP and (d) PU-PP.

2114 M. Hietala et al. / Composites: Part A 42 (2011) 2110–2116

from the length and width measurements determined using imageanalysis software.

As can be seen from Table 4, the wood particle samples from thecomposites in which dried wood chips were used (UD-PP, PD-PP)had similar aspect ratios as wood flour typically used in wood–polymer composites. The use of undried wood chips as raw mate-rial resulted in significantly greater wood particle aspect ratiosafter compounding. The wood chip pre-treatment slightly in-creased the wood particle aspect ratio, which may explain the im-proved mechanical properties of the composites with pre-treatedwood chips. The wood particles with the highest aspect ratio wereextracted from the composite made with pre-treated and undriedwood chips (PU-PP), which also had the highest flexural strengthand impact properties (Table 3).

Fig. 6. Details from the fracture surfaces of the wood chip-PP composites. (a) UD-PP, (b) UU-PP, (c) PD-PP and (d) PU-PP.

Table 4The average lengths, widths and aspect ratios of the extracted wood particles.

Material Length, L(l)a (mm) Width (lm) Aspect ratio (l/d)

UD-PP 0.3 46.3 7.9UU-PP 0.6 34.1 23.1PD-PP 0.3 38.3 8.9PU-PP 0.7 35.7 25.3

a Length weighted average length.

Fig. 7. Wood particles after matrix removal. (a) UD-PP, (b) UU-PP, (c) PD-PP and (d) PU-PP.

M. Hietala et al. / Composites: Part A 42 (2011) 2110–2116 2115

4. Conclusions

In this work the effect of chemical pre-treatment and moisturecontent (dried and undried wood chips) in manufacture of woodchip-PP composites was studied with an aim to create more indi-vidual wood fibres during the processing as well as improve thecomposites’ mechanical properties. This study showed that chem-ical pre-treatment with sodium sulphite improved the composites’mechanical properties and that the wood particle aspect ratio wasincreased in the extrusion process.

The use of pre-treated wood chips enhanced the flexural prop-erties of the wood chip-PP composites. The use of undried woodchips resulted in improved flexural strength in comparison withthe composites with dried wood chips, but the flexural moduluswas decreased. Study of the aspect ratios of wood particles showedthat the use of pre-treated and undried wood chips resulted in thehighest aspect ratio after compounding. In general, aspect ratioswere considerably higher when undried wood chips were used asraw material instead of dried wood chips.

The overall results are promising. The study shows that it ispossible to use wood chips as raw material for the productionwood–polymer composites and that the wood chips can be defi-brated during the composite manufacturing process. When a sim-ple chemical pre-treatment is used to soften the wood a moredefibrated wood structure can be achieved.

Acknowledgements

The authors wish to thank Borealis Polymers for supplying thepolypropylene used in the study, and Mr. Runar Långström at Swe-rea SICOMP AB for his help with the compression moulding of thespecimens.

References

[1] Oksman K, Bengtsson M. Wood fibre thermoplastic composites: processing,properties and future developments. In: Fakirov S, Bhattacharyya C, editors.Handbook of engineering biopolymers: homopolymers, blends andcomposites. Munich: Carl Hanser Verlag; 2007. p. 655–71.

[2] Clemons C. Raw materials for wood–polymer composites. In: Oksman Niska K,Sain M, editors. wood–polymer composites. Cambridge: Woodhead PublishingLimited; 2008. p. 1–22.

[3] Stark NM. Wood fiber derived from scrap pallets used in polypropylenecomposites. Forest Prod J 1999;49(6):39–46.

[4] Stark NM, Rowlands RE. Effects of wood fiber characteristics on mechanicalproperties of wood/polypropylene composites. Wood Fiber Sci 2003;35(2):167–74.

[5] Caulfield DF, Clemons C, Jacobson RE, Rowell RM. Wood thermoplasticcomposites. In: Rowell RM, editor. Handbook of wood chemistry and woodcomposites. Boca Raton: CRC Press; 2005. p. 365–78.

[6] Dalväg H, Klason C, Strömvall HE. The efficiency of cellulosic fillers in commonthermoplastics II: filling with processing aids and coupling agents. Int J PolymMater 1985;11(1):9–38.

[7] Maldas D, Kokta BV, Daneault C. Influence of coupling agents and treatmentson the mechanical properties of cellulose fiber–polystyrene composites. J ApplPolym Sci 1989;37(3):751–75.

[8] Raj RG, Kokta BV, Daneault C. Use of wood fibers in thermoplastics. VII. Theeffect of coupling agents in polyethylene–wood fiber composites. J Appl PolymSci 1989;37(4):1089–103.

[9] Felix JM, Gatenholm P. The nature of adhesion in composites of modifiedcellulose fibers and polypropylene. J Appl Polym Sci 1991;42(3):609–20.

[10] Ichazo MN, Albano C, González J, Perera R, Candal MV. Polypropylene/woodflour composites: treatments and properties. Compos Struct 2001;54(2–3):207–14.

[11] Zaini MJ, Fuad MYA, Ismail Z, Mansor MS, Mustafah J. The effect of fillercontent and size on the mechanical properties of polypropylene/oil palm woodflour composites. Polym Int 1996;40(1):51–5.

[12] Stark NM, Berger MJ. Effect of particle size on properties of wood–flourreinforced polypropylene composites. In: Proceedings of 4th internationalconference on woodfiber-plastic composites, Madison; May, 1997. p. 134–43.

[13] Chen HC, Chen TY, Hsu CH. Effects of wood particle size and mixing ratios ofHDPE on the properties of the composites. Holz Roh Werkst2006;64(3):172–7.

[14] Bouafif H, Koubaa A, Perré P, Cloutier A. Effects of fiber characteristics on thephysical and mechanical properties of wood plastic composites. Compos Part A2009;40(12):1975–81.

[15] Migneault S, Koubaa A, Erchiqui F, Chaala A, Englund K, Wolcott MP. Effects ofprocessing method and fiber size on the structure and properties of wood–plastic composites. Compos Part A 2009;40(1):80–5.

[16] Yemele MCN, Koubaa A, Cloutier A, Soulounganga P, Wolcott M. Effect of barkfiber content and size on the mechanical properties of bark/HDPE composites.Compos Part A 2010;41(1):131–7.

[17] Renner K, Kenyó C, Móczó J, Pukánszky B. Micromechanical deformationprocesses in PP/wood composites: particle characteristics, adhesion,mechanisms. Compos Part A 2010;41(11):1653–61.

[18] Balasuriya PW, Ye L, Mai Y-W. Mechanical properties of wood flake–polyethylene composites. Part I: effects of processing methods and matrixmelt flow behaviour. Compos Part A 2001;32(5):619–29.

[19] Bledzki AK, Faruk O. Wood fibre reinforced polypropylene composites: effectof fibre geometry and coupling agent on physico-mechanical properties. ApplCompos Mater 2003;10(6):365–79.

[20] Hietala M, Niinimäki J, Oksman K. Processing of wood chip–plastic composites:effect on wood particle size, microstructure and mechanical properties. PlastRubber Compos 2011;40(2):49–54.

[21] Krogerus B, Fagerholm K, Löytynoja L. Analytical fractionation of pulps by tubeflow. Pap Puu–Pap Tim 2003;85(4):209–13.

[22] Laitinen O, Löytynoja L, Niinimäki J. Tube flow fractionator: a simple methodfor laboratory fractionation. Pap Puu–Pap Tim 2006;88(6):1–5.

[23] Lindholm CA, Kurdin J. Chemimechanical pulping. In: Sundholm J, editor.Papermaking science and technology. Book 5. MechanicalPulping. Jyväskylä: Fapet Oy; 1999. p. 223–49.

2116 M. Hietala et al. / Composites: Part A 42 (2011) 2110–2116

Paper IV

Bionanocomposites of thermoplastic starch and cellulose nanofibersmanufactured using twin-screw extrusion

Maiju Hietala a,b, Aji P. Mathew a, Kristiina Oksman a,⇑aDivision of Materials Science, Department of Engineering Sciences and Mathematics, Luleå University of Technology, SE 97187, Luleå, Swedenb Fibre and Particle Engineering Laboratory, Department of Process and Environmental Engineering, FI-90014 University of Oulu, Finland

a r t i c l e i n f o

Article history:Received 7 September 2012Received in revised form 24 October 2012Accepted 26 October 2012Available online xxxx

Keywords:Cellulose nanofibersThermoplastic starchBionanocompositesTwin-screw extrusionMechanical properties

a b s t r a c t

The aim of this study was to investigate if cellulose nanofiber (CNF) gels with high watercontents can be processed to nanocomposites with starch powder using continuoustwin-screw extrusion and to improve the mechanical properties and moisture sensitivityof thermoplastic starch. Nanocomposites with 0, 5, 10, 15 and 20 wt.% cellulose nanofibercontent were prepared. The characterization methods were conventional tensile testing,UV/Vis spectroscopy, scanning electron microscopy and moisture absorption. The cellulosenanofiber gel with high water content was mixed with starch powder, fed to the extruderas powder, performing the gelatinization of starch as well as the mixing of CNF in one step.The microscopy study showed that the CNF aggregated during the process and that thescrew configuration needs to be more distributive and dispersive to get homogeneousmaterial. The results showed that the addition of CNF improved the mechanical propertiesand had a positive effect on moisture uptake of the thermoplastic starch. Also, the translu-cency of the TPS/CNF composite films remained, even with high CNF content (20 wt.%).

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Interest in utilizing the unique properties of nanocellu-lose in various applications has been growing continuouslysince the first publications on fibrillation of wood pulp intocellulose nanofibers in the 1980s [1,2]. In addition to itsmechanical properties, the main benefit of nanosized cellu-lose is that it can be considered a sustainable material dueto its renewability, biodegradability and abundance in nat-ure [3,4].

Cellulose nanofibers are typically separated from ligno-cellulosic plants, such as wood and agricultural crops,using mechanical treatment [5]. Usually, lignin is also re-moved from the plant cell wall prior to fibrillation usingchemical treatment [5–7]. Depending on the source andseparation method used, cellulose nanofibers are typically20–40 nm in diameter and several micrometers long [3,8].

Unlike the cellulose nanowhiskers/crystals prepared byacid hydrolysis, cellulose nanofibers contain both theamorphous and crystalline regions of cellulose and cantherefore create entangled networks [5,9].

Because of the specific characteristics of cellulosenanofibers, such as the high surface area and aspect ratio,it is of great interest to use them to enhance the propertiesof polymers. However, the suitable methods and matricesfor composite manufacturing are somewhat limited dueto the fact that nanocellulose is hydrophilic and forms irre-versible aggregates when dried [3,10]. For this reason, themost successful methods used in the preparation of cellu-lose nanocomposites have been solution casting of diluteslurry of matrix and nanofibers [5,11–14] and resinimpregnation of a dried nanofiber network [15–18]. Forlarger scale production these methods are, however, slowand expensive. Extrusion compounding is one of the mostpromising methods for industrial processing due to easyscale-up and the possibility of further molding of the mate-rials [9,19,20]. Yet there are only few published studies onextrusion processing of cellulose nanocomposites

0014-3057/$ - see front matter � 2012 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

⇑ Corresponding author. Tel.: +46 920 493371; fax: +46 920 491399.E-mail addresses: maiju.hietala@ltu.se (M. Hietala), aji.mathew@ltu.se

(A.P. Mathew), kristiina.oksman@ltu.se (K. Oksman).

European Polymer Journal xxx (2012) xxx–xxx

Contents lists available at SciVerse ScienceDirect

European Polymer Journal

journal homepage: www.elsevier .com/locate /europol j

Please cite this article in press as: Hietala M et al. Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured usingtwin-screw extrusion. Eur Polym J (2012), http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

reinforced with cellulose nanofibers [20–22], which is whymore research is needed.

Thus, the aim of this work was to study the extrusionprocessing of bionanocomposites of thermoplastic starchand cellulose nanofibers. The processing as well as thenanofibers’ dispersion and effect on the mechanical, opticaland moisture absorption properties of the composites werestudied. Starch was chosen as the matrix polymer, as it canbe processed into thermoplastic starch in the presence ofwater [23], and therefore the cellulose nanofibers can beused without drying in the compounding step. Basically,native starch can be converted into thermoplastic starchunder heat and shear in the presence of plasticizer (typi-cally water and a polyol, such as glycerol or sorbitol)[24]. Under these conditions the starch granules aredestructurized, plasticized and melted, forming a materialwhich has a similar behavior to that of thermoplastics[24,25]. Earlier studies, in which the solution casting meth-od has been mainly used, have also shown that addition ofcellulose nanofibers improves the mechanical propertiesand moisture sensitivity of starch matrix [8,11,14,25–27].

2. Materials and methods

2.1. Materials

Softwoodwoodflour (ScandinavianWoodFiberAB,Orsa,Sweden)with a particle size range of 200–400 lmwas usedas a starting material for cellulose nanofibers. Sodiumchlorite (Sigma–Aldrich, Germany) and glacial acetic acid(Merck, Darmstadt, Germany) were used in the delignifica-tion of wood flour. A potato starch, ELIANE™ 100 (AVEBE,Veendam, Netherlands) with high amylopectin content(>99%) was used as matrix, D-Sorbitol (Merck, Darmstadt,Germany)was used as plasticizer for starch, and stearic acid(VWR, Leuven, Belgium) was used as a lubricant.

2.2. Preparation of nanofibers

Following the previous works of Abe et al. 2007,Iwamoto et al. 2008 and Gong et al. 2011 [28–30], thewood flour was delignified using acidified sodium chloritesolution using the Jayme-Wise method [31,32]. 250 g ofwood flour (dry mass) was treated in a flask containing3500 mL deionized water with 5 mL acetic acid and33.5 g of sodium chlorite at 70–75 �C. Additions of acetic

acid and sodium chlorite were continued at 2 h intervalsuntil the wood became white, which in this case required11 additions of acetic acid and sodium chlorite. The woodwas washed twice during the treatment and at the endwith minimum of 60 L of deionized water. Before washing,the wood was left in the acidified chlorite solution for 12 hat 70–75 �C. The total treatment time of the wood flour inacidified chlorite solution was 58 h.

The cellulose nanofibers (CNF) were obtained from del-ignified wood flour through mechanical fibrillation. A3 wt.% suspension was prepared from the delignified woodflour and deionized water, dispersed homogenously usinga laboratory shear mixer (Silverson L4RT, SilversonMachine Ltd., England), and passed twice through anultrafine grinder, a super masscolloider MKCA6-3 (MasukoSangyo Co, Ltd., Japan), using rotational speed of 1440 rpm.The total grinding time was 16 min. After grinding the drymatter content of the CNF suspension was increased bycentrifuging to 12 wt.%.

Fig. 1 shows the visual appearance of wood flour in dif-ferent stages of the preparation of cellulose nanofibersfrom macro- to nanoscale. As Fig. 1a and b show, the origi-nal wood flour consisted of small wood particles, and afterremoval of lignin by the sodium chlorite treatment, indi-vidual wood fibers were obtained. The ultrafine grindingtreatment of the bleached fibers resulted in a thick, viscousgel, and transmission electron microscopy (Fig. 1c) con-firmed that nanoscale fibers were obtained.

2.3. Processing of the materials

The TPS/CNF composites and neat TPS samples weremelt-compounded using a co-rotating twin-screw extru-der (ZSK-18 MEGALab, Coperion W&P, Stuttgart,Germany) with an L/D ratio of 40. The extruder barrelwas equipped with two atmospheric vents and vacuumventilation in purpose to remove the vaporized waterfrom the material. Feeding of the materials was donemanually due to the small amounts of the prepared pre-mixes (200 g). The extrusion set-up together with thescrew configuration and the temperature profile is shownin Fig. 2. The materials were extruded using a die with arectangular cross-section of 5 � 20 mm. The extrusionset-up used was similar to an earlier study made in ourlaboratory on thermoplastic starch reinforced with cellu-lose nanocrystals [33].

Fig. 1. Photographs and electron microscopy (SEM and TEM) images of cellulosic materials in preparation of cellulose nanofibers: (a) wood flour, (b)delignified wood flour, and (c) cellulose nanofibers after ultrafine grinding.

2 M. Hietala et al. / European Polymer Journal xxx (2012) xxx–xxx

Please cite this article in press as: Hietala M et al. Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured usingtwin-screw extrusion. Eur Polym J (2012), http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

Before extrusion, premixes of starch, sorbitol, stearicacid and cellulose nanofibers (12 wt.% concentration inwater) were made with a Waring laboratory blender usingshort blending pulses of 3 � 3–5 s. All the premixes were ina powdery state, despite the fact that the cellulose nanofi-ber suspension had high water content in order to avoidthe formation of CNF aggregates in the composites. TheCNF contents in the composites were 5, 10, 15 and20 wt.% based on the dry weight. The amount of sorbitolwas 30 wt.% and the amount of stearic acid was 1 wt.%based on dry starch. Stearic acid was added to preventmaterial sticking to the screws and clogging the die.

A speed of 200 rpm was used in the experiments, as itwas found to be high enough to transport the materialand to be able to have the vacuum venting on in the extru-der in order to remove moisture from the material. Higherrotational speeds, on the other hand, reduce the residencetime in the extruder, affecting the gelatinization of starchand the evaporation of water. In general, there were noproblems during the compounding process of the materi-als. The cellulose nanofiber content mainly affected theextrudates’ moisture content, as the materials with higherCNF content were stickier after extrusion due to higherwater content. Because of this, the extruded materialswere left to dry in room temperature and humidity(21 �C, 22% RH) for one week before compression molding.

2.4. Specimen preparation

After the extrusion, 10 g pieces of the extruded materi-als were compression molded into thin films with a thick-ness of �0.3 mm using a LPC-300 Fontijne Grotnes hotpress (Vlaardingen, Netherlands). The material was placedbetween polyethylene terephthalate films and metalplates, preheated in the press without pressure at 115 �Cfor 3 min, pressed with 10.5 MPa for 3 min at 115 �C, andthen cooled to 25 �C for 10 min under the same pressure.

2.5. Characterization

2.5.1. Mechanical testingFor tensile testing rectangular strips with the size of

80 � 5 � 0.3 mm were cut from the compression moldedfilms. All test specimens were conditioned for at leasttwo weeks in a desiccator of 53% relative humidity (RH)in room temperature before testing. The specimens were

kept in the desiccator until the moment they were goingto be tested, so as to minimize the variations in the mois-ture content. The tensile properties of the materials weretested using an Instron 4411 tensile testing machine witha 500 N load cell, gauge length of 40 mm and testing speedof 4 mm/min. At least six replicates of each material weretested.

2.5.2. Optical propertiesTransparency of the TPS and the TPS-CNF films was

studied using UV/Visible light spectroscopy. A PerkinElmer UV/Vis Spectrometer Lambda 2S (Überlingen,Germany) was used to measure the light transmittance ofthe films in a light wavelength area from 300 to1000 nm. The scan speed used in the analysis was240 nm/min, and three replicates of each material weremeasured.

2.5.3. MicroscopyScanning electron microscopy (JSM-6460, Jeol Ltd,

Japan; Zeiss ULTRA Plus, Carl Zeiss SMT AG, Germany)and transmission electron microscopy (HT-7700, Hitachi,Japan) were used to study the microstructure of cellulosicmaterials as well as the fracture surfaces of the thermo-plastic starch and the composites films. To create the frac-ture surfaces, specimens were first frozen under liquidnitrogen and then fractured. Specimens for SEMwere sput-ter-coated with a thin layer of gold or platinum beforeobservation.

2.5.4. Moisture absorptionThe effect of cellulose nanofibers on the kinetics of water

absorption of the specimens was studied according to theASTM D 5229 standard by exposing the samples to 98% rel-ative humidity and determining the water uptake as a func-tion of time until the moisture equilibrium content wasreached. Specimenswere cut from the compressionmoldedfilms, and the specimen size was 15 � 15 � 0.3 mm. Thefilm thickness 2 L was therefore considered thin enoughfor the molecular diffusion to be one-dimensional. Beforeabsorption experiments, specimens were dried overnightin an oven at 100 �C and then weighed. The average weightof the dried specimen was 72 mg (instead of the recom-mended 5.0 g in the standard due to lack of space). Theoven-dry samples were placed in a desiccator with a 98%RH. At specific time intervals the samples were removed

Fig. 2. Extrusion set-up used in the compounding of TPS/CNF composites.

M. Hietala et al. / European Polymer Journal xxx (2012) xxx–xxx 3

Please cite this article in press as: Hietala M et al. Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured usingtwin-screw extrusion. Eur Polym J (2012), http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

from the desiccator andweighed until an equilibriummois-ture value was reached. Five replicates of each materialwere used in the analysis. The moisture absorption valueswere calculated by taking the average of five values, andthe moisture diffusion coefficient was determined usingthe slope of the initial linear part of themoisture absorptioncurve, which was plotted using the averaged moistureabsorption values.

2.6. Statistical analysis

One-way analysis of variance (ANOVA) followed byTukey–Kramer multiple comparison tests with a 0.05significance level was used to analyze the results frommechanical testing.

3. Results and discussion

3.1. Mechanical properties of the materials

Typical stress–strain curves for the TPS/CNF films areshown in Fig. 3, and the results obtained from the tensiletesting are summarized in Table 1. From Table 1 it canbe seen that the strength of the thermoplastic starch filmsincreased with the addition of cellulose nanofibers. Thestrength increased with the cellulose nanofiber contentup to 10 wt.% of CNF, but the higher nanofiber concentra-tions did not increase the strength significantly. The reason

for this can be nanofiber agglomerates with higher CNFcontents, as dispersion of nanosized reinforcement in thematrix polymer is the main problem in melt compoundingof nanocomposites [9,19]. Nevertheless, with the additionof 10 wt.% of CNF, the strength of a neat thermoplasticstarch nearly doubled (8.8 and 16.4 MPa). The Young’smodulus increased linearly with the increasing cellulosenanofiber content, and the material containing 20 wt.% ofCNF had the highest modulus of elasticity. When comparedto the neat thermoplastic starch, the modulus increasedfrom 455 to 1317 MPa (190%) with the addition of20 wt.% cellulose nanofibers. As expected, the elongationat break reduced with the addition of cellulose nanofibers(Table 1, Fig. 3).

3.2. Optical properties

Transparency can be used as an indirect measure of thesize and dispersion of cellulose nanofibers in the matrix. Ifthe cellulose reinforcement is not in nanoscale (non-fibrillated or aggregated), the transparency is lower dueto the increased light scattering [16,34]. Fig. 4 shows theresults from the UV/Vis spectroscopy for the TPS andTPS/CNF films. The addition of cellulose nanofibers reducedthe light transmittance of the prepared films, indicatingthat the dispersion of CNF was not homogeneous in thecomposite films. At 550 nm light wavelength, which isapproximately in the center of the visible light spectrum,the transmittance was reduced by 27%, 28%, 41% and 54%with 5, 10, 15 and 20 wt.% addition of CNF, respectively,when compared to the neat TPS matrix.

The transparency of the films was also examined visu-ally against a background image (Fig. 5). From Fig. 5 itcan be seen that all the films can be considered translu-cent, and up to 15 wt.% of CNF, there is no great reductionin the clarity of the films. In addition, no visible aggregatesof cellulose nanofibers can be seen by the eye in any of thefilms.

3.3. Microscopy

The SEM images from the fracture surfaces of the ther-moplastic starch (TPS) and TPS/CNF composites with 5, 10,15 and 20 wt.% of CNF are shown in Fig. 6. The images weretaken with a magnification of 500 to show the overview ofthe fracture surfaces. From Fig. 6 it can be seen that the

Fig. 3. Typical stress–strain curves of the TPS and TPS/CNF films.

Table 1Mechanical properties of the films of thermoplastic starch (TPS) andthermoplastic starch reinforced with cellulose nanofibers (CNF) togetherwith standard deviations.

Material Tensile strength*

(MPa)Young’s modulus*

(MPa)Elongation atbreak* (%)

TPS 8.8 ± 0.7a 455.3 ± 39.2a 23.0 ± 5.5a

5%CNF 11.8 ± 0.6b 622.9 ± 34.8b 8.9 ± 2.9b

10%CNF 16.4 ± 1.1c 914.9 ± 125.0c 8.8 ± 2.0b

15%CNF 16.8 ± 2.1c 1056.0 ± 99.9c 3.0 ± 0.9c

20%CNF 17.5 ± 2.1c 1317.0 ± 134.7d 1.9 ± 0.4c

* Means marked with the same superscript letter within the samecolumn are not significantly different at 5% significance level based on theANOVA and Tukey–Kramer pairwise comparison test. Fig. 4. UV/Vis transmittance spectra for the TPS and TPS/CNF films.

4 M. Hietala et al. / European Polymer Journal xxx (2012) xxx–xxx

Please cite this article in press as: Hietala M et al. Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured usingtwin-screw extrusion. Eur Polym J (2012), http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

dispersion of cellulose nanofibers has not been uniformand aggregates of CNF can be found in all composite mate-rials. Due to the increasing amount of cellulose nanofibersin the composites with 15 and 20 wt.% of CNF the amountof aggregates also increases, which is most likely the rea-son why the tensile strength did not increase in the caseof these materials.

The cellulose nanofibers were used in 12 wt.% watersuspension in the composite preparation in order to avoidthe formation of nanofiber aggregates. However, as theSEM images show, it seems that this water content wasinadequate, as aggregates were formed. It is also possiblethat the water evaporated too fast from the material inextrusion, causing the aggregation of CNF. In addition,modification of the screw configuration to more shearingcould favor uniform dispersion of cellulose nanofibers.

3.4. Moisture absorption

The results from the moisture absorption study, themoisture equilibrium contents and the moisture diffusioncoefficients, are shown in Table 2. As Table 2 shows, therewere no significant differences between the equilibriummoisture contents of the materials despite the increasingamount of cellulose nanofibers, as reported elsewhere forTPS/CNF composites [11,25]. Similar type of behavior wasobserved by Mathew and Dufresne 2002 when studyingthe moisture absorption of sorbitol plasticized thermoplas-tic starch reinforced with cellulose nanowhiskers [35].When sorbitol is used as the plasticizer, it is believed to re-strict the moisture uptake of thermoplastic starch morethan glycerol due to higher chain length and fewer end hy-droxyl groups [35]. Because sorbitol was also used as the

Fig. 5. Visual appearance of the TPS and TPS/CNF films.

Fig. 6. Fracture surfaces of the TPS and TPS/CNF composites with 5, 10, 15 and 20 wt.% of cellulose nanofibers.

M. Hietala et al. / European Polymer Journal xxx (2012) xxx–xxx 5

Please cite this article in press as: Hietala M et al. Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured usingtwin-screw extrusion. Eur Polym J (2012), http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

starch plasticizer in this study, it is most probably the rea-son why there were no significant differences between themoisture equilibrium contents of the TPS and TPS/CNFcomposites.

The moisture diffusion coefficients indicate thatthe addition of cellulose nanofibers reduced the moisturediffusion of thermoplastic starch (Table 2). The moisturediffusion coefficient was greatest for the neat TPS, and itwas decreasing with the increasing amount of cellulosenanofibers. Therefore, we can say that the addition ofcellulose nanofibers restricts the moisture diffusion inthermoplastic starch. The reduction of moisture absorptionof two hydrophilic materials, starch and cellulose, can beexplained by several factors: (1) moisture penetration isreduced by the good interfacial adhesion between celluloseand starch [26,36]; (2) cellulose is less hygroscopic thanstarch due to higher degree of crystallinity, which is whythe moisture absorption reduces with addition of CNF[11,26], and (3) formation of fibrous network of celluloseprevents the swelling of the starch matrix and thus reducesthe moisture penetration [25,36].

4. Conclusions

The aim of this work was to study the extrusion pro-cessing of cellulose nanocomposites and the properties ofthe prepared composites. Cellulose nanofibers were usedin wet state (12 wt.% water suspension) in the processingof the composites together with a thermoplastic starchmatrix. The composites were successfully compoundedusing a twin-screw extruder, and the study shows that itis possible to use cellulose nanofibers with high water con-tent in extrusion of bionanocomposites.

Mechanical testing of the nanocomposites showed thatthe tensile strength and modulus of the starch matrixwere improved, and the moisture sensitivity was reducedwith the addition of cellulose nanofibers. The tensilemodulus increased linearly with the increasing nanofibercontent, but the strength properties were highest for thematerial with 10 wt.% of cellulose nanofibers. Scanningelectron microscopy study revealed aggregates of cellu-lose nanofibers in the composites, especially in the caseof composites with 15 and 20 wt.% of CNF, which explainswhy there was no improvement in the strength propertiesof these composites. Also, the UV/Vis spectroscopyindicated the presence of CNF aggregates, as the lighttransmittance of the composite films was reduced byaddition of nanofibers. However, the films maintainedtheir translucency, even with 20 wt.% of cellulose nanofi-ber content.

Acknowledgements

The authors would like to thank the EuropeanCommission (contract number NMP2-LA-2008-210037WOODY) for financial support, AVEBE, Netherlands forkindly providing the potato starch used in the study, andDr. Mehdi Jonoobi for his help with TEM study.

References

[1] Turbak AF, Snyder FW, Sandberg KR. Microfibrillated cellulose, a newcellulose product: properties, uses, and commercial potential. J ApplPolym Sci: Appl Polym Symp 1983;37:815–23.

[2] Herrick FW, Casebier RL, Hamilton JK, Sandberg KR. Microfibrillatedcellulose: morphology and accessibility. J Appl Polym Sci: ApplPolym Symp 1983;37:797–813.

[3] Siró I, Plackett D. Microfibrillated cellulose and new nanocompositematerials: a review. Cellulose 2010;17(3):459–94.

[4] 4. Spence K, Habibi Y, Dufresne A. Nanocellulose-Based Composites.In: Kalia S, Kaith BS, Kaur I, editors. Cellulose Fibers: Bio- and nano-polymer composites. green chemistry and technology, SpringerBerlin Heidelberg, 2011. p. 179-213.

[5] Azizi Samir MAS, Alloin F, Paillet M, Dufresne A. Tangling effect infibrillated cellulose reinforced nanocomposites. Macromolecules2004;37(11):4313–6.

[6] Hubbe MA, Rojas OJ, Lucia LA, Sain M. Cellulosic nanocomposites: areview. BioRes 2008;3(3):929–80.

[7] Chen W, Yu H, Liu Y, Chen P, Zhang M, Hai Y. Individualization ofcellulose nanofibers from wood using high-intensity ultrasonicationcombined with chemical pretreatments. Carbohydr Polym2011;83(4):1804–11.

[8] Svagan AJ, Azizi Sami MAS, Berglund L. Biomimetic polysaccharidenanocomposites of high cellulose content and high toughness.Biomacromolecules 2007;8(8):2556–63.

[9] Oksman K, Mathew AP, Sain M. Novel bionanocomposites:processing, properties and potential applications. Plast, RubberCompos 2009;38(9/10):396–404.

[10] Klemm D, Kramer F, Moritz S, Lindström T, Ankerfors M, Gray D,et al. Nanocelluloses: a new family of nature-based materials.Angew Chem Int Ed 2011;50(24):5438–66.

[11] Dufresne A, Vignon MR. Improvement of starch film performancesusing cellulose microfibrils. Macromolecules 1998;31(8):2693–6.

[12] Malainine ME, Mahrouz M, Dufresne A. Thermoplasticnanocomposites based on cellulose microfibrils from Opuntia ficus-indica parenchyma cell. Compos Sci Technol 2005;65(10):1520–6.

[13] Bhatnagar A, Sain M. Processing of cellulose nanofiber-reinforcedcomposites. J Reinf Plast Compos 2005;24(12):1259–68.

[14] Alemdar A, Sain M. Biocomposites from wheat straw nanofibers:morphology, thermal and mechanical properties. Compos SciTechnol 2008;68(2):557–65.

[15] Nakagaito AN, Yano H. The effect of morphological changes frompulp fiber towards nanoscale fibrillated cellulose on the mechanicalproperties of high-strength plant fiber based composites. Appl PhysA 2004;78(4):547–52.

[16] Yano H, Sugiyama J, Nakagaito AN, Nogi M, Matsuura T, Hikita M,et al. Optically transparent composites reinforced with networks ofbacterial nanofibers. Adv Mater 2005;17(2):153–5.

[17] Iwamoto S, Nakagaito AN, Yano H, Nogi M. Optically transparentcomposites reinforced with plant fiber-based nanofibers. Appl PhysA 2005;81(6):1109–12.

[18] Nakagaito AN, Yano H. Novel high-strength biocomposites based onmicrofibrillated cellulose having nano-order-unit web-like networkstructure. Appl Phys A 2005;80(1):155–9.

[19] Oksman K, Mathew AP, Bondeson D, Kvien I. Manufacturing processof cellulose whiskers/polylactic acid nanocomposites. Compos SciTechnol 2006;66(15):2776–84.

[20] Mathew AP, Chakraborty A, Oksman K, Sain M. The structure andmechanical properties of cellulose nanocomposites prepared bytwin screw extrusion. In: Sain M, Oksman K, editors. Cellulosenanocomposites: processing, characterization and properties,Oxford: Oxford University Press; 2006. p. 114-131.

[21] Jonoobi M, Harun J, Mathew AP, Oksman K. Mechanical properties ofcellulose nanofiber (CNF) reinforced polylactic acid (PLA) preparedby twin screw extrusion. Compos Sci Technol 2010;70(12):1742–7.

[22] Gong G, Pyo J, Mathew AP, Oksman K. Tensile behavior, morphologyand viscoelastic analysis of cellulose nanofiber-reinforced (CNF)polyvinyl acetate (PVAc). Composites Part A 2011;42(9):1275–82.

Table 2Moisture equilibrium contents and moisture diffusion coefficients for thethermoplastic starch (TPS) and TPS/cellulose nanofiber (CNF) films.

Material Moisture equilibriumcontent (%)

Moisture diffusioncoefficient (cm2/s)

TPS 53.9 ± 1.3 1.66 � 10�9

5%CNF 51.6 ± 2.5 1.50 � 10�9

10%CNF 51.4 ± 2.1 1.43 � 10�9

15%CNF 51.8 ± 2.7 1.31 � 10�9

20%CNF 52.3 ± 3.8 1.27 � 10�9

6 M. Hietala et al. / European Polymer Journal xxx (2012) xxx–xxx

Please cite this article in press as: Hietala M et al. Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured usingtwin-screw extrusion. Eur Polym J (2012), http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

[23] Stepto RFT. The processing of starch as a thermoplastic. MacromolSymp 2003;201(1):203–12.

[24] Avérous L, Halley PJ. Biocomposites based on plasticized starch.Biofuels, Bioprod Biorefin 2009;3(3):329–43.

[25] Dufresne A, Dupeyre D, Vignon MR. Cellulose microfibrils frompotato tuber cells: processing and characterization of starch–cellulose microfibril composites. J Appl Polym Sci2000;76(14):2080–92.

[26] Kaushik A, Singh M, Verma G. Green nanocomposites based onthermoplastic starch and steam exploded cellulose nanofibrils fromwheat straw. Carbohydr Polym 2010;82(2):337–45.

[27] Savadekar NR, Mhaske ST. Synthesis of nanocellulose fibers andeffect on thermoplastics starch based films. Carbohydr Polym2012;89(1):146–51.

[28] Abe K, Iwamoto S, Yano H. Obtaining cellulose nanofibers with auniform width of 15 nm from wood. Biomacromolecules2007;8(10):3276–8.

[29] Iwamoto S, Abe K, Yano H. The effect of hemicelluloses on wood pulpnanofibrillation and nanofiber network characteristics.Biomacromolecules 2008;9(3):1022–6.

[30] Gong G, Mathew AP, Oksman K. Strong aqueous gels of cellulosenanofibers and nanowhiskers isolated from softwood flour. Tappi J2011;10(2):7–14.

[31] Wise LE, Murphy M, D’Addieco AA. Chlorite holocellulose, itsfractionation and bearing on summative wood analysis and onstudies on the hemicelluloses. Paper Trade J 1946;122(2):35–43.

[32] Cullen LE, MacFarlane C. Comparison of cellulose extraction methodsfor analysis of stable isotope ratios of carbon and oxygen in plantmaterial. Tree Physiol 2005;25(5):563–9.

[33] Visakh PM, Mathew AP, Oksman K, Thomas S. Starch-basedbionanocomposites: processing and properties. In: Habibi Y, LuciaLA, editors. Hoboken. New Jersey: John Wiley & Sons, Inc.; 2012. p.287–306.

[34] Besbes I, Vilar MR, Boufi S. Nanofibrillated cellulose from Alfa,Eucalyptus and Pine fibres: preparation, characteristics andreinforcing potential. Carbohydr Polym 2011;86(3):1198–206.

[35] Mathew AP, Dufresne A. Morphological investigation ofnanocomposites from sorbitol plasticized starch and tunicinwhiskers. Biomacromolecules 2002;3(3):609–17.

[36] Svagan AJ, Hedenqvist MS, Berglund L. Reduced water vapoursorption in cellulose nanocomposites with starch matrix. ComposSci Technol 2009;69(3–4):500–6.

M. Hietala et al. / European Polymer Journal xxx (2012) xxx–xxx 7

Please cite this article in press as: Hietala M et al. Bionanocomposites of thermoplastic starch and cellulose nanofibers manufactured usingtwin-screw extrusion. Eur Polym J (2012), http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

Paper V

1

One-step extrusion processing of green biocomposites: compounding,

fibrillation efficiency and fiber dispersion

Maiju Hietalaa,b, Pierre Rolloa, Kaarina Kekäläinenb, Kristiina Oksmana,*

aDivision of Materials Science, Department of Engineering Sciences and Mathematics, Luleå

University of Technology, SE- 97187, Luleå, Sweden

bFibre and Particle Engineering Laboratory, Department of Process and Environmental

Engineering, University of Oulu, FI-90014 Oulu, Finland

*Corresponding author: email kristiina.oksman@ltu.se (K. Oksman) Tel.: +46 920 493371,

fax: +46 920 491399.

E-mail addresses:

maiju.hietala@ltu.se (M. Hietala), pierol-1@student.ltu.se (P. Rollo),

kaarina.kekalainen@oulu.fi (K. Kekäläinen), kristiina.oksman@ltu.se (K. Oksman)

2

Abstract

The aim of this study was to examine if twin-screw extrusion process can be used to fibrillate

cellulose fibers into micro- or nanosize fibers and at the same time prepare green

bionanocomposites of thermoplastic starch with 10 wt% fiber content. The effect of the

processing set-up on micro/nanofibrillation and fiber dispersion/distribution in starch was

studied. Two different types of cellulose fibers were used; namely, bleached wood fibers and

TEMPO-oxidized cellulose fibers, in order to examine the effect of fiber treatment on

fibrillation efficiency. In addition, a composite with cellulose nanofibers was prepared to

examine the nanofiber distribution and dispersion in the starch as well as to compare the

properties with the composites containing bleached fibers and TEMPO-oxidized fibers.

Microscopic methods (optical microscopy and scanning electron microscopy) were used to

examine how the extrusion process affected the fiber size as well as their distribution and

dispersion in the starch matrix. As indirect measures of the fibrillation and fiber dispersion in

the composites, their mechanical properties, transparency and moisture uptake were

measured.

Keywords:

Biocomposites; Nanocomposite; Cellulose fibers; Cellulose Nanofibers; Thermoplastic starch;

Twin-screw extrusion; Microscopy

3

1 Introduction

Renewability, biodegradability, abundance in nature, non-toxicity, high mechanical

properties, high specific surface area; these are some of the features that give nanocellulose

such great potential for use in a wide variety of applications ranging from green packaging to

medical materials. Depending on the source and isolation method used, cellulose nanofibers

are typically 10-100 nm in diameter and several micrometers long [Herrick et al. 1983,

Besbes et al. 2011] and, as opposed to the cellulose nanowhiskers prepared by the acid

hydrolysis method, they contain both the amorphous and crystalline regions of cellulose

[Oksman et al. 2009].

Cellulose nanofibers are usually separated mechanically from lignocellulosic plants,

such as wood and agricultural crops, after removal of lignin and other plant cell wall

substances chemically [Azizi Samir et al. 2004, Gong et al 2011]. The most typical methods

currently used in the laboratory -scale preparation of cellulose nanofibers are high-pressure

homogenization [Turbak et al. 1983, Nakagaito & Yano 2004, Seydibeyoglu & Oksman

2008] and ultra-fine grinding [Taniguchi & Okamura 1998, Iwamoto et al. 2005, Jonoobi et

al. 2012]. However, in these methods the amount of energy needed to produce nanocellulose

is rather high. Consequently, one of the main obstacles hindering the use of cellulose

nanofibers in commercial applications is the high energy consumption needed to isolate

nanofibers [Spence et al. 2011]. Hence, several researchers have focused on studying ways to

reduce the energy consumption used in the isolation process. Some of the most successful

methods have been pre-treating cellulose chemically [Saito et al. 2006, Saito et al. 2007,

Wagberg et al. 2008) or enzymatically prior mechanical nanofibrillation treatment

[Janardhnan & Sain 2006, Henriksson et al. 2007, Pääkkö et al. 2007]. In addition, there are

studies showing that less-conventional methods, such as the twin-screw extrusion, can be used

in defibration of wood fibers from wood particles [Hietala et al. 2011a,b], or even as energy-

efficient methods in isolation of cellulose micro or nanofibers from wood pulp [Lee et al.

2009, Nakagaito et al. 2010].

Processing native starch granules together with a plasticizer into thermoplastic starch

under heat and shear is already a well-known method [Stepto 2003, Averous & Halley 2009],

and in our earlier study we have shown that preparation of thermoplastic starch can be done in

a single extrusion step as compounding of cellulose nanocomposites [Hietala et al. 2012].

Thus, in this study the aim was to combine the processing of starch into thermoplastic starch

4

as well as the composite compounding and nanofibrillation of cellulose in one extrusion step

in order to manufacture green biocomposites of thermoplastic starch and cellulose fibers.

To study the effect of fiber treatment of the nanofibrillation, two different types of

cellulose fibers were used as the cellulosic raw material; namely, bleached softwood fibers

and TEMPO-oxidized hardwood fibers. The TEMPO-oxidized fibers were used, as the

TEMPO-oxidation method has been shown to greatly reduce the mechanical energy needed

for nanofibrillation, even to the extent that a blender [Saito et al. 2006] or a magnetic stirrer

[Saito et al 2007] can be used to isolate nanofibers after the oxidation treatment. In principle,

negatively charged carboxylate groups are introduced to the surface of cellulose microfibrils,

creating repulsive forces between them and, thus, weakening the structure of the cellulose

fiber in the TEMPO-oxidation method [Saito et al. 2006, Saito et al 2007].

In this study, the ability to use twin-screw extrusion process to fibrillate cellulose

fibers into micro- or nanosize fibers and at the same time prepare green bionanocomposites of

thermoplastic starch with 10wt% fiber content was studied. Microscopic methods (optical

microscopy and scanning electron microscopy) were used to examine the fiber size after

extrusion and fiber distribution in the composites. As indirect measurements of the fiber size,

dispersion and distribution, the properties of the prepared composites were analyzed by means

of tensile testing, UV/Vis spectroscopy and moisture uptake analysis and compared to those

of a composite reinforced with 10 wt% of cellulose nanofibers.

2 Materials and methods

2.1 Materials

Softwood wood flour (Scandinavian Wood Fiber AB, Orsa, Sweden) with a particle size

range of 200-400 µm was used as a starting material for bleached cellulose fibers and

cellulose nanofibers (Fig. 1). Sodium chlorite (Sigma-Aldrich, Germany) and glacial acetic

acid (Merck, Germany) were used in the delignification of wood flour. Never-dried bleached

hardwood kraft pulp (Betula pendula) was used to prepare the TEMPO-oxidized cellulose

fibers (Fig. 1). 2,2,6,6-tetramethylpiperidine-1-oxyl radical (TEMPO), sodium hypochlorite,

sodium bromide and sodium hydroxide used in the TEMPO-oxidation were purchased from

Sigma-Aldrich (Germany).

5

Cellulose nanofibers (CNF) were obtained from delignified (bleached) softwood fibers

through mechanical fibrillation, using an ultrafine grinder, a super masscolloider MKCA6-3

(Masuko Sangyo Co, Ltd., Japan), at a rotational speed of 1440 rpm. The total grinding time

was 16 minutes.

A potato starch, ELIANETM 100 (AVEBE, Netherlands) with high amylopectin content

(>99%) was used as matrix; D-Sorbitol (Merck, Germany) was used as plasticizer for starch,

and stearic acid (VWR, Leuven, Belgium) was used as a lubricant.

Fig. 1. Used raw materials: a) cellulose nanofibers, b) TEMPO-oxidized fibers, c) bleached

wood fibers.

2.2 Preparation of the bleached wood fibers

Following the same procedure used in our previous work [Hietala et al. 2012], the wood flour

was delignified with acidified sodium chlorite solution using the Jayme-Wise method to

obtain bleached wood fibers. 250 g of wood flour (dry mass) was treated in a flask containing

3500 mL deionized water with 5 mL acetic acid and 33.5 g of sodium chlorite at 70-75°C.

Additions of acetic acid and sodium chlorite were continued at 2-hour intervals until the wood

became white, which in this case required 11 additions of acetic acid and sodium chlorite. The

wood was washed twice during the treatment and at the end with a minimum of 60L of

deionized water. Before washing, the wood was left in the acidified chlorite solution for 12h

at 70-75°C. The total treatment time of the wood flour in acidified chlorite solution was 58h.

6

2.3 Preparation of TEMPO-oxidized cellulose fibers

The TEMPO-mediated oxidation was performed under alkaline conditions according to the

procedure described by Saito et al. 2006 and 2007. A 50 g cellulose sample was diluted to 1%

consistency and 0.1 mmol/g of TEMPO and 1 mmol/g of NaBr were added to the suspension.

The reaction was started by adding 10% NaClO solution and the pH was adjusted to 10-10.5

and maintained at this level with 0.5 M NaOH during the oxidation. Total amount of NaClO

used was 10 mmol/g. The reaction was finished after 5.5 hours and the oxidized sample was

washed with deionized water until the conductance of the filtrate reached a value under 10

µC/cm. The carboxyl content of the oxidized sample was 1.3 mmol/g, and it was determined

using conductometric titration in duplicate according to Araki et al. 2001 and Katz et al. 1984.

2.4 Composite processing

The neat thermoplastic starch and cellulose fiber-reinforced samples were melt-compounded

using a co-rotating twin-screw extruder (ZSK-18 MEGALab, Coperion W&P, Stuttgart,

Germany) with an L/D ratio of 40. The extruder barrel was equipped with two atmospheric

vents to remove the vaporized water during the process. Feeding of the materials was done

manually due to the small amounts of the prepared premixes (150-200 g). The extrusion set-

up is shown in Fig. 2. A die with a rectangular cross-section of 5 mm x 20 mm was used in

the experiments.

Fig. 2. The extrusion set-up used in the experiments.

The screw configuration used in the experiments is shown in Fig. 3. The screw configuration

was designed for high shear forces in order to facilitate the fibrillation of nanofibers. When

compared to the regular screw configuration used in our laboratory for composite

7

compounding, this configuration consisted of several extra kneading elements, reversed screw

elements (red elements in Fig. 3) and distributing elements.

Fig. 3 The screw configuration used in the experiments.

Before extrusion, premixes of starch, sorbitol, stearic acid and 10 wt% cellulose fibers were

prepared using a Waring laboratory blender with short blending pulses of 3 x 3-5 s. Due to the

low dry matter content of the TEMPO-oxidized fiber suspension (3.4 wt%), moisture content

of the premix with TEMPO-oxidized fibers was 74 wt%; therefore, the water content of the

other premixes was adjusted to the same level. The sorbitol content was 30 wt% and the

content of stearic acid was 0.7 wt%, both based on the dry weight of starch. The stearic acid

was added as a processing aid, preventing the compound from sticking to the screws or

clogging the die. The processing speed was set to 300 rpm because high screw speed affects

the distribution of the fibers in the matrix positively. On the other hand, increased processing

speed will reduce the residence time, affecting the gelatinization of starch and also the

evaporation of water; therefore, a speed higher than 300 rpm was not used.

In general, there were no problems during the compounding process of the composite

materials; however, the high water content in the premixes resulted in low viscosity melt and

it was therefore not possible to use vacuum ventilation, which would have been a more

efficient way to remove the moisture from the extrudates. The moisture content of the

extruded materials was reduced after the process by drying the materials in room temperature

and humidity during 2-3 days before compression molding of the sheets/films.

2.5 Compression molding

After drying, the extruded materials were compression molded to a thickness of ~0.2 mm

using an LPC-300 Fontijne Grotnes hot press (Vlaardingen, Netherlands). A 7-g piece of the

material was placed between polyethylene terephthalate films and metal plates, preheated in

contact mode without pressure at 110°C for 3 minutes, pressed with 7.4 MPa for 3 minutes at

110°C, and then cooled to 25°C for 10 minutes under the same pressure.

8

2.6 Characterization methods

2.6.1 Microscopy

Scanning electron microscopy (JSM-6460, Jeol Ltd, Japan) was used to study the fracture

surfaces of the neat TPS and the composite films. To create the fracture surfaces, specimens

were first frozen under liquid nitrogen and then fractured. Specimens for SEM were sputter-

coated with a thin layer of gold before observation. Optical microscopy (Dialux 20, Leitz,

Germany and ECLIPSE MA200, Nikon, Japan) was used to visually observe the distribution

and size of the cellulose fibers in the composite films.

2.6.2 Fiber characteristics after extrusion

After extrusion, the thermoplastic starch matrix was extracted from the composites to be able

to examine how the extrusion process affected the size and morphology of the fibers using

optical microscopy. A small amount of the composite was mixed with water, and the mixture

was heated and kept at 90-95°C for approximately 4h under stirring with a magnetic stirrer.

The solution was then filtrated in order to collect the fibrous material.

2.6.3 Mechanical testing

For the tensile testing, rectangular strips measuring 80 mm x 5 mm x 0.2 mm were cut from

the compression-molded films. All test specimens were conditioned for at least 10 days in a

desiccator at 50% relative humidity (RH) in room temperature before testing. The specimens

were kept in the desiccator until the moment they were going to be tested, to minimize the

variations in the moisture content. The tensile properties of the materials were tested using an

Instron 4411 tensile testing machine with a 500 N load cell, gauge length of 40 mm and

testing speed of 4 mm/min. At least six replicates of each material were tested.

9

2.6.4 Optical properties

The transparency of the neat TPS and the composite films was studied using a Perkin Elmer

UV/Vis Spectrometer Lambda 2S (Überlingen, Germany). The light transmittance of the films

was measured in a light wavelength area from 300 to 1000 nm at a scanning speed of 240

nm/min. Three replicates of each material were measured.

2.6.5 Moisture uptake

The moisture uptake of the prepared materials was studied by exposing the samples to 98%

relative humidity. Specimens were cut from the compression-molded films, and the specimen

size was 40 mm x 40 mm x 0.4 mm. Four replicates of each material were used in the

analysis. First, the specimens were dried overnight in an oven at 65°C and then weighed. The

average weight of the dried specimen was 0.7 g. The oven-dry samples were placed in a

desiccator with 98% RH, and after 16 days the specimens were removed from the desiccator

and weighed. The water uptake was determined as the gain of weight during this time. 16

days was expected to be enough for the material to reach the moisture uptake at equilibrium.

Due the degradation of the neat TPS film it was difficult to continue the experiment any

longer.

2.6.6 Statistical analysis

One-way analysis of variance (ANOVA) followed by Tukey-Kramer multiple comparison

tests with a 0.05 significance level was used to analyze the results from mechanical testing.

10

3 Results and discussion

3.1 Microscopy of the composites

The SEM images of the fracture surfaces of the thermoplastic starch (TPS) and the

composites with 10% of cellulose nanofibers, TEMPO-oxidized fibers and bleached cellulose

fibers are shown in Figs. 4 and 5. Fig. 4 shows an overview of the fractured surfaces and more

details can be seen in Fig. 5. From Figs. 4c) and 4d) it can be seen that the fibers in the

composites were several micrometers in size, indicating that the shear forces in twin-screw

extrusion have not been enough to fibrillate the bleached wood fibers or TEMPO-oxidized

fibers into nanofibers. In both cases, the fibers seen in the fractured surfaces are much larger

in size in comparison with those in the composite containing cellulose nanofibers (Fig. 4 b).

When comparing the fracture surfaces of the composites with bleached wood fibers (Fig. 4d)

and TEMPO-oxidized fibers (Fig. 4c), the TEMPO-oxidized fibers appear to be smaller.

However, the reason for this is most likely the TEMPO-oxidation treatment as well as the

different source of fibers (birch fibers). Birch fibers typically have diameters of 20-36 µm,

whereas softwood fibers such as spruce and pine have average diameters ranging from 25 µm

to 45 µm [Smook 1999, p. 15].

Fig. 4. Fracture surfaces of a) neat thermoplastic starch (TPS), b) TPS reinforced with

cellulose nanofibers (CNF), c) TPS reinforced with TEMPO-oxidized fibers (TEMPO), and d)

TPS reinforced with bleached wood fibers (BL.WF).

11

Fig. 5. Details of the fracture surfaces of a) neat thermoplastic starch (TPS), b) TPS reinforced

with cellulose nanofibers (CNF), c) TPS reinforced with TEMPO-oxidized fibers (TEMPO),

and d) TPS reinforced with bleached wood fibers (BL.WF).

From Fig. 5 more details from the fracture surfaces of the composites can be seen. In the case

of TPS reinforced with cellulose nanofibers (CNF), some aggregation of nanofibers can be

seen (Fig. 5b), even though the dispersion appears to be rather good (Fig. 4.). In Fig. 5c it can

be seen that some gaps exist between the TEMPO-oxidized fibers and TPS matrix, perhaps

due to the swelling of fibers as a result of the TEMPO-oxidation treatment, which could

impair the mechanical properties of this material. When comparing the fractured surface of

the composite with TEMPO-oxidized fibers (5c) and bleached wood fibers (Fig. 5d), it is

clear that the bleached wood fibers have greater diameters.

To examine the morphology and the dispersion of the cellulose fibers in the TPS matrix more

closely, optical microscopy was used. The optical microscopy images of the composite films

are presented in Fig. 6. The distribution of fibers was rather good in all the composites,

though in the material made with 10% bleached wood fibers some less homogenous areas

could be seen (Fig. 6c). The observation made from the SEM images can be confirmed from

the optical microscopy images: the bleached wood fibers and TEMPO-oxidized fibers have

not been fibrillated into nanofibers and, also in the film of TPS reinforced with cellulose

nanofibers, some non-fibrillated fibers can be seen.

12

Fig. 6. Optical microscopy images of the TPS films reinforced with 10wt% of cellulose fibers:

a) cellulose nanofibers, b) TEMPO-oxidized fibers, and c) bleached wood fibers.

3.2 Fiber characteristics after extrusion

After extrusion, the thermoplastic starch matrix was removed from the composites to be able

to examine the size and morphology of the fibers more closely. Fig. 7 shows the optical

microscopy images of the bleached wood fibers and TEMPO-oxidized fibers before and after

twin-screw extrusion. As Fig. 7 shows, the fiber length and diameter do not seem to be

affected much in the case of bleached cellulose fibers. The TEMPO-oxidized fibers, on the

other hand, appear to be much shorter when compared to the fibers before extrusion,

indicating that the aspect ratio would be lower than with bleached wood fibers.

A possible explanation for not obtaining nanofibers using these extrusion conditions can be

the high water content of the used premixes (74%) resulting in low viscosity of the starch-

fiber melt inside the extruder, as the shear stress in a twin-screw extruder is affected by the

viscosity of the melt, shear rate, shear stress and the duration of the stress (Kirchhof 2008, p.

167).

13

Fig. 7. Bleached wood fibers and TEMPO-oxidized fibers before compounding with starch

and after compounding and removal of the TPS matrix. The images from TEMPO-oxidized

fibers were taken using a polarizing filter.

3.3 Mechanical properties of the materials

Typical stress-strain curves of the films of neat thermoplastic starch and TPS reinforced with

10% of cellulose nanofibers (CNF), TEMPO-oxidized fibers (TEMPO) and bleached wood

fibers (BL.WF) are shown in Fig. 8, and the results obtained from the tensile testing are

summarized in Table 1. From Table 1 it can be seen that the strength and the elastic modulus

of the thermoplastic starch films was increased with the addition of all three types of cellulose

fibers used in the study (10 wt% fiber content in the composites). The modulus was increased

by 27-34 % and the strength by 12-16 % from the neat thermoplastic starch. However, quite

unexpectedly, the type of cellulose fiber (bleached wood fibers, TEMPO-oxidized fibers or

cellulose nanofibers) used in the manufacturing of the composites did not have a significant

effect on the mechanical properties.

14

Fig. 8. Typical stress-strain curves of the prepared materials.

Table 1. Mechanical properties of the films of thermoplastic starch (TPS) and thermoplastic

starch reinforced with 10 wt% of cellulose nanofibers (CNF), TEMPO-oxidized fibers

(TEMPO) and bleached wood fibers (BL.WF) together with standard deviations.

Material Tensile strength*

(MPa)

Young’s modulus*

(MPa)

Elongation at break*

(%)

TPS 24.2 ± 1.3a 1.36 ± 0.06a 2.6 ± 0.3a

CNF 27.2 ± 2.0b 1.83 ± 0.11b 2.2 ± 0.4a

TEMPO 28.1 ± 1.5b 1.72 ± 0.06c 2.4 ± 0.3a

BL.WF 27.9 ± 1.9b 1.79 ± 0.05bc 2.6 ± 0.4a

*Means marked with the same superscript letter within the same column are not significantly

different at 5% significance level based on the ANOVA and Tukey-Kramer pairwise

comparison test.

3.4 Optical properties

Transparency can be used as an indirect measure of the size and dispersion of cellulose fibers

in the matrix, since, if the reinforcement is not in nanoscale (non-fibrillated cellulose fibers,

aggregated nanofibers), the light transmittance of the material decreases due to the increased

light scattering [Yano et al. 2005, Besbes et al. 2011]. Fig. 9 shows the results from the

15

UV/Vis spectroscopy of the films of neat TPS and TPS reinforced with cellulose nanofibers,

TEMPO-oxidized fibers and bleached wood fibers. From Fig. 9 it can be seen that the light

transmittance of the neat TPS films was very high, and it was reduced greatly with the

addition of cellulose fibers. The thermoplastic starch reinforced with 10% cellulose

nanofibers (CNF) had the highest transparency of the composite films. The films with 10 wt%

of TEMPO-oxidized fibers and bleached cellulose fibers had very similar transparencies,

indicating that these films contain more non-fibrillated cellulose fibers than the film with

cellulose nanofibers. The large difference in the transparency of the neat TPS film and the

composites can also be partly explained by the fact that the neat TPS film had a very smooth

surface in comparison with the films reinforced with cellulose fibers, thus reducing the

scattering of light and increasing the light transmittance.

Fig. 9. The light transmittance of neat TPS and TPS/cellulose fiber composites with 10 wt%

of cellulose nanofibers (CNF), TEMPO-oxidized fibers (TEMPO), and bleached wood fibers

(BL.WF).

The transparency of the TPS and TPS/cellulose fiber films was also examined visually against

a background image, as shown in Fig. 10. From Fig. 10 it can be seen that all the films are

nearly transparent, despite the fact that the reinforcement is not in nanoscale in materials

containing TEMPO-oxidized and bleached wood fibers. In addition, no visible aggregates of

cellulose fibers or nanofibers can be seen in any of the films. The visual transparency is

16

slightly lower in the case of TPS reinforced with bleached wood fibers (Fig. 10d) and

TEMPO-oxidized fibers (Fig. 10c) when compared to the film reinforced with cellulose

nanofibers (Fig. 10b).

Fig. 10. Visual appearance of the films: a) thermoplastic starch, b) TPS with 10 wt% of

cellulose nanofibers, c) TPS with 10 wt% TEMPO-oxidized fibers, and d) TPS with 10 wt%

bleached wood fibers.

3.5 Moisture uptake

The effect of different types of cellulose fibers on the moisture absorption of the

thermoplastic starch was determined by placing the specimen in 98 % RH for 16 days. The

results from the moisture uptake analysis are shown in Table 2. As Table 2 shows, there were

no significant differences between the equilibrium moisture contents of any the materials,

despite the addition of cellulose fibers and nanofibers. This kind of behavior can be explained

by the use of sorbitol as the plasticizer, as sorbitol is thought to restrict the moisture uptake of

thermoplastic starch more than glycerol due to higher chain length and fewer end hydroxyl

groups [Mathew et al. 2002].

17

Table 2. Moisture uptake of the neat thermoplastic starch (TPS) and TPS reinforced with

different types of cellulose fibers (10 wt%).

Material Moisture uptake (%)

TPS 40.5 ± 2.8

CNF 37.6 ± 0.6

TEMPO 37.9 ± 0.6

BL.WF 38.9 ± 1.4

4 Conclusions

In this study the fibrillation of cellulose fibers into micro- or nanosize fibers and the

preparation green bionanocomposites in one twin-screw extrusion step was studied. The effect

of the processing set-up and the fiber type on micro/nanofibrillation and fiber

dispersion/distribution in starch was studied. The microscopy studies showed that the

nanofibrillation of cellulose fibers was not achieved, though the size of the TEMPO-oxidized

fibers was decreased. Tensile testing, light transmittance and moisture uptake analyses were

used as indirect measures of the fiber size in the composites. The mechanical properties of

thermoplastic starch were enhanced by the addition of cellulose fibers; however, the type of

cellulose fiber (bleached wood fibers, TEMPO-oxidized fibers or cellulose nanofibers) did not

have any significant effect on the composites’ mechanical properties. The light transmittance

was lowest for the materials containing bleached wood fibers and TEMPO-oxidized fibers,

confirming the fact that the fibers were not nanofibrillated during the extrusion. The moisture

uptake of the materials was not affected by the addition of cellulose fibers.

The dispersion and distribution of fibers was good in all the composites, indicating that

extrusion processing of cellulose nanocomposites can be done with high water content.

However, the low viscosity resulting from high water content may explain why the

nanofibrillation of cellulose fibers was not successful.

18

Acknowledgements

The authors would like to thank the European Commission (contract number NMP2-LA-

2008-210037 WOODY) for financial support and AVEBE, Netherlands for kindly providing

the potato starch used in the study.

References

Avérous, L., & Halley, P. J. (2009). Biocomposites based on plasticized starch. Biofuels,

Bioproducts and Biorefining, 3(3), 329-343.

Araki, J., Wada, M., & Kuga, S. (2001). Steric Stabilization of a Cellulose MicroCrystal

Suspension by Poly(ethylene glycol) Grafting. Langmuir 17(1), 21-27.

Azizi Samir, M. A. S., Alloin, F., Paillet, M., & Dufresne, A. (2004). Tangling effect in

fibrillated cellulose reinforced nanocomposites. Macromolecules, 37(11), 4313-4316.

Besbes, I., Vilar, M. R., & Boufi, S. (2011). Nanofibrillated cellulose from alfa, eucalyptus

and pine fibres: Preparation, characteristics and reinforcing potential. Carbohydrate Polymers,

86(3), 1198-1206.

Gong, G., Mathew, A. P., & Oksman, K. (2011). Strong aqueous gels of cellulose nanofibers

and nanowhiskers isolated from softwood flour. TAPPI Journal, 10(2), 7-14.

Henriksson, M., Henriksson, G., Berglund, L. A., & Lindström, T. (2007). An

environmentally friendly method for enzyme-assisted preparation of microfibrillated cellulose

(MFC) nanofibers. European Polymer Journal, 43(8), 3434-3441.

Herrick, F. W., Casebier, R. L., Hamilton, J. K., & Sandberg, K. R. (1983). Microfibrillated

cellulose: Morphology and accessibility. Journal of Applied Polymer Science: Appl. Polym.

Symp, 37, 797-813.

Hietala, M., Mathew, A. P., & Oksman, K. (2012). Bionanocomposites of thermoplastic

starch and cellulose nanofibers manufactured using twin-screw extrusion. Eur Polym J,

http://dx.doi.org/10.1016/j.eurpolymj.2012.10.016

19

Hietala, M., Niinimäki, J., & Oksman, K. (2011). The use of twin-screw extrusion in

processing of wood: The effect of processing parameters and pretreatment. BioResources,

6(4), 4615-4625.

Hietala, M., Samuelsson, E., Niinimäki, J., & Oksman, K. (2011). The effect of pre-softened

wood chips on wood fibre aspect ratio and mechanical properties of wood–polymer

composites. Composites Part A: Applied Science and Manufacturing, 42(12), 2110-2116.

Iwamoto, S., Nakagaito, A., Yano, H., & Nogi, M. (2005). Optically transparent composites

reinforced with plant fiber-based nanofibers. Applied Physics A: Materials Science &

Processing, 81(6), 1109-1112.

Janardhnan, S., & Sain, M. (2006). Isolation of cellulose microfibrils-an enzymatic approach.

BioResources, 1(2), 176-188.

Jonoobi, M., Mathew, A. P., & Oksman, K. (2012). Producing low-cost cellulose nanofiber

from sludge as new source of raw materials. Industrial Crops and Products, 40, 232-238.

Katz, S., Beatson, R. P., & Scallan, A. M. (1984). The determination of strong and weak

acidic groups in sulfite pulps. Svensk Papperstidning 87(6), 48-53.

Kirchhof, J. (2008). Mixing and Dispersing: Principles. In: Kohlgrüber, K. & Wiedmann, W.

Co-Rotating Twin-Screw Extruders. Fundamentals, Technology, and Applications. Carl

Hanser Verlag, Munich, Germany. p. 160-179.

Lee, S., Teramoto, Y., & Endo, T. (2009). Enzymatic saccharification of woody biomass

micro/nanofibrillated by continuous extrusion process I – effect of additives with cellulose

affinity. Bioresource Technology, 100(1), 275-279.

Mathew, A. P., & Dufresne, A. (2002). Morphological investigation of nanocomposites from

sorbitol plasticized starch and tunicin whiskers. Biomacromolecules, 3(3), 609-617.

Nakagaito, A. N., & Yano, H. (2004). The effect of morphological changes from pulp fiber

towards nano-scale fibrillated cellulose on the mechanical properties of high-strength plant

fiber based composites. Applied Physics A: Materials Science & Processing, 78(4), 547-552.

Nakagaito, A.N., Sato, S., Sato, A., Yano, H. (2010). Cellulose Nanofiber-reinforced

Unsaturated Polyester as a Potential Substitute for Glass Fiber-reinforced Plastics.

20

Proceedings of the TAPPI International Conference on Nanotechnology for the Forest Product

Industry, 27-29th September, Espoo, Finland.

Oksman, K., Mathew, A. P., & Sain, M. (2009). Novel bionanocomposites: Processing,

properties and potential applications. Plastics, Rubber and Composites, 38(9/10), 396-404.

polyurethane and micro fibrillated cellulose. Composites Science and Technology, 68(3–4),

908-914.

Pääkkö, M., Ankerfors, M., Kosonen, H., Nykänen, A., Ahola, S., Österberg, M., et al.

(2007). Enzymatic hydrolysis combined with mechanical shearing and high-pressure

homogenization for nanoscale cellulose fibrils and strong gels. Biomacromolecules, 8(6),

1934-1941.

Saito, T., Kimura, S., Nishiyama, Y., & Isogai, A. (2007). Cellulose nanofibers prepared by

TEMPO-mediated oxidation of native cellulose. Biomacromolecules, 8(8), 2485-2491.

Saito, T., Nishiyama, Y., Putaux, J., Vignon, M., & Isogai, A. (2006). Homogeneous

suspensions of individualized microfibrils from TEMPO-catalyzed oxidation of native

cellulose. Biomacromolecules, 7(6), 1687-1691.

Smook, G. A. (1999). Handbook for pulp and paper technologists (2nd ed.) Angus Wilde

Publications Inc.

Spence, K., Venditti, R., Rojas, O., Habibi, Y., & Pawlak, J. (2011). A comparative study of

energy consumption and physical properties of microfibrillated cellulose produced by

different processing methods. Cellulose, 18(4), 1097-1111.

Stepto, R. F. T. (2003). The processing of starch as a thermoplastic. Macromolecular

Symposia, 201(1), 203-212.

Taniguchi, T., & Okamura, K. (1998). New films produced from microfibrillated natural

fibres. Polymer International, 47(3), 291-294.

Turbak, A. F., Snyder, F. W., & Sandberg, K. R. (1983). Microfibrillated cellulose, a new

cellulose product: Properties, uses, and commercial potential. J. Appl. Polym. Sci.: Appl.

Polym. Symp., 37, 815-823.

21

Wagberg, L., Decher, G., Norgren, M., Lindstrom, T., Ankerfors, M., & Axnas, K. (2008).

The build-up of polyelectrolyte multilayers of microfibrillated cellulose and cationic

polyelectrolytes. Langmuir, 24(3), 784-795.

Yano, H., Sugiyama, J., Nakagaito, A. N., Nogi, M., Matsuura, T., Hikita, M., et al. (2005).

Optically transparent composites reinforced with networks of bacterial nanofibers. Advanced

Materials, 17(2), 153-155.