ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2015

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1215

Elasticity of Cellulose NanofibrilMaterials

GABRIELLA JOSEFSSON

ISSN 1651-6214ISBN 978-91-554-9135-2urn:nbn:se:uu:diva-240250

Dissertation presented at Uppsala University to be publicly examined in Polhemssalen,Ångströmlaboratoriet, Lägerhyddsvägen 1, Uppsala, Friday, 13 February 2015 at 09:00 forthe degree of Doctor of Philosophy. The examination will be conducted in English. Facultyexaminer: Reader Karin de Borst (School of Engineering, University of Glasgow).

AbstractJosefsson, G. 2015. Elasticity of Cellulose Nanofibril Materials. Digital ComprehensiveSummaries of Uppsala Dissertations from the Faculty of Science and Technology 1215. 60 pp.Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-9135-2.

The demand for renewable load-carrying materials is increasing with increasing environmentalawareness. Alternative sources for materials manufacturing and design have to be investigated inorder to replace the non-biodegradable materials. The work presented in this thesis investigatesstructure-property relations of such renewable materials based on cellulose nanofibrils.Cellulose is the most abundant polymer on earth and exists in both ordered and disorderedphases, where the ordered crystalline cellulose shows excellent mechanical properties. Thecelluloses nanofibril is composed of partly crystalline cellulose where the stiff crystal regions,or crystallites, are orientated in the axial direction of the fibrils. The cellulose nanofibrils have ahigh aspect ratio, i.e. length to diameter ratio, with a diameter of less than 100 nm and a lengthof some micrometres. In the presented work, different properties of the cellulose nanofibril werestudied, e.g. elastic properties, structure, and its potential as a reinforcement constituent. Theproperties and behaviour of the fibrils were studied with respect to different length scales, fromthe internal structure of the cellulose nanofibril, based on molecular dynamic simulations, to themacroscopic properties of cellulose nanofibril based materials. Films and composite materialswith in-plane randomly oriented fibrils were produced. Properties of the cellulose nanofibrilbased materials, such as stiffness, thickness variation, and fibril orientation distribution, wereinvestigated, from which the effective elastic properties of the fibrils were determined. Thestudies showed that a typical softwood based cellulose nanofibril has an axial stiffness of around65 GPa. The properties of the cellulose nanofibril based materials are highly affected by thedispersion and orientation of the fibrils. To use the full potential of the stiff fibrils, well dispersedand oriented fibrils are essential. The orientation distribution of fibrils in hydrogels subjectedto a strain was therefore investigated. The study showed that the cellulose nanofibrils have highability to align, where the alignment increased with increased applied strain.

Keywords: Cellulose nanofibrils, Elastic properties, Micromechanics, Composites

Gabriella Josefsson, Department of Engineering Sciences, Applied Mechanics, 516, UppsalaUniversity, SE-751 20 Uppsala, Sweden.

© Gabriella Josefsson 2015

ISSN 1651-6214ISBN 978-91-554-9135-2urn:nbn:se:uu:diva-240250 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-240250)

To my family

Preface

The work presented in this thesis was carried out at Uppsala University in the division of Applied Mechanics. During my PhD studies, I have had the privilege to be surrounded by wonderful people that have helped me and supported me.

I would like to express my deep gratitude to my supervisor Professor Kristofer Gamstedt for giving me the opportunity to do this thesis. You have a fantastic ability to see the positive side of everything, which is something that has helped me in my work. Thank you for always supporting me and for making me believe in my research.

I wish to thank my co-supervisor Associate Professor Urmas Valdek. You are a wonderful person, and it has been a pleasure to assist you in the courses you are conducting. Thank you for recommending me and believing in me for this position. It is thanks to you I started my PhD studies.

I would like to acknowledge the co-authors of the appended papers. Spe-cial thanks to Dr. Gary Chinga-Carrasco at PFI and Dr. Fredrik Berthold at Innventia for fruitful collaborations.

To Maria Melin, Maria Skoglund, Ingrid Ringård, Jonatan Bagge, Ylva Johansson, and to all administrative staff at Department of Engineering Sci-ences, thank you for all your help.

I also wish to thank my beloved colleagues. Alexey Vorobyev, during the last years we have spent almost every day with each other, at work and on the dance floor, and I have enjoyed every single one of them. Thank you for being so kind and helpful, and for always supporting me. Associate Profes-sor Ivón Hassel, since the day you started at Uppsala University, I have had more fun at work than one can ever imagine. You spread so much positive energy, which has helped me to do a better work. Fengzhen Sun, everyone should have a roommate like you. You are so kind and caring, and I appreci-ate all the interesting and crazy discussions we have had. Dr. Thomas Joffre, thank you for all the fun we have had at research trips and the advice and help in the beginning of my studies. Elias Garcia Urquia, thank you for cre-ating many moments of laughter. Amra Battini, thank you for your kindness and care. Professor Emeritus Bengt Lundberg, thank you for all the interest-ing discussions and for your kind help with the thesis. Associate Professor Göran Lindström, thank you for your care and support, and thank you for all the fun discussions we have had. Peter Bergkvist, thank you for all the great activities that has strengthened the group. Jinxing Huo, thank you for all

your kind help. Professor Per Isaksson, thank you for your care. Dr. Nico van Dijk, thank you for always wanting to help. To all of the current and former colleagues at the Division of Applied Mechanics, thank you for con-tributing to such a friendly and warm environment. It has been an honour to be a part of the group.

Martin Löfqvist, you have always supported me, believed in me and en-couraged me. You have always taken the time to listen to me, and giving me advice. Thank you for always being there for me.

Finally, I would like to thank my mother, my father, and my sister for, during my whole life, believing in me and supporting me. You made me think that nothing is impossible, and that I am able to do whatever I want to do.

Gabriella Josefsson Uppsala 2015

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Josefsson, G., Tanem, B.S., Li, Y., Vullum, P.E., Gamstedt, E.K. (2013) Prediction of elastic properties of nanofibrillated cellulose from micromechanical modeling and nano-structure characterization by transmission electron microscopy. Cellu-lose, 20(2):761-770

II Josefsson, G., Berthold, F., Gamstedt, E.K. (2014) Stiffness contribution of cellulose nanofibrils to composite materials. In-ternational Journal of Solids and Structures, 51(5):945-953

III Josefsson, G., Chinga-Carrasco, G., Gamstedt, E.K. Elastic models coupling the cellulose nanofibril to the macroscopic film level. Submitted.

IV Mikczinski, M.R., Josefsson, G., Chinga-Carrasco, G., Gam-stedt, E.K., Fatikow, S. (2014) Nanorobotic testing to assess the stiffness properties of nanopaper. IEEE Transactions on Robot-ics, 30(1):115-119

V Josefsson, G., Chinga-Carrasco, G., Gamstedt, E.K. Thickness

variability of cellulose nanofibril films: Measurement and im-plications for mechanical characterization. Submitted.

VI Josefsson, G., Ahvenainen, P., Ezekiel Mushi, N., Gamstedt, E.K. Fibril orientation redistribution induced by stretching of cellulose nanofibril hydrogels. Submitted.

Reprints were made with permission from the respective publishers.

Contents

1. Introduction ............................................................................................... 11 1.1 Structure of wood ............................................................................... 11 1.2 Cellulose nanofibrils .......................................................................... 12 1.3 CNF based materials .......................................................................... 14 1.4 Aims of the thesis ............................................................................... 15 

2. Theory ....................................................................................................... 17 2.1 Composite materials ........................................................................... 17 

2.1.1 Mori-Tanaka and self-consistent model ..................................... 17 2.1.2 Hashin micromechanical model ................................................. 18 

2.2 Films and fibre networks .................................................................... 21 2.2.1 Classical laminate theory ............................................................ 21 2.2.2 Various micromechanical models ............................................... 23 

3. Experimental procedure ............................................................................ 25 3.1 Material preparation ........................................................................... 25 

3.1.1 CNF films ................................................................................... 25 3.1.2 CNF composites.......................................................................... 26 

3.2 Experimental measurements ............................................................... 26 3.2.1 Tensile testing ............................................................................. 26 3.2.2 X-ray diffraction ......................................................................... 26 3.2.3 Electron microscopy ................................................................... 27 3.2.4 Differential scanning calorimetry ............................................... 27 

4. Results and discussion .............................................................................. 28 4.1 Elastic properties of cellulose nanofibrils .......................................... 28 4.2 Properties of cellulose nanofibril materials ........................................ 37 

5. Conclusions ............................................................................................... 48 

7. Outlook ..................................................................................................... 50 

6. Sammanfattning på svenska ...................................................................... 52 

8. References ................................................................................................. 55 

Abbreviations

Abbreviations Full name BC Bacterial cellulose nanowhisker CLT Classical laminate theory CNF Cellulose nanofibril DSC Differential scanning calorimetry HRTEM High-resolution transmission electron microscopy MCC Microcrystalline cellulose PLA Poly(lactic acid) SEM Scanning electron microscopy TEMPO 2,2,6,6-tetramethylpiperidine-1-oxyl XRD X-ray diffraction Symbol Description

Components of a strain-concentration tensor described by A

Components of the stiffness matrix of the material m

Young’s modulus of the isotropic material m Young’s modulus of the anisotropic material m in the i direction Components of the Eshelby tensor

Shear modulus of the isotropic material m Shear moduli of the anisotropic material m

Bulk modulus of material m Poisson ratios of material m

Components of the reduced stiffness matrix of the material m

Density of the material m Components of the compliance matrix of the material m

Invariants of the reduced stiffness matrix of material m Volume fraction of the material m

11

1. Introduction

Wood has been used during many thousands of years due to its availability and good mechanical properties, from the beginning predominantly as con-struction building blocks but later on also in other products like paper or as reinforcement in composite materials [1-3]. The focus in modern time has been on wood fibre products but as technology develops, more demands are put on material development to produce stiffer, stronger and lighter materials of higher quality. In other words, wood based products have to be upgraded to match the requirements of today. A promising product extracted from wood is the cellulose nanofibril (CNF), also called nanofibrillated cellulose (NFC). CNFs are small cellulose fibres with diameter less than 100 nm and length of some micrometre [4]. They are highly crystalline fibrils and have promising mechanical properties [5]. Numerous studies have been carried out in order to understand the structure and properties of the CNF and to be able to use it in new renewable materials [6, 7]. However, the small size of the CNFs causes problems when investigating their mechanical properties. The study carried out in this thesis focuses on determination of the elastic properties of the CNF with different methods and the purpose of using the fibrils as reinforcement in new stiff materials. Depending on the application area, different properties of the CNF should be screened. The present work contributes to the study of the effective elastic properties of CNF. The struc-ture of CNF based materials was also investigated, which has significant impact on the strength and stiffness of the materials.

In this introduction chapter, some general information of the CNF and its origin is presented followed by more specific aims of the thesis.

1.1 Structure of wood Wood has a complex ultrastructure and a multiphase design. The wood is composed mainly of tracheids, or fibres, that are oriented along the trees and are attached to one another with a lignin rich region referred as the middle lamella. The wood fibres are hollow tubes were the fibre cell wall has a lam-inate structure consisting of several layers called primary cell wall (P), outer layer of the secondary cell wall (S1), the middle layer of the secondary cell wall (S2) and the inner layer of the secondary cell wall (S3) [8, 9]. All cell wall layers have the same building blocks, namely cellulose microfibrils

12

embedded in lignin and hemicellulose. The microfibrils are organized in different ways depending on the cell wall layer [10]. In the primary layer, the microfibrils are randomly oriented while in the secondary layers they are oriented in specific angles. The microfibrils consist of bundles of elementary fibrils which constitutes the smallest building block of the wood structure. The fibrils are composed of ordered and disordered cellulose and have a composite structure with the stiff cellulose crystallites acting as reinforce-ment in a surrounding matrix of disordered cellulose [11]. The crystallites, with an axial stiffness of around 150 GPa [12, 13] are oriented in the axial direction of the microfibrils, which is the reason for the good mechanical properties of the microfibrils and the wood fibres. A schematic image show-ing the structure of wood in different length scales is shown in Figure 1 [9, 11].

Figure 1. Schematic image visualizing the structure of wood in different length scales. In the image, a fibre and the fibre cell wall layers P, S1, S2, and S3 are shown. The microfibrils are constructed of elementary fibrils composed of crystal-line and disordered cellulose.

1.2 Cellulose nanofibrils Cellulose nanofibrils (CNF) can be extracted from trees and other plants by fibrillation of the cellulose fibres where the microfibrils in the fibre cell wall

Crystalline cellulose

Disordered cellulose

P

S1

S2

S3

Wood fibre

Microfibril

Elementary fibril

13

are liberated into CNFs. The first effective procedure to disintegrate fibres into CNFs was developed by Turbak et al. [14] in 1983. In the presented procedure, a suspension of pulp fibres was passed through a high pressure homogenizer with a small opening in which the suspension was subjected to a high pressure drop. The pressure drop introduces sudden shearing and im-pact forces in the fibres and by subjecting the fibre suspension to several pressure drops, the CNFs are liberated [15, 16]. The mechanical treatment results in CNFs consisting of aggregated elementary fibrils with a diameter 10-100 nm. The use of chemical pretreatment can help liberating the CNFs into fibrils with more homogenous diameters. It also reduces the energy con-sumption during the mechanical treatment down to less than 1% of the ener-gy required without pretreatment [17]. During chemical pretreatment, charged entites are added on the fibril surface resulting in repulsion forces between the fibrils which facilitate the separation of the microfibrils to CNFs [18]. By pretreatment of the pulp with carboxymethylation, CNFs with a diameter of 5-15 nm were achieved [19]. The most common used pretreat-ment is 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) mediated oxidation which is a powerful pretreatment resulting in fibrils with a diameter down to 3-5 nm, corresponding to the width of an elementary fibril, and a length in the micrometre scale [18, 20]. Depending on the treatment used to liberate the CNFs, the fibrils will be constructed of elementary fibrils or aggregates of elementary fibrils. The elementary fibrils are organized in ordered crystal-line regions and disordered regions acting as chains along the fibril [17, 21, 22]. A high-resolution transmission microscopy (HRTEM) image of a CNF is presented in Figure 2.

Figure 2. HRTEM image of a CNF.

Given the small size of fibrils, it is difficult to measure the mechanical prop-erties of the CNFs directly. Only a few papers report measurements per-

14

formed on single cellulose nanofibrils to determine the mechanical proper-ties, e.g. Cheng and Wang [23, 24], Iwamoto et al. [25] and Guhados et al. [26] who used atomic force microscopy to measure the elastic modulus in bending of cellulose nanofibrils from different raw materials. The values obtained were 84 GPa for CNFs from wood, 78 GPa for bacterial cellulose fibrils, 93-98 for fibrils from Lyocell and around 150 GPa for the highly crystalline tunicate fibrils. Recently, Sun et al. [27] modelled the structure of CNFs and fibre cell walls which resulted in a CNF Young’s modulus of 65 GPa. The structure and the elastic properties of the CNF were investigated in Paper I. A more common way to characterize the mechanical performance of CNFs is to do quantitative measurements of CNF based materials, such as films or composite materials reinforced by CNF [28-30]. The mechanical properties of the CNF based materials can be determined and are strongly affected by the mechanical properties of the constituent nanofibrils.

1.3 CNF based materials The demand for renewable and biodegradable materials is rapidly increasing with growing environmental awareness of customers and consumers. This has led to investigations of alternative constituents for materials manufactur-ing. Prompted by the impressive property improvements of composites rein-forced by nanosized fibres, such as carbon nanotubes [31] and exfoliated nanoclay [32], the use of CNFs in composite materials is of great interest. The CNF has high stiffness and high aspect ratio (i.e. length to diameter ratio), which are beneficial properties for reinforcements in materials. In comparison to for example carbon nanotubes, cellulose nanofibrils have a functional surface, available for surface modifications to control the fibril–matrix interface and the dispersion of the fibrils in polymers [33, 34]. The abundance of hydroxyl groups on the cellulose fibril surface facilitates inter-action with polar matrix systems and formation of a stiffening network [35]. Many studies of CNF based materials are presented in the literature, and one of the most common way to utilize the CNFs is by casting CNF films [36, 37]. Different properties of CNF films were studied in Papers III-VI. These films typically have a thickness of 20-100 um, high transparency, and a stiffness and strength of 10-15 GPa and 100-200 MPa, respectively. By sur-face modification and coating of the CNFs, the films can be used as loud-speaker membrane [38], oil barrier for packaging laminates [19], and in elec-trical devices [39-41]. For the electrical devices, CNFs are generally coated with a conducting polymer, and films of the coated fibrils are produced. CNF films can be further used to produce composite materials by immersing the films with a polymer such as phenol-formaldehyde [42] and neat acryl resin [43]. Other methods to produce CNF reinforced composite materials is

15

hot press moulding [44, 45], where the CNFs and a polymer are mixed and pressed together with high pressure and high temperature, or through extru-sion [46]. Composites produced by hot press moulding were studied in Paper II. The CNF based composites have better mechanical properties than the matrix material alone and show increased stiffness and strength with in-creased fibril content [44, 46].

Most of the CNF based materials contain fibrils having a randomly in-plane orientation. However, many structural components are subjected to uniaxial stresses and it is therefore of importance to develop manufacturing methods where the fibril orientation can be controlled. This subject has gained increased interest during the recent years, and different methods to manufacture materials with oriented fibrils have been presented, e.g. by wet-spinning [47], wet extrusion [48] and cold-drawing [49]. In Paper VI, the structure and fibril orientation of cold-drawn films were investigated. With fibrils in an aligned configuration, the mechanical properties of the compo-nents are radically improved.

1.4 Aims of the thesis The overall aim of the thesis was to gain knowledge about the elastic behav-iour of CNFs and CNF based materials. In Papers I-III, the elastic properties of CNFs were determined by modelling of the CNF structure (Paper I) and back calculation of the CNF effective elastic properties from CNF based materials (Papers II and III). Structural properties, which affect the elastic behaviour of CNF based materials, were investigated in Papers IV-VI. The specific aims of each paper are stated below. Paper I. The aim of the study was to estimate the elastic properties of a sin-gle CNF by modelling its internal structure in terms of disordered and crys-talline cellulose. Paper II. The aim of the study was to investigate the behaviour of different cellulose fibrils in composite materials. The structure and stiffness of the composites were studied as well as the effective stiffness of the fibrils. Paper III. The aim of the study was to provide a tool to estimate and com-pare the elastic properties of CNFs, produced from different raw materials and by different methods, from the elastic properties of CNF films. Paper IV. The aim of the study was to investigate the elastic properties of CNF films using different measuring methods.

16

Paper V. The aim of the study was to map the thickness variation of CNF films and to provide a tool to determine more accurate estimations of stiff-ness and strength of CNF based materials. Paper VI. The aim of the study was to produce films of partly oriented CNFs and to understand the alignment mechanism of the CNFs in films.

17

2. Theory

2.1 Composite materials Many micromechanical models have been developed in order to relate the elastic properties of a composite material to those of the constituents. In this section, some of the micromechanical models used in the thesis are present-ed.

2.1.1 Mori-Tanaka and self-consistent model Mori-Tanaka’s effective medium theory [50] was developed in 1973 for non-dilute composite materials. It relates the average stress of the matrix to the transformation strain of the inclusions. The theory is based on Eshelby’s inclusion theory [51] which treats a single ellipsoidal inclusion in an infinite matrix. Mori-Tanaka’s theory was reformulated by Benveniste [52] in 1987 for the purpose of using the theory to calculate the effective stiffness of a composite material, and that formulation has been used in Papers I and II. According to Benveniste, the stiffness matrix of a composite material, with aligned inclusions of prolate or oblate shape surrounded by a matrix materi-al, can be expressed by . (1)

Here the summation convention applies and denotes components of the

stiffness tensor of the matrix material, denotes components of the stiff-ness tensor of the inclusions, is volume fraction of the inclusions and are components of the strain-concentration tensor. From the Mori-Tanaka theory, the strain-consternation tensor can be expressed as a function of the strain-concentration tensor derived by Eshelby for a single inclusion in an infinite matrix as ,

(2)

where is the inverse of , (3)

18

and is the inverse of 1 . (4)

denotes components of the identity tensor, are components of the

Eshelby tensor [51] and are components of the compliance tensor of the matrix material. The Eshelby tensor for an isotropic matrix material is a function of the inclusion aspect ratio and the elastic constants of the matrix material and can be found in the paper written by Qiu and Weng [53].

When the stiffness properties of the composite material have been deter-mined, they can be used in a self-consistent scheme for a more accurate es-timation of the elastic properties of the composite. The derived stiffness ten-sor of the composite material is then used as input in a new strain-

concentration tensor given by the inverse of , (5)

i.e. , (6)

where is the compliance tensor of the composite material and can be derived by taking the inverse of the stiffness tensor. In the self-consistent scheme, the matrix material, surrounding the inclusions, is regarded as hav-ing the properties of the composite material. For a composite material with aligned fibres, the over-all properties are transversely isotropic and the Eshelby tensor used in the self-consistent scheme has to be calculated for a transversely isotropic matrix. The calculation of the Eshelby tensor is de-scribed by Chou et al. [54] and Lin and Mura [55]. When the new strain-concentration tensor is substituted in Eq. (1), an improved stiffness tensor of the composite material can be calculated. To get a more accurate estimation of the stiffness of the composite, the procedure is repeated, using the new stiffness tensor estimated with the self-consistent scheme. The iteration con-tinues until the values of the stiffness tensor of the composite material con-verge.

2.1.2 Hashin micromechanical model Hashin [56] developed many micromechanical models for calculations of the elastic properties of composites with different kinds of inclusions. One of the models is used in Paper II and describes the elastic properties of a statistical-ly isotropic composite material with arbitrary internal phase geometry. In Paper II, the model was used to calculate the elastic properties of a semicrys-

19

talline polymer containing polycrystalline aggregates but it can also be used to calculate the elastic properties of particle reinforced composites. Exact solutions of the elastic properties of such materials are not presented, but relatively close upper and lower bounds of the bulk and shear moduli are given. The bounds of the bulk modulus are expressed as

1 33 4

(7)

and

1 33 4

, (8)

where and are the bulk moduli of the amorphous phase and polycrys-talline phase, respectively, and and are the volume fractions and and are the shear moduli of these phases. In a similar way, the bounds of the shear modulus of the composite material are given by

1 6 25 3 4

(9)

and

1 6 25 3 4

. (10)

By taking the average of the upper and lower bounds of the bulk and shear moduli separately, the elastic parameters of the semicrystalline material can be estimated as

2

(11)

and

2. (12)

20

For an isotropic material, the Young’s modulus can be expressed by the bulk modulus and shear modulus by 9

3. (13)

When the Hashin micromechanical model for semicrystalline materials is used, the isotropic properties of the polycrystalline aggregates have to be calculated from the properties of the anisotropic elementary crystals that together build up these aggregates. Although the single crystals are aniso-tropic, the resulting property of crystalline aggregates is isotropic since it is assumed that the crystals have a random orientation distribution in space. In 1928, Voigt [57] derived the properties of an aggregate of crystals by aver-aging the relations expressing the stress in a single crystal in terms of the given strain. One year later, Reuss [58] derived the properties of an aggre-gate of crystals by averaging the relations expressing the strain in terms of the given stress. These two approaches result in theoretical upper and lower bounds of the properties of the polycrystals. Hill [59] found from experi-mental measurements that the moduli of the crystal aggregates lie in between the values obtained by Voigt and Reuss. By taking the average of the Voigt and Reuss models, the isotropic properties of the crystal aggregates can be estimated. The Voigt shear and bulk moduli of a polycrystal constructed of orthotropic crystals are given by

, 115

15

(14)

and , 1

929

, (15)

where are elements in the stiffness matrix of the orthotropic crystal. The Reuss shear and bulk moduli are given by

, 15

4 4 3

(16)

and

21

, 12

, (17)

where are elements in the compliance matrix of the orthotropic crystal.

By taking the average of the Voigt and Reuss moduli, the isotropic Young’s modulus of the crystal aggregate can be obtained as

93

. (18)

The isotropic constants for the crystal aggregates can be used in the Hashin model for calculations of the stiffness moduli of partly crystalline materials.

2.2 Films and fibre networks Fibres bonded together can form a network or a dense film. Numerous mi-cromechanical models have been developed to relate the fibre properties to those of a network or a film. In the thesis, different such micromechanical models have been used (Papers I and II) and evaluated (Paper III) to relate the CNF properties to the film properties.

2.2.1 Classical laminate theory 2.2.1.1 Two-dimensional laminate theory The classical laminate theory (CLT) (e.g. Tsai [60]) can be used to calculate the stiffness of a laminate constructed from hypothetical unidirectional plies of multiple angles. The stiffness of the laminate is calculated by averaging the in-plane stiffnesses of the unidirectional plies. All in-plane elastic con-stants are taken into account, and the stiffness matrices of the off-axis plies can be calculated by coordinate transformation. CLT was used in Papers I and III. For a laminate composite with randomly orientated fibres, or nano-fibrils, the isotropic in-plane Young’s modulus can be calculated by

. (19)

Here and are invariants of the reduced stiffness matrix of the hypothetical unidirectional plies given by 3

838

14

12

(20)

22

and 1

818

34

12

, (21)

where are components of the reduced stiffness matrix of the plies. They

can be expressed in terms of the elastic constants of the plies as

1,

(22)

1,

(23)

,

(24)

1, (25)

and , (26) where is the longitudinal Young’s modulus, is the transverse Young’s modulus, and are the Poisson ratios, and is the shear modulus.

For a network or film of fibres, as studied in Papers I and III, the proper-ties of the hypothetical unidirectional plies will be directly proportional to the properties of the fibres, i.e. the ply elastic constants are replaced by the fibre elastic constants. However, the porosity in the film has to be taken into account and the Young’s modulus of the film should be scaled with the ratio of the density of the film and density of the fibres as

(27)

where and are invariants of the reduced stiffness matrix of the fibres.

2.2.1.2 Three-dimensional laminate theory Laminate theory in three dimensions can be used to calculate all elastic con-stants in the stiffness matrix for a laminated structure of hypothetical plies of

23

unidirectional fibres. A 3D laminate theory for sub-laminate homogenization has been presented by Whitcomb and Noh [61] and was used in Paper II for small films, called platelets, of CNFs with random in-plane orientation, i.e. films with transversely isotropic properties. On the basis of this theory, a closed analytical form of the constants in the stiffness matrix was presented in Paper II for a film of fibres, or CNFs, with random in-plane orientation. The elements in the stiffness matrix of the film are expressed in terms of the elements of the stiffness matrix of the fibre or cellulose nanofibril as 3

814

38

12

18

, (28)

, (29)

1

834

18

12

18

, (30)

, (31)

1

212

, (32)

2

, (33)

and

2. (34)

2.2.2 Various micromechanical models In Paper III, different micromechanical models that relate film properties to CNF properties were evaluated. The models will be presented briefly in this chapter.

2.2.2.1 Tsai-Pagano model Tsai and Pagano [62] developed a micromechanical model that can be used to calculate the stiffness of a film, or a composite, with in-plane randomly orientated fibres, or CNFs. By scaling the Tsai-Pagano model with the ratio of the film density and fibre density, to account for the porous structure of

24

films, the Young’s modulus of a film can be expressed in terms of the Young’s moduli of in the longitudinal direction, , and in the transverse direction, , of the fibres. It is given by

3

858

. (35)

2.2.2.2 Cox model Cox [63] derived a micromechanical model for a network of fibres. He as-sumed that the fibres are infinitely long and can carry load only at the fibre ends. Also, the flexural stiffness is neglected so that the fibres can carry load only in tension. The Young’s modulus of the fibre network, in terms of the longitudinal Young’s modulus of the fibre, is given by

1

3, (36)

where the ratio of the film density and fibre density accounts for the porous structure of the film.

2.2.2.3 Krenchel model Krenchel [64] developed a micromechanical model for composites by intro-ducing an efficiency factor of reinforcement that depends on the orientation of the fibre in the composite. He assumed that the fibres are infinitely long, extending from one side of the composite to the other, and that the flexural stiffness of the fibres can be neglected. By scaling the Krenchel model with the ratio of the film density and fibre density, to account for the porous struc-ture of films, the model can be used to determine the stiffness of films of fibres, or CNFs. For a random orientation of the fibres, the Young’s modulus of the film is given by

3

8.

(37)

25

3. Experimental procedure

In this chapter, the procedures to produce CNF based materials, as well as the experimental methods used in the studies, are briefly described. Specific details on which equipment that were used and experimental setups can be found in the appended papers. Only the general principles of the used exper-imental methods are outlined in the following.

3.1 Material preparation

3.1.1 CNF films Films can be produced from CNFs extracted by different methods from vari-ous renewable raw materials. The CNFs can be fibrillated by mechanical treatment or by a combination of mechanical and chemical treatment. How-ever, the film preparation is typically the same, irrespective of CNF type. The films studied in this thesis were all produced from dilute CNF suspen-sions with a nanofibril consistency of 0.1 wt%. The suspensions were thor-oughly stirred and poured in funnels with a bottom covered by a filtering membrane. In Paper VI, vacuum filtration was used to speed up the process. After filtration, the films were further dried in an oven or by a rapid sheet dryer to remove excess water. The films studied in Papers III-V were pro-duced from different types of CNFs extracted from hardwood and softwood pulp, of which some of the pulps were pretreated by TEMPO mediated oxi-dation.

In Paper VI, films of partly aligned fibrils were produced from CNFs ex-tracted from TEMPO pretreated softwood pulp. The procedure for the film preparation is the same as for films of randomly oriented CNFs, except for one additional step. From the vacuum filtering process, hydrogel films con-taining 70-90% water are formed. These hydrogel films were carefully peeled off the membrane, cut in 10 mm stripes, and subjected to a strain of 35-50%. The stripes were thereafter directly dried in the strained configura-tion using a rapid sheet dryer.

26

3.1.2 CNF composites In Paper II, composites of poly(lactic acid) (PLA) reinforced by different types of cellulose fibrils were produced. CNFs from wood, bacterial cellu-lose nanowhiskers (BC), and microcrystalline cellulose (MCC) were used as reinforcement. The composites were produced in two steps. In the first step, PLA powder was added to the dilute suspensions of fibrils and thoroughly stirred. The suspensions of PLA/fibrils were then further diluted with dH2O and sonicated for three minutes. The well mixed suspensions were poured in funnels and vacuum filtered, resulting in pre-formed puck shaped cakes. The pre-formed cakes were placed in a vacuum oven for complete drying. In the second step, the pre-formed cakes were placed inside a thick metal frame and hot pressed for 20 minutes using a dial pressure of 150 bars and a tem-perature of 185 °C. The chosen temperature is well above the melting point of PLA, Tm = 150 °C, but still lower than the degradation temperature of the fibrils [45]. The final composite disks had a diameter of 52 mm and a thick-ness of 1.5-2 mm. From the disks, dog-bone shaped samples were cut using a water jet.

3.2 Experimental measurements

3.2.1 Tensile testing The mechanical properties of a material can be characterized by performing a tensile test. During a tensile test, the sample is subjected to a force and the corresponding elongation of the sample is measured. With the sample di-mensions, a relation between stress and strain is obtained. The initial linear part of the stress-strain curve can be used to determine the Young’s modulus of the material in that direction. Tensile tests were performed in Papers II-IV.

3.2.2 X-ray diffraction X-ray diffraction (XRD) can be used to measure the internal structure of a material. By exposure of the sample to X-rays, the diffraction from the sam-ple is measured. The diffraction from the sample gives information on e.g. the sample constituents, crystallinity and crystal orientation in the sample. XRD was used in Papers I and VI.

27

3.2.3 Electron microscopy There are many types of electron microscopes, e.g. scanning electron micro-scope (SEM) and transmission electron microscope (TEM). The TEM can be run in high-resolution mode (HRTEM) that allows imaging of the atomic structure. Electron microscopes use electrons to bombard the sample and measure the interaction of the electrons with the sample. The information gathered can be used to form a high-resolution image of the sample. Electron microscopes were used in Papers I-V.

3.2.4 Differential scanning calorimetry Differential scanning calorimetry (DSC) is normally used to examine the properties of polymers. A sample of the polymer and a reference sample are subjected to an increase in temperature, and the difference in energy required to increase the sample and the reference sample is measured. The infor-mation can be used to determine a number of characteristic properties such as the glass transition temperature, crystallization temperature, and degree of crystallinity of the sample. DSC was used in Paper II.

28

4. Results and discussion

The studies presented in the thesis are all related to the structure and me-chanical performance of the cellulose nanofibril. The structure of CNF-based materials, such as films and composite materials, has been carefully investi-gated and the stiffness of the CNF has been estimated in different ways to ensure reliability of the results. In this chapter, the main results of the thesis are presented.

4.1 Elastic properties of cellulose nanofibrils Modelling and quantitative measurements are tools for determining the elas-tic properties of the small size CNFs. In Paper I, the elastic properties of the CNF were investigated with a modelling approach, in which the internal structure of the CNF was assumed to be composed of crystalline and amor-phous cellulose. The CNF was examined with HRTEM and XRD to gather information about the internal structure of the nanofibril. It showed that the CNF has a composite structure with the cellulose crystallites acting as rein-forcement in a surrounding matrix of disordered cellulose. The crystallites are oriented in the axial direction of the fibril, and an aspect ratio of around 1.8 was observed for the investigated type of CNFs. The crystallinity of the fibril was estimated to 69%. Two HRTEM images of CNFs are shown in Figure 3.

Based on the information on the nanofibril structure, the elastic properties of the CNF were modelled as a unidirectional short-fibre reinforced compo-site. A self-consistent Mori-Tanaka model [50] was used, and the properties of the constituents, i.e. crystalline and amorphous cellulose, were taken from results of molecular-dynamic simulations of cellulose structures found in the literature [12, 65]. The modelling showed that the CNF has a longitudinal Young’s modulus of 65 GPa.

The stiffness of the CNF is highly dependent on the crystallite aspect ra-tio. Peura et al. [66] measured the crystallite aspect ratio in native wood and found values around 7-9, which is much higher than what is obtained in Pa-per I. The reduction of crystallite aspect ratio is a result of the defibrillation process, which affects both the crystallinity and the crystallite size [67].

29

10 nm10 nm

Figure 3. HRTEM images showing the internal structure of the CNFs.

This indicates that a milder defibrillation process could result in even higher stiffness of the CNF than that estimated in Paper I. To illustrate the effect of crystallite aspect ratio and degree of crystallinity, the Young’s modulus was plotted as a function of the aspect ratio for three values of the degree of crys-tallinity, see Figure 4.

Figure 4. Young’s modulus of the CNF as a function of the aspect ratio α plotted for three values of the degree of crystallinity. The dashed line indicates the value ob-tained from the rule of mixture.

For low aspect ratios, particularly in the range of 1 (corresponding to a spherical crystallite) to 10, the Young’s modulus increases rapidly with in-creased aspect ratio. For a crystallite aspect ratio of 10, the Young’s modulus of the CNF is only 8% lower than predicted by the rule of mixture, corre-sponding to α → ∞.

30

In order to investigate the dependency of the CNF stiffness to the proper-ties of amorphous cellulose, the Young’s modulus of the CNF was plotted as a function of the Young’s modulus of the amorphous cellulose, see Figure 5.

Figure 5. Young’s modulus of the CNF as a function of the Young’s modulus of the amorphous cellulose plotted for three values of the degree of crystallinity.

In Paper II, composite materials were produced to investigate the effec-tive stiffness of different types of cellulose fibrils. The elastic properties of the composites were measured, and the effective stiffness of the reinforce-ment fibrils was calculated using a micromechanical model linking the com-posite stiffness and the fibril stiffness. The mixed experimental and analyti-cal model showed that the CNF has a longitudinal Young’s modulus of 65 GPa, which supports the results in Paper I. Two other types of reinforce-ments, in addition to CNFs, were also studied, namely bacterial cellulose nanowhiskers (BC) and microcrystalline cellulose (MCC). The PLA-matrix composites were produced by hot press moulding with a fibril content of 10 wt%.

To develop the micromechanical model, relating the stiffness of the fibril to that of the composite, information of the composite structure had to be gathered. The PLA matrix was investigated with DSC to determine the de-gree of crystallinity, and the composite structures were studied by SEM. At a first glance, the BC and CNFs were believed to be well dispersed and have a random in-plane orientation in the composites. Investigation of cross sec-tions of the composites revealed, however, that the fibrils had agglomerated and formed micrometre scale platelets. A SEM image of the cross section of a composite reinforced by BC, where a platelet of agglomerated fibrils is clearly visible, is shown in Figure 6.

31

Figure 6. Cross section of a composite reinforced by BC. The image shows a platelet of agglomerated fibrils embedded in the PLA matrix.

Several images of the fracture cross section of the composites were studied and the aspect ratio of the fibril platelets was determined by measuring the thickness and the diameter of the platelets. Five measurements of the platelet aspect ratio were done for each type of composite. It showed that the CNFs formed thick platelets with an aspect ratio α of 0.15, while the BC formed thin platelets with a smaller aspect ratio with a value of 0.02. Images of cross sections of composites, reinforced by BC and CNFs, are shown in Figure 7.

Figure 7. SEM images of cross sections of composites reinforced by (a) CNFs and (b) BC. It can be seen in the images that CNFs form thicker platelets than the BCs.

MCC consists of highly crystalline cellulose and can be produced from wood pulp by acid hydrolysis, where the amorphous cellulose is removed. The polycrystalline aggregates are assumed to have arbitrary geometry with sta-tistically isotropic properties and an aspect ratio around 1 [68]. A SEM im-age showing a MCC particle with a diameter in the 10-20 µm range can be seen in Figure 8.

32

Figure 8. SEM image showing a MCC particle.

With the structural information of the composites, a micromechanical model based on three steps was developed. A schematic illustration of the model-ling steps is shown in Figure 9. The idea of the model was to link the elastic properties of the composite to that of the different constituents. With such model, the elastic properties of the composites can be divided into the con-tributing stiffness of the PLA matrix, the platelets, and finally into the fibrils. In order to divide the elastic properties of the composite, into PLA and plate-lets contributions, a homogenization of the PLA was performed as an initial step. This is described in STEP I. PLA is a semicrystalline polymer consist-ing of crystallites embedded in a matrix in the amorphous state [69]. To ob-tain the homogenized elastic properties of the PLA, Hashin micromechanical model [56] was used. In STEP II, partition of the elastic properties of the composite, into the contributing elastic properties of the homogenized PLA matrix and the platelets, is described. It was assumed that the platelets of agglomerated fibrils could be considered as oblate spheroids, embedded in the PLA, with their axis of revolution perpendicular to the plane of the com-posite, as can be seen in the schematic image in Figure 10.

With this assumption, a self-consistent Mori Tanaka model [50] was used. The model expresses the stiffness of a composite as a function of the stiff-ness of aligned spheroids, of oblate or prolate shape. In STEP III, the elastic properties of platelets were divided into the contributing elastic properties of the fibrils. For that, 3D laminate theory was used.

When applying all three steps in the micromechanical model, the elastic properties of the composites can be described as a function of the elastic properties of the fibrils. For the purpose of using the model for back-calculation to estimate the contributing stiffness of the fibrils, some assump-tions regarding the elastic properties of the fibrils needed to be made.

33

Figure 9. Schematic illustration of the linked modelling scales, used to relate the stiffness of fibrils to that of the macroscopic composites.

, , , → , , ,

, , , , → ,

STEP I Homogenization of semi-crystalline polymer matrix Calculation of the elastic proper-ties of a matrix containing crys-talline aggregates:

or

, → ,

STEP II Partition of the compo-site The flat platelets of fibril aggre-gate can be considered as oblate inclusions and a partition of the composite into fibril aggregate and matrix can be done:

→

STEP III Partition of the plate-lets Division of the platelets of fibrils into single fibrils:

34

Figure 10. Schematic image of a composite containing the flat platelets of agglomer-ated fibrils that can be considered as oblate spheroids lying in the plane of the com-posite.

To relate the composite stiffness to one characteristic parameter of the fi-brils, anisotropy ratios of the fibril elastic moduli with respect to the fibril longitudinal Young’s modulus were used. The fibrils were assumed to have transversely isotropic properties and the anisotropy ratios were based on results of micromechanical modelling of a single CNF, characterized by HR-TEM, as described in Paper I. These ratios are given by

0.36, (38)

0.055, (39)

and

0.14. (40)

The Poisson ratios used, also taken from the modelling, were = 0.24,

= 0.09 and = 0.32. With these ratios, the composite stiffness became a function of only the longitudinal Young’s modulus of the fibrils. As input data in the micromechanical model, the elastic properties of the composites and pure PLA, measured by tensile test, were used. The elastic properties of PLA crystallites were taken from molecular dynamic simulation presented by Lin et al. [70] The back-calculated effective stiffness values of the rein-forcing fibrils are given in Table 1.

Table 1. Young’s moduli of the composites and the different types of fibrils.

Fibril type Composite Young’s modulus [GPa]

Fibril Young’s modulus [GPa]

CNF 5.00 65.1 BC 5.78 61.4 MCC 4.05 38.3

35

In Paper III, the effective stiffness of CNFs were calculated from elastic properties of CNF films. The paper investigates numerous mechanical mod-els that can be used for the purpose of relating the elastic properties of films to that of CNFs. The models investigated were classical laminate theory (e.g. Tsai [60]), the Tsai-Pagano model [62], the Cox model [63] and the Krenchel model [64]. For each model, the assumptions made to derive the expression were discussed, and the different models were compared in rela-tion to the underlying assumptions. In classical laminate theory (CLT), all in-plane elastic constants of the fibrils are taken into account, and all types of in-plane loading of the constituent fibrils are considered. The Tsai-Pagano model relates the Young’s modulus of a CNF film to the axial and transverse Young’s moduli of the CNFs. The shear modulus of the CNFs is assumed to be a function of the transverse Young’s modulus of the CNFs, and is thereby indirectly related to the film stiffness. The expressions of the Young’s modu-lus derived by Cox and Krenchel include only the axial Young’s modulus. To graphically show the influence of the assumptions made in the deriva-tions of the different models, the Young’s modulus of a CNF film was calcu-lated as a function of the transverse Young’s modulus of the CNFs. A plausible value of the longitudinal Young’s modulus, = 50 GPa, was cho-sen. In CLT, three different shear moduli were used, = /5, = /10, and = 0. The predicted Young’s modulus of a film, calculated with the different models as a function of the transverse Young’s modulus of the CNF, is shown in Figure 11.

Figure 11. Young’s modulus of a CNF film as a function of the transverse Young’s modulus of the CNF, calculated by different models.

36

It can be seen that for small transverse Young’s moduli and shear moduli, the models gives similar results. As the transverse modulus and shear modu-lus increase, the difference between the models increases.

Considering the models investigated in Paper III, CLT is most likely the model that best capture the mechanical behaviour of dense films since it includes the in-plane anisotropic behaviour of the CNFs. CLT was therefore used to back-calculate the effective stiffness of the CNFs from the stiffness of films. Films of three types of CNFs were manufactured. The CNFs were extracted from softwood pulp and hardwood pulp, where some of the soft-wood pulp was TEMPO pretreated. The in-plane isotropic Young’s modulus of the films was measured, and CLT was used to determine the effective axial Young’s modulus of the CNFs. As CLT uses all in-plane elastic con-stants of the CNFs, anisotropy ratios were assumed in the same way as in Paper II. With the anisotropy ratios, the Young’s modulus of the films could be expressed as a function of only the longitudinal Young’s modulus of the CNFs. The Young’s moduli of the CNFs determined by back-calculation from the stiffness of the produced films, as well as from stiffness values of films found in the literature, are presented in Table 2. The back-calculated values of the CNF stiffness are lower than the CNF stiffness values found in Paper II and III. There are also scatter in the stiffness values, which could be an effect of differences in moisture content. The ambient conditions affect-ing the moisture absorption was not uniform and sometimes not even docu-mented in the cited studies. Nevertheless, the presented approach gives stiff-ness values in the right order of magnitude and could probably be used to rank the stiffness of the constitutive fibrils in a wide variety of films.

Table 2. Stiffness of fibrils, back-calculated from stiffness of different films. The fibrils are extracted from different kinds of wood pulp with the use of different pre-treatments and processing methods. SW and HW stand for softwood and hardwood, respectively.

Raw material / pretreatment

Film Young’s modulus [GPa]

Porosity [%]

Fibril Young’s modulus [GPa]

Reference

SW / Non 9.7 24 20 Paper III HW / Non 8.5 29 25 Paper III SW / TEMPO 12.7 0 27 Paper III SW / enzymatic

14 10.5 33 [71]

SW / Non 16 1.33 34 [29] SW / enzymatic

14.7 19 38 [37]

SW / N/A 13.5 28 39 [28] SW / Non 15.7 45.9 61 [36]

37

The values of the CNF stiffness, estimated by modelling of the internal structure of a CNF (Paper I) and by back-calculation from elastic properties of CNF based materials (Papers II and III), were all in the same range and in the same order of magnitude. A value of 65 GPa was established in both Paper I and Paper II. The stiffness values obtained in the study in Paper III are somewhat lower with one exception of 61 GPa. Young’s moduli of the CNF, obtained in Papers I-III, are summarised in Table 3.

Table 3. CNF stiffness values estimated in Papers I-III by different methods.

Method CNF Young’s modulus [GPa]

Paper

Modelling of the CNF structure

65 Paper I

Back-calculation from elastic properties of composites

65 Paper II

Back-calculation from Elastic properties of films

20-61 Paper III

4.2 Properties of cellulose nanofibril materials In Paper IV-VI, properties related to mechanical performance of CNF ma-

terials were investigated. In Paper IV, the stiffness of CNF films were de-termined by out-of-plane stretching with a nanorobotic mounted in a SEM. Films were manufactured by CNFs from three different raw materials, soft-wood pulp, hardwood pulp and TEMPO pretreated softwood pulp. For the nanorobotic stretching, the films were cut in small strips with a width of 2 mm and mounted in the nanorobotic device. The small strips were clamped in two ends and subjected to a force in the middle of the sample by moving the samples towards a force sensor. The force sensor tip had an area of 50 µm × 50 µm. The experimental setup is shown in Figure 12. The subjected force and the movement of the films were registered. The information was used to model the behaviour of the films using finite element method (FEM) simulation in order to estimate the stiffness of the films. In the simulation, the symmetry of the sample was used so that only one quarter of the sample was considered, as shown in Figure 13. The FEM model was then adjusted to fit the nanorobotic data by adjusting the Young’s modulus of the CNF films. For comparison, conventional tensile tests were performed on CNF film strips. The Young’s modulus determined from the FEM simulation and the Young’s modulus measured from tensile test are shown in Figure 14.

38

Figure 12. Experimental setup inside the SEM. The schematic (a) shows the stretched CNF film (exaggerated), which is clamped at both ends. Images (b) and (c) show the film and the tip sensor at different magnifications.

Figure 13. FEM model showing the displacement field of a deformed CNF film. The scale bar shows the total displacement of the film in metre. The blue area denotes the most indented area of the film sample.

Figure 14. Young’s modulus of the films made of CNFs from hardwood (HW CNF film) and from softwood (SW CNF film) estimated from tensile test (white) and nanorobotic out-of-plane stretching test (grey).

39

When investigating mechanical properties of materials, the thickness of the samples has to be carefully measured. The thicknesses of the films studied in Paper IV were determined from SEM images. The images were taken of the cross sections of the films, and the average film thickness was determined. In tensile testing, the expressions for strength and Young’s modulus, in terms of load and other properties, are inversely proportional to the sample thickness. In flexural testing, strength and stiffness are inversely proportional to the cube of the sample thickness. When dealing with CNF films, accurate measurements of the films, having a variable thickness, are therefore of im-portance. Many papers dealing with CNF films measure the film thickness with a micrometer screw gauge [43, 72, 73]. The thicknesses that are meas-ured in such ways are the maximum thicknesses, resulting in a significant error when determining mechanical properties of the films with a high varia-bility of local thickness, such as unsmooth CNF films. To illustrate the dif-ference between the thickness that is measured using a micrometre screw gauge and the average thickness, that can be determined from SEM images of cross sections, a binarised SEM image of a cross section of a CNF film is shown in Figure 15. In the image, the average thickness is drawn with green dashed lines and the maximum thickness is drawn with red dashed lines.

Figure 15. Cross section of a CNF film. The green dashed lines show the mean thickness of the film measured from the SEM image. The red dashed lines show the thickness that would have been measured with a micrometer screw gauge on this part of the sample.

This difference in thickness values obtained by different measuring methods, and the importance of accurate measurements of film thicknesses, were in-vestigated in Paper V. In the study presented in In Paper V, the statistical distribution of CNF film thicknesses was investigated by studying SEM images of cross sections of films. Ten images of eight different films, pro-duced from CNFs fibrillated with different methods and from different raw materials, were examined. The CNFs used to produce the films were fibril-lated from TEMPO pretreated and non-pretreated softwood and hardwood pulp, subjected to three and five homogenization passes. The samples were named from the raw materials used and the fibrillation process of the CNFs. Table 4 summarises the codes of the films produced from the non-pretreated

40

pulp, and Table 5 summarises the codes of the films produced from the TEMPO pretreated pulp.

Table 4. Sample codes for the different films produced from non-pretreated pulp.

Table 5. Sample codes for the different films produced from TEMPO pretreated pulp.

The thickness of the films was measured from the SEM images by counting the number of pixels through the thickness along the films. In total 26 000 thicknesses were measured for each sample, with a pixel resolution of 50 nm × 50 nm. The thicknesses were also measured with a micrometer screw gauge. The thickness distributions of the CNF films can be seen in Figure 16 for the films of CNFs made from non-pretreated pulp, and in Figure 17 for the films of CNFs made from TEMPO pretreated pulp. In each histogram, the mean thickness is marked with a green dashed line, and the thickness measured with the micrometer screw gauge is marked with a red dashed line. The coefficient of variation, CV, was calculated for the thickness distribu-tion of each film. It can be seen in Figure 16 and Figure 17 that the pretreat-ment results in a more homogenous thickness, irrespective of the pulp origin. Additional passes through the homogenizer reduces the thickness variability further. It can also be seen in these figures that the thickness measured with the micrometer screw gauge is greatly overestimated compared to the mean thickness of the films obtained from the SEM images. The roughness of the film surface is substantial, and when determining the mechanical properties of materials, such as strength and stiffness, a significant error will be intro-duced if the extreme value of the maximum thickness is used instead of the average thickness. By calculating the ratio between the average thickness obtained from the SEM images, and the maximum thickness measured with a micrometre screw gauge, a scale factor was determined which could be used to estimate the mean thickness from measurements with a micrometre screw gauge. These scale factors are presented in Table 6.

Non-pretreated raw material

3 passes of homogenization

5 passes of homogenization

Softwood SW-NP-3 HW-NP-3 Hardwood HW-NP-3 HW-NP-3

TEMPO pretreated raw material

3 passes of homogenization

5 passes of homogenization

Softwood SW-TP-3 SW-TP-5 Hardwood HW-TP-3 HW-TP-5

41

Figure 16. Probability density functions of the thickness of the films made of non-pretreated CNFs, together with the mean thickness value and the thickness measured with micrometer screw gauge. The distributions are plotted for the samples made of (a) softwood pulp, SW-NP-3, and (b) hardwood pulp, HW-NP-3, subjected to three homogenization passes, and from (c) softwood pulp, SW-NP-5, and (d) hardwood pulp, HW-NP-5, subjected to five homogenization passes.

(c)

(a) (b)

(d)

42

Figure 17. Probability density functions of the thickness of films made of TEMPO pretreated CNFs, together with the mean thickness value and the thickness measured with micrometer screw gauge. The distributions are plotted for the samples made of (a) softwood pulp, SW-TP-3, and (b) hardwood pulp, HW-TP-3, subjected to three homogenization passes, and from (c) softwood pulp, SW-TP-5, and (d) hardwood pulp, HW-TP-5, subjected to five homogenization passes.

(a) (b)

(c) (d)

43

Table 6. Ratios between the mean thickness obtained from the SEM images, and the maximum thickness measured with the micrometre screw gauge.

In Table 7, the ratios for the mean thickness and maximum thickness ob-tained from the SEM images are presented. It can be seen that the ratios for the films of non-pretreated CNFs are small, ranging from 0.64 to 0.70. This means that the difference between the mean thickness and the maximum thickness is significant. The ratios for the TEMPO pretreated films ranging from 0.73 to 0.87 with the largest value corresponding to the films of hard-wood CNFs homogenized with five passes.

Table 7. Ratios between the mean thickness and the maximum thickness of the films taken from the SEM images.

Pretreatment 3 passes of homogenization 5 passes of homogenization Softwood Hardwood Softwood Hardwood Non 0.64 0.64 0.65 0.70 TEMPO 0.73 0.72 0.81 0.87

To investigate the thickness variation further, various distribution functions were fitted to the thickness data. The distributed functions used were the Gamma distribution, the Weibull distribution, the lognormal distribution, extreme value distribution type I, and the normal distribution (e.g. Bury [74]). For the three first mentioned distributions, the three-parameter case was used to be able to shift the distribution and to avoid limiting the distribu-tion to start at zero. No conclusive statements could be made on which dis-tribution function is the most suitable to represent the thickness variations. It could, however, be confirmed that the often used normal distribution were consistently the distribution function showing the worst goodness of fit. The distributions that best described the thickness distribution of the various films are given in Table 8.

Table 8. The best fit of the distributions for the thickness variation of different film.

Pretreatment 3 passes of homogenization 5 passes of homogenization Softwood Hardwood Softwood Hardwood Non Gamma Gamma Gamma Weibull TEMPO Lognormal Lognormal Lognormal Weibull

The studies in Papers II-V are all focusing on the properties of CNF based materials with a random in-plane orientation of the fibrils. However, a more

Pretreatment 3 passes of homogenization 5 passes of homogenization Softwood Hardwood Softwood Hardwood Non 0.57 0.51 0.86 0.71 TEMPO 0.82 0.72 0.91 0.65

44

advantageous use of the stiff fibrils would be in an aligned configuration. To be able to control the mechanical performance of the CNF based materials, a technique to orient the nanofibrils is essential. In the last years many re-searchers have focused on producing materials with aligned CNFs using different methods such as wet-spinning [47], wet extrusion [48] and extru-sion followed by hot drawing [75]. In Paper VI, films of oriented CNFs were produced by cold-drawing, and the orientation of the fibrils were analysed using XRD. Cold-drawing has been used by Gindl and Keckes [76] for films of microcrystalline cellulose and by Sehaqui et al. [49] for films of CNFs. Both studies showed films with high degree of fibril alignment and improved mechanical properties, achieving a Young’s modulus of 33 GPa. The study carried out in Paper VI focuses on the realignment that occurs during the cold-drawing and quantification of the orientation distribution of the CNFs. The films were subjected to increasing strain, ranging from 35% to 50%, and the degree of alignment of the fibrils were studied. When cold-drawing a never dried CNF film, or hydrogel, a reorientation of the CNFs will occur. In the hydrogels, the nanofibrils are lying in the water matrix and are not fully bonded to each other. When a strain is applied, the nanofibrils will reorient and align towards the direction of the applied strain.

The XRD measurements were performed on the dried stretched films and provided information of the crystal orientation in films. For cellulose crys-tals, the 200-diffraction peak can be used to determine the orientation of the crystals. As the c axis of these crystals is oriented along the direction of the CNFs, the orientation of the cellulose crystals corresponds to the orientation of the CNFs [77]. The XRD patterns were carefully analysed and the 200-diffraction peak investigated. An example of a XRD pattern, showing the 200-diffraction peak, the axes of the scattering angle 2θ, and the azimuthal angel φ, are shown in Figure 18. The XRD was measured from a sample subjected to 44% strain.

By integration of the intensity on a ring corresponding to scatter from crystalline cellulose, the orientation distribution of the CNFs was plotted as a function of the azimuthal angle. The orientation distributions of the CNFs for the different samples are shown in Figure 19.

A simple model for reorientation of fibrils in a hydrogel matrix, subjected to a strain in one direction, was developed. For a fibril originally oriented with an angle from the x-axis, an applied strain in the x-direction will re-sult in a new fibril angle ′ from the x-axis. According to the presented model, the new fibril angle can be expressed as a function of the applied strain and by the in-plane Poisson ratio by

tan1

1tan . (41)

45

Figure 18. XRD pattern from a CNF film subjected to a strain of 44%. The image shows the difference in intensity by the different colours, and the axes of the scatter-ing angle 2 and the azimuthal angle .

Figure 19. Orientation distribution of the CNFs in the films subjected to various strain levels.

2

200

46

With the model, the orientation distribution of fibrils in hydrogel composites subjected to the same amount of strain as in the experimental procedures was plotted, which can be seen in Figure 20.

Figure 20. Orientation distribution of CNFs in a hydrogel composite subjected to various strain levels, according to the presented model.

As can be seen in the figure, the CNF alignment, predicted by the model, is not as strong as that seen in the experimental results. All of the probability distribution curves are flatter than those observed by XRD measurements of CNF orientation in real films. The difference could probably be explained by additional CNF orientation mechanisms, as well as the difference in local and average strain. The geometrical model is in principle based on an ideal-ised elastic situation, where the CNFs are tilted and the matrix is deformed according to the global strain. In practice, there is a flow in the liquid matrix, which contributes to additional alignment. Furthermore, the experimental strain was measured by the displacement of the end clamps, which gives the average strain of the sample. The XRD measurements were conducted on a spot in the middle of the sample where the local strain can be higher than the average strain subjected to the sample.

To study the degree of orientation with a single parameter, the full width at half maximum (FWHM) of the unimodal bell-shaped distributions gath-ered from the experiment were plotted as a function of the applied strain. The plot is shown in Figure 21. Being based on experimental orientation distribution functions determined by XRD, there is a remarkable linear cor-

47

relation between the FWHM and the applied strain, with a minute deviation for the samples subjected to 44% strain. The high degree of linear correlation indicates that an orientation-strain relationship can be formulated, preferably based on the physical deformation mechanisms.

Figure 21. Full width at half maximum of the orientation distribution peaks as a function of the applied strain.

48

5. Conclusions

The work presented in this thesis investigates different phenomena of the CNF with respect to different length scales. The phenomena studied are all related to the elasticity of the CNF, with the aim to gain more knowledge about CNFs for materials manufacturing. CNFs are nanosized fibrils with high aspect ratio and high stiffness. Direct measurements of elastic proper-ties of the CNF are challenging why this thesis present some alternative methods to investigate the elastic properties of single CNFs. The first meth-od was a modelling approach where the internal structure of the CNF was modelled in terms of crystalline and disordered cellulose. A self-consistent Mori-Tanaka scheme was used to model the elastic properties of the CNF, which resulted in an axial stiffness of 65 GPa. The axial stiffness of the CNF is highly dependent on the crystallite aspect ratio and much could be won if gentler defibrillation processes were developed that do not significantly re-duce the relative crystallite length. Another method used to investigate the elastic properties of CNFs was micromechanical back-calculation, where the effective stiffness of CNFs was estimated from the stiffness of materials reinforced by CNFs. Films and composite materials were produced, and the structure and elastic properties of the materials were investigated. Suitable micromechanical models were chosen, and the effective stiffness of the CNFs was calculated. From the composite stiffness, a CNF stiffness of 65 GPa was obtained. From the stiffness of the films, a somewhat lower CNF stiffness was observed, with a value of 20-27 GPa.

All estimates of the elastic properties of the CNF are in the same order of magnitude. The axial stiffness of 65 GPa was established from two different methods which indicates that it may be a reliable value. However, there are many natural deviations in the composition of the CNFs, such as difference in crystallinity and crystallite aspect ratio, which affects the elastic proper-ties. With these natural deviations, it is not possible to adopt one single value of the CNF stiffness. The value of 65 GPa can only be regarded as an esti-mate for a typical CNF.

Films are often produced to quantitatively measure the elastic properties of the cellulose nanofibrils. The CNF films have commonly a thickness of some tens of micrometre, random in-plane orientated fibrils, and a rough surface. The mechanical properties of the film are determined from e.g. ten-sile tests and the dimensions of the film sample. Typically, the tensile prop-erties are accurately measured, while the thickness dimensions of the films

49

are measured with a fast and relatively inaccurate method, such as with a micrometer screw gauge. The film surface is typically rough, and when measuring the thickness with such method, the maximum thickness is ob-served. As the mechanical properties of the film are proportional to the in-verse of the film thickness, accurate measurements of the film thickness is of importance. In this thesis, different properties related to the mechanical per-formance were investigated. The thickness variability of the films was quan-tified by studying SEM images of film cross sections. The thickness of films of CNFs, produced from different raw materials and by different defibrilla-tion processes, was investigated. It showed that the thickness measured with a conventional micrometer screw gauge could be as much as 95% higher than the mean thickness of the film. Scale factors were suggested that can be used to scale the mechanical properties of films when using a micrometer screw gauge.

In the last study presented in the thesis, films of partly aligned fibrils were produced and analysed. Hydrogel films with randomly oriented cellulose nanofibrils were subjected to a controlled strain aiming to align the fibrils in the films. The hydrogels were then dried and the fibril orientation distribu-tion was analysed. It showed a strain induced reorientation of the fibrils and that the orientation increased with the increased strain. To compare the de-gree of orientation as a function of the applied strain, the full width at half maximum of the orientation probability curves were plotted with respect to the applied strain and a high linear correlation was found.

50

7. Outlook

CNF based materials have high potential as a load carrying material. The CNFs have low density, excellent mechanical properties, and are extracted from renewable resources. Furthermore, the CNFs have a functional surface available for modifications to enable control of engineering properties aim-ing for a certain application. In order to commercialize and make practical use of CNF based materials, a deeper knowledge of the CNF mechanical behaviour is necessary. The list below highlights some of the issues that needs further investigation.

For CNF composite materials, the fibril-matrix adhesion has impact on

the mechanical performance as it controls the stress transfer between the fibrils and the matrix. To use the full capacity of the stiff fibrils, deeper investigation is needed in order to understand and to be able to control the fibril-matrix interface.

Structural components are often subjected to uniaxial loads. It is there-fore of importance to develop stable manufacturing methods where the fibril orientation can be controlled to optimize the mechanical perfor-mance in the loading direction.

Theoretically, the advantage of using nanosized fibrils as reinforcement in composite materials, in comparison to for example reinforcement of millimetre sized fibres, is the high surface area of the fibrils. In compo-site materials, the stiff fibres induce a stiffened immobilized region of the polymer matrix material around the fibres. For composite materials reinforced by nanosized fibril, this effect will be more pronounced and result in stiffer composites. It is essential to find manufacturing tech-niques to produce composites with fully dispersed fibrils and to prevent agglomeration of the fibrils in the composite.

CNFs are highly hydrophilic and the moisture uptake affects the me-chanical properties of the CNFs. To use the CNFs in structural compo-nent, a decrease of hydrophilicity needs to be achieved. A decrease of hydrophilicity would also enhance the compatibility of the CNFs with conventional hydrophobic polymers, such as polyethylene and polypro-pylene.

The defibrillation process of the CNFs has high energy consumption and the production of CNFs, from low-cost raw materials such as wood, is still expensive. The defibrillation process is rough and causes degrada-

51

tion of the crystal regions of the CNFs. More efforts needs to be made on the defibrillation process to lower the energy consumption and, espe-cially, to maintain the stiff structure of the CNFs.

During the last decade, much attention has been given by the research community to the chemistry and manufacturing of laboratory scale batches of CNF based materials. It has of course been useful to show what properties could be achieved with this relatively new class of mate-rial. Nevertheless, up-scaling to pilot plants and industrial production remains a main limitation. In this respect, processing and mechanical engineering needs to complement the wide ongoing materials chemistry research. Meanwhile, mechanics can also prove itself more useful in material development, in identifying the nano- and microstructural fea-tures that contribute the desired properties of the material. The present thesis has been a contribution in this domain. Further developments can make material mechanics much more useful in rationalized development of materials, as compared with empirical or qualitative studies of the ef-fects of materials composition and processing parameters on the result-ing material properties.

52

6. Sammanfattning på svenska

Med ökande miljömedvetenhet växer intresset för förnyelsebara material och alternativ till de oljebaserade plasterna blir alltmer efterfrågade. Även om de flesta plaster, eller polymerer, som används i komponenter är oljebaserade, utgör cellulosa den mest förekommande polymeren i världen. Cellulosa kan ha både oordnade strukturer och ordnade, så kallad kristallina, strukturer. Den kristallina cellulosan har mycket bra mekaniska egenskaper och är den komponent som ger träd och växter dess stabilitet och styvhet. Studien som presenteras i denna avhandling undersöker de elastiska egenskaperna hos cellulosananofibriller, förkortat CNF, i syfte att använda dessa för att produ-cera styva miljövänliga material.

CNF är den minsta strukturella beståndsdelen i trä med en diameter på under 100 nm och en längd på några mikrometer. Fibrillerna är lokaliserade i träfiberns cellvägg. Träfibrernas cellvägg består av flera lager där fibrillerna ligger orienterade i olika riktningar för att maximera de mekaniska egen-skaperna hos fibrerna. Genom att bearbeta träfibrerna mekaniskt och ke-miskt kan CNF frigöras från cellväggen. CNF har en kompositstruktur där regioner av kristallin cellulosa, kallad kristalliter, ligger inbäddade i en ma-tris av oordnad cellulosa, se Figur XX. Dessa kristalliter har en axiell elasti-citetsmodul på ca 150 GPa och bidrar i huvudsak till fibrillernas höga styv-het. På grund av CNFs storlek är det dock svårt att direkt uppskatta dess mekaniska egenskaper, så som styvhet och styrka, och alternativa metoder för att bestämma de elastiska egenskaperna måste användas. För att kunna använda CNF för materialtillverkning behövs mer kunskap om dess elastiska egenskaper vilket var syftet med studierna i artiklarna I-III.

Ett av de vanligaste sätten att använda CNF för materialtillverkning är att producera filmer, också kallat nanopapper. Filmerna produceras lätt genom att filtrera en utspädd lösning av CNF och vatten, en process som likar den vid tillverkning av papper. Filmernas mekaniska egenskaper kan testas ge-nom till exempel ett dragprov. Filmernas egenskaper är starkt kopplade till orienteringen av fibrillerna och porositeten. Filmers egenskaper så som styv-het och struktur studerades i artiklarna IV-VI.

På följande sidor ges en kort sammanfattning av varje artikel och dess resul-tat.

53

I artikel I studerades CNFs inre struktur med högupplöst elektronmikroskopi och röntgendiffraktion. Mätningarna visade tydligt CNFs inre komposit-struktur samt kristalliternas storlek, omfattning och orientering. Den oord-nade cellulosan antogs vara amorf cellulosa och de elastiska egenskaperna hos en CNF modellerades med en mikromekanisk modell. De elastiska egen-skaperna hos den kristallina och amorfa cellulosan hämtades från litteraturen där egenskaperna tagits fram genom molekylärdynamiksimulering. Från modelleringen uppskattades elasticitetsmodulen för CNF till 65 GPa.

I artikel II producerades kompositer med CNF, bakteriell cellulosa och mikrokristallin cellulosa som förstärkning. Kompositernas elastiska egen-skaper testades och de bidragande elastiska egenskaperna hos fibrillerna tillbakaberäknades med en mikromekanisk modell. Som matrismaterial an-vändes den nedbrytbara polymeren polylaktid (PLA) som producerats av stärkelse från majs. Fibrillernas spridning och orientering studerades med svepelektronmikroskop och användes som indata i den mikromekaniska modellen för att beräkna de bidragande elastiska egenskaperna hos fibriller-na. Den beräknade effektiva styvheten för CNF uppgick till 65 GPa, vilket stödjer resultatet i artikel I. Den effektiva styvheten för bakteriella cellulosan och mikrokristallina cellulosan uppgick till 61 respektive 38 GPa.

I artikel III användes CNF, som tagits fram från olika träslag och genom olika bearbetningar, för att producera filmer. Filmernas elastiska egenskaper testades och flera mekaniska modeller som kopplar filmernas elastiska egen-skaper till egenskaperna hos CNF utvärderades. Den modell som bäst ansågs representera nätverket av fibriller och tar hänsyn till fibrillernas anisotropa egenskaper var klassisk laminatteori. Från filmernas elastiska egenskaper tillbakaberäknades de elastiska egenskaperna hos CNF. Elastiska egenskaper hos filmer presenterade i litteraturen användes också i studien. En styvhet på 20-61 GPa uppskattades för CNF vilket är något lägre än de värden som erhölls i artiklarna I och II.

I artikel IV producerades filmer och de elastiska egenskaperna testades ge-nom böjprov i en nanorobot. Filmer, med en storlek på ca 1 × 2 mm, monte-rades i nanoroboten och utsattes för en last i mitten av filmen. Utböjningen mättes som funktion av den pålagda lasten och informationen användes för att göra en finit elementmodell som anpassades för att uppskatta filmernas elastiska egenskaper. Filmerna testades också genom dragprov och de upp-mätta elastiska egenskaperna jämfördes för de två metoderna. Elasticitets-modulen som uppskattats genom finita elementmodellen stämde bra överens med den elasticitetsmodulen som uppmättes genom dragprov. Elasticitets-modulen för filmerna uppmättes till 8-10 GPa med båda metoderna.

54

I artikel V undersöktes tjockleksvariationen hos filmer producerade av CNF. När man beräknar de mekaniska egenskaperna hos ett material med inform-ation från ett dragprov, är egenskaperna omvänt proportionerliga mot materialprovets laterala dimensioner. Vid böjprov är egenskaperna omvänt proportionerliga mot materialprovets laterala dimensioner i kubik. Det är därför av stor vikt att mäta bredd och tjocklek på ett korrekt sätt. Dock pre-senteras många arbeten i litteraturen där snabba och osäkra mätningar av filmtjockleken har utförts. Ett vanligt sätt att mäta tjockleken är genom an-vändning av en mikrometerskruv, vilket generellt mäter den maximala film-tjockleken. Vid beräkning av de mekaniska egenskaperna används då den maximala tjocklek vilket resulterar i felaktiga uppskattningar av filmernas egenskaper. Filmer av CNF har skrovlig yta och har ofta en tjocklek på några tiotals mikrometer. I studien mättes filmers tjockleksvariation genom att studera elektronmikroskopbilder av filmtvärsnitt. De visade att den max-imala tjockleken hos filmen kan vara upp till 95 % större än medeltjockle-ken.

I artikel VI producerades filmer av delvis orienterade CNF. Filmerna utsattes för en kontrollerad töjning, innan filmerna torkades, i syfte att omorientera fibrillerna. Filmerna utgjordes då av en så kallad hydrogel och hade ett vat-teninnehåll på 70-90 %. Hydrogelerna torkades i den töjda konfigurationen och fibrillernas orientering studerades genom röntgendiffraktion. En enkel mikromekanisk modell presenterades i syfte att försöka förklara fibrillernas omorientering som funktion av den pålagda töjningen. Mätningarna visade att fibrillernas orientering ökade med ökad töjning och att orienteringen var mer påtaglig än modellen förutspådde. För att studera förhållandet mellan graden av orientering som funktion av den pålagda töjningen, plottades bredden vid halva höjden av sannolikhetskurvorna mot den pålagda töjning-en. Graden av orientering och den pålagda töjningen visade ett starkt linjärt samband.

55

8. References

1. Bledzki, A. and Gassan, J. (1999) Composites reinforced with cellulose based fibres. Progress in polymer science, 24(2):221-274

2. Neagu, R. C., Gamstedt, E. K., and Berthold, F. (2006) Stiffness contribution of various wood fibers to composite materials. Journal of composite materials, 40(8):663-699

3. Saheb, D. N. and Jog, J. (1999) Natural fiber polymer composites: a review. Advances in polymer technology, 18(4):351-363

4. Chinga-Carrasco, G. (2011) Cellulose fibres, nanofibrils and microfibrils: The morphological sequence of MFC components from a plant physiology and fibre technology point of view. Nanoscale Research Letters, 6(1):1-7

5. Isogai, A. (2013) Wood nanocelluloses: fundamentals and applications as new bio-based nanomaterials. Journal of wood science, 59(6):449-459

6. Eichhorn, S. J., Dufresne, A., Aranguren, M., Marcovich, N. E., Capadona, J. R., Rowan, S. J., et al. (2010) Review: current international research into cellulose nanofibres and nanocomposites. Journal of Materials Science, 45(1):1-33

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

8. Fahlén, J. and Salmén, L. (2003) Cross-sectional structure of the secondary wall of wood fibers as affected by processing. Journal of materials science, 38(1):119-126

9. Barnett, J. R. and Bonham, V. A. (2004) Cellulose microfibril angle in the cell wall of wood fibres. Biological Reviews, 79(2):461-472

10. Barnett, J. and Bonham, V. A. (2004) Cellulose microfibril angle in the cell wall of wood fibres. Biological reviews, 79(2):461-472

11. Lavoine, N., Desloges, I., Dufresne, A., and Bras, J. (2012) Microfibrillated cellulose–Its barrier properties and applications in cellulosic materials: A review. Carbohydrate polymers, 90(2):735-764

12. Eichhorn, S. J. and Davies, G. R. (2006) Modelling the crystalline deformation of native and regenerated cellulose. Cellulose, 13(3):291-307

13. Tanaka, F. and Iwata, T. (2006) Estimation of the Elastic Modulus of Cellulose Crystal by Molecular Mechanics Simulation. Cellulose, 13(5):509-517

56

14. Turbak, A. F., Snyder, F. W., and Sandberg, K. R. (1983) Microfibrillated cellulose, a new cellulose product: properties, uses, and commercial potential. Journal of Applied Polymer Science, 37:815-27

15. Herrick, F. W., Casebier, R. L., Hamilton, J. K., and Sandberg, K. R., "Microfibrillated cellulose: morphology and accessibility," in J Appl Polym Sci: Appl Polym Symp (United States), 1983.

16. Nakagaito, A. N. and 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, 78(4):547-552

17. Klemm, D., Kramer, F., Moritz, S., Lindström, T., Ankerfors, M., Gray, D., et al. (2011) Nanocelluloses: A New Family of Nature‐Based Materials. Angewandte Chemie International Edition, 50(24):5438-5466

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

19. Aulin, C., Gällstedt, M., and Lindström, T. (2010) Oxygen and oil barrier properties of microfibrillated cellulose films and coatings. Cellulose, 17(3):559-574

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

21. Xu, P., Donaldson, L. A., Gergely, Z. R., and Staehelin, L. A. (2007) Dual-axis electron tomography: a new approach for investigating the spatial organization of wood cellulose microfibrils. Wood Science and Technology, 41(2):101-116

22. Habibi, Y., Lucia, L. A., and Rojas, O. J. (2010) Cellulose nanocrystals: chemistry, self-assembly, and applications. Chemical reviews, 110(6):3479-3500

23. Cheng, Q., Wang, S., and Harper, D. P. (2009) Effects of process and source on elastic modulus of single cellulose fibrils evaluated by atomic force microscopy. Composites Part A Applied Science and Manufacturing, 40(5):583-588

24. Cheng, Q. and Wang, S. (2008) A method for testing the elastic modulus of single cellulose fibrils via atomic force microscopy. Composites Part A: Applied Science and Manufacturing, 39(12):1838-1843

25. Iwamoto, S., Kai, W., Isogai, A., and Iwata, T. (2009) Elastic modulus of single cellulose microfibrils from tunicate measured by atomic force microscopy. Biomacromolecules, 10(9):2571-2576

57

26. Guhados, G., Wan, W., and Hutter, J. L. (2005) Measurement of the Elastic Modulus of Single Bacterial Cellulose Fibers Using Atomic Force Microscopy. Langmuir, 21(14):6642-6646

27. Sun, Z., Zhao, X., Wang, X., and Ma, J. (2014) Multiscale modeling of the elastic properties of natural fibers based on a generalized method of cells and laminate analogy approach. Cellulose, 21(3):1135-1141

28. Sehaqui, H., Allais, M., Zhou, Q., and Berglund, L. A. (2011) Wood cellulose biocomposites with fibrous structures at micro- and nanoscale. Composites Science and Technology, 71(3):382-387

29. Yano, H. and Nakahara, S. (2004) Bio-composites produced from plant microfiber bundles with a nanometer unit web-like network. Journal of Materials Science, 39(5):1635-1638

30. Fujisawa, S., Okita, Y., Fukuzumi, H., Saito, T., and Isogai, A. (2011) Preparation and characterization of TEMPO-oxidized cellulose nanofibril films with free carboxyl groups. Carbohydrate Polymers, 84(1):579-583

31. Thostenson, E. T., Ren, Z., and Chou, T.-W. (2001) Advances in the science and technology of carbon nanotubes and their composites: a review. Composites science and technology, 61(13):1899-1912

32. Luo, J.-J. and Daniel, I. M. (2003) Characterization and modeling of mechanical behavior of polymer/clay nanocomposites. Composites Science and Technology, 63(11):1607-1616

33. Dzenis, Y. A. (2008) Structural nanocomposites. Science, 319:419-420

34. Lönnberg, H., Fogelström, L., Berglund, L., Malmström, E., and Hult, A. (2008) Surface grafting of microfibrillated cellulose with poly(ε-caprolactone) – Synthesis and characterization. European Polymer Journal, 44(9):2991-2997

35. Rusli, R. and Eichhorn, S. J. (2008) Determination of the stiffness of cellulose nanowhiskers and the fiber-matrix interface in a nanocomposite using Raman spectroscopy. Applied Physics Letters, 93:033111

36. Syverud, K. and Stenius, P. (2009) Strength and barrier properties of MFC films. Cellulose, 16(1):75-85

37. Henriksson, M., Berglund, L. A., Isaksson, P., Lindström, T., and Nishino, T. (2008) Cellulose nanopaper structures of high toughness. Biomacromolecules, 9(6):1579-1585

38. Galland, S., Andersson, R. L., Salajková, M., Ström, V., Olsson, R. T., and Berglund, L. A. (2013) Cellulose nanofibers decorated with magnetic nanoparticles–synthesis, structure and use in magnetized high toughness membranes for a prototype loudspeaker. Journal of Materials Chemistry C, 1(47):7963-7972

58

39. Nyström, G., Razaq, A., Strømme, M., Nyholm, L., and Mihranyan, A. (2009) Ultrafast all-polymer paper-based batteries. Nano letters, 9(10):3635-3639

40. Nyström, G., Mihranyan, A., Razaq, A., Lindström, T., Nyholm, L., and Strømme, M. (2010) A nanocellulose polypyrrole composite based on microfibrillated cellulose from wood. The Journal of Physical Chemistry B, 114(12):4178-4182

41. Carlsson, D. O., Nyström, G., Zhou, Q., Berglund, L. A., Nyholm, L., and Strømme, M. (2012) Electroactive nanofibrillated cellulose aerogel composites with tunable structural and electrochemical properties. Journal of Materials Chemistry, 22(36):19014-19024

42. Nakagaito, A. and Yano, H. (2008) Toughness enhancement of cellulose nanocomposites by alkali treatment of the reinforcing cellulose nanofibers. Cellulose, 15(2):323-331

43. Iwamoto, S., Nakagaito, A. N., and Yano, H. (2007) Nano-fibrillation of pulp fibers for the processing of transparent nanocomposites. Applied Physics A, 89(2):461-466

44. Suryanegara, L., Nakagaito, A. N., and Yano, H. (2009) The effect of crystallization of PLA on the thermal and mechanical properties of microfibrillated cellulose-reinforced PLA composites. Composites Science and Technology, 69(7):1187-1192

45. Nakagaito, A. N., Fujimura, A., Sakai, T., Hama, Y., and Yano, H. (2009) Production of microfibrillated cellulose (MFC)-reinforced polylactic acid (PLA) nanocomposites from sheets obtained by a papermaking-like process. Composites Science and Technology, 69(7–8):1293-1297

46. Jonoobi, M., Harun, J., Mathew, A. P., and 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

47. Iwamoto, S., Isogai, A., and Iwata, T. (2011) Structure and mechanical properties of wet-spun fibers made from natural cellulose nanofibers. Biomacromolecules, 12(3):831-836

48. Walther, A., Timonen, J. V., Díez, I., Laukkanen, A., and Ikkala, O. (2011) Multifunctional high‐performance biofibers based on wet‐extrusion of renewable native cellulose nanofibrils. Advanced Materials, 23(26):2924-2928

49. Sehaqui, H., Ezekiel Mushi, N., Morimune, S., Salajkova, M., Nishino, T., and Berglund, L. A. (2012) Cellulose nanofiber orientation in nanopaper and nanocomposites by cold drawing. ACS applied materials & interfaces, 4(2):1043-1049

50. Mori, T. and Tanaka, K. (1973) Average stress in matrix and average elastic energy of materials with misfitting inclusions. Acta Metallurgica, 21(5):571-574

59

51. Eshelby, J. D. (1957) The determination of the elastic field of an ellipsoidal inclusion, and related problems. Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences, 241(1226):376-396

52. Benveniste, Y. (1987) A new approach to the application of Mori-Tanaka's theory in composite materials. Mechanics of Materials, 6(2):147-157

53. Qiu, Y. P. and Weng, G. J. (1990) On the application of Mori-Tanaka's theory involving transversely isotropic spheroidal inclusions. International Journal of Engineering Science, 28(11):1121-1137

54. Chou, T.-W., Nomura, S., and Taya, M. (1980) A self-consistent approach to the elastic stiffness of short-fiber composites. Journal of Composite Materials, 14(3):178-188

55. Lin, S. C. and Mura, T. (1973) Elastic fields of inclusions in anisotropic media (II). physica status solidi (a), 15(1):281-285

56. Hashin, Z. (1983) Analysis of composite materials. Journal of Applied Mechanics, 50(2):481-505

57. Voigt, W. (1928) Lehrbuch der kristallphysik. Teubner, Leipzig, 58. Reuss, A. and Angnew, Z. (1929) A calculation of the bulk modulus

of polycrystalline materials. Math Meth, 9:55 59. Hill, R. (1952) The elastic behaviour of a crystalline aggregate.

Proceedings of the Physical Society. Section A, 65(5):349 60. Tsai, S. W., Theory of composites design: Think composites Dayton,

1992. 61. Whitcomb, J. and Noh, J. (2000) Concise derivation of formulas for

3D sublaminate homogenization. Journal of Composite Materials, 34(6):522-535

62. Tsai, S. W. and Pagano, N. J., "Invariant properties of composite materials," DTIC Document1968.

63. Cox, H. (1952) The elasticity and strength of paper and other fibrous materials. Brit. J. Appl. Phys. , 3(3):72

64. Krenchel, H., Fibre reinforcement: Alademisk forlag, 1964. 65. Chen, W., Lickfield, G. C., and Yang, C. Q. (2004) Molecular

modeling of cellulose in amorphous state. Part I: model building and plastic deformation study. Polymer, 45(3):1063-1071

66. Peura, M., Müller, M., Vainio, U., Sarén, M.-P., Saranpää, P., and Serimaa, R. (2008) X-ray microdiffraction reveals the orientation of cellulose microfibrils and the size of cellulose crystallites in single Norway spruce tracheids. Trees - Structure and Function, 22(1):49-61

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

60

68. Mathew, A. P., Oksman, K., and Sain, M. (2005) Mechanical properties of biodegradable composites from poly lactic acid (PLA) and microcrystalline cellulose (MCC). Journal of Applied Polymer Science, 97(5):2014-2025

69. Lim, L. T., Auras, R., and Rubino, M. (2008) Processing technologies for poly(lactic acid). Progress in Polymer Science, 33(8):820-852

70. Lin, T., Liu, X.-Y., and He, C. (2010) Ab initio elasticity of poly(lactic acid) crystals. The Journal of Physical Chemistry B, 114(9):3133-3139

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

72. Plackett, D., Anturi, H., Hedenqvist, M., Ankerfors, M., Gällstedt, M., Lindström, T., et al. (2010) Physical properties and morphology of films prepared from microfibrillated cellulose and microfibrillated cellulose in combination with amylopectin. Journal of applied polymer science, 117(6):3601-3609

73. Stelte, W. and Sanadi, A. R. (2009) Preparation and characterization of cellulose nanofibers from two commercial hardwood and softwood pulps. Industrial & engineering chemistry research, 48(24):11211-11219

74. Bury, K. V., Statistical Models in Applied Science: John Wiley & Sons Inc, 1975.

75. Jalal Uddin, A., Araki, J., and Gotoh, Y. (2011) Toward “strong” green nanocomposites: polyvinyl alcohol reinforced with extremely oriented cellulose whiskers. Biomacromolecules, 12(3):617-624

76. Gindl, W. and Keckes, J. (2007) Drawing of self‐reinforced cellulose films. Journal of applied polymer science, 103(4):2703-2708

77. Lichtenegger, H., Muller, M., Paris, O., Riekel, C., and Fratzl, P. (1999) Imaging of the helical arrangement of cellulose fibrils in wood by synchrotron X-ray microdiffraction. Journal of applied crystallography, 32(6):1127-1133

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally throughthe series Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-240250

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2015