Surface Modification of Nanocellulose towards Composite Applications

Assya Boujemaoui

Doctoral Thesis

KTH Royal Institute of Technology, Stockholm 2016 Department of Fibre and Polymer Technology

Akademisk avhandling som med tillstånd av Kungliga Tekniska Högskolan i Stockholm framlägges till offentlig granskning för avläggande av teknisk doktorsexamen fredagen den 22:e april 2016, kl. 10.00 i sal F3, Lindstedtsvägen 26, KTH, Stockholm. Avhandlingen försvaras på engelska.

Copyright © 2016 Assya Boujemaoui All rights reserved

Paper I © 2012 American Chemical Society Paper IV © 2015 Elsevier Ltd

TRITA-CHE Report 2016:12 ISSN 1654-1081 ISBN 978-91-7595-888-0

To my family

Abstract

The desire to develop high-value end-products derived from renewable resources is continuously growing as a result of environmental awareness and the depletion of fossil resources. In this context, nanocelluloses have gained great interest during recent decades owing to their renewability, abundancy and remarkable physical and mechanical properties. The aim of the present work was to investigate new strategies for surface modification and functionalization of nanocelluloses and their subsequent incorporation in polymer-host matrices. Nanocomposites of cellulose nanofibrils (CNF) and polycaprolactone (PCL) were produced by employing CNF nanopaper (NP) as a template and surface-initiated ring-opening polymerization (SI-ROP) of ε-caprolactone (ε-CL). SI-ROP of ε-CL from filter paper (FP), which has a low surface area compared with NP, was also carried out for comparison. A larger amount of PCL was grafted from NP than from FP as a result of more available hydroxyl groups. The grafted NP had stronger mechanical properties than a neat PCL film. Cellulose nanocrystal (CNC)-reinforced polyvinyl acetate (PVAc) nanocomposites were also investigated. CNC were modified via “SI-reversible addition- fragmentation chain transfer and macromolecular design via the interchange of xanthate” (SI-RAFT/MADIX) polymerization of vinyl acetate (VAc). The PVAc- grafted CNC and pristine CNC were incorporated into a PVAc matrix via solvent casting. The resulting nanocomposites exhibited improved mechanical performance than the unmodified CNC due to the greater compatibility between the nanoreinforcing-agent and the matrix. It is generally agreed that covalent grafting is superior to physical adsorption for the modification of a reinforcing agent. However, this hypothesis has never been thoroughly investigated. In the present work, CNC was modified either through covalent grafting or through physical adsorption of poly(butyl methacrylate) (PBMA). The two surface modification approaches were compared by incorporating the modified CNC in a PCL matrix via extrusion. Both methods resulted in improved mechanical performance than that of pure PCL or PCL containing unmodified CNC. However, covalent grafting gave the best mechanical performance even at high relative humidity. Functionalized CNC (F-CNC) were obtained through a versatile methodology employing organic acids bearing a functional group: double bond, triple bond, atom transfer radical polymerization (ATRP) initiator and thiol were employed for the simultaneous acid hydrolysis and esterification of cellulose fibers. This provided a facile route for the preparation of F-CNC. Sammanfattning

Strävan efter att utveckla förädlade produkter från förnyelsebara råvaror ökar stadigt som en konsekvens av ökande miljömedvetenhet och dessutom en hotande utarmning av fossila resurser. Med detta som bakgrund har intresset för nanocellulosa ökat markant under de senaste decennierna eftersom de är förnyelsebara, finns att tillgå i stor mängd, och har mycket bra fysikaliska och mekaniska egenskaper. Syftet med detta arbete var att undersöka nya strategier för ytmodifiering och funktionalisering av nanocellulosor och dess inkorporering i polymera matriser. Nanokompositer av cellulosa nanofibriller (CNF) och polykaprolakton (PCL) framställdes genom att CNF nanopapper (NP) användes som ett startmaterial från vilken ε-kaprolakton (ε-CL) polymeriserades med ringöppningspolymerisation (SI-ROP). Som jämförelse ympades även ε-CL från filterpapper (FP) som har mindre ytarea jämfört med NP med SI-ROP. Resultatet av polymerisationerna var att större mängd av polykaprolakton (PCL) ympades från NP jämfört med FP, som en konsekvens av fler tillgängliga hydroxylgrupper. Det ytmodiferade NP hade bättre mekaniska egenskaper jämfört med en ren PCL- film. Nanokompositer av cellulosananokristaller (CNC) och polyvinylacetat (PVAc) undersöktes också. CNC modifierades via “SI-reversible addition-fragmentation chain transfer and macromolecular design via the interchange of xanthate” (SI- RAFT/MADIX) för polymerisation av vinylacetat (VAc). Både PVAc-ympade CNC och omodifierade CNC inkorporerades i en matris av PVAc via lösningsmedelsgjutning. De resulterande nanokompositerna uppvisade bättre mekaniska egenskaper jämfört med omodifierade CNC på grund av förbättrad kompatibilitet mellan nanokristallerna och matrisen. Man har antagit att kovalent ympning är en överlägsen metod för modifiering av ett förstärkande element jämfört med fysikalisk adsorption, men denna hypotes har aldrig undersökts ordentligt. I denna del av avhandlingen har CNC modifierats endera genom kovalent ympning eller fysikalisk adsorption av poly(butylmetakrylat) (PBMA). De två ytmodifieringsmetoderna jämfördes genom att modifierad CNC inkorporerades i en PCL-matris via extrudering. Båda metoderna gav förbättrad mekanisk prestanda jämfört med ren PCL och PCL innehållande omodifierad CNC, men kovalent ympning gav bäst prestanda även vid hög relativ fuktighet. Funktionell CNC (F-CNC) framställdes genom en användbar metod som baseras på organiska syror med en funktionell grupp: alken, alkyn, tiol eller en intitiator för ”atomöverföringsradikalpolymerisation” (ATRP) initiator. F-CNC erhålls genom att hydrolysen av cellulosafibrer utförs genom att använda en kombination av sur hydrolys och förestring. Detta är en enkel och mycket användbar metod för att framställa F-CNC med en rad olika funktionaliteter. List of papers

This thesis is a summary of the following papers:

I “Facile Preparation Route for Nanostructured Composites: Surface-Initiated Ring-Opening Polymerization of ε- Caprolactone from High-Surface-Area Nanopaper”, Boujemaoui, A., Carlsson, L., Malmström, E., Lahcini, M., Berglund, L., Sehaqui, H., Carlmark, A. ACS Applied Materials and Interfaces 2012, 4, 3191−3198

II “RAFT/MADIX Polymerization of Vinyl Acetate on Cellulose Nanocrystals for Nanocomposite Applications”, Boujemaoui, A., Mazières, S., Malmström, E., Destarac, M., and Carlmark, A. Submitted.

III “Polycaprolactone Nanocomposites Reinforced with Cellulose Nanocrystals Surface-modified via Covalent Grafting or Physical Adsorption – a Comparative Study”, Boujemaoui, A., Cobo Sanchez, C., Engström, J., Fogelström, L., Carlmark, A., and Malmström, E. Manuscript.

IV “Preparation and Characterization of Functionalized Cellulose Nanocrystals”, Boujemaoui, A., Mongkhontreerat, S., Malmström, E., and Carlmark, A. Carbohydrate Polymers, 2015, 115, 457–464.

This thesis also contains unpublished results.

My contribution to the appended papers:

I Most of the experimental work, part of the analysis and the preparation of the manuscript

II Most of the experimental work, analysis and preparation of the manuscript

III Part of the experimental work, analysis and major part of the preparation of the manuscript

IV Most of the experimental work, analysis and preparation of the manuscript

Other scientific contributions not included in this thesis:

V “Thermoresponsive Cryogels Reinforced with Cellulose Nanocrystals” Larsson, E., Boujemaoui, A., Malmström, E., Carlmark, A. RSC Advances (2015), 5(95), 77643-77650

VI “Dendritic hydrogels: From Exploring Various Crosslinking Chemistries to Introducing Functions and Naturally Abundant Resources” Mongkhontreerat, S., Andrén, O. C. J., Boujemaoui, A., Malkoch M. Journal of Polymer Science, Part A: Polymer Chemistry (2015), 53(21), 2431-2439

VII “Copper-based Dye-sensitized Solar Cells with Quasi-Solid Nanocellulose Composite Electrolytes”. Willgert, M.; Boujemaoui, A., Malmström, E., Constable, E. C., Housecroft, C. E. Submitted

VIII “Functionalized Cellulose Nanocrystals, a Method for the Preparation thereof and use of Functionalized Cellulose Nanocrystals in Composites and for Grafting”, Malmström, E., Carlmark, A., Boujemaoui, A. PCT/SE2013/051276, WO2014070092 A1. (2014)

Table of contents

1 PURPOSE OF THE STUDY ...... 1 2 INTRODUCTION...... 2

2.1 CELLULOSE ...... 2 2.1.1 Hierarchical structure of cellulose ...... 2 2.1.2 Cellulose nanofibrils (CNF) ...... 4 2.1.3 Cellulose nanocrystals (CNC) ...... 4 2.1.4 Nanocellulose-reinforced composites ...... 6 2.2 SURFACE MODIFICATION ...... 6 2.2.1 Covalent grafting ...... 7 2.2.2 Physical adsorption ...... 9 2.3 POLYMERIZATION TECHNIQUES ...... 10 2.3.1 Ring-opening polymerization (ROP) ...... 10 2.3.2 Reversible‐deactivation radical polymerization (RDRP) ...... 12 2.3.2.1 Atom transfer radical polymerization (ATRP) ...... 13 2.3.2.2 Reversible addition-fragmentation chain transfer and macromolecular design via the interchange of xanthates (RAFT/MADIX) polymerization ...... 14 2.4 PROCESSING OF NANOCELLULOSE COMPOSITES ...... 17 3 EXPERIMENTAL ...... 19

3.1 MATERIALS ...... 19 3.2 CNF NANOPAPER ...... 19 3.3 CELLULOSE NANOCRYSTALS ...... 20 3.4 PREPARATION OF FUNCTIONALIZED CNC (F-CNC)...... 20 3.5 SURFACE-INITIATED POLYMERIZATION ...... 21 3.5.1 SI-ROP of PCL from cellulose substrates ...... 21 3.5.2 SI-RAFT/MADIX of VAc from HCl-CNC ...... 22 3.5.3 SI-ATRP of PBMA from H2SO4-CNC ...... 24 3.6 PHYSICAL ADSORPTION OF MICELLES AND LATEX PARTICLES ...... 25 3.7 FILM PREPARATION ...... 25

4 RESULTS AND DISCUSSION ...... 27

4.1 SURFACE MODIFICATION OF NANOCELLULOSE ...... 27 4.1.1 SI-polymerization from nanocellulose ...... 27 4.1.1.1 SI-ROP of ε-caprolactone from CNF nanopaper ...... 27 4.1.1.2 SI-RAFT/MADIX of vinyl acetate from HCl-CNC ...... 31 4.1.1.3 SI-ATRP of butyl methacrylate from H2SO4-CNC ...... 35 4.1.2 Physical adsorption of micelles and latex particles on H2SO4-CNC ...... 36 4.1.3 Functionalized CNC ...... 38 4.2 CHARACTERIZATION OF NANOCOMPOSITES ...... 42 4.2.1 PCL grafted CNF nanopaper ...... 42 4.2.2 PVAc reinforced with PVAc grafted CNC ...... 45 4.2.3 PCL reinforced with covalently grafted or physisorbed PBMA- modified CNC ...... 47 4.2.4 F-CNC-based hydrogels ...... 51 5 CONCLUSIONS ...... 53 6 FUTURE WORK ...... 55 7 ACKNOWLEDGEMENTS ...... 57 8 REFERENCES ...... 59

Abbreviations

1H NMR Proton nuclear magnetic resonance 2-BPA 2-Bromopropanoic acid 2-BPB 2-Bromopropionyl bromide 2-PyA 2-Propynoic acid 3-MPA 3-Mercaptopropionic acid 4-PA 4-Pentenoic acid AA Acrylic acid AIBN 2,2′-Azobis(2-methylpropionitrile) AFM Atomic force microscopy ATRA Atom transfer radical addition ATRP Atom transfer radical polymerization BET Brunauer−Emmett−Teller BIB α-Bromoisobutyryl bromide BnOH Benzyl alcohol BPB Bromopropionyl bromide -CL ε-Caprolactone CNC Cellulose nanocrystals CNF Cellulose nanofibrils CNW Cellulose nanowhiskers CRP Controlled radical polymerizations CO2 Carbon dioxide CTA Chain transfer agent εbreak Compressive strain at break CuBr Copper(I) bromide CuBr2 Copper(II) bromide ÐM Molar-mass dispersity DCC N,N′-Dicyclohexylcarbodiimide DCM Dichloromethane DMA Dynamic mechanical analysis DMAP 4-(Dimethylamino)pyridine DMF Dimethylformamide DPn Degree of polymerization DLD Dendritic-linear-dendritic DLS Dynamic light scattering DR13-N3 Azide functionalized disperse red 13 DR13-SH Thiol functionalized disperse red 13 DSC Differential scanning calorimetry DTNB 5,5’‐Dithiobis(2‐nitrobenzoic acid) DVS Dynamic vapor sorption E Young’s modulus E’ Storage modulus EBIB Ethyl α-bromoisobutyrate ETOAc Ethyl acetate EtOH Ethanol F-CNC Functionalized cellulose nanocrystals FE-SEM Field emission-scanning electron microscopy FP Filter paper FRP Free radical polymerization FT-IR Fourier transform infrared spectrometry H2SO4 Sulfuric acid HCl Hydrochloric acid HMTETA 1,1,4,7,10,10-Hexamethyltriethylenetetramine I Initiator KEX Potassium ethyl xanthogenate MADIX Macromolecular design via the interchange of xanthates MCC Microcrystalline cellulose MeOH Methanol MFC Microfibrillated cellulose MH Microwave heating Mn Number average molar mass Mn,th Theoretical number average molar mass Mw Weight average molar mass MMA Methyl methacrylate n-BMA n-Butyl methacrylate NCC Nanocrystalline cellulose NFC Nanofibrillated cellulose NH Normal heating NHS N-Hydroxysuccinimide NP Nanopaper PB Poly(1,2-butadiene) PBA Poly (n-butyl acrylate) PBMA Poly(n-butyl methacrylate) PCL Polycaprolactone PEG Poly(ethylene glycol) PDEGMA Poly(di(ethylene glycol) methyl ether methacrylate) PLA Polylactide PMMA Poly(methyl methacrylate) Poly{6-[4-(4-methoxyphenylazo)phenoxy] hexyl PMMAZO methacrylate} PNIPAAM Poly(N-isopropylacrylamide) PS Polystyrene PVAc Polyvinyl acetate qPDMAEMA Quaternized poly(dimethylaminoethylmethacrylate) RAFT Reversible addition-fragmentation chain transfer RDRP Reversible‐deactivation radical polymerization RH Relative humidity ROP Ring-opening polymerization R-X Alkyl halide RT Room temperature SEC Size exclusion chromatography Sn(Oct)2 Tin 2-ethylhexanoate SI Surface-initiated TEA Triethylamine TGA Thermogravimetric analysis Tc Crystallization temperature Tg Glass transition temperature Tm Melting temperature THF Tetrahydrofuran VAc Vinyl acetate Wh Cellulose whiskers Xc Degree of crystallinity XPS X-ray photoelectron spectroscopy XRD X-ray diffraction analysis

————— Purpose of the study —————

1 Purpose of the study

The increasing environmental awareness in combination with the diminishing fossil resources has motivated the development of bio-based products derived from renewable resources. In this context, nanocelluloses, obtained from cellulose fibers, have emerged as a “green” alternatives to reinforcing agents for the design of nanocomposites, owing to their remarkable mechanical properties. However, the inherent hydrophilic character of nanocelluloses restricts their dispersion within a hydrophobic host matrix, and thus affects the mechanical properties of the nanocomposite. The hydrophilicity of nanocelluloses can, however, be reduced through various surface modification techniques to improve their compatibility with a hydrophobic matrix. The overall purpose of the work described in this thesis was to investigate new strategies for the surface modification and functionalization of nanocelluloses and their subsequent incorporation in host matrices. The surface modification has been conducted either through covalent grafting of polymers, employing controlled polymerization techniques or via the physical adsorption of preformed block copolymers. The modified nanocelluloses were utilized as reinforcing agents in polymer host matrices and the performance of the resulting nanocomposite was evaluated. The simultaneous functionalization and preparation of functionalized cellulose nanocrystals has also been studied.

1 ————— Introduction —————

2 Introduction

2.1 Cellulose

Cellulose is a biopolymer found mainly in trees and in plants such as cotton, hemp, jute and flax. It can also be produced by living organisms such as bacteria, algae and tunicate sea animals.1 Cellulose is synthesized in high purity by cotton and by some bacteria (> 90 %), whereas it is embedded with other biopolymers, i.e., lignin and hemicelluloses, in plants and in wood cell walls.2 In wood, the cellulose content varies from 30 % up to 45 % depending on the species,3 and isolation of the cellulose fibers is therefore required prior to their utilization. The total production of cellulose per year is estimated to be over 7.5×1010 tons,4 of which around 1.8×109 tons is industrially extracted from wood.5 The main cellulose-based products include textiles, paper and cardboard.4

2.1.1 Hierarchical structure of cellulose

Cellulose is a linear polysaccharide consisting of -1,4-D-glucosidic repeating units (Figure 1). The degree of polymerization (DPn) of native cellulose varies depending on its source and can be as high as 15000.1, 2, 5 Each repeating unit contains two anhydroglucose units (AGU). In each AGU there is one primary hydroxyl (OH) group at the carbon 6 (C6) position, and two secondary OH groups at the C2 and C3 positions. Owing to these OH groups, cellulose has a hydrophilic character, although the strong inter‐ and intramolecular hydrogen bonds restrict the solubility of cellulose in water.

Figure 1. Chemical structure of the cellulose repeating unit.

2 ————— Introduction —————

The biosynthesis of cellulose by plants and wood is still not fully understood. However, it has been suggested that cellulose chains are synthesized by protein complexes (also called rosettes) present in the cell wall. These rosettes contain six “lobes” and each of them may synthesize six cellulose chains. Once the chains are formed, they co-crystallize to form 36-chain nanofibrils (also referred to as elementary fibrils) that are 3-5 nm wide. These nanofibrils are composed of both crystalline and less ordered amorphous regions, and they are combined into bundles to form microfibrils (10-60 nm), which further assemble to build up cellulose fibers (10-30 m).2, 4, 6-8 The hierarchical structure of cellulose is represented in Figure 2.

Figure 2. Hierarchical structure of cellulose from a molecular to a micro-scale. The schematic picture is adapted from Postek et al.7

Cellulose fibers are the main load-bearing component in trees and plants due to the high modulus of its crystalline part, which can reach 140 GPa.9 In combination with its low density, cellulose has a potential reinforcing capability comparable to that of inorganic and synthetic fibers such as aramid and glass.1 However, isolation of cellulose in the nano-size range, i.e., nanocellulose, is required in order to take full advantage of its inherent properties.

The two types of cellulose nanofibers that can be liberated from cellulose fibers are described in the two next sections.

3 ————— Introduction —————

2.1.2 Cellulose nanofibrils (CNF)

An aqueous suspension of cellulose nanofibrils has a gel-like texture, and it is composed of nanofibrils and microfibrils 5-60 nm wide and several m long (Figure 3, b). The extraction was first achieved in the early 1980’s by subjecting a wood pulp slurry to high shear forces10. High- pressure homogenizers (also termed microfluidizers) and ultra-fine friction grinders are the mechanical devices usually employed to disintegrate the wood fibers and liberate the cellulose nanofibrils. The homogenization process requires 5 to 10 passes depending on the composition and thermal history of the wood pulp. Therefore, the energy consumption for the production of cellulose nanofibrils is high, ranging from 25000 kWh t-1 11 up to 70000 kWh t-1 12, 13 In order to reduce the production cost of cellulose nanofibrils, a number of surface modification pre-treatments have been developed introducing either negatively-charged carboxylic groups through 2,2,6,6- tetramethylpiperidine-1-oxyl (TEMPO)-mediated oxidation,14 or positively-charged trimethylammonium groups through surface quaternization.15 Carboxymethylation of cellulose surface with chloroacetic acid has been also performed.16 A mild enzyme hydrolysis pre-treatment is an environment-friendly approach to facilitate the fiber disintegration process.17 In this work, CNF was prepared through enzymatic pre-treatment and mechanical disintegration utilizing a microfluidizer. Several terms have been used for cellulose nanofibrils (CNF) such as microfibrillated cellulose (MFC) and nanofibrillated cellulose (NFC). In 2011, the term CNF was recommended by a TAPPI workshop on international standards for nanocellulose based on the ISO TC-229 nomenclature protocols.2, 18 Hence, this terminology will be used throughout this thesis.

2.1.3 Cellulose nanocrystals (CNC)

In the early 1950’s, Rånby et al. were the first to report the preparation of a stable colloid suspension of cellulose nanocrystals via sulfuric acid 19-22 (H2SO4) hydrolysis. Analysis of the dried suspension by electron microscopy revealed the presence of both individualized and aggregated rod-like particles of a few hundred nanometers.19-22

4 ————— Introduction —————

During the acid hydrolysis of cellulose fibers, the amorphous domains are more susceptible to degradation/hydrolysis while the crystalline regions remain more or less intact. The yield, size, morphology and crystallinity of cellulose nanocrystals (CNC) strongly depend both on the cellulose source and on the hydrolysis conditions, i.e. time, temperature and acid concentration/pH.8 Tunicates are the favorable source for the preparation of CNC owing to the high aspect ratio of their crystalline domains comparable to that of CNF. However, the high cost of harvesting and their limited availability are the main drawbacks for CNC production. Generally, wood and cotton are the key resources due to their natural abundance and high cellulose content, respectively. The CNC from these sources are 3-70 nm wide and 100-300 nm long (Figure 3, a).2, 8

The acid hydrolysis of cellulose fibers with H2SO4 introduces negatively- charged sulfate groups on the CNC surface. These groups promote the dispersion and colloidal stability of the CNC suspension via electrostatic repulsion. However, the thermostability of CNC is compromised.23 Other mineral acids such as hydrochloric (HCl), phosphoric and nitric acids have also been utilized.24 Recently, a combination of HCl and organic acids was also studied in order to introduce other functional groups onto the CNC surface via simultaneous hydrolysis and esterification.25 The terms cellulose whiskers (Wh), nanocrystalline cellulose (NCC) and cellulose nanowhiskers (CNW) have also been employed over the past decades to describe cellulose nanocrystals, CNC.2 However, since CNC is the terminology recommended by TAPPI workshop on international standards for nanocellulose,2, 18 this terminology will be used throughout this thesis.

Figure 3. TEM images of (a) CNC and (b) CNF extracted from bleached eucalyptus pulp. Reprinted with permission from reference26. Copyright (2013) American Chemical Society.

5 ————— Introduction —————

2.1.4 Nanocellulose-reinforced composites

Nanocelluloses, i.e. CNF and CNC, have attracted great attention in recent decade owing to their renewability, nano-size dimension, large surface area and remarkable reinforcing capability. These nanoparticles have a vast field of applicability; but they are typically explored as a reinforcing agent for nanostructured composites. It has been shown that the addition of a small volume fraction of nanocellulose (< 5 %) results in a nanocomposite material with a significantly stronger mechanical performance than that of an unreinforced matrix.2 The mechanical and thermal properties of a nanocomposite are usually critically affected by the extent of the interaction between the nanocomposite components, the distribution and the dispersity of the nanoparticles within the host matrix. One challenge encountered in the preparation of nanocellulose-reinforced composites is to achieve a good dispersion of the hydrophilic nanocellulose within the hydrophobic host matrix. This difference in hydrophilicity between nanocellulose and the host matrix results in poor compatibility and leads to the formation of nanoparticles aggregates. Consequently, surface modification of nanocelluloses is often required to modify their surface chemistry and thus improve their compatibility with the hydrophobic matrix.4, 11

The following section summarizes the surface modification approaches which have been described in the literature with regard to nanocellulose, with an emphasis on surface modification with polymers.

2.2 Surface modification

The surface properties of nanocelluloses can be altered since a reasonably large amount of reactive hydroxyl groups are available on their surfaces. Furthermore, surface charges can easily be incorporated (cationic or anionic) during nanocellulose production, making modification through electrostatic interaction possible. Several different surface-modified nanocelluloses have been prepared either by covalent grafting or by physical adsorption.4, 27-32

6 ————— Introduction —————

Low molecular weight molecules have been covalently attached to nanocellulose surface through various chemical reactions such as esterification, silylation and etherification.2, 4, 33 Esterification is the most commonly employed reaction for the hydrophobization of cellulose. In this process, the reaction of cellulose hydroxyl groups with an acid anhydride,34-36 an organic acid,25, 37-39 or an acyl halide40, 41 leads to an ester group, as shown in Figure 4.

Figure 4. Esterification of cellulose.

The physical adsorption approach has also been employed for the adsorption of surfactants onto the cellulose surface through secondary interactions. A surfactant is usually an amphiphilic compound that contains both a hydrophilic and a hydrophobic part, which means that, the surfactant may adsorb to cellulose via its hydrophilic part while the hydrophobic tail is exposed and thus decreases the surface tension of the nanocellulose. For this purpose, the non-ionic amphiphilic sorbitan monostearate surfactant has been employed.42, 43 The physical adsorption can be further enhanced via electrostatic interaction where a charged amphiphilic compound is adsorbed onto an oppositely charged surface. Quaternary ammonium salts have mainly been utilized for surface modification of nanocellulose. 44, 45, 46

Surface modification of nanocellulose can also be achieved with macromolecules, i.e. polymers. In this case, increased adhesion between the grafted nanocellulose and the host matrix can be achieved by the formation of polymer entanglements.47

2.2.1 Covalent grafting

Polymers can be covalently grafted onto nanocellulose surfaces by means of various polymerization techniques;2, 4, 31 either through “grafting-from” or “grafting-to” techniques (Figure 5). The grafting-from method, also denoted surface-initiated (SI) polymerization, is based on the initiation

7 ————— Introduction —————

and subsequent propagation of the monomer through reactive moieties on the surface.28 However, in this case, the characterization of the grafted polymer in terms of its weight average molar mass (Mw) and molar-mass dispersity (ÐM) can be difficult, requiring the formed polymer to be cleaved off the surface. One method to achieve this is by acid or enzymatic hydrolysis of the cellulose, followed by the characterization of the polymer grafts left intact.28, 48 However, this procedure can only be applied for polymers with a strong resistance to the hydrolysis conditions.28 It has recently been shown that the addition of a sacrificial initiator to the SI-polymerization medium generates free polymer chains 49-52 which have Mw and ÐM similar to the grafted chains. It can therefore be assumed that the properties are similar for the grafted and free forming polymer. The main drawback of this approach is the formation of a large amount of free polymer and the often tedious isolation and purification of the grafted nanocelluloses, but it is possible to assess the

Mw and ÐM of the grafted chains by determining the Mw and ÐM of the free polymer, and this approach has been used in the present work.

In the “grafting-to” approach, pre-formed polymers with a known Mw and

ÐM are covalently attached to the surface, where the chain-end of the pre- formed polymer is designed to react with reactive moieties on the cellulose. Unlike “grafting-from”, this approach requires only a low amount of the pre-formed polymer. However, steric hindrance and low diffusion of bulky polymer chains are usually the limiting factors to achieve high grafting densities.28, 31, 48, 53-55

8 ————— Introduction —————

Figure 5. Schematic illustration of the “grafting-from” and “grafting-to” approaches from a solid substrate (rectangle).

2.2.2 Physical adsorption

The physical adsorption of macromolecules has been also reported in the literature as an alternative route for the surface modification of cellulose.2, 56 Positively charged polymers, i.e. polyelectrolytes, can be adsorbed onto the oppositely charged cellulose surface based on the gain in entropy and the release of counter ions.57, 58 Furthermore, block copolymers bearing a positively charged polyelectrolyte and a segment with a hydrophobic character or having the desired functionality have been also employed in order to alter the surface properties of nanocelluloses (Figure 6). The polyelectrolyte segment acts as an anchoring block and insures the adsorption of the block copolymer. Quaternized poly(dimethylaminoethylmethacrylate) (qPDMAEMA)-based block copolymers, bearing a positively charged polyelectrolyte and hydrophobic segments, are one of the block copolymers frequently reported in the literature for the surface modification of nanocelluloses.56 Relevant examples are: qPDMAEMA- b-polycaprolactone (PCL),59 poly(1,2-butadiene) (PB)-b-qPDMAEMA,60 and qPDMAEMA-b-poly(di(ethylene glycol) methyl ether methacrylate) (PDEGMA))61.

9 ————— Introduction —————

Figure 6. Schematic illustration of the physical adsorption of a block copolymer onto an oppositely charged substrate (rectangle).

Control over the surface modification of nanocelluloses through covalent grafting and physical adsorption of polymers requires the synthesis of well-tailored polymers and block copolymers. This can be achieved via controlled polymerization techniques.

2.3 Polymerization techniques

2.3.1 Ring-opening polymerization (ROP)

The ring-opening polymerization (ROP) of cyclic esters was reported for the first time by Carothers et al. in the 1930’s,.62 Since this first report, the polymerization technique has been applied to other cyclic monomers such as siloxanes, cyclic carbonates and ethers.63 ROP is usually triggered by an alcohol as initiator, such as benzyl alcohol (BnOH), although primary or secondary amines may also be employed.64, 65 The key reason for utilizing BnOH is that it is possible to determine the end-group functionality of the polymer via proton nuclear magnetic resonance (1H- NMR), due to the distinct shift of the benzyl hydrogens, and thereby 63 calculate the DPn. Various types of catalysts have been studied for ROP; metal-based, organic compounds and enzymes,66-69 but tin 2- ethylhexanoate (Sn(Oct)2) is the most commonly employed catalyst due to its low cost in combination with its excellent performance, coupled to 64, 70 the fact that it is FDA approved. To date, the mechanism of Sn(Oct)2- catalyzed ROP is not exactly known. However, Penzcek et al. have suggested a coordination-insertion mechanism (Scheme 1).70 The catalyst is first transformed into a metal alkoxide (pre-initiation step) followed by coordination-insertion of the first monomer (initiation). Subsequently,

10 ————— Introduction —————

propagation occurs through coordination-insertion of monomers. Finally, the polymerization reaction may be terminated via protonation. It is worth mentioning that termination can also occur via inter- and intramolecular transesterifications (Scheme 2), which may broaden the

ÐM. The extent of the transesterification reactions is greatly dependent on the polymerization conditions such as temperature and monomer conversion.71 A somewhat different mechanism has been proposed by Kricheldorf et al. where the polymerization is initiated via complexation of both the monomer and the initiator with Sn(Oct)2 followed by propagation through coordination-insertion of the monomers.72 For both proposed mechanisms, it has been shown that the polymer is only formed 72 from one arm of the Sn(Oct)2 catalyst.

Scheme 1. Coordination-insertion of ROP of -CL proposed by Penzcek et al. employing 70 Sn(Oct)2 and an alcohol as catalyst and initiator, respectively.

ROP is both a water- and oxygen-sensitive reaction. Traces of water can also act as initiating sites, and this reduces the control over the polymerization.73 Therefore, dried reactants and an inert atmosphere in the reaction vessel are required. Pärssinen et al. have recently shown that ROP of lactones and lactides can be conducted in air at high temperature (70-140 °C) utilizing a titanium alkoxide catalyst such as titanium n- 74 butoxide (Ti(On-Bu)4), see Figure 7. Noteworthy, these titanium-based compounds act both as initiator and catalyst for ROP, so that no co- initiator is needed. Moreover, Kricheldorf et al. confirmed by 1H-NMR

11 ————— Introduction —————

that the polymer is formed from all four arms of the catalyst, which means that the monomer-to-initiator/catalyst ratio should be divided by 72 four in order to determine the average DPn.

Scheme 2. Intra- and intermolecular transesterification reactions.

Figure 7. Chemical structure of tin 2-ethylhexanoate (Sn(Oct)2) and titanium n-butoxide

Ti(On-Bu)4 catalysts.

SI-ROP of lactones and lactides from nanocelluloses has been widely reported in the literature.29 In this case, the native hydroxyl groups on the cellulose surface act as initiators for the SI-polymerization. For example, PCL,75 polylactide (PLA),76 and PCL-b-PLA77 have been successfully grafted from CNC and CNF. The grafted polymer may act as compatibilizer between nanocellulose and a hydrophobic host matrix such as PCL and PLA.78 These aliphatic polyesters are also of great interest owing to their biodegradability and biocompatibility.

2.3.2 Reversible‐deactivation radical polymerization (RDRP)

Free radical polymerization (FRP) is widely employed for the industrial production of polymers due to its simplicity. However, control over the reaction is restricted, often resulting in a high ÐM, branching and an uncontrollable molecular weight. Moreover, accurate control over the

12 ————— Introduction —————

end‐group functionality of a polymer is challenging;79 therefore, better controlled polymerization techniques are required. Controlled radical polymerizations (CRP) have emerged as robust and powerful techniques for the design and development of complex and well- defined macromolecular architectures.80 CRP are based on a reversible‐ deactivation mechanism (examples are discussed in more detail below), and the terminology CRP has therefore recently been replaced by the term reversible‐deactivation radical polymerization (RDRP).81

2.3.2.1 Atom transfer radical polymerization (ATRP)

Atom transfer radical polymerization (ATRP) is the most frequently studied and utilized RDRP. Its development dates back to the 1940’s when Kharasch et al. investigated the addition reaction of carbon tetrachloride to olefins catalyzed by organic peroxide.82 Based on Kharasch addition, atom transfer radical addition (ATRA) was reported in 1988 by Curran for the synthesis of 1:1 adducts of alkyl halides and alkenes, catalyzed by transition-metal complexes.83 A few years later, Matyjaszewski et al.84, 85 and Sawamoto et al.86 independently developed ATRP in light of ATRA. Since then, it has been shown that ATRP is a powerful tool for the polymerization of a wide range of monomers, such as (meth)acrylates,87-89 meth(acrylamides),90, 91 styrenics,92, 93 and 94, 95 acrylonitriles, resulting in polymers with a narrow ÐM and controlled 96 Mw.

ATRP is based on the dynamic equilibrium between dormant and active species according to the mechanism shown in Scheme 3, as suggested by Matyjaszewski et al.96 Firstly, the transition metal catalyst abstracts the halogen atom from the alkyl halide initiator (R‐X), resulting in homolytic cleavage of the R-X bond. The halogen radical is transferred to the n n+1 catalyst complex (Mt /L) forming the deactivated species X-Mt /L. The active radicals (R•) can then react with a monomer molecule forming the • • active radical species (P n). The active species P n continues to propagate through the consecutive addition of monomers and deactivation. In this process, the rate constant of deactivation must be higher than the propagation rate, i.e. the equilibrium must be shifted to the dormant side, in order to maintain a low concentration of the active radicals. As a consequence, termination reactions are drastically reduced in comparison

13 ————— Introduction —————

with FRP. The main disadvantage of ATRP is its sensitivity to oxygen, which may lead to undesired termination. ATRP should therefore be conducted under an inert atmosphere.

Scheme 3. General mechanism based on the dynamic equilibrium between dormant and active species in ATRP.

In a pioneering work, Carlmark and Malmström have reported the successful SI-ATRP of methyl acrylate from cellulose filter paper.53 Since then, several reports of SI-ATRP on various cellulose substrates have been published.28, 30 Various monomers, macromolecular architectures and functionalities have been targeted. Noteworthy, the SI-ATRP from cellulose requires the immobilization of an ATRP-initiating moiety on the surface prior to polymerization. Examples of polymers grafted from nanocellulose via SI-ATRP include: PS,41, 97 poly(methyl methacrylate) (PMMA),98 poly (n-butyl acrylate) (PBA),98 PDMAEMA, Poly{6-[4-(4- methoxyphenylazo) phenoxy] hexyl methacrylate} (PMMAZO),99 and poly(N-isopropylacrylamide) (PNIPAAM).100 Interestingly, nanocelluloses have also been modified though the physical adsorption of amphiphilic macromolecules synthesized by ATRP or by a combination of ATRP and ROP. Relevant examples are PDMAEMA-b-PS,101 PBA-b- PDMAEMA,60 PDMAEMA-b-PDEGMA,61 and PDMAEMA-b-PCL59 block copolymers.

2.3.2.2 Reversible addition-fragmentation chain transfer and macromolecular design via the interchange of xanthates (RAFT/MADIX) polymerization

Reversible addition-fragmentation chain transfer (RAFT) is a RDRP technique discovered in 1998 by Rizzardo et al.102, 103 At the same time, Zard et al. developed the RDRP titled macromolecular design via the

14 ————— Introduction —————

interchange of xanthates polymerization (MADIX).104 Both polymerization techniques are based on the reversible chain transfer reaction induced by the use of thiocarbonylthio compounds (Figure 8), which act as chain transfer agents (CTA). The CTA is composed of the main functional group, i.e. thiocarbonylthio functionality, together with a stabilizing group, Z, and a reactive group, R. RAFT and MADIX follow the same mechanism and differ only by the CTA employed.105 In MADIX, the CTA contains a xanthate functionality, which means that the stabilizing Z-group is an (O-R´) group (Figure 8). MADIX is therefore usually referred to as RAFT/MADIX. The mechanism of RAFT/MADIX employing xanthate as CTA is illustrated in Scheme 4.105

Figure 8. General examples of RAFT agent and MADIX chain transfer agent (CTA) chemical structures.

As with conventional free radical polymerization, RAFT/MADIX requires the addition of a free radical initiator (I), such as 2,2′-azobis(2- methylpropionitrile) (AIBN). However, the molar ratio I/CTA is usually kept low (around 0.1) in order to ensure good control over the polymerization.105 Radicals are first generated from the initiator and subsequently added to the monomer to form short oligomeric • • macroradicals (Pn ). In the second step, the Pn can react with the CTA generating dormant species and reactive radicals R•, which may re- • initiate and add to monomers, and thus create other macroradicals (Pm ). When all the transfer agents are consumed, an equilibrium of addition- fragmentation chain transfer is established between the active species (growing chains) and the dormant species (chains bearing a thiocarbonylthio moiety at one end). This equilibrium is responsible for controlling the polymerization. However, as with all RDRP techniques, the irreversible termination reactions between radicals cannot be totally avoided, and thus a small fraction of terminated chains is generated.

15 ————— Introduction —————

Scheme 4. Mechanism of RAFT/MADIX polymerization employing the xanthate chain transfer agent.

RAFT/MADIX is a versatile polymerization technique capable of polymerizing a wide range of monomers including vinyl ester monomers.106, 107 Moreover, these techniques can be conducted in bulk,108, 109 in solution,109 110 or in emulsion,109, 111, 112 which enlarges their field of applicability.

SI-RAFT/MADIX requires the immobilization of CTA on the surface prior to SI-polymerization either via Z-group or R-group approach (Figure 9).105, 113 In the Z(OR)-group approach, the xanthate-moieties are permanently attached to the surface during SI-RAFT/MADIX, while the propagating macroradicals grow at the nexus of the surface. Consequently, steric hindrance may restrict the access of the growing chains to the CTA functionality as the molar mass increases. On the other hand, in the R-group approach, the propagating radicals are positioned at the surface whereas the mediating xanthate-moieties are leaving it. Therefore, the chains grow directly from the surface, and thus a higher grafting density is obtained than with the Z-group approach.

16 ————— Introduction —————

Figure 9. Schematic illustration of Z- and R-group approaches in SI-RAFT/MADIX adapted to a cellulose surface as an example.

Despite the fact that RAFT and RAFT/MADIX are versatile polymerization techniques, few studies have used SI-RAFT and SI- RAFT/MADIX with cellulose as the surface,105 and no study has yet described SI-RAFT/MADIX from nanocelluloses. Relevant work on SI- RAFT/MADIX from polysaccharides are the grafting of poly(vinyl acetate) (PVAc) from methyl cellulose and hydroxypropyl cellulose via the Z-group approach,114 and PVAc, PS, PBA and PS-b-poly(4-vinylbenzyl chloride) from wood fibers (containing 75 wt% of cellulose and hemicellulose, and 25 wt% of lignin) via the R-group approach.115

2.4 Processing of nanocellulose composites

Nanocelluloses have been mainly investigated for nanocomposite applications due to their remarkable mechanical properties. Indeed, cellulose is the load-bearing component that provides strength and rigidity to the cells of higher plants. Various methods of incorporating nanocelluloses into polymeric matrices have been considered, such as melt-compounding, solvent casting and in-situ polymerization.2, 8 Among these methods, solvent casting is the most commonly utilized method for research purposes. The nanocellulose is mixed together with the host polymer matrix in a suitable solvent, and the mixture is then cast in a recipient, usually an aluminum dish. The nanocomposite film is then formed by evaporation of the solvent. One drawback associated with this approach is the need to remove large quantities of solvent.

17 ————— Introduction —————

In view of their economic availability and large-scale production, melt- processing techniques such as extrusion are the most appropriate processing procedures for the industrial production particularly of CNC- reinforced nanocomposites. However, the major drawback of processing CNCs by means of thermal-mechanical compounding is their low thermal stability due to sulfate groups present on the surface, which generate corrosive species upon heating, and thus induce cellulose chain degradation.24 Another approach consists of the initial formation of a CNF network followed by the addition of the polymer matrix. This method allows the production of high-content cellulose nanostructured composites. However, combining a high fraction of CNF network with a hydrophobic matrix, as presented in Paper I, has been largely unexplored. A CNF network in the form of nanopaper, can be prepared via vacuum filtration of an aqueous suspension of CNF followed by solvent exchange and supercritical carbon dioxide (CO2) drying. This process allows solvent removal without degradation of cellulose as the critical temperature and pressure of CO2 are 31°C and 7.4 MPa, respectively. The CNF network, having an internal surface area as high as 480 m2 g-1, is thus preserved.116

18 ————— Experimental —————

3 Experimental

An overview of the experimental procedures employed in this work is given here. Detailed information regarding materials and instrumentation could be found in the appended papers (I-IV).

3.1 Materials

ε-Caprolactone (ε-CL), benzyl alcohol (BnOH), n-butyl methacrylate (n-

BMA), vinyl acetate (VAc), titanium n-butoxide (Ti(On-Bu)4), tin 2- ethylhexanoate (Sn(Oct)2), 2-bromopropionic acid (BPA), 3- mercaptopropionic acid (3-MPA), 4-pentenoic acid (4-PA), 2-propynoic acid (2-PyA), 37 % hydrochloric acid (HCl), copper sulphate, sodium ascorbate, polycaprolactone (PCL) (Mn 80000 g/mol, ÐM<2), polyvinyl acetate (PVAc) (Mn 80000 g/mol, ÐM<2) α-bromoisobutyryl bromide (BIB), ethyl α-bromoisobutyrate (EBIB), triethylamine (TEA), 1,1,4,7,10,10-hexamethyltriethylenetetramine (HMTETA), copper(I) bromide (CuBr), copper(II) bromide (CuBr2), 4-(dimethylamino)pyridine

(DMAP), sulfuric acid (H2SO4), potassium ethyl xanthogenate (KEX), methyl 2-bromopropionate, 2-bromopropionyl bromide (BPB), N,N′- dicyclohexylcarbodiimide (DCC), 2,2′-azobis(2-methylpropionitrile) (AIBN), acetone, ethyl acetate (EtOAc) N,N-dimethylformamide (DMF), toluene, ethanol (EtOH), dichloromethane (DCM), tetrahydrofuran (THF) and methanol (MeOH) were purchased from Sigma-Aldrich. Filter paper (FP) (Whatman No. 1) was either ground utilizing a coffee grinder, or cut into pieces 2.5 х 3 cm2 in size, washed with acetone and methanol and then dried in a vacuum oven at 50 °C for 24 hours prior to use.

3.2 CNF nanopaper

CNF aqueous suspension was diluted to 0.1 wt% and filtered through a 0.65 μm membrane with vacuum filtration. The “cake” formed was then solvent-exchanged to methanol and dried using CO2 supercritical point dryer (Tousimis) to obtain high surface area CNF nanopaper.

19 ————— Experimental —————

3.3 Cellulose nanocrystals

CNCs were prepared via acid hydrolysis utilizing either hydrochloric acid

(HCl) or sulfuric acid (H2SO4) leading to uncharged (HCl-CNC) and negatively charged CNC (H2SO4-CNC), respectively.

HCl-CNC was prepared according to Weder et al.117 Briefly, ground filter paper (10 g) was dispersed in HCl (300 mL, 3 M) in a round-bottomed flask and immersed in an oil bath at 110 °C for 90 minutes under rigorous magnetic stirring. Thereafter, the reaction mixture was diluted ten times with deionized water and the HCl-CNC was thoroughly washed through repeated filtration/centrifugation and re-dispersion until a neutral pH was reached. Subsequent ultrasonication treatment was conducted to ensure dispersion of the nanofibers prior to freeze drying.

118 H2SO4-CNC was prepared as described in the literature. Sulfuric acid (64 wt%, 175 mL) was added to ground filter paper (20 g) placed in a round-bottomed flask equipped with a magnetic stirrer. The reaction flask was immersed in an oil bath at 45 °C for 45 min. Thereafter, the reaction mixture was diluted 10 times with deionized water and washed by repeated centrifugation dispersion. The H2SO4-CNC was further dialyzed against deionized water for 10 days, sonicated for 30 min, and finally filtered through a pore 1 glass filter to remove residual microfibers.

H2SO4-CNC was 229 ± 5 nm in size as estimated by dynamic light scattering (DLS), and their charge was 263 ± 6 μeq/g as determined by polyelectrolyte titration.

3.4 Preparation of functionalized CNC (F-CNC)

All functionalized CNCs were prepared utilizing the same method, as follows. In a typical procedure, of ground filter paper (0.50 g) and deionized water (11.25 mL) were added to a round-bottomed flask (50 mL) equipped with a magnetic stirrer and immersed in an ice/water bath. To this mixture, HCl (3.75 mL, 37 %) was added drop wise and the flask was then immersed in a preheated oil bath at 110 ˚C for 15 min. Thereafter, hydrolyzed cellulose fibers were filtered and washed with deionized water through a glass filter (pore size 1) until a neutral pH was reached. The collected cellulose fibers were further hydrolyzed with a

20 ————— Experimental —————

functional acid (10 mL) for 4 hours at 110 °C. This functionalized CNC (F- CNC) was washed and collected as reported for HCl-CNC. The F-CNCs were denoted: 2-PyA-CNC, 4-PA-CNC, 2-BPA-CNC and 3-MPA-CNC for 2-propynoic acid, 4-pentenoic acid, 2-bromopropanoic acid and 3- mercaptopropionic acid hydrolysis, respectively. A schematic illustration of the preparation of the F-CNCs is given in Figure 10.

Figure 10. Illustration of the procedure for preparing functionalized cellulose nanocrystals (F-CNC) via acid hydrolysis of cellulose fibers

3.5 Surface-initiated polymerization

3.5.1 SI-ROP of PCL from cellulose substrates

The surface modification of cellulose substrates, filter paper (FP) and nanopaper (NP) was carried out according to Figure 11. A known mass of cellulose substrate was placed in an E-flask equipped with a magnetic stirrer together with the ε-CL monomer (20.6 g, 181 mmol) and a catalytic amount of the catalyst Ti(On-Bu)4 (35.4 mg, 0.10 mmol). The mixture was degassed by vacuum for 30 min and the E-flask was then immersed in an oil bath at 120 °C (normal heating (NH)) or was opened and conditioned under microwave irradiation (MH) of 180 W using a domestic microwave oven. For NH, samples were continuously withdrawn to determine the extent of monomer conversion by 1H-NMR. When the targeted conversion was reached, the reaction was terminated by cooling the E-flask in an ice bath and adding THF to the mixture. The

21 ————— Experimental —————

free PCL was precipitated in cold MeOH, filtered and dried in a vacuum oven at 50 °C. Residual non-grafted PCL was removed via ultrasonication of the cellulose substrate three times in THF (50 mL) for 10 min, and thereafter by Soxhlet extraction with THF for 24 h. The grafted filter paper was dried under vacuum at 50 °C overnight, whereas the grafted nanopaper was solvent-exchanged to methanol and subsequently dried with supercritical CO2. ROP with tin 2-ethylhexanoate catalyst was conducted in a manner similar to that with Ti(On-Bu)4. Briefly, the monomer ε-CL (20.6 g, 180.5 mmol) and cellulose substrate were placed in an E-flask equipped with a magnetic stirrer. The catalyst Sn(Oct)2 (0.4 g, 2 wt% of ε-CL) and the co- initiator benzyl alcohol (47.2 mg, 0.44 mmol) were added to the reaction flask under a flow of argon. Thereafter, the flask was degassed by 3 vacuum/argon cycles and then treated under the same conditions as described previously. The grafted cellulose substrate and the free polymer were treated as previously described.

Figure 11. Grafting of ε-CL from cellulose substrates, filter paper or CNF nanopaper, via surface-initiated ROP.

3.5.2 SI-RAFT/MADIX of VAc from HCl-CNC

Prior to SI-RAFT/MADIX, the CTA was immobilized on HCl-CNC. Two approaches were investigated (Figure 12); a one-step approach via esterification (Method 1), and a two-step approach (Method 2) where a bromo-ester group was first attached to CNC followed by reaction with KEX. Detailed information regarding the experimental procedures can be found in Paper II.

22 ————— Experimental —————

Figure 12. Immobilization of CTA agent.

SI-RAFT/MADIX polymerization from CNC-CTAn (n = 1 or 2 refers to Method 1 or Method 2, respectively) was conducted according to

Figure 13. First, CNC-CTAn (0.20 g) was placed in a round-bottomed flask (25 mL) and the VAc monomer (10.0 g, 116 mmol), EtOAc (10.0 g), free

CTA (amount depending on the targeted DPn), and AIBN (0.1 n(CTA) ) were then added. The reaction mixture was degassed under a flow of argon for 30 min and the reaction vessel was then immersed in a preheated oil bath at 60 °C. The reaction mixture was run to high conversion and finally terminated by cooling, opening the system and adding THF. The reaction mixture was diluted with THF, and CNC was isolated and washed by Soxhlet extraction with THF for 24 hours. The free polymer was concentrated by evaporation under reduced pressure prior to further analysis. Blank reactions with unmodified HCl-CNC were also carried out under the same conditions.

23 ————— Experimental —————

Figure 13. SI-RAFT/MADIX of VAc from cellulose nanocrystals via surface-initiated RAFT/MADIX.

3.5.3 SI-ATRP of PBMA from H2SO4-CNC

Prior to SI-ATRP on H2SO4-CNC, the ATRP initiator was immobilized on the CNC surface (Figure 14) according to the procedure described in the literature41 with a slight modification. For detailed information, see Paper III.

Figure 14. Immobilization of ATRP initiator and grafting of n-BMA from cellulose nanocrystals via surface-initiated ATRP.

In a typical procedure of SI-ATRP from CNC, the ATRP initiator- immobilized CNC (CNC-Br) (0.50 g) was dispersed in toluene (25 g) in a round-bottomed flask (100 mL) equipped with a magnetic stirrer and the dispersion was sonicated using a sonication bath for 5 min. Thereafter, BMA (25.0 g, 176 mmol) was added and the reaction flask was kept in an ice/water bath, and EBIB (137 mg, 0.70 mmol) and HMTETA (203 mg, 0.88 mmol) were then added to the mixture. The reaction flask was degassed by one vacuum/argon cycle after which Cu(I)Br (101 mg, 0.70 mmol) and Cu(II)Br (39.3 mg, 0.18 mmol) were added under a flow of

24 ————— Experimental —————

argon. Finally, the round-bottomed flask was sealed with a rubber septum, degassed with 2 vacuum/argon cycles and immersed in an oil bath pre-heated to 80 °C. The monomer conversion was monitored by withdrawing 1H-NMR samples. The reaction was terminated at a conversion of about 70 % by exposing the reaction mixture to air and diluting it with DCM. The polymer-grafted CNCs (CNC-g-PBMA(S or L)) were separated and purified from the free polymer by dispersing the reaction mixture in DCM and filtering. The free polymer was passed through neutral aluminum oxide to remove copper and was then precipitated in cold methanol, decanted and finally dried under vacuum at 50 °C overnight. The grafted CNC-g-PBMA(S or L) were purified individually by Soxhlet extraction with THF for 24 h to remove any unbonded free polymer, and thereafter with MeOH for 24 h to remove any residual copper.

3.6 Physical adsorption of micelles and latex particles

The physical adsorption of P(DMAEMA-co-MAA)-b-PBMA latex particles and PDMAEMA-b-PBMA micelles was performed according to the following procedure. Typically, an aqueous suspension of H2SO4-CNC was diluted with deionized water (0.01 wt%) to which a dispersion of P(DMAEMA-co-MAA)-b-PBMA latex particles or PDMAEMA-b-PBMA micelles was added slowly under rigorous stirring. Thereafter, the mixture was stirred for an additional hour followed by subsequent freeze- drying overnight. The physisorbed amount of PBMA on CNC was similar to the grafted amount of PBMA on CNC-g-PBMA(S or L). The samples were denoted CNC-m-PBMA (S or L) and CNC-l-PBMA(S or L) for micelles and latex particles, respectively. The letters S and L in the sample names stand for short (S) and long (L) PBMA chain length. Detailed information regarding the synthesis of micelles and latex particles can be found in Paper III.

3.7 Film preparation

PVAc reinforced with CNC-g-PVAc (Paper II) was prepared via solvent casting. CNC-g-PVAc was first dispersed in THF by sonication for 30 s, and PVAc was then added and the mixture was stirred overnight to ensure full dissolution of PVAc. Thereafter, the mixture was solvent-cast

25 ————— Experimental —————

in aluminum cups and left to dry at 50 °C until constant weight was reached. The nanocomposite films were further hot pressed, under 150 kN at 70 ºC for 10 min, to ensure homogeneous thickness of the samples (ca. 130 µm).

For the comparison study (Paper III), the films of pure PCL and PCL reinforced with unmodified CNC or modified CNC were prepared according to the following procedure: PCL (5 g) together with o, 0.5, 1, or 3 wt% CNCs were mixed in a twin mini-extruder operating at 100 rpm, 110 ºC for 6 min. The extruded material was further hot-pressed into 130 µm thick films under 200 kN at 80 ºC for 10 min.

.

26 ————— Results and discussion —————

4 Results and discussion

The main focus of this work was to investigate new strategies for the surface modification of nanocelluloses in order to increase their dispersability and compatibility with hydrophobic matrices. The first part of this chapter describes the surface modification of nanocelluloses by SI- polymerization of various monomers employing controlled polymerization techniques, namely ROP, RAFT/MADIX and ATRP, and by physical adsorption of amphiphilic block copolymers (Papers I-III), together with the simultaneous preparation and functionalization of CNC (Paper IV). The second part of this chapter deals with the properties of the modified nanocellulose embedded in a hydrophobic polymer matrix (Papers I-III).

4.1 Surface modification of nanocellulose

4.1.1 SI-polymerization from nanocellulose The controlled polymerization by SI-ROP, SI-RAFT/MADIX and SI- ATRP from nanocelluloses (CNF nanopaper and CNC), and the characterization of the resultant materials is here described.

4.1.1.1 SI-ROP of ε-caprolactone from CNF nanopaper

ROP is an oxygen- and moisture-sensitive reaction. Traces of water can also initiate the polymerization, which lowers the final molecular weight and broadens the molar-mass dispersity. The main challenge of SI-ROP from nanocellulose is therefore the removal of all water. Conventionally, solvent exchange into an organic solvent can be performed in order to replace and remove water from an aqueous CNF suspension, but this involves tedious multi-step processes, and drying of CNF by simple solvent evaporation leads to the formation of a dense CNF film with irreversibly aggregated nanofibrils. These CNF aggregates are unable to redisperse, especially in an organic solvent. Herein, CNF nanopaper (NP) with a high specific surface area (304 m2/g), where a network of individual nanofibrils was prepared via filtration and subsequent CO2 supercritical drying, was employed as a template for the preparation of nanocomposites by SI-ROP of ε-CL. SI-ROP from filter paper (FP) has

27 ————— Results and discussion —————

previously been studied.119-121 This substrate exhibits a specific surface area as low as 0.59 m2/g, and FP is therefore a suitable substrate for comparison with NP.

Recently, Pärssinen et al. have shown that ROP can be conducted in air 74 with a titanium alkoxide catalyst. Thus, Ti(O-nBu)4 was considered in this study as an appropriate catalyst for comparison with Sn(Oct)2, the most commonly employed catalyst for ROP.

Two modes of heating for the polymerization were also tested, normal heating with an oil-bath (NH) and microwave heating (MH). Microwave heating was investigated since several successful studies reported ROP of ε-CL under MH.122, 123 In all the systems, the reactions were performed targeting both high and low monomer conversions (Table 1). The extent of the SI-ROP polymerization was determined by gravimetric analysis (Table 1) and also by Fourier transform infrared spectrometry (FT-IR) (Figure 15). The samples were denoted (FP or NP)-(NH or MH)- and the weight percentage of PCL in the grafted substrate (% grafted PCL), e.g. FP-NH-33.

Comparison between high- and low-surface-area substrates In order to investigate the grafting ratio, the FP and NP were weighed before and after grafting and the percentage of PCL in the composite was calculated. As predicted, the amount of PCL grafted from NP was substantially higher than that grafted from FP for both methods of heating and for both catalysts, due to the higher specific surface area of the former and thus more available OH groups. It was found that 64 wt%

PCL could successfully be grafted onto the NP using Ti(On-Bu)4 under normal heating in the presence of air and 79 wt% using Sn(Oct)2 under an inert atmosphere. This is remarkably high and is expected to give favorable mechanical properties to the nanocomposite.

28 ————— Results and discussion —————

Table 1. Amounts of PCL grafted from cellulose substrates before and characterization of the free PCL.

Sample % Sample conv. Reaction Mn,th Mn name in grafted ÐM name (%)a time (g/mol) (g/mol)c

Paper I PCLb Catalyst

FP-NH-33 FP-Ti-33 33 33 12.5 h 16 300 10 700 1.16

FP-NH-50 FP-Ti-50 78 50 22.0 h 38 600 25 800 1.95

4 NP-NH-50 NP-Ti-50 32 50 19.3 h 15 800 15 000 1.15

Bu) NP-NH-64 NP-Ti-64 70 64 29.0 h 34 600 27 900 1.45 - n FP-MH-10 _* 40 10 5.5 min 19 800 37 900 2.16

Ti(O FP-MH-12 _* 88 12 7.0 min 43 500 56 000 2.20 NP-MH-21 _* 30 21 5.5 min 14 800 28 100 1.96 NP-MH-44 _* 81 44 7.0 min 40 100 30 200 2.89 FP-NH-2 FP-Sn-2 35 2 25 min 16 500 15 700 1.16

FP-NH-3 FP-Sn-3 77 3 45 min 36 300 36 700 1.43 2 NP-NH-61 NP-Sn-61 45 61 45 min 21 200 20 300 1.13 NP-NH-79 NP-Sn-79 81 79 60 min 38 100 39 400 1.59

Sn(Oct) FP-MH-3 _* 77 3 3.0 min 36 300 39 900 2.20

NP-MH-33 _* 93 33 3.0 min 43 800 26 700 2.36

d FP-NH-4 FP-b 1 4 22.0 h _ _ _ NP-NH-8 NP-b 2 8 29.0 h _ _ _

FP-MH-2 _* 2 2 7.0 min _ _ _ blank

reaction NP-MH-2 _* 1 2 7.0 min _ _ _ adetermined by proton NMR bpercentage of PCL in the grafted cellulose substrate, cnumber average molar mass of free polymer determined by chloroform size exclusion chromatography (SEC) dno catalyst was used. *not included in Paper I

Comparison between Sn(Oct)2 and Ti(O-nBu)4

Table 1 shows that the FP grafted with PCL using Ti(On-Bu)4 as catalyst gave a significantly higher amount of grafted PCL than FP grafted using

Sn(Oct)2, at the same time, the reaction was much faster with Sn(Oct)2 than with Ti(On-Bu)4. Interestingly, both catalysts led to high amount of PCL grafted from NP which could be explained by the larger surface area of NP than the dense FP.

Noteworthy, the number average molar mass (Mn) of free PCL obtained with Sn(Oct)2 was almost the same as the theoretical value (Mn,th) under

NH, while the difference between the Mn and the Mn,th values was more pronounced with Ti(On-Bu)4. This could be due to favored

29 ————— Results and discussion —————

transesterification reactions at high temperature and the longer reaction time with Ti(On-Bu)4 than with Sn(Oct)2.

Comparison between normal heating and microwave heating SI-ROP employing MH was conducted in a shorter time, maximum 7 min, for high conversions than with NH, which required up to 29 h. The characterization of the free PCL formed in bulk from ROP of both FP and

NP under NH (Table 1) shows that the Mn of free PCL increased with increasing extent of conversion. This is reasonable since these parameters are linearly proportional to each other in a controlled polymerization.

Interestingly, the ÐM values obtained with MH are higher than these obtained with NH, indicating that the reaction is less controlled when a domestic microwave oven is used. The domestic microwave oven leads to heterogeneous heating, and consequently to an uneven polymerization.

Figure 15. The 1800-1600 cm-1 and 3200-3600 cm-1 regions of the FT-IR spectra of cellulose substrates, (a-c) normal heating (NH), (d-f) microwave heating (MH) and (g) full spectra of unmodified FP and NP, Ref-FP and Ref-NP.

30 ————— Results and discussion —————

The FT-IR spectra of the unmodified and the grafted cellulose substrates (Figure 15) confirmed the results of the gravimetric analysis, and supported the above discussion. The intensity of the broad band corresponding to OH groups on cellulose, located at about 3300 cm-1, decreased with increasing PCL content, and the intensity of the band characteristic of carbonyl group at around 1730 cm-1, which is present in PCL and absent in cellulose, increased with increasing amount of grafted PCL.

SI-ROP has attracted great interest for the surface modification of cellulose as it is initiated by the naturally occurring hydroxyl groups on the cellulose backbone. However, this polymerization technique is limited to cyclic monomers such as lactones and lactams. To circumvent this issue, controlled radical polymerizations, capable of synthesizing a wide range of monomers, have also been employed.28, 30 One highly interesting RDRP method, SI-RAFT/MADIX has not however yet been widely studied for the modification of cellulose, and the next section summarizes the results of SI-RAFT/MADIX of VAc conducted on HCl-CNC.

4.1.1.2 SI-RAFT/MADIX of vinyl acetate from HCl-CNC

Unlike SI-ROP, which is initiated by naturally occurring OH groups on the cellulose backbone, SI-RAFT/MADIX requires, in theory, the immobilization of a chain transfer agent (CTA) on the surface prior to the SI-polymerization. In this particular study, a CTA was attached to the CNC surface employing either a one-step esterification method (Method 1) or a two-step approach (Method 2). In Method 1, the CTA was attached to CNC through esterification with an organic acid bearing the CTA functionality, whereas in Method 2, a secondary bromo-ester group was first immobilized, followed by substitution of the bromine with potassium ethyl xanthogenate (KEX) (Figure 12, section 3.5.2). Successful immobilization of the CTA on CNC by both methods was confirmed by X- ray photoelectron spectroscopy (XPS) and FT-IR spectroscopy. An increase in the O-C=O bond contribution to the C1s peak, and a slight increase in sulfur content were observed by XPS, and a low intensity of the carbonyl band at 1730 cm-1 was detected by FT-IR. The 2-((ethoxycarbonothioyl)thio)propanoic acid employed for the preparation of CNC-CTA1 via a single-step approach has a carboxylic

31 ————— Results and discussion —————

functionality (COOH) at one end. The esterification reaction of 2- ((ethoxycarbonothioyl)thio)propanoic acid with hydroxyl groups on cellulose forms an ester functionality (Figure 12). The presence of an ester group and the absence of the COOH contribution to the C1s peak in the CNC-CTA1 indicate that CTA is covalently attached to the CNC surface, and that the sample contains no unbound CTA. However, no conclusion could be drawn with regard to the CNC-CTA2 sample prepared via a two- step approach. In an attempt to monitor the immobilization of CTA on the surface of CNC-CTA2, the second step of the immobilization process was followed by UV spectrometry, taking advantage of the strong absorbance of KEX at 295 nm. The results showed that the concentration of KEX in the solution fluctuated and that the amount of reacted KEX was higher than the theoretical number of available OH groups on the cellulose. This was found for both bromo-ester-immobilized cellulose and unmodified cellulose, which indicates that KEX has a strong affinity to cellulose, and thus that the CTA was not fully covalently attached in the CNC-CTA2 sample.

CTA-immobilized CNC; CNC-CTA1 and CNC-CTA2 from the single-step and two-step reactions respectively were further utilized for SI- RAFT/MADIX of VAc. Table 2 summarizes the characteristics of the free polymer formed in bulk for DP 57 and DP 230. The samples were denoted

CNCCTAx (x corresponds to 1 or 2 depending on the method utilized)-g-

PVAcDP (DPn = 57 or 230). The samples from blank reactions were named

CNC-blank-PVAcDPn. The kinetic study of all the systems (Figure 16) shows that the presence of CNC had no strong influence on the

RAFT/MADIX of VAc and that the ÐM of the free polymer was below 1.2 for conversions lower than 80 %.

32 ————— Results and discussion —————

Table 2. Monomer conversion (conv.), theoretical molecular weight (Mn,th), number- average molecular weight (Mn) and molar-mass dispersity (ÐM) of the free PVAc. c Conv. Mn,th Mn ÐM (%)a (g mol-1)b (g mol-1)c

PVAc57 83 3 780 4 590 1.23

CNC-blank-PVAc57 85 3 880 5 160 1.10

CNCCTA1-g-PVAc57 80 3 630 5 000 1.10

CNCCTA2-g-PVAc57 85 3 880 5 840 1.11

PVAc230 96 18 900 27 420 1.44

CNC-blank-PVAc230 83 16 500 31 210 1.13

CNCCTA1-g-PVAc230 81 16 270 28 650 1.11

CNCCTA2-g-PVAc230 82 15 930 27 890 1.12 abased on 1H-NMR bcalculated from monomer conversion cdetermined by DMF-SEC with PMMA standards

Figure 16. Mn and ÐM of the free polymer as function of the degree of conversion during the polymerization of vinyl acetate without CNC (), with CNC (), CNC-CTA1 () and

CNC-CTA2 (); (a) DPn target 57 and (b) DPn target 230.

The characterization of CNC-g-PVAcDPn by FT-IR revealed that a higher degree of grafting was achieved with CNC-CTA1 than with CNC-CTA2 (Figure 17). This could be related to the CTA amount attached to the CNC. A larger amount of CTA may lead to a more extensive grafting of PVAc. These results were further confirmed by thermogravimetric analysis

(TGA). CNCCTA1-g-PVAcDPn showed a lower degradation temperature (Td) than the counterparts prepared from CNC-CTA2. A larger content of PVAc will generate more acetic acid upon heating and this will consequently catalyze the thermal degradation of the sample.

33 ————— Results and discussion —————

The FT-IR spectrum of the blank sample, CNC-blank-PVAcDPn, also revealed the presence of a low intensity of the carbonyl band, which indicated that an undesirable side-reaction of chain transfer occurred due to the presence of AIBN, as previously reported by others.124 The TGA thermogram of CNC-blank-PVAcDPn showed a trend similar to that of the pristine CNC. Thus, only a small amount of monomer, oligomer, or polymer was attached to this sample than with CNCCTAn-g-PVAcDPn.

Figure 17. FT-IR spectra from wavenumber 1790 to 1690 cm-1 of CNC-blanks (CNC-blank-

PVAcDPn ) and CNC-grafts (CNCCTAn-g-PVAc57 and CNCCTAn-g-PVAc230).

The surface modification of nanocelluloses, by means of SI-controlled polymerization, has been widely employed to tailor surface properties. However, the major drawback of this methodology is usually the formation of a large amount of free polymer, and immobilization of initiating species on the surface is usually required prior to the SI- polymerization, often requiring harsh reaction conditions not suitable for industrial up-scaling. A potential alternative method for surface modification of nanocelluloses is physical adsorption of an amphiphilic block copolymers. This approach as has been previously demonstrated by our group and by others,59-61, 101 and it is generally performed under mild conditions in water. A comparative study of the two surface modification methods in terms of the resulting nanocomposite properties has, to the

34 ————— Results and discussion —————

best of our knowledge, not been attempted previously. The next two sections deal with the surface modification of CNC by means of SI-ATRP of PBMA from H2SO4-CNC and physical adsorption of a PBMA-based amphiphilic block copolymer to H2SO4-CNC.

4.1.1.3 SI-ATRP of butyl methacrylate from H2SO4-CNC

As in the case of the SI-RAFT/MADIX, SI-ATRP also requires the immobilization of an initiator on the CNC surface. In this case, a tertiary ATRP initiator was attached to the surface through an esterification reaction utilizing an acyl bromide bearing an ATRP moiety (Figure 14, section 3.5.3). The resulting CNC-Br was further employed for SI-ATRP of BMA targeting two DPns. The DPn target was chosen to be either below or above the critical chain entanglement of PBMA chains, which is at about DP 400.125 Table 3 summarizes the characteristics of the free PBMA polymers formed in bulk during the SI-ATRP. The amount of PBMA attached to CNCs was estimated by means of gravimetric analysis and was 4 wt% and 28 wt% for DP 110 and DP 487, respectively.

Table 3. Theoretical molecular weight (Mn,th), molecular weight and molar-mass dispersity

(ÐM) of free PBMA.

-1 -1 Mn,th (g mol ) Mn (g mol ) ÐM

PBMA110 15 640 17 480 1.11

PBMA487 69 170 80 260 1.09

FT-IR, contact angle determination and TGA were utilized to characterize the grafted CNCs, CNC-g-PBMA(S or L), where S and L corresponds to DP 110 and DP 487, respectively, (denoted CNC-Graft(S or L) in paper III). The successful grafting was confirmed by the presence of a band at about 1730 cm-1 corresponding to the carbonyl functionality present in PBMA. The contact angle of CNC-g-PBMA(S, L) was 90° compared to 34° for unmodified CNC. Furthermore, the degradation temperature of CNC increased considerably upon grafting, as an effect of shielding of the charged sulfate groups on the CNC. The sulfate groups are reported in the literature to be responsible for catalyzing the thermal degradation of 24 H2SO4-CNC upon heating due to the generation of corrosive species.

35 ————— Results and discussion —————

4.1.2 Physical adsorption of micelles and latex particles on

H2SO4-CNC

Micelles and latex particles consisting of PBMA and a cationically charged PDMAEMA-based block were synthesized through RDRP polymerization techniques (see paper III for more details). The characteristics of the PBMA-based micelles and latex particles are summarized in Table 4. In order to make a reliable comparison between the two methodologies of surface modification, i.e. covalent grafting and physical adsorption, the adsorbed amount of micelles or latex particles was targeted to be similar to that of the PBMA covalently attached to CNC as illustrated in

Scheme 5. Furthermore, the DPns of the PBMA blocks were similar to those grafted from the CNC.

Table 4. Theoretical molecular weight (Mn,th), molecular weight and molar-mass dispersity

(ÐM) of PDMAEMA-b-PBMA and P(DMAEMA-co-MAA)-b-PBMA.

Polymer Mn,th Mn ÐM

(g mol-1) (g mol-1)

PDMAEMA27 (Macroinitiator) 3 930 4 220 1.14

PDMAEMA27-b-PBMA93 17 100 28 500 1.06

PDMAEMA27-b-PBMA523 78 610 89 000 1.20 Micelles

P(DMAEMA-co-MAA)25 (MacroRAFT) 4 200* - -

P(DMAEMA-co-MAA)25-b-PBMA160 25 910 48 400 1.78

P(DMAEMA-co-MAA)25-b-PBMA684 101 380 223 100 1.39 Latex *estimated from gravimetric measurements

36 ————— Results and discussion —————

Scheme 5. Surface modification of CNC via covalent grafting and physical adsorption approaches.

In this thesis, the micelle- and the latex-adsorbed CNC samples are denoted CNC-m-PBMA(S or L) and CNC-l-PBMA(S or L), respectively. Their denotation in Paper III was CNC-Micelle(S or L) and CNC-Latex(S or L), respectively.

The successful modification of CNC through physical adsorption was confirmed, as in the case of CNC-g-PBMA(S or L), by FT-IR, contact angle determination and TGA. Interestingly, the CNC-m-PBMA showed both a higher contact angle and a greater thermal stability than the CNC- l-PBMA. This could be due to less spreading of latex particles on the surface than of micelles. This assumption was supported by an analysis of the annealed samples of both CNC-m-PBMA and CNC-l-PBMA by TGA. The annealing is expected to enhance the spreading of PBMA blocks on the surface, resulting in a better shielding of the sulfate groups. The thermal stability of CNC-l-PBMA increased upon annealing but remained stable for CNC-m-PBMA. The contact angles of water on both micelle- and the latex-adsorbed CNC samples were lower than the corresponding CNC-g-PBMA.

37 ————— Results and discussion —————

4.1.3 Functionalized CNC

As previously mentioned, immobilization of a functional or cross-linkable group on CNC is usually necessary prior to SI-polymerization or crosslinking reactions. For this purpose, tedious functionalization, washing and multi-step purification procedures are usually employed. In an attempted to circumvent this problem, the production of functionalized CNC (F-CNC) by means of simultaneous acid hydrolysis and esterification was investigated.

A series of F-CNC were produced through acid hydrolysis employing an organic acid bearing a functional group, such as a double or triple bond, ATRP initiator, or thiol. A prehydrolysis step of cellulose fibers utilizing HCl was found to considerably shorten the reaction time.

The morphology of the F-CNC series together with HCl-CNC produced with HCl was investigated by atomic force microscopy (AFM) (Figure 18). All the F-CNCs had a rod-like shape, 13-18 nm wide and 175-249 nm long, depending on the organic acid employed. No clear correlation was observed between the size of F-CNC and the pKa value of the organic acid used. F-CNCs were further analyzed by X-ray diffraction analysis (XRD) to assess their crystallinity. The starting material, i.e. filter paper, had a crystallinity of 71 % while the F-CNCs had a crystallinity of up to 81 % as a consequence of removal of the most disordered regions in the cellulose fiber. F-CNCs prepared with 2-propynoic acid had the lowest crystallinity of 65 %, which could be explained by the greater peeling effect when 2- propynoic acid was used. Successful surface modification of F-CNC was confirmed by FT-IR, XPS and elemental analysis. A low-intensity but noticeable band, ascribed to the carbonyl group at about 1730 cm-1, was observed in the FT-IR spectra.

38 ————— Results and discussion —————

Figure 18. 5 × 5 m AFM images of functionalized CNCs obtained utilizing: (A) HCl (HCl- CNC), (B): 2-propynoic acid (2-PyA-CNC), (C) 2-bromopropionic acid (2-BPA-CNC), (D) 3- mercaptopropionic acid (3-MPA-CNC) and (E) 4-pentenoic acid (4-PA-CNC).

Figure 19. Contributions of O-C=O, O-C-O, O-C, C-C bonds to C1s in the deconvoluted signal of unmodified cellulose nanocrystals (HCl-CNC) and F-CNCs.

39 ————— Results and discussion —————

In the deconvoluted C1s peak of the XPS (Figure 19), a larger O-C=O bond contribution was observed for all F-CNCs than for HCl-CNC, originating from the ester group formed due to the reaction of hydroxyl groups on cellulose with the functional acids. The atomic percentages of bromine (Br) and sulfur (S) were found to be 0.3 % and 1.0 %, respectively, which correspond to weight percentages of 1.75 % and 2.42 %, respectively. This was further confirmed by elemental analysis where the Br and S contents were 2.86 % and 2.17 % for 2-BPA-CNC and 3-MPA-CNC, respectively. The difference in the Br percentage between the results of the XPS and of the elemental analysis could be explained by the sensitivity of Br to XPS analysis.126 The degree of substitution (DS) of F-CNC calculated from elemental analysis results varied between 0.06 and 0.27. No correlation was found between the pKa values of the organic acids and the DS. An explanation of this variation in DS might be the dissimilarity in viscosity and swelling ability of cellulose in the various acids.

In order to investigate the efficiency of the functional groups immobilized on the F-CNC surface, these nanoparticles for further modified. For instance, 2-BPA-CNC was employed to graft poly(methyl methacrylate) (PMMA) through SI-ATRP. Again, FT-IR spectra after grafting showed an increase in the intensity of the carbonyl band, originating from the ester group in PMMA. The contribution of the carbonyl functionality to the C1s peak, assessed by XPS analysis, also increased considerably, thus confirming the PMMA grafting. Copper-catalyzed Huisgen cycloaddition was employed to attach an azide-functionalized disperse red 13 (DR13-N3) on 2-PyA-CNC. This method has been reported to be an efficient and selective “click reaction” for azides and alkynes.127 After purification/washing, the modified CNC had a red color which indicated successful attachment of DR13-N3. The amount of the attached dye was estimated by UV-vis analysis to be as high as 335 mol g-1. The DS calculated from this value was lower than the DS obtained by elemental analysis. A probable explanation for this difference could be the steric hindrance caused by the bulky structure of DR13-N3 (Figure 20). The amount of attached DR13-N3 was higher than the values reported in the literature for rhodamine B isothiocyanate (RBITC) and fluorescein-5’-isothiocyanate (FITC), where the average

40 ————— Results and discussion —————

amounts of dye attached to CNCs were 2.1 mol g-1 and 2.8 mol g-1, respectively.128

Figure 20. Chemical structure of azide-functionalized disperse red 13 (DR13-N3).129

Similarly, the versatile thio-ene reaction was employed to attach thiol- functionalized disperse red (DR13-SH) onto 4-PA-CNC. The CNCs turned red after the reaction; but unmodified CNC employed as a blank sample also exhibited a red color, indicating significant non-specific sorption of the dye onto CNC. Hence, the amount of the ene-functionality on 4-PA- CNC was unfortunately not quantified through the thio-ene reaction with DR13-SH.

Thiol moieties present on the 3-MPA-CNC surface can be quantified by 5,5’‐dithiobis(2‐nitrobenzoic acid) (DTNB), also known as Ellman’s reagent. The reaction between a thiol and DTNB generates 5‐thio‐2‐ nitrobenzoic acid (TNB), which is a chromophore that absorbs light at 412 nm (Figure 21). Thus, the TNB content can be assessed by UV-vis spectroscopy. In this case, the measured absorbance was utilized to calculate the amount of thiols present on 3-MPA-CNC, which was found to be 766 mol g-1.

Figure 21. Reaction scheme of Ellman’s reagent with a compound bearing a thiol functionality.

41 ————— Results and discussion —————

4.2 Characterization of nanocomposites

4.2.1 PCL grafted CNF nanopaper

PCL grafted NP- and FP-composites prepared by SI-ROP (section 4.1.1.1) were characterized by differential scanning calorimetry (DSC), field emission-scanning electron microscopy (FE-SEM), Brunauer−Emmett−Teller (BET) method, dynamic vapor sorption (DVS) and dynamic mechanical analysis (DMA).

The degree of crystallinity (Xc), the crystallization temperature (Tc) and the melting temperature (Tm), assessed by DSC (Table 5), increased with increasing grafted amount of PCL, which is in accordance with the results reported by Hult et al.75 However, the samples prepared under microwave irradiation showed a deviation, probably due to the inhomogeneity of the grafting.

Table 5. Results of the DSC analysis of grafted cellulose substrates.

a Cellulose Denotation Tc (°C) Tm (°C) Xc (%) Mn (g/mol) substrate in Paper I FP-NH-33 FP-Ti-33 27.1 52.9 39.3 10 700 FP-NH-50 FP-Ti-50 32.6 57.2 47.8 25 800 NP-NH-50 NP-Ti-50 27.1 51.6 27.7 15 000 NP-NH-64 NP-Ti-64 30.3 56.0 35.1 27 900 FP-MH-10 _b 25.4 52.5 13.3 35 900 FP-MH-12 _b 24.5 49.8 21.6 56 000 NP-MH-21 _b 24.9 50.1 29.1 28 100 NP-MH-44 _b 23.2 46.9 11.1 30 200 NP-NH-61 NP-Sn-61 16.6 41.2 19.9 20 300 NP-NH-79 NP-Sn-79 27.1 53.7 34.7 39 400 aaverage number molar mass of free polymer determined by chloroform SEC. bnot included in Paper I

The grafting of PCL on cellulose substrates resulted in a smoother surface texture than that of the unmodified substrates, as can be seen in the FE- SEM micrographs (Figure 22). Moreover, the thickness of cellulose fibers/nanofibrils increased with increasing amount of grafted PCL. This had a significant impact on the pore size, porosity and density, as shown

42 ————— Results and discussion —————

in Table 6, and strongly indicates that it should be possible to control these parameters by varying the amount of grafted PCL.

Figure 22. SEM micrographs of unmodified and grafted cellulose nanopaper and filter paper.

Table 6. Specific surface area, pore size and porosity measured by the BET method, and density of the unmodified and grafted cellulose substrates. Specific Pore size Density Porosity Sample surface area (kg/m3) (m2/g) (nm) (%) FP 0.59 * 550 62 FP-NH-50 0.39 * 670 47 NP 304 35.8 380 74 NP-NH-50 52 31.8 600 52 NP-NH-64 28 28.1 685 47 *Out of the measurement range of the equipment

Cellulose is very hygroscopic. The moisture uptake of the unmodified NP and of the PCL-grafted NP (NP-NH-50) was determined by DVS at three different relative humidities (25 %, 50 % and 80 % RH), and it was found that the moisture uptake increased with increasing RH for both substrates, but that the moisture uptake by NP-NH-50 was 60 % less than that of the pristine NP.

43 ————— Results and discussion —————

Figure 23. Storage modulus vs temperature for neat PCL, reference substrates (FP-NH-4 and NP-NH-8) and for the grafted substrates (FP-NH-50 and NP-NH-50).

Mechanical properties of the PCL composites utilizing Ti(On-Bu)4 under NH and containing similar amount of grafted PCL (FP-NH-50 and NP-NH-50), unmodified cellulose substrates, and the neat PCL were evaluated by DMA. The curves of storage modulus versus temperature are shown in Figure 23.

PCL is a semi-crystalline polymer with a glass transition temperature of about -60 °C, and it therefore has a relatively low storage modulus (E´) of 225 MPa at room temperature (RT), but the grafted substrates show a much higher E´ at RT; 800 MPa and 700 MPa for the grafted NP-NH-50 and FP-NH-50, respectively. Starting from the melting temperature of PCL at 60 °C, the neat film undergoes a dramatic decrease in the storage modulus before it finally breaks at 75 °C. The grafted substrates did not break, however, but retained good mechanical properties at high temperatures as an effect of fiber/nanofiber network. For instance, the E´ modulus at 200 °C was 85 MPa and 325 MPa for the grafted FP and the grafted NP, respectively. Below 75 °C, the mechanical properties of FP- NH-50 and NP-NH-50 are comparable, but the decrease in the E´ at high temperatures is more pronounced for FP-NH-50 than for NP-NH-50. This is expected, as the large contact area between the CNF in the NP results in a stronger network, capable of resisting high temperatures.

44 ————— Results and discussion —————

4.2.2 PVAc reinforced with PVAc grafted CNC

Unmodified CNC, CNC-blank-PVAc230, CNC-g-PVAcDPn (prepared via SI- RAFT/MADIX from CNC-CTA1) were incorporated in a PVAc matrix by means of solvent casting. The CNC content in the nanocomposite films was either 0, 0.5, 1, 3, or 5 wt%. The characterization of PVAc and the CNC-g-PVAc/PVAc nanocomposite by UV-vis spectroscopy showed that the film transparency decreased with increasing CNC content, but that this decrease was less pronounced in the nanocomposites containing CNC-g-PVAc than unmodified CNC, which indicates better a dispersion of the nano-reinforcing agent as a result of the surface modification.

FE-SEM micrographs of the cross-section of cryo-fractured films of PVAc and PVAc nanocomposites containing 3 wt% of CNC, CNC-blank-

PVAc230, or CNC-g-PVAcDPn are shown in Figure 24. In general, the roughness of the film cross-section increased upon addition of CNC nanoparticles, showing that the resistance to crack propagation was higher in the PVAc nanocomposites than in the neat PVAc film.

Figure 24. SEM cross-section images of cryo-fractured films of unfilled PVAc, CNC/PVAc,

CNC-blank-PVAc/PVAc and CNC-g-PVAc/PVAc for DP 57 and DP 230. Nanocomposite films containing 3 wt% of CNCs.

The mechanical properties of PVAc and PVAc nanocomposites were evaluated by tensile testing. The Young’s modulus (E) is shown in Figure 25 and the values of E, stress at break and strain at break are given

45 ————— Results and discussion —————

in Paper II. The mechanical properties of neat PVAc film are relatively low at RT as a consequence of its low glass transition temperature (Tg) of ca. 26 °C and its amorphous nature. The incorporation of CNC nanoparticles generally increased the Young’s modulus and tensile strength of PVAc with increasing content of nano-reinforcing agent. This increase was significantly higher with CNC-g-PVAcDPn than with unmodified CNC.

Figure 25. Young’s modulus of PVAc and CNC/PVAc, CNC-blank-PVAc230/PVAc and CNC- g-PVAcDPn/PVAc, with different CNC contents, from left to right 0.5, 1, 3 and 5 wt%. The figures on the columns show the percentage increase in Young’s modulus compared with neat PVAc.

The strain at break of PVAc decreased on addition of the nano-reinforcing agent, especially for a CNC content > 3 wt%. This decrease in film ductility could be related to the rigid nature of the nanoparticles and to the probable formation of CNC a percolating network with high contents of the nano-reinforcing agent. Interestingly, the nanocomposite with

CNC-blank-PVAc230 showed the greatest increase of Young’s modulus, up to 195 % compared to the neat PVAc film. This could be related to the

46 ————— Results and discussion —————

difference in surface modification mechanisms in the preparation of

CNC-blank-PVAc230 and CNC-g-PVAcDPn, where the former is believed to have been prepared through an undesired chain transfer side-reaction while the latter was prepared by SI-RAFT/MADIX.

No clear trend was observed when the DPn of the grafted PVAc chains was varied.

4.2.3 PCL reinforced with covalently grafted or physisorbed PBMA-modified CNC

PCL-nanocomposite films were produced with unmodified CNC (nCNC/PCL) or with the differently modified CNC (nCNC-g-PBMA(S or L)/PCL, nCNC-m-PBMA(S or L)/PCL, or nCNC-l-PBMA(S or L)/PCL, where the number n shows the weight percentage of CNC in the nanocomposite). These were characterized by UV-vis spectroscopy, tensile testing and their cryo-fractured cross-sections were studied by FE-SEM.

The transparency of a neat PCL film with a thickness of 130 m is low due to its semi-crystalline nature. The addition of CNC significantly increased the transparency even at a content of 3 wt%. Interestingly, the CNC modified with long PBMA chains showed the highest transmittance. This is in accordance with the literature, where good compatibility between the nano-reinforcing agent and the matrix has been shown to increase the matrix transparency.130

The FE-SEM micrographs of the cross-section of cryo-fractured PCL- nanocomposite films (Figure 26) showed that the morphology of the PCL nanocomposite films containing modified CNC was similar to that of unfilled PCL film, but large aggregates were observed in the 3CNC/PCL sample. This was expected because of the poor compatibility between the hydrophilic CNC and the hydrophobic PCL matrix, supporting the finding that modification is necessary if superior composite properties are to be achieved.

47 ————— Results and discussion —————

Figure 26. Cross-section FE-SEM images of cryo-fractured PCL, 3CNC/PCL, 3CNC-g- PBMA(L)/PCL, 3CNC-m-PBMA(L)/PCL and 3CNC-l-PBMA(L)/PCL.

The mechanical properties of PCL and PCL-based nanocomposite films are shown in Figure 27. At room temperature and 50 % RH, the neat PCL film was able to undergo a large extension of about 1200 %. Its Young’s modulus and strength were relatively modest 280 MPa and 25 MPa, respectively, which is characteristic of a semi-crystalline polymer. The Young’s modulus increased up to 28 % when unmodified CNC was added to the PCL matrix, but its strain at break was considerably reduced, decreasing from 1200 % to 79 % with the addition of 3 wt% CNC. The addition of modified CNC increased the Young’s modulus and strength of PCL without reducing its ductility. The strength and strain at break of nanocomposites with modified CNC were up to 600 % and 1214 % higher respectively than those of the corresponding nanocomposites with pristine CNC. The length of the PBMA chains had no clear effect on the modulus, but both the strength and strain at break were greater with the shortest PBMA chains.

48 ————— Results and discussion —————

Figure 27. Mechanical properties at 50 % RH of PCL, CNC/PCL and CNC-g, m, or l- PBMA(S, L)/PCL with from left to right 0.5, 1 and 3 wt% of CNC. The figures in the columns show the percentage change compared with PCL reinforced with unmodified CNCs.

49 ————— Results and discussion —————

In general, CNC-g-PBMA(S or L)/PCL nanocomposite exhibited better mechanical properties than the CNC-m or l-PBMA(S or L)/PCL counterparts, and the CNC-l-PBMA(S or L)/PCL showed the lowest performance. All CNC-m or l-PBMA(S or L)/PCL showed a better mechanical performance than the corresponding nanocomposites with pristine CNC. The difference observed between CNC-g-PBMA(S or L)/PCL and CNC-m, l-PBMA(S or L)/PCL could be related to a better surface coverage of the grafted CNC compared with micelles- and latex- physisorbed CNC (see section 4.1.2). The difference in adsorption mechanism between micelles and latex particles could explain the difference observed in the mechanical performance of the corresponding nanocomposite.

At high humidity (98 % RH), the PCL strength increased considerably while the other parameters, i.e. Young’s modulus and strain at break, remained in the same range as at 50 % RH (Figure 28). This could be explained by the formation of strong hydrogen bonds between the carbonyl groups on the PCL backbone and the adsorbed moisture. The strain at break and strength of PCL nanocomposites containing 1 wt% of CNC-m or l-PBMA(S) decreased on conditioning at 98 % RH, but they were maintained by the CNC-g-PBMA(S)/PCL film. These results can be explained by the weakened interface between PCL matrix and the reinforcing agent as a consequence of moisture adsorption by the hydrophilic PDMAEMA blocks present in both the micelles and latex particles.

50 ————— Results and discussion —————

Figure 28. Mechanical properties of PCL, CNC/PCL and CNC-g, m, or l-PBMA(S)/PCL with 1 wt% CNC content at 50 % RH (pink) and 98 % RH (purple).

4.2.4 F-CNC-based hydrogels

Studies regarding the incorporation of F-CNC into nanocomposite materials (Paper V and VI) are not included in this thesis, but it is worth mentioning briefly the work that has been conducted in these studies.

In Paper V, thermo-responsive cryogels of poly(N-isopropylacrylamide) (PNIPAAm) reinforced with acrylate-functionalized CNC (AA-CNC) were prepared through free radical polymerization. AA-CNC was prepared in a manner similar to that of the other F-CNCs, reported in the experimental section, by utilizing acrylic acid. Non-reinforced PNIPAAm and reinforced PNIPAAm with unmodified CNC (HCl-CNC) cryogels were

51 ————— Results and discussion —————

also prepared for comparison. The CNC content in the cryogel was 1, 2 and 5 wt%. The unreinforced PNIPAAm cryogel exhibited a compressive strain at break (εbreak) at about 50 %. The thermo-responsiveness and the

εbreak of all the reinforced cryogels were similar to or exceeded that of the non-reinforced cryogel. Interestingly, PNIPAAm reinforced with 1 wt% of

AA-CNC had the highest εbreak, above 90 %. This strongly indicates that AA-CNC nanoparticles were covalently attached to the PNIPAAM matrix through their cross-linkable functionality.

The scope of the study presented in Paper IV was to synthesis a library of functional dendritic-linear-dendritic (DLD) block copolymers. In an attempt to produce bio-based hydrogels, 3-MPA-CNC or bovine serum albumin were crosslinked in 1:1 molar ratio with DLD block copolymers through thiol-ene, thiol-yne, or amine-N-Hydroxysuccinimide (NHS) ester chemistries. The swelling capacity of a CNC-based hydrogel was found to be about 250 %, which is less than that of the neat hydrogel. This could be explained by the crystalline and rigid nature of the CNC nanoparticles combined with the high content of these, up to 55 wt% of the hydrogel’s dry content.

52 ————— Conclusions —————

5 Conclusions

In the work described in this thesis, new strategies for the surface modification of nanocelluloses have been studied, mainly by means of covalent grafting or by the physical adsorption of polymers utilizing different controlled polymerization techniques (ROP, ATRP and RAFT/MADIX polymerizations).

SI-ROP of ε-CL from CNF nanopaper provides a new and facile methodology for the design of bionanocomposites avoiding tedious and multi-step solvent exchange processes, and this bottom-up technique makes it possible to prepare nanocomposites with a high content of well- dispersed nanoparticles within the matrix. It is believed that the porosity and density of the bionanocomposite can be tuned by varying the grafted amount of PCL. Interestingly, the resulting bionanocomposites exhibited stronger mechanical properties than the unfilled PCL especially at elevated temperature, and their moisture uptake was much lower than that of the unmodified NP.

SI-RAFT/MADIX polymerization of vinyl acetate has been explored for surface modification of CNC for the first time. This technique is a robust and versatile RDRP capable of polymerizing a wide range of “difficult” monomers. The immobilization of CTA on CNC prior to polymerization was attempted via two approaches. The single-step esterification approach is recommended, as it led to a larger amount of grafted PVAc than with the two-step method. This is probably due to the non-specific binding of potassium xanthogenate salt to cellulose, and to cleavage of the bromo-ester group during the immobilization process. The reproducibility was relatively low for both methods. The PVAc reinforced with modified CNC nanocomposites gave a better transparency and mechanical performance than the materials containing unmodified CNC and neat PVAc film. The CNC obtained from the blank reaction was also modified, probably through undesirable chain-transfer side reactions, and their corresponding nanocomposites exhibited the highest Young’s modulus with contents higher than 3 wt%. Hence, the immobilization of CTA may not be necessary in the case of SI-RAFT/MADIX polymerization of VAc.

53 ————— Conclusions —————

Surface modification of nanocelluloses has been studied primarily through the covalent grafting or physical adsorption of polymers. However, comparative studies between the two methods have not been attempted previously. For reliably comparison, a similar amount and DPn of the polymer, i.e. PBMA, was attached to the CNC surface by the two methods. The modified CNCs were incorporated in PCL matrix and the Young’s modulus of PCL increased after sole addition of unmodified CNCs but the strain was remarkably reduced, especially with a high reinforcing content. However, the strain and strength of the nanocomposites based on modified CNCs were much greater than for those based on unmodified CNCs. The covalently-grafted CNCs nanocomposites showed a better performance than the physically- adsorbed CNCs, and the mechanical performance of the latter were considerably reduced at 98 % RH while the performance of covalently- grafted CNCs nanocomposites was unchanged.

The industrial utilization of CNCs is hampered by the required multi-step surface modification and functionalization, but simultaneous acid hydrolysis and Fisher esterification has been reported for the preparation of CNCs bearing butyrate or acetate groups. Inspired by this, the simultaneous preparation and functionalization of CNCs was carried out employing functional organic acids. In this manner, several functionalities have been introduced to the CNC surface, such as double bond, triple bond, thiol and a secondary ATRP initiator. The F-CNCs were utilized for further modification, such as SI-ATRP and immobilization of disperse red dye through ‘click chemistry’. CNCs bearing either acrylate or a thiol functionality were also employed, in other separate publications, for the preparation of thermoresponsive cryogels and DLD-based hydrogels.

54 ————— Future work —————

6 Future work

In this work, new approaches for the surface modification and functionalization of nanocelluloses have been investigated. However it would be valuable to apply these techniques for the design of new value- added products based on nanocelluloses.

SI-ROP of ε-CL from CNF nanopaper template has been applied for the preparation of truly nanostructured composites. These materials could probably be utilized as scaffolds in biomedical applications. Therefore, it would be interesting to study the cell viability, toxicity and degradability of these materials.

Although it may be challenging to attain reproducible results for the attachment of CTA on CNC, it would be interesting to further investigate, in depth, the immobilization mechanism. Furthermore, the polymerization of VAc in the presence of pristine CNC, without prior surface modification, may lead to the preparation of reinforced PVAc with improved performance. The versatility of SI-RAFT/MADIX could be further explored with other interesting monomers, such as 1-vinyl-2- pyrrolidinone, diallyldimethylammonium chloride and N- isopropylacrylamide.

SI-polymerization resulted in significantly better mechanical properties of a CNC-reinforced PCL matrix than the physical adsorption approach. However, it would be interesting to study the relationship between the charge density on the CNC and the amount of physisorbed polymer, and its consequent effect on the properties of the nanocomposite.

Acid hydrolysis of cellulose fibers employing functional organic acids makes possible the simultaneous functionalization and preparation of CNCs. The acidic conditions should be optimized for each acid used, and it would be interesting to investigate the relationship between the viscosity and pKa of the organic acid and the size and DS of the CNC. It would also be valuable to explore the post-functionalization of F-CNCs by means of various chemistries, such as click chemistry, SI-polymerization and crosslinking.

55

56

————— Acknowledgements —————

7 Acknowledgements

Wilhelm Beckers Jubileumsfond and the Swedish Research Council (VR) are acknowledged for financial support.

During the past few years, there were a lot of memorable moments that I shared with so many wonderful people who accompanied me, helped me and supported me throughout these years. For that I will be always grateful and I would like to say: THANK YOU!

First of all, I would like to express my sincere gratitude to my supervisors Assoc. Prof. Anna Carlmark and Prof. Eva Malmström for accepting me as PhD student and giving me the opportunity to work on such interesting projects. Thank you for your excellent scientific guidance, support and encouragement throughout these years. It has been a pleasure and an honor to work with you.

I would also like to greatly thank all my co-authors of the present work including Prof. Lars Berglund, Prof. Mohammed Lahcini, Prof. Mathias Destarac, Assist Prof. Stéphane Mazière, Dr. Houssine Sehaqui, Dr. Linn Carlsson, Dr. Linda Fogelström, Dr. Surinthra Mongkhontreerat, M.Sc. Carmen Cobo Sanchez and M.Sc. Joackim Engström, for excellent and fruitful collaborations. Many thanks to all other people I have worked with during my time as Ph.D. student.

All the seniors at the Fibre and Polymer Technology (FPT) department are thanked for their interesting and inspiring courses and also for creating a comfortable working atmosphere within FPT. Inger, Mia, Inga, Barbara, Vera, Mona, Bosse, Karin and Kjell, thank you for your kind administrative, technical and computer assistance.

All current and former colleagues and friends at Polymer factory and at FPT especially at the division of coating technology are greatly thanked for all the fun, great time and help during these years. Thank you also for your priceless friendship and for the interesting and amusing scientific and personal discussions.

57

————— Acknowledgements —————

All members of P3R group at Toulouse University are thanked for being very friendly and helpful. Thank you for making my stay pleasant! My friends outside KTH, whom are all over the world, are thanked for all the fun and the nice moments we have shared.

My heartfelt thanks are to my parents to whom I owe “everything”. No words can express my deep love and my infinite gratitude. Thank you for encouraging me, believing in me, and for teaching me that no goal can be achieved without commitment and perseverance. I would also like to thank my brother Said and my sister Asma. You have been there for me all the time no matter the distance between us. I love you so much! Karima, thank you for being very nice and always ready to help. You are like a little sister to me! Many thanks to my family-in-law, particularly Saida and her family, for all the nice time we have had.

Last but not least, I would like to thank my best friend and husband Anas for standing by my side during the ups and downs throughout these years. Looking forward to the next chapter of our life together!

58

8 References

1. Siqueira, G.; Bras, J.; Dufresne, A., Polymers 2010, 2, (4), 728. 2. Dufresne, A., Nanocellulose, From Nature to High Performance Tailored Materials. 2012. 3. Rowell, R. M., Handbook of wood chemistry and wood composites. CRC press: 2012. 4. Habibi, Y., Chemical Society Reviews 2014, 43, (5), 1519-1542. 5. Eichhorn, S. J.; Baillie, C. A.; Zafeiropoulos, N.; Mwaikambo, L. Y.; Ansell, M. P.; Dufresne, A.; Entwistle, K. M.; Herrera-Franco, P. J.; Escamilla, G. C.; Groom, L.; Hughes, M.; Hill, C.; Rials, T. G.; Wild, P. M., Journal of Materials Science 2001, 36, (9), 2107-2131. 6. Zimmermann, T.; Pöhler, E.; Geiger, T., Advanced Engineering Materials 2004, 6, (9), 754-761. 7. Michael, T. P.; András, V.; John, D.; Natalia, F.; Bin, M.; Ryan, W.; Arvind, R.; Robert, J. M.; Ronald, S.; Theodore, H. W.; James, B., Measurement Science and Technology 2011, 22, (2), 024005. 8. Habibi, Y.; Lucia, L. A.; Rojas, O. J., Chemical reviews 2010, 110, (6), 3479-3500. 9. Sakurada, I.; Nukushina, Y.; Ito, T., Journal of Polymer Science 1962, 57, (165), 651-660. 10. Turbak, A. F. S., F. W.; Sandberg, K. R., ournal of Applied Polymer Science: Applied Polymer Symposium 1983, 37, 815-827. 11. Klemm, D.; Kramer, F.; Moritz, S.; Lindström, T.; Ankerfors, M.; Gray, D.; Dorris, A., Angewandte Chemie International Edition 2011, 50, (24), 5438-5466. 12. Chauve, G.; Bras, J., Handbook of Green Materials. World Scientific 2014, 233-252. 13. Eriksen, O.; Syverud, K.; Gregersen, O., Nordic Pulp & Paper Research Journal 2008, 23, (3), 299-304. 14. Saito, T.; Nishiyama, Y.; Putaux, J.-L.; Vignon, M.; Isogai, A., Biomacromolecules 2006, 7, (6), 1687-1691. 15. Pei, A.; Butchosa, N.; Berglund, L. A.; Zhou, Q., Soft Matter 2013, 9, (6), 2047-2055. 16. Wågberg, L.; Decher, G.; Norgren, M.; Lindström, T.; Ankerfors, M.; Axnäs, K., Langmuir 2008, 24, (3), 784-795. 17. Henriksson, M.; Henriksson, G.; Berglund, L. A.; Lindström, T., European Polymer Journal 2007, 43, (8), 3434-3441.

59

18. Arlington, USA, June 9, 2011. 2011. 19. Rånby, B. G.; Banderet, A.; Sillén, L. G., Acta Chem. Scand. 1949, 3, 649-650. 20. Rånby, B.; Ribi, E., Experientia 1950, 6, (1), 12-14. 21. Rånby, B. G., Discussions of the Faraday Society 1951, 11, 158- 164. 22. Rånby, B. G., Tappi 1952, 35, (2), 53-58. 23. Roman, M.; Winter, W. T., Biomacromolecules 2004, 5, (5), 1671-1677. 24. Camarero Espinosa, S.; Kuhnt, T.; Foster, E. J.; Weder, C., Biomacromolecules 2013, 14, (4), 1223-1230. 25. Braun, B.; Dorgan, J. R., Biomacromolecules 2008, 10, (2), 334- 341. 26. Xu, X.; Liu, F.; Jiang, L.; Zhu, J. Y.; Haagenson, D.; Wiesenborn, D. P., ACS Applied Materials & Interfaces 2013, 5, (8), 2999-3009. 27. Missoum, K.; Belgacem, M.; Bras, J., Materials 2013, 6, (5), 1745. 28. Malmström, E.; Carlmark, A., Polymer Chemistry 2012, 3, (7), 1702-1713. 29. Carlmark, A.; Larsson, E.; Malmström, E., European Polymer Journal 2012, 48, (10), 1646-1659. 30. Carlmark, A., Macromolecular Chemistry and Physics 2013, 214, (14), 1539-1544. 31. Roy, D.; Semsarilar, M.; Guthrie, J. T.; Perrier, S., Chemical Society Reviews 2009, 38, (7), 2046-2064. 32. Missoum, K.; Martoïa, F.; Belgacem, M. N.; Bras, J., Industrial Crops and Products 2013, 48, 98-105. 33. Klemm, D.; Heublein, B.; Fink, H.-P.; Bohn, A., Angewandte Chemie International Edition 2005, 44, (22), 3358-3393. 34. Lin, N.; Huang, J.; Chang, P. R.; Feng, J.; Yu, J., Carbohydrate Polymers 2011, 83, (4), 1834-1842. 35. Jonoobi, M.; Harun, J.; Mathew, A. P.; Hussein, M. Z. B.; Oksman, K., Cellulose 2010, 17, (2), 299-307. 36. Rodionova, G.; Lenes, M.; Eriksen, Ø.; Gregersen, Ø., Cellulose 2011, 18, (1), 127-134. 37. Lee, K.-Y.; Quero, F.; Blaker, J. J.; Hill, C. A.; Eichhorn, S. J.; Bismarck, A., Cellulose 2011, 18, (3), 595-605.

60

38. Berlioz, S.; Molina-Boisseau, S.; Nishiyama, Y.; Heux, L., Biomacromolecules 2009, 10, (8), 2144-2151. 39. Espino-Pérez, E.; Domenek, S.; Belgacem, N.; Sillard, C.; Bras, J., Biomacromolecules 2014, 15, (12), 4551-4560. 40. de Menezes, A. J.; Siqueira, G.; Curvelo, A. A.; Dufresne, A., Polymer 2009, 50, (19), 4552-4563. 41. Morandi, G.; Heath, L.; Thielemans, W., Langmuir 2009, 25, (14), 8280-8286. 42. Kim, J.; Montero, G.; Habibi, Y.; Hinestroza, J. P.; Genzer, J.; Argyropoulos, D. S.; Rojas, O. J., Polymer Engineering & Science 2009, 49, (10), 2054-2061. 43. Rojas, O. J.; Montero, G. A.; Habibi, Y., Journal of Applied Polymer Science 2009, 113, (2), 927-935. 44. Padalkar, S.; Capadona, J. R.; Rowan, S. J.; Weder, C.; Won, Y.- H.; Stanciu, L. A.; Moon, R. J., Langmuir 2010, 26, (11), 8497-8502. 45. Xhanari, K.; Syverud, K.; Chinga-Carrasco, G.; Paso, K.; Stenius, P., Cellulose 2011, 18, (2), 257-270. 46. Salajkova, M.; Berglund, L. A.; Zhou, Q., Journal of Materials Chemistry 2012, 22, (37), 19798-19805. 47. Habibi, Y.; Dufresne, A., Biomacromolecules 2008, 9, (7), 1974- 1980. 48. Roy, D.; Guthrie, J. T.; Perrier, S., Macromolecules 2005, 38, (25), 10363-10372. 49. Barsbay, M.; Güven, O.; Stenzel, M. H.; Davis, T. P.; Barner- Kowollik, C.; Barner, L., Macromolecules 2007, 40, (20), 7140-7147. 50. Hansson, S.; Antoni, P.; Bergenudd, H.; Malmström, E., Polymer Chemistry 2011, 2, (3), 556-558. 51. Pyun, J.; Jia, S.; Kowalewski, T.; Patterson, G. D.; Matyjaszewski, K., Macromolecules 2003, 36, (14), 5094-5104. 52. Roy, D.; Guthrie, J. T.; Perrier, S., Soft Matter 2008, 4, (1), 145- 155. 53. Carlmark, A.; Malmström, E., Journal of the American Chemical Society 2002, 124, (6), 900-901. 54. Carlmark, A.; Malmström, E. E., Biomacromolecules 2003, 4, (6), 1740-1745. 55. Habibi, Y.; Goffin, A.-L.; Schiltz, N.; Duquesne, E.; Dubois, P.; Dufresne, A., Journal of Materials Chemistry 2008, 18, (41), 5002- 5010.

61

56. Hatton, F. L.; Malmström, E.; Carlmark, A., European Polymer Journal 2015, 65, 325-339. 57. Gärdlund, L.; Wågberg, L.; Gernandt, R., Colloids and Surfaces A: Physicochemical and Engineering Aspects 2003, 218, (1–3), 137-149. 58. Wagberg, L., Nordic Pulp & Paper Research Journal 2000, 15, (5), 586-597. 59. Utsel, S.; Bruce, C.; Pettersson, T.; Fogelström, L.; Carlmark, A.; Malmström, E.; Wågberg, L., ACS Applied Materials & Interfaces 2012, 4, (12), 6796-6807. 60. Wang, M.; Olszewska, A.; Walther, A.; Malho, J.-M.; Schacher, F. H.; Ruokolainen, J.; Ankerfors, M.; Laine, J.; Berglund, L. A.; Österberg, M.; Ikkala, O., Biomacromolecules 2011, 12, (6), 2074-2081. 61. Larsson, E.; Sanchez, C. C.; Porsch, C.; Karabulut, E.; Wågberg, L.; Carlmark, A., European Polymer Journal 2013, 49, (9), 2689-2696. 62. Carothers, W. H.; Dorough, G. L.; Natta, F. J. v., Journal of the American Chemical Society 1932, 54, (2), 761-772. 63. Dubois, P.; Coulembier, O.; Raquez, J.-M.; DegÃ, P., Handbook of ring-opening polymerization. Wiley Online Library: 2009. 64. Kowalski, A.; Libiszowski, J.; Biela, T.; Cypryk, M.; Duda, A.; Penczek, S., Macromolecules 2005, 38, (20), 8170-8176. 65. Clark, L.; Cushion, M. G.; Dyer, H. E.; Schwarz, A. D.; Duchateau, R.; Mountford, P., Chemical Communications 2010, 46, (2), 273-275. 66. Labet, M.; Thielemans, W., Chemical Society Reviews 2009, 38, (12), 3484-3504. 67. Penczek, S.; Cypryk, M.; Duda, A.; Kubisa, P.; Słomkowski, S., Progress in Polymer Science 2007, 32, (2), 247-282. 68. Kamber, N. E.; Jeong, W.; Waymouth, R. M.; Pratt, R. C.; Lohmeijer, B. G. G.; Hedrick, J. L., Chemical reviews 2007, 107, (12), 5813-5840. 69. Heise, A.; Duxbury, C. J.; Palmans, A. R., Handbook of Ring- Opening Polymerization 2009, 59. 70. Penczek, S.; Duda, A.; Kowalski, A.; Libiszowski, J.; Majerska, K.; Biela, T., Macromolecular Symposia 2000, 157, (1), 61-70. 71. Biela, T.; Duda, A.; Penczek, S., Macromolecular Symposia 2002, 183, (1), 1-10. 72. Kricheldorf, H. R.; Kreiser-Saunders, I.; Boettcher, C., Polymer 1995, 36, (6), 1253-1259.

62

73. Storey, R. F.; Sherman, J. W., Macromolecules 2002, 35, (5), 1504-1512. 74. Pärssinen, A.; Kohlmayr, M.; Leskela, M.; Lahcini, M.; Repo, T., Polymer Chemistry 2010, 1, (6), 834-836. 75. Lönnberg, H.; Fogelström, L.; Berglund, L.; Malmström, E.; Hult, A., European Polymer Journal 2008, 44, (9), 2991-2997. 76. Goffin, A.-L.; Raquez, J.-M.; Duquesne, E.; Siqueira, G.; Habibi, Y.; Dufresne, A.; Dubois, P., Biomacromolecules 2011, 12, (7), 2456- 2465. 77. Goffin, A.-L.; Habibi, Y.; Raquez, J.-M.; Dubois, P., ACS Applied Materials & Interfaces 2012, 4, (7), 3364-3371. 78. Lönnberg, H.; Fogelström, L.; Zhou, Q.; Hult, A.; Berglund, L.; Malmström, E., Composites Science and Technology 2011, 71, (1), 9-12. 79. Hawker, C. J.; Frechet, J. M. J.; Grubbs, R. B.; Dao, J., Journal of the American Chemical Society 1995, 117, (43), 10763-10764. 80. Matyjaszewski, K.; Spanswick, J., Materials Today 2005, 8, (3), 26-33. 81. Jenkins, A. D.; Jones, R. G.; Moad, G., Pure and Applied Chemistry 2009, 82, (2). 82. Kharasch, M. S. J., Elwood V.; Urry, W. H., Science 1945, 102, 128. 83. Curran, D. P., Synthesis 1988, (07), 489-513. 84. Wang, J.-S.; Matyjaszewski, K., Journal of the American Chemical Society 1995, 117, (20), 5614-5615. 85. Wang, J.-S.; Matyjaszewski, K., Macromolecules 1995, 28, (23), 7901-7910. 86. Kato, M.; Kamigaito, M.; Sawamoto, M.; Higashimura, T., Macromolecules 1995, 28, (5), 1721-1723. 87. de Cuendias, A.; Le Hellaye, M.; Lecommandoux, S.; Cloutet, E.; Cramail, H., Journal of Materials Chemistry 2005, 15, (32), 3264-3267. 88. Jankova, K.; Bednarek, M.; Hvilsted, S., Journal of Polymer Science Part A: Polymer Chemistry 2005, 43, (17), 3748-3759. 89. Haddleton, D. M.; Jasieczek, C. B.; Hannon, M. J.; Shooter, A. J., Macromolecules 1997, 30, (7), 2190-2193. 90. Jewrajka, S. K.; Mandal, B. M., Macromolecules 2003, 36, (2), 311-317. 91. Teodorescu, M.; Matyjaszewski, K., Macromolecules 1999, 32, (15), 4826-4831.

63

92. Kajiwara, A.; Matyjaszewski, K.; Kamachi, M., Macromolecules 1998, 31, (17), 5695-5701. 93. Jankova, K.; Hvilsted, S., Macromolecules 2003, 36, (5), 1753- 1758. 94. Matyjaszewski, K.; Mu Jo, S.; Paik, H.-j.; Gaynor, S. G., Macromolecules 1997, 30, (20), 6398-6400. 95. Matyjaszewski, K.; Jo, S. M.; Paik, H.-j.; Shipp, D. A., Macromolecules 1999, 32, (20), 6431-6438. 96. Matyjaszewski, K., Macromolecules 2012, 45, (10), 4015-4039. 97. Yi, J.; Xu, Q.; Zhang, X.; Zhang, H., Polymer 2008, 49, (20), 4406-4412. 98. Lacerda, P. S. S.; Barros-Timmons, A. M. M. V.; Freire, C. S. R.; Silvestre, A. J. D.; Neto, C. P., Biomacromolecules 2013, 14, (6), 2063- 2073. 99. Xu, Q.; Yi, J.; Zhang, X.; Zhang, H., European Polymer Journal 2008, 44, (9), 2830-2837. 100. Zoppe, J. O.; Habibi, Y.; Rojas, O. J.; Venditti, R. A.; Johansson, L.-S.; Efimenko, K.; Österberg, M.; Laine, J., Biomacromolecules 2010, 11, (10), 2683-2691. 101. Utsel, S.; Carlmark, A.; Pettersson, T.; Bergström, M.; Malmström, E. E.; Wågberg, L., European Polymer Journal 2012, 48, (7), 1195-1204. 102. Moad, G.; Rizzardo, E.; Thang, S. H., Synthesis of dithioester chain transfer agents and use of bis(thioacyl) disulfides or dithioesters as chain transfer agents. In Google Patents: 1999. 103. Chiefari, J.; Chong, Y. K.; Ercole, F.; Krstina, J.; Jeffery, J.; Le, T. P. T.; Mayadunne, R. T. A.; Meijs, G. F.; Moad, C. L.; Moad, G.; Rizzardo, E.; Thang, S. H., Macromolecules 1998, 31, (16), 5559-5562. 104. Biadatti, T.; Charmot, D.; Corpart, P.; Michelet, D.; Zard, S., Method for block polymer synthesis by controlled radical polymerisation. In Google Patents: 1998. 105. Perrier, S.; Takolpuckdee, P., Journal of Polymer Science Part A: Polymer Chemistry 2005, 43, (22), 5347-5393. 106. Harrisson, S.; Liu, X.; Ollagnier, J.-N.; Coutelier, O.; Marty, J.- D.; Destarac, M., Polymers 2014, 6, (5), 1437. 107. Taton, D.; Destarac, M.; Zard, S. Z., Macromolecular Design by Interchange of Xanthates: Background, Design, Scope and Applications.

64

In Handbook of RAFT Polymerization, Wiley-VCH Verlag GmbH & Co. KGaA: 2008; pp 373-421. 108. Zhu, J.; Zhou, D.; Zhu, X.; Chen, G., Journal of Polymer Science Part A: Polymer Chemistry 2004, 42, (10), 2558-2565. 109. Kanagasabapathy, S.; Sudalai, A.; Benicewicz, B. C., Macromolecular Rapid Communications 2001, 22, (13), 1076-1080. 110. Severac, R.; Lacroix-Desmazes, P.; Boutevin, B., Polymer International 2002, 51, (10), 1117-1122. 111. Monteiro, M. J.; Sjöberg, M.; van der Vlist, J.; Göttgens, C. M., Journal of Polymer Science Part A: Polymer Chemistry 2000, 38, (23), 4206-4217. 112. Monteiro, M. J.; de Barbeyrac, J., Macromolecules 2001, 34, (13), 4416-4423. 113. Beija, M.; Marty, J.-D.; Destarac, M., Progress in Polymer Science 2011, 36, (7), 845-886. 114. Fleet, R.; McLeary, J. B.; Grumel, V.; Weber, W. G.; Matahwa, H.; Sanderson, R. D., European Polymer Journal 2008, 44, (9), 2899-2911. 115. Tastet, D.; Save, M.; Charrier, F.; Charrier, B.; Ledeuil, J.-B.; Dupin, J.-C.; Billon, L., Polymer 2011, 52, (3), 606-616. 116. Sehaqui, H.; Zhou, Q.; Ikkala, O.; Berglund, L. A., Biomacromolecules 2011, 12, (10), 3638-3644. 117. van den Berg, O.; Schroeter, M.; Capadona, J. R.; Weder, C., Journal of Materials Chemistry 2007, 17, (26), 2746-2753. 118. Beck-Candanedo, S.; Roman, M.; Gray, D. G., Biomacromolecules 2005, 6, (2), 1048-1054. 119. Larsson, E.; Pendergraph, S. A.; Kaldeus, T.; Malmström, E.; Carlmark, A., Polymer Chemistry 2015, 6, (10), 1865-1874. 120. Pendergraph, S. A.; Klein, G.; Johansson, M. K. G.; Carlmark, A., RSC Advances 2014, 4, (40), 20737-20743. 121. Funabashi, M.; Kunioka, M., Macromolecular Symposia 2005, 224, (1), 309-322. 122. Chang, P. R.; Zhou, Z.; Xu, P.; Chen, Y.; Zhou, S.; Huang, J., Journal of Applied Polymer Science 2009, 113, (5), 2973-2979. 123. Ebner, C.; Bodner, T.; Stelzer, F.; Wiesbrock, F., Macromolecular Rapid Communications 2011, 32, (3), 254-288. 124. Misra, B. N.; Jassal, J. K.; Pande, C. S., Journal of Polymer Science: Polymer Chemistry Edition 1978, 16, (1), 295-297.

65

125. Vinogradov, G. V.; Dreval, V. E.; Borisenkova, E. K.; Kurbanaliev, M. K.; Zabugina, M. P., Journal of Polymer Science Part B: Polymer Physics 1986, 24, (9), 1971-1981. 126. Rieke, P.; Baer, D.; Fryxell, G.; Engelhard, M.; Porter, M., Journal of Vacuum Science & Technology A 1993, 11, (4), 2292-2297. 127. Hein, C. D.; Liu, X.-M.; Wang, D., Pharmaceutical research 2008, 25, (10), 2216-2230. 128. Nielsen, L. J.; Eyley, S.; Thielemans, W.; Aylott, J. W., Chemical Communications 2010, 46, (47), 8929-8931. 129. Mongkhontreerat, S.; Walter, M. V.; Andrén, O. C. J.; Cai, Y.; Malkoch, M., Advanced Functional Materials 2015, 25, (30), 4837- 4843. 130. Cobo Sánchez, C.; Wåhlander, M.; Taylor, N.; Fogelström, L.; Malmström, E., ACS Applied Materials & Interfaces 2015, 7, (46), 25669-25678.

66