Current international research into cellulose as a functional nanomaterial for advanced applications

This review paper provides a recent overview of current international research that is being conducted into the functional properties of cellulose as a nanomaterial. A particular emphasis is placed on fundamental and applied research that is being undertaken to generate applications, which are now becoming a real prospect given the developments in the field over the last 20 years. A short introduction covers the context of the work, and definitions of the different forms of cellulose nanomaterials (CNMs) that are most widely studied. We also address the terminology used for CNMs, suggesting a standard way to classify these materials. The reviews are separated out into theme areas, namely healthcare, water purification, biocomposites, and energy. Each section contains a short review of the field within the theme and summarizes recent work being undertaken by the groups represented. Topics that are covered include cellulose nanocrystals for directed growth of tissues, bacterial cellulose in healthcare, nanocellulose for drug delivery, nanocellulose for water purification, nanocellulose for thermoplastic composites, nanocellulose for structurally colored materials, transparent wood biocomposites, supercapacitors and batteries.


Introduction
The world is facing a very near and present crisis in terms of climate change and the threat to life. A dramatic reduction in global greenhouse gas emissions is needed, and in doing so fossil fuels require significant replacement. Linked to this is a decarbonization of our materials cycle. Our continued reliance on fossil fuels, and in particular oil, for the production of plastics is simply not sustainable. To this end we need to use sources of materials that are renewable, sustainable and have at least the potential to be placed into a closed-loop recycling system. These criteria are often used as a justification for the use of cellulose, which is derived from a renewable resource, i.e., plants, which in addition sequester carbon dioxide from the Earth's atmosphere for its production and can be potentially returned to the Earth at end-of-life. Nevertheless, this perfect view of cellulose is not yet realized in practice, and we are somewhat far yet in reaching this goal.
Cellulose is a carbohydrate polymeric material, containing carbon (C), hydrogen (H) and oxygen (O). It also belongs to a broader class of natural polymeric materials called polysaccharides, some of which have similar structures to cellulose, but also include other atomic groups like nitrogen (N) (e.g., chitin [1]). We would like to address early on a mistake in the previous review [2] where it stated that the repeat unit of cellulose is cellobiose. It is in fact glucose [3], and cellulose is rather unique among carbohydrates in that it can be both synthesized from, and hydrolyzed to, monosaccharides [4]. In synthesis, glucose monomers are polymerized into long chains, forming anhydroglucose units joined via β-1,4 glycosidic linkages, as shown in Fig. 1.
The crystalline forms of cellulose are numerous. For the purposes of this article, the two sub-allomorphs of cellulose type I-cellulose Iα and Iβ [5]are perhaps the most relevant in that they appear to different degrees in the various types of native cellulose from plants, some animals (tunicates), and bacterial forms. The crystal structures of these allomorphs have been determined with great accuracy, in particular their complex and extensive hydrogen bonding [6,7]. The hydrogen bonding is often quoted as the reason for the high axial stiffness of cellulose, although this is perhaps sometimes overstated, and quite likely hydrophobic interactions between the planes of the pyranose rings play a role too, but most importantly limit solubility [8]. The recalcitrance of cellulose to common solvents has long been an issue for its processing, and it is perhaps for this reason, among many, that the material has been traditionally 'structured' by top-down processing, or bottom-up chemical modification or biosynthesis. Recent attempts to self-assemble cellulose-like polymers from enzymatically generated oligomers offer perhaps new ways of producing nanomaterials [9,10], but these approaches are in their infancy, and not yet suitable for application.
The main forms of cellulose nanomaterials (CNMs) covered in this review are cellulose nanocrystals (CNCs), cellulose nanofibrils (CNFs) and bacterial cellulose (BC); we address the terminology used here in the ''Cellulose nanomaterial terminology'' section. CNCs are typically produced via the acid hydrolysis of plant-based and other forms of cellulose to produce rod-like particles (see Fig. 2a), although many other production routes are possible. The production and properties of CNCs have been previously reviewed, and readers are referred to that publication for more details [13]. CNFs are typically produced by the mechanical fibrillation of plant cellulose, either via processes such as homogenization, grinding, or excessive beating of pulp. This generates fibrillar materials (see Fig. 2b), and the reader is referred to a previous publication on this subject (where the material is termed 'microfibrillated cellulose') [14]. Finally, BC is typically produced by the gram-negative bacterium Glucanobacter xylinum and forms a reticulated network of fibrils, (see Fig. 2c), similar to CNF, but with 'juncture' points between the fibrils. Again, the reader is referred to a previous review of BC, and the other forms of CNFs, for more details [15].
Cellulose is the world's most used material, and it has been exploited for many centuries as wood for fuel and construction, as fibers to produce paper, and textile materials for clothing, ropes, sails and other applications. Our most intimate connection with the material is with wood, and in fact the very word 'material' derives from an old Latin word māteria for 'trunk of a tree', which itself derives from the word māter which means 'mother'; many other languages include similar words such as Moeder (Dutch), Maman (French), Māṁ (ਮ) (Punjabi), Mama (Swahili), Maa (媽) (Cantonese), and Mutter (German). The use of this word provides further connection to 'mother earth' a concept that is contained in many indigenous languages and belief systems. Perhaps this connection ought to be regained since we are seeking to protect our planet collectively through sustainability, and through the use of mother nature's materialcellulose.
A review in this journal published in 2010 [2] quoted the English poet Chaucer. It seems apposite to now recall the words of the Japanese Haiku author Matsuo Bashō (1644 -1694) "butt of the tree see in it the cut end today's moon" highlighting the mysteries of wood, or perhaps the old Kenyan proverb that 'Sticks in a bundle are unbreakable.' This proverb might both suggest collaboration is a good thing and that we have had a long history of understanding the mechanical properties of wood. Certainly, our relationships with wood are ancient, and intimately associated with our languages, as already described. Wood itself is also probably the oldest composite, and it is well known that it possesses a hierarchical structure, which has been well-documented in the literature [16]. The review we present to you contains work that both deconstructs the woody and plant materials into nanomaterials, but also addresses the use and modification of wood itself, making use of its own inherent nanostructure. We therefore present to you a collection of international research on nanocellulose and its application in a variety of fields. It is truly staggering how this material has grown in interest, with publications on 'nanocellulose', and citations to them, growing dramatically over the last decade.
Perhaps it is true to say that many real applications are still yet to materialize. Charreau et al. [17] have, however, recently published data on the numbers of published patents, and their growth since 2010, when the last review in this series was published [2]. This growth in the patent literature is truly dramatic, and nanocellulose is certain to make an impact into many application areas. This review aims to cover research that is a prelude to, and underpins, applications of nanocellulose. Several target areas are covered in the review, namely healthcare, water security, composites, and energy. Recent research in these target areas is discussed, focusing on applications, but also the fundamental research itself that is, and needs to be undertaken to underpin this translation to real products. Each section in this review highlights some work being undertaken by a selected number of Figure 2 Example electron microscope images of cellulose nanomaterials (CNMs); TEM images of a cellulose nanocrystals (CNCs) obtained via acid hydrolysis of microcrystalline cellulose [11] b cellulose nanofibrils (CNFs) [2] and c a scanning electron microscope image of bacterial cellulose (BC) [12]. Figure 2a international groups, but also contexts this with current work in the field.

Cellulose nanomaterial terminology
Since 'nanocellulose' emerged onto the scene as a material there has been a proliferation of the terminology used to describe these materials. We will aim to be consistent in our description of nanocellulose and thereby conform to standards that have been recently laid out in another comprehensive review of techniques to analyze what should collectively be called cellulose nanomaterials (CNMs) [18]. This is the acronym we will use to refer to the different forms of 'nanocellulose'. We will use the terms cellulose nanocrystals (CNCs) and cellulose nanofibrils (CNFs) to refer to the rod-like and fibrillar cellulosic materials. Confusingly CNCs have also been called whiskers, needles and nanocrystalline cellulose (NCC), and we will avoid such terms. Bacterial cellulose (BC) will also be used as a term, referring to the fibrillar material produced by the gram-negative bacteria Glucanobacter xylinum. We will conform to the international organization for standardization (ISO) on the terminology used for CNMs [19] where possible and to otherwise revert to commonly used terms.

Introduction to nanocellulose in healthcare
Nanocellulose has found suitability in various healthcare applications, for example, tissue engineering and drug delivery, as well as diagnostic devices, wound healing, coatings, drug screening and biosensing [20,21]. Some key enablers allowing nanocellulose to be used in healthcare applications are its biocompatibility and relatively low cost, as well as its versatility with respect to both the variety of forms available and its ability to be chemically modified. While purity and uniformity of different nanocellulose types (CNFs, CNCs, BC) are dependent on the starting source and production method, they generally offer a reliable, chemically defined, and robust nanomaterial. Plant-based biomaterials can solve some issues present with animal-derived biomaterials where poorly defined, complex mixtures, and high variability from batch to batch suggest advantages of using exogenous components in biomedical applications.
All types of nanocellulose have been evaluated extensively in tissue engineering and drug delivery applications and this section aims to summarize some of the newest advances and trends in the area. For applications in healthcare, BC stands out for its purity (no hemicelluloses, lignin, etc.) and its ability to be manipulated following biotechnological production pathways; in some product development areas like implants, wound healing and cell cultivation/encapsulation, BC has surpassed CNFs and CNCs as reviewed elsewhere [22]. Additionally, we highlight how different architectures like micelles, spheres, patterned surfaces, and 3D-printed shapes, that are based on nanocellulose building blocks, offer advantages such as large surface area, high porosity and enhanced interactions with drugs and cells. While nanocellulose has been shown to have low toxicity [23], there are key variables that have been reported to affect toxicity including size, morphology, crystallinity, surface chemistry and stability [24]. Despite the broad consensus on the low toxicity of the different types of nanocellulose, further studies are recommended to evaluate this property specific to each targeted application, as well as long-term bioaccumulation in the body.
Other important aspects of nanocellulose for healthcare applications are its in vivo biodegradability and bioactivity [23,25,26]. High crystallinity celluloses possess low biodegradability and, thus, may limit their use in some bio-applications. Significant efforts have been made to improve biodegradability through chemical modification [27][28][29] or through higher-order arrangement where the structure degrades but not the nanocellulose itself [30]. It is also known that pure nanocellulose lacks bioactivity [25]. Consequently, many efforts have focused on chemical and physical surface modification routes to provide CNF, CNC, and BC scaffolds with improved bioactivity [25,26]. Their effect on cell culture parameters has been studied and compared to non-modified nanocellulose forms, demonstrating significant progress.
Owing to the rich chemical, structural, and morphological diversity of nanocellulose, a vast amount of research and commercial translation toward medical applications has been reported. In this section, up-to-date information on the use of nanocellulose for tissue engineering and drug delivery applications is presented. Specifically, the use of CNFs, CNCs and BC as tissue engineering scaffolds for in vitro cell culture and the effect of surface modification to render nanocellulose bioactive are summarized. This includes the introduction of chemical moieties at the surface of nanocellulose to provide negative or positive surface charges, and the use of amino acids, proteins, and growth factors, to enhance cell uptake/adsorption at the scaffold surface and promote cell adhesion, growth, proliferation, and specific cell morphologies. Additionally, the effect of nano-and microscale anisotropy within nanocellulose scaffolds on cell response and aligned growth is discussed. Nanocellulose offers multiscale control in both 2D and 3D environments, creating new opportunities within tissue engineering. As for drug delivery, the role of nanocellulose as a drug carrier, co-stabilizer, or release modulator in various forms including sheets/films, nanoparticles, and micelles is reported. The use of various types of nanocellulose-based drug delivery systems is presented including examples with hydrophilic and hydrophobic drugs as well as various administration routes such as oral, transdermal, local and triggeredrelease.
Franck Quero (University of Chile, Chile): tailoring surface chemistry of nanocellulose scaffolds for in vitro cell culture Cells modify their behavior dependent on the cues that they perceive from their microenvironment [31][32][33]. One strategy to render the surfaces of several types of nanocellulose more biocompatible and bioactive is through surface modification [25,26]. Chemical moieties, biomolecules, bio-oligomers and biomacromolecules can be introduced physically or chemically at the surface of nanocellulose to provide cell signals. These include positive or negative surface charges as well as cell receptors such as the amino acid sequence arginine-glycine-aspartic acid. These signals aim at favoring cell uptake/adsorption at the surface of nanocellulose scaffolds, and their effect on cell adhesion, growth, proliferation, and morphology is typically studied.
Research involving the use of BC and CNCs for tissue engineering applications was reviewed by Dugan et al. in 2013 [34]. The surface modification of BC with bioactive peptide sequences using cellulosebinding domains was discussed as a strategy to enhance its bioactivity. The surface charge of CNCs has also been mentioned as a critical factor to provide their surface with biocompatibility and bioactivity. It is known that mammalian cells possess a net negative charge. As a result, materials with positively charged surfaces could potentially favor cell uptake by electrostatic attraction. On the other hand, materials with negatively charged surfaces would minimize cell uptake due to electrostatic repulsion. The first in vitro study of CNCs with living cells was by Roman et al., where the potential of CNCs as carriers in targeted drug delivery applications was demonstrated [35]. Mahmoud et al. then modified the surface of CNCs with fluorescein isothiocyanate or alternatively rhodamine B isothiocyanate, providing them with negative and positive surface charges, respectively [36]. CNCs modified with positive surface charges were found to be uptaken by human embryonic kidney 293 cells, whereas those modified with negative surface charges were not significantly taken up by both cell types at a physiological pH. The results were explained in terms of cell/material surface electrostatic interactions [36]. Cell uptake mechanisms are, however, more complex, and other scaffold features need to be considered including local nanofiber alignment among others, which are discussed in detail below.
The research group led by Ferraz at Uppsala University explored the effect of nanocellulose surface charge on human dermal fibroblast (HDF) cell culture 2D film scaffolds. In their first work, cationic CNFs and Cladophora nanocellulose were obtained by glycidyltrimethylammonium chloride condensation, whereas anionic CNFs and Cladophora nanocellulose were obtained by carboxymethylation and TEMPOoxidation, respectively [37]. The results revealed that anionic CNF films possessed greater cytocompatibility than non-modified and cationic CNF films. On the other hand, anionic Cladophora films better promoted cell adhesion and viability compared to non-modified and cationic Cladophora films. The improved cell adhesion of HDF onto anionic Cladophora films was attributed to local nanofiber alignment.
In a subsequent study, non-modified, anionic, and cationic CNF films were evaluated as 2D scaffolds to direct monocyte/macrophage (MM) responses in the absence or presence of lipopolysaccharide [38]. The results suggested that MM cultured onto anionic CNF films experienced activation toward a proinflammatory phenotype. Non-modified CNF films, however, promoted a mild activation of THP-1 monocyte cells, whereas cationic CNF films behaved J Mater Sci (2022) 57:5697-5767 as a bioinert material. None of the materials were able to directly activate the MM toward an anti-inflammatory response.
In a third study by the same group, the effect of the negative surface charge density of TEMPO-oxidized Cladophora nanocellulose on the response of HDF and human osteoblastic cells was investigated [39]. From a carboxyl group amount≥260 μmol g −1 , equivalent to a threshold ζ-potential value of -36 mV, TEMPOoxidized Cladophora nanocellulose was found to be cytocompatible, demonstrating that bioinert nanomaterials can be turned bioactive by adjusting the magnitude of their surface charge density.
More recently, three primary works have studied the effect of surface charge on cell culture. Films composed of non-modified, anionic, and cationic CNFs were obtained by an evaporation-induced droplet-casting method [40]. Non-modified and cationic CNFs resulted in 2D surfaces with higher degrees of local nanofiber orientation compared to anionic CNFs. With respect to cell viability and proliferation, anionic and cationic CNF surfaces were found to perform similarly compared to a positive control surface. Although the use of fibronectin coating slightly enhanced cell response for all 2D surfaces, uncoated anionic and cationic CNF surfaces were found to support cell growth. Cationic CNF surfaces, along with the presence of CNF alignment, were found to guide cell growth toward a specific orientation direction [40]. In a subsequent study by Pajorova et al. [41], cellulose mesh 3D scaffolds were coated with either cationic, anionic or a 1:1 mixture of cationic and anionic CNFs. Cell adhesion, proliferation, spreading, and morphology were studied by seeding the 3D scaffolds with either HDF or adiposederived stem cells (ADSC). Anionic CNFs promoted the proliferation of both HDF and ADSC, whereas cationic CNFs enhanced the adhesion of ADSC. The cationic and anionic CNF mixture resulted in promoting combined benefits arising from each CNF type [41]. Lastly, CNFs, CNCs and TEMPO-CNFs with variations in total surface charge were investigated as coatings for cell culture [42]. TEMPO-CNFs with a total surface charge of 1.14 mmol g −1 were found to provide the highest cell viability and adhesion compared to the mechanically isolated CNFs without chemical pre-treatment, and CNCs, from which HDF cells were unable to adhere, leading to low viability [42].
Another strategy to render nanocellulose bioactive is by binding either amino acids, peptides, or proteins onto its surface. A first study covalently bound amino acids to the surface of commercial cellulose filter membranes [43]. Cationic amino acids including lysine and arginine were found to enhance cell adhesion, whereas anionic as well as small amino acids significantly reduced cell adhesion. In subsequent work, the surface of α-cellulose fibrous networks was modified by covalent grafting of hydrophilic, aliphatic and aromatic amino acids onto their surface by esterification [44]. Aromatic amino acids, and in particular tryptophan, resulted in enhanced fibroblast cell spreading. Immobilization of collagen peptides onto the surface of dialdehyde BC was proposed by Wen et al. [45] and found to promote enhanced fibroblast cell adhesion and attachment compared to non-modified BC. Another investigation by Barud et al. [46] used silk fibroin proteins to modify the surface of BC, forming a sponge-like scaffold. This was found to facilitate the attachment and growth of L-929 cells, where proteins acted as cell receptors [46]. The research group led by Franck Quero at the University of Chile have produced protein-functionalized cellulose fibrils from the tunic of Pyura chilensis by a top-down approach [47]. As illustrated in Fig. 3, the CNFs were used to produce films, which were subsequently evaluated as 2D scaffolds to culture mouse skeletal C2C12 myoblast cells [47]. Membranes having 3.1% residual proteins at their surface were found to promote higher cell density and spreading as well as a more orientated shape cell morphology compared to membranes constituted of bleached CNFs. In another work by Zhang et al. [48], the real-time cell adsorption of cells onto non-modified and modified CNF films was monitored by multi-parametric surface plasmon resonance. The presence of either human recombinant laminin-L521 (natural protein of the extracellular matrix) or poly-Llysine at the surface of the CNF films resulted in enhanced attachment of human hepatocellular carcinoma cells compared to non-modified CNF films.
Another method to render the surface of CNFs bioactive has been to modify their surface by polyion complex formation between negatively charged TEMPO-oxidized CNFs and positively charged basic fibroblast growth factors (bFGFS) [49]. In this way, Liu et al. mimicked the interactions that naturally exist between bFGFS and heparin sulfate in the extracellular matrix. The scaffold was found to release controlled amount of bFGFS, which was regulated by both CNF surface chemistry and enzymatic deconstruction of the scaffolds. This resulted in significantly enhanced fibroblast cell proliferation [49].
Instead of tailoring the surface chemistry of nanocellulose, other features have been reported to control cell response including multiscale topographical and multicomponent approaches.
These aspects and their specifics are presented in the next subsection and could potentially be translated toward new commercial products in the near future.
Marcus Johns and Emily Cranston (University of British Columbia, Canada): nanocellulose for directional multiscale tissue engineering Three-dimensional biophysical and biochemical interactions between cells and scaffolds modulate the cell response in terms of proliferation, migration, differentiation, deposition of extracellular matrix proteins, and-ultimately-cell survival. These interactions are dependent on cell surface receptors and, when cell attachment occurs, are regulated by integrin pairs that have a defined nanometric spacing between them [51]. The control of cell alignment via topographical features has been known since the late 1980s/early 1990s [52][53][54]. Microscale features with step changes[10 µm and spacings[2 µm inhibit cell migration and spreading [55], whilst nanoscale features with dimensions\70 nm and spacings between 70 and 300 nm disrupt focal adhesions [56]. Thus, modification of the tissue engineering scaffold over multiple length scales is significant in defining the response of the cells.
Since all nanocellulose types are anisotropic, being fiber or filament-like with diameters at the nanoscale (e.g., BC and CNF), or spindle-like nanoparticles with all dimensions at the nanoscale (e.g., CNCs), the design of tissue scaffolds with, at least, short-range directionality and nano/micro features becomes possible. Furthermore, long-range directionality and mechanical anisotropy to direct cell response can be achieved via topographical patterning of the scaffold (e.g., soft lithography, 3D printing, templating, electrospinning) or through directed assembly to induce nanoparticle alignment.
In 2010, Dugan et al. [93] were the first to report that the nanoscale structure of CNCs could align myoblast muscle cells (Fig. 4a&b), despite being orders of magnitude smaller than the cells themselves. Straightforward 2D surfaces with spin-coated (radially oriented) CNCs and no other components were sufficient to demonstrate this effect [93]. Similarly, the anisotropic deposition of CNCs onto titanium surfaces guided fibroblast proliferation, Figure 4 a Myotubes stained for myosin one week after induction of differentiation on aligned CNC surface. Arrow: approx. degree of CNC alignment, scale bar: 250 µm; b histogram of myoblast orientation relative to approx. radial axis of CNCs; c representative confocal image of human dermal fibroblast adhering to BC substrate with grating microtopography. Orientation of grating reported in top left corner; d histogram reporting the cell alignment distribution relative to grating. Error bar:±S.E.; e myoblast cell adhesion on uniaxially microscale wrinkled POEGMA-CNC hydrogel sheets with nanoscale parallel-orientated fibers. Scale bar: 50 µm; f histogram of cell orientation angle relative to wrinkles. a, b Reprinted (adapted) with permission from [93] (Copyright American Chemical Society, 2010). c, d Reprinted (adapted) with permission from [97] (Copyright American Chemical Society, 2015). e, f Reprinted (adapted) with permission from [109] (Copyright Elsevier, 2021). S.E.=Standard Error.
offering new opportunities for integrating implants [94]. Although these features appear too small to enable cell attachment according to earlier literature, it has been recently demonstrated that integrin clusters can form across nanofeatures so long as the feature spacing is\110 nm [95]. It is reasonable to expect that these integrin clusters arrange themselves in line with the features, enabling cell alignment. This work in simple 2D nanocellulose films set the stage for more complex 2D scaffolds and higher dimensionality structures with optimized physicochemical properties and a variety of topographies.
Soft lithographical techniques have been utilized in tissue engineering for almost 20 years [96], however, the first reported use with nanocellulose was demonstrated on BC by Bottan et al. [97] in 2015 (Fig. 4c&d). Since then, the patterning of BC is typically achieved using polydimethylsiloxane with designed structural features[1 µm [97][98][99][100]. Spacings of 10 µm between features have been used to direct growth of fibroblasts [97,99], neuronal cells [98,100] and muscle cells [100], whilst having a more limited impact on keratinocytes [97,99]. Importantly, structural control of fibroblast growth via BC scaffolds can influence scar formation in vivo [101]. Jin et al. demonstrated that structured pure BC scaffolds with 10 µm stripes reduced fibroblast proliferation, which limited collagen accumulation. The reduction in collagen, in turn, reduced scar contraction and limited hypertrophic scar formation [101]. Likewise, Boni et al. [99] produced similarly structured BC scaffolds impregnated with silk sericin. The structural features enabled alignment of fibroblasts whilst the silk sericin enhanced fibroblast and keratinocyte proliferation in vitro. However, the structural features limited collagen deposition compared to structurally unmodified BC, which they suggested enhanced the potential to limit fibrosis and scar formation [99]. Thus, soft lithographical modification of BC offers opportunities to produce inexpensive scaffolds and wound dressings with improved healing properties.
Despite previous publications describing increased cell alignment with decreased feature spacing of BCbased materials [97], reports of directly controlling BC 'ribbon' alignment are limited. Wang et al. [102] have recently demonstrated that alignment could be achieved via wet-drawn stretching of the BC film. The gelatin impregnated, aligned films exhibited enhanced mechanical properties and significantly improved fibroblast alignment in vitro, with further enhancement achieved via electric field stimulation [102].
Alternatively, one may produce aligned fibers via electrospinning. He et al. produced aligned regenerated cellulose fibers (with ca. 200 nm diameters) loaded with CNCs and spun from lithium chloride/ dimethyl acetamide solution [103]. This composition and processing led to improved tensile properties of the fibers and enabled aligned growth of dental follicle cells [103]. The same scaffolds impregnated with bone morphogenic protein-2 promoted osteogenic differentiation of mesenchymal stem cells in vitro, while anisotropic fiber orientation promoted cell alignment, no significant differences in biomarkers (alkaline phosphatase activity, calcium content) were observed between aligned and unaligned fibers [104]. However, in vivo, the aligned fibers enabled aligned collagen deposition and new cortical bone growth on the external face of the implant. These responses were not observed in the unaligned scaffolds, nor in a similar study using aligned poly(L-lactic acid) nanofibers. As such, the enhanced response in the CNC loaded sample was attributed to the improved mechanical properties observed by the incorporation (and alignment) of CNCs in the scaffold [104].
The improvement of mechanical properties and development of nanoscale anisotropy in multicomponent electrospun and 3D-printed scaffolds via CNC/CNF inclusion has only recently begun to be explored [105][106][107][108][109][110][111]. For example, De France et al. reported on the facile production of '2.5D' poly(oligoethylene glycol methacrylate) (POEGMA)/CNC scaffolds that enabled microscale control of 'wrinkled' features, similar to those produced using soft lithography, via controlled thermal shrinkage while independently controlling the alignment of the nanoscale features via electrospinning [109]. Uniaxially (micro) wrinkled features with parallel-orientated (nano) fibers resulted in aligned myoblast growth (Fig. 4e&f), whereby CNC content was used to control the compressive modulus and protein uptake [109]. Similarly, Huang et al. [110] demonstrated that they could align oxidized BC along the direction of extrusion in simple printed structures, which influenced the orientation of lung epithelial stem cells. While research into the nanoscale alignment of nanocellulose in 3D-printed architectures through control of shear is still in the initial stages [112][113][114], it offers opportunities for the development of scaffolds that exhibit photoresponsive mechanical J Mater Sci (2022) 57:5697-5767 properties [115], or change shape depending on the degree of hydration [116].
Nanocellulosic 3D structures with directed cell growth due to anisotropic microscale topographies have been achieved via directional freezing to produce aligned pores within cryogels [117,118] (as opposed to the anisotropy resulting from aligned nanocellulose itself). The effect of pore morphology on cell response is less well defined than that of 2D ridges/grooves. For example, Karageorgiou and Kaplan reported that the optimal pore size to promote osteogenesis was [300 µm [119], while the optimal pore diameter for neuronal cells has often been reported to be \100 µm [120][121][122]. However, strategies to control pore size and morphology in nanocellulosic cryogels are known, including regulation of the gel composition [123-125]; freeze-casting temperature and/or rate [123,126]; sol pH [127]; and nanocellulose morphology [126]. Furthermore, Tetik et al. [128] have recently incorporated directional freezing with 3D printing, which enabled the production of controlled 3D geometries with aligned micropores.
Unfortunately, there are limited comparative studies investigating cell responses to different nanocellulose morphologies. Kummala et al. [42] have recently examined the response of dermal fibroblasts on different nanocellulose types (i.e., CNCs, a predominantly microfibrillated CNF, and two predominantly nanofibrillated CNFs). While true 'nano'-CNFs supported cell attachment and proliferation, limited cell response was observed on the microfibrillated CNF and CNC surfaces. However, as the authors note, one must consider all the variables involved-including topographical features, surface chemical group type and degree of modification, fibril dimensions, tensile properties, and cell type-to be able to draw conclusions as to the cell response. This requires modeling to isolate cell response to specific parameters, such as the regression modeling performed by Johns et al. [148], and must also take the growth media into consideration as selective molecular adsorption will mediate cell-surface interactions [41,118].
We conclude that topographically anisotropic nanocellulose scaffolds are promising biobased biomaterials for enabling cell alignment in both 2D and 3D, impacting-for example-wound healing in vitro and stem cell differentiation in vivo. Control of the topographic features at both the microscale and nanoscale through patterning and nanocellulose alignment offers new opportunities in regulating cellular response. Combining these structures with stimuli-dependent chemistry may present future opportunities in dynamic/bio-responsive tissue engineering.
Mudrika Khandelwal (IIT Hyderabad, India) and Anu Sebastian (CIPET: IPT-Kochi, India): nanocellulose-based drug delivery systems Cellulose has a long history of application in pharmaceutics owing to ease of availability and a good compaction property, primarily as an excipient in oral formulations. Nanocellulose has been shown to play a variety of roles and offer several advantages in drug delivery applications, for example, for its release modulation, as a drug carrier, its improved mechanical properties, better compaction, and appropriate rheological modification [20,23,[149][150][151]. It has been shown that the addition of nanocellulose can control the release of incorporated drugs to significantly reduce consumption [152][153][154][155][156]. Nanofibers offer an additional advantage in terms of mechanical support and improvement of shelf life by improving oxidative stability [29]. Novel drug delivery systems such as triggered and targeted forms have emerged using different types of nanocellulose [157,158]. Most interestingly, nanocellulose allows easy incorporation of multiscale therapeutic agents such as nanoparticles, drug molecules, supramolecular organization and as a template for the production of other drug carriers [159][160][161][162].
In this subsection, the most up-to-date and relevant findings in the literature are summarized in Table 1, describing the nanocellulose type and source, composition or formulation, mode of drug delivery, and key outcomes. The ability of nanocellulose to act as a drug delivery modulator may be attributed to various reasons: aggregation of nanocellulose, interactions between drug molecules and cellulose hydroxyl groups, as well as cellulose's ability to modulate the microstructure and morphology of composite materials. It has been shown that nanocellulose can be successfully used for both water-soluble as well as poorly water-soluble drugs [24,151,163]. Nanocellulose can be modified, and in the case of BC, to hasten as well as delay the release of a drug and offer a combination of release profiles [164][165][166][167]. Nanocellulose has been utilized for various routes of drug delivery-oral, transdermal, local. There are a few recent reviews on nanocellulose in drug delivery [24,151,163,168]; however, an important emerging concept of triggered or actuated drug delivery remains less discussed [24,151,163,168].

Suitability of nanocellulose for drug delivery
Given the wide variety of nanocellulose types, it can be used in various ways in different drug delivery systems. For example, nanocellulose can be the carrier for the drug or act as a delivery modulating agent. Furthermore, it is important to recognize that nanocellulose can be obtained or processed in the form of nanoparticles, microparticles, tablets, aerogels, hydrogels, and membranes, enabling varied modes of drug delivery [169]. The large surface area and high density of surface hydroxyl groups on nanocellulose make it conducive for hydrophilic drug delivery specifically. For hydrophobic drug delivery, functionalization or the production of composites or hybrid materials has been shown to be useful [152,170,171].
Despite many advantages, nanocellulose also has some limitations-namely, moisture sensitivity and low thermal stability [172]. Significant progress has been made to improve these properties through pre-treatments and surface modifications. Aggregation can be a challenge and may be overcome by deploying electrostatic effects and steric stabilization mechanisms [163]. To summarize, surface modification pathways have emerged as an important step to optimizing nanocellulose in drug delivery to 1) carry the drug; 2) make it suitable for delivering hydrophobic drugs; 3) prevent aggregation; 4) improve processability (i.e., by enhancing thermal stability); and 5) improve shelf life by decreasing moisture sensitivity [162].

Nanocellulose for various routes of drug delivery
Delivery systems must be developed to administer drugs using the most suitable route. Different types of nanocellulose have been used to deliver various classes of drugs including anticancer, anti-inflammatory, analgesics and antibiotics following oral, transdermal, implantable, and local delivery routes, as summarized in Table 1. While oral drug delivery is the most common mode of drug administration, it suffers from challenges such as the need for high and frequent doses, and side effects, which need to be addressed. Nanocellulose has been used for the delivery of, for example, indomethacin, salbutamol sulfate, diclofenac, ampicillin, and ranitidine. It has also been demonstrated that the addition of nanocellulose to these formulations offers several benefits such as an increase in the dissolution rate and oral bioavailability, high drug entrapment efficacy, enabling sustained and controlled drug release, prolonged drug release in fasted state-simulated stomach fluid, and good mechanical and viscoelastic properties [154,[173][174][175][176].
Nanocellulose has shown great potential in transdermal as well as topical drug delivery systems where the drugs are administered through the skin to achieve therapeutic concentrations. Thus, it allows the drug to by-pass the gastrointestinal tract and liver metabolism and enables an effect at lower doses. Nanocellulose has enabled release modulation, high drug entrapment, good permeation rates with drugs such as berberine compounds, diclofenac, providoneiodine, hydroquinone, ceftriaxone, and crocin [177][178][179][180][181]. It is important to note that transdermal delivery works better for small molecule drugs.
Another popular approach is implantable systems or local drug delivery systems that release the drug at or near the target site, increasing the effectiveness Oral delivery High drug loading is achieved at pH 3 due to electrostatic interactions Surface grafting of PEI with CNF results in sustained drug release pH and temperature-dependent drug release [190] Plant CNC CNC, starch, vitamin B12 Oral delivery CNC gives a retardant effect when combined with starch Drug release rate is approximately 2.9 times slower than starch microparticles [191]  The film incorporated with 4% GNP-NC exhibits improved thermo-mechanical properties, water vapor permeability, drug encapsulation efficiency, and transparency [193] BC BC, salbutamol sulfate (SS) Oral delivery BC capsule shells as an alternative to gelatin-based shells allows immediate release Adding release retardant polymer to the core of the capsule sustains the drug delivery [194] BC BC, poloxamer, octenidine Wound dressing Drug delivery was sustained over up to 8 days Addition of poloxamers-induced octenidine loaded micelle formation leading to a biphasic and controlled release profile [195] BC and Plant CNF BC, CNF, high amylose starch (RS), pectin (P) Methotrexate (MTX), Colonic delivery A better controlled release of MTX is observed from RS/P-NFC film due to its lower porosity [196] BC Poly(N-methacryloyl glycine), BC, diclofenac sodium salt (DCF)

Oral and transdermal delivery
Controlled and pH-responsive drug delivery is observed where drug delivery is fast in the intestinal pH [197] Plant CNC CNC, oxidized CNC (OCNC), chitosan nanoparticles (CHNP), repaglinide (RPG)

Oral delivery
Drug release is dependent on medium pH where enhanced RPG release is observed at pH 1-2 compared to that at pH 6-8 Increasing the amount of CNC or OCNC results in slower release of drug with more pronounced effect in case of OCNC [176] Plant CNF Hydroxypropyl methylcellulose (HPMC), CNF, ketorolac tromethamine Transdermal delivery HPMC, which is reported to have a faster drug release on contact with biological fluids, attains a slow and steady release on adding CNF Drug release is decreased with the increase of CNF concentration in the composites [198] Plant CNC Alginate, CNC, ampicillin Oral delivery Presence of CNC enhances the release profile of Ampicillin from alginate due to the free space [175] BC Nanostructured lipid carriers (NLCs-NH), doxorubicin Intratumoral delivery BC-NLCs-NH films revealed sustained drug release, high drug loading Significant decrease in the tumor-to-control ratio of tumor volume ex vivo, with no side effects [199]  Wound dressing Composite shows thermal stability excellent mechanical strength Excellent antimicrobial activity against broad spectrum bacteria making them promising material in chronic wound dressing [206] and lowering the required dose. Nanocellulose has been particularly beneficial with local delivery of diclofenac, doxorubicin, and methotrexate [165,182]. Another local delivery route explored with nanocellulose is ocular drug delivery, which takes advantage of nanocellulose gel properties [153]. Nanocellulose has also been explored to produce stimuli-responsive materials for smart/triggered drug delivery systems. These materials are sensitive to specific stimuli which can be a change in humidity, pH, and light, or the application of an electric or a magnetic field. For example, multi pH/near-infrared responsive polydopamine/CNF composites with calcium ions as crosslinkers have been developed for drug delivery and wound healing applications [183]. The drug tetracycline hydrochloride could be released in a controlled fashion on exposure to Near Infrared (NIR) radiation or lower pH conditions. Such a system also offers a synergetic effect on wound healing and is advantageous as it is easy to fabricate while providing multi-response and sustained release of the drug. CNF-sodium alginatebased gel macrospheres were developed for intestinal targeted delivery of probiotics, thereby protecting them from the acidic conditions of the stomach [184]. A novel dual stimuli responsive drug delivery system of aminated nanodextran and carboxylated nanocellulose deposited on the surface of modified graphene oxide was prepared by layer-by-layer assembly [158]. It showed that curcumin can destroy HCT116 cells upon exposure to NIR radiation. In this case, the drug was loaded into the nanocomposite based on hydrogen bonding or π-π stacking and was released faster in an acidic environment than at an intestinal pH [158].
A dual responsive hydrogel based on BC and sodium alginate which reacts to changes in pH and electric stimuli has been developed [157]. The model drug, ibuprofen, showed an enhanced release in a neutral or alkaline medium and in the presence of  [208] an electric stimulus [157]. Another BC-based drug delivery system was fabricated by chemical oxidative polymerization of BC and polyaniline [185]. This resulted in a pH-electro sensitive hydrogel that showed a slower release of the drug berberine hydrochloride in acidic conditions thereby protecting the drug before it is released into the small intestine for drug absorption, and it also showed accelerated release on application of an electric potential [185].

Commercialization of healthcare products incorporating nanocellulose
Several companies have been commercializing medical grade products based on BC (e.g., Membracel® by Vuelopharma, Epi Nouvelle ? ® naturelle by JeNaCell GmbH) and CNFs derived from tunicates (e. g., Ocean TuniCell®) and wood pulp (e.g., UPM Biomedicals). Most of these products are based on non-surface-modified nanocellulose but UPM Biomedicals now sells a cell culture media product named GrowDex®-A that consists of surface-modified CNFs with proteins or peptides. UPM Biomedicals also produces FibDex®, an advanced CNF wound care dressing that has proven to provide efficient wound healing in skin graft donor sites [209]. CELLINK have also developed a series of bioinks based on CNFs and alginate for the 3D printing of tissue engineering cell scaffolds. As such, commercial opportunities exist to move forward with the use of surface-modified nanocellulose toward new commercial products. Surrounding wound dressings with transdermal drug delivery, application, product development and commercialization of nanocellulose continues to advance significantly. Some commercial products include Biofill®, Bioprocess®, Suprasorb X?PHMB® and Xcell®. Nanocellulose-based drug delivery products such as Gengiflex® membranes for dental implants exist; however, there is a need to push clinical trials and commercialization of nanocellulose further [23,149]. For widespread acceptance of nanocellulose in drug delivery, a better understanding of the influence and regulation of drug release, interactions between drug molecules and nanocellulose, as well as possible reduction or destruction of drug activity and structure is required. In addition, toxicity needs to be further assessed, likely on a caseby-case basis [210].

Introduction to nanocellulose and water purification
Although water is a basic need for all 7.9 billion of us, access to clean portable water remains a significant challenge globally. The UN estimates that over one million people in developing countries do not have access to clean drinking water and that up to 159 million people around the world consume untreated water from surface water sources [211]. Yet this water is often contaminated by toxic levels of heavy metals, dyes, and hydrocarbons.
This section presents advances in the application of cellulose in water treatment. We classify the use of cellulose for water treatment applications into the following three categories: (i) as the active agent i.e., adsorbent, (ii) as a support for adsorbents or catalysts, and (iii) for the enhancement of photocatalyst performance through reduction of the band gap energy.
The use of cellulose as an adsorbent support takes advantage of the physical properties of cellulose, in particular, its porosity [219]. Pores within and between fibers can act as nucleation and deposition sites for adsorbents, reducing their attrition/ loss, thus increasing the performance and lifespan. Iron oxides, for instance, are excellent adsorbents for arsenic, whose contamination of groundwater poses serious health effects to over 40 million people worldwide [220]. However, recovery from treated water poses a significant challenge to the use of iron oxides for this purpose. Recent work has shown that iron-oxide-cellulose composites can overcome this challenge, increasing both the performance and lifespan of the adsorbent [221]. A similar rationale has been employed in the design of catalysts for advanced oxidative processes, to allow for reuse/recycling of catalysts [222]. Further, it has also been suggested that cellulose increases catalyst performance by providing settling sites for radicals and pollutants thus increasing their interaction, and the subsequent degradation of the latter [223], in addition to preventing catalyst aggregation of the catalyst, increasing access to active sites on the catalyst surface [224].
A third way by which cellulose has been applied in water treatment has been for the enhancement of the performance of photocatalysts. In the presence of graphitic structures (e.g., cellulose char), electronhole recombination after exposure of the catalyst to irradiation is hindered, as electrons are instead quickly captured by the cellulose char [225]. Holes created in the conduction band (CB) because of this electron capture result in a higher probability of the formation of •OH radicals, while electrons in the biochar combine with oxygen in solution to form superoxide radicals (•O 2¯) . Cellulose therefore acts as a sustainable and renewable graphitic source, improving catalyst performance. Indeed, Fu et al. [226] found that for the oxidative degradation of orange II by peroxymonosulfate, the graphitization degree had a greater influence than surface area and pore volume of a MnFe 2 O 4 -based catalyst.

Anita Etale (University of Bristol, UK): nanocellulose for water purification
Using locally available agricultural waste biomass including hemp, and corn stover, various approaches for the preparation of heavy metal adsorbents have been explored. In one study, cellulose-supported iron oxides were prepared and applied for the removal of As(III), As(V) and Cr(VI) ions. Notwithstanding that arsenic is considered carcinogenic even at very low concentrations, over 200 million people in 40 countries around the world are exposed to drinking water with As levels above World Health Organisation guidelines (10 µg L −1 ) [227]. Although As(V) predominates (as H 2 AsO 4 − and HAsO 4 2− ) in oxidizing conditions, in regions of previous gold mining activity which are often characterized by reducing conditions and high sulfate concentrations, arsenic exists predominantly in the more toxic As(III), as H 3 AsO 3 0 and H 2 AsO 3 − . Nevertheless, because of slow redox transformations, both forms often occur in either redox environment [228]. Anita Etale's group has explored two strategies for the use of cellulose in water treatment applications. The first involves using cellulose as a support material for iron oxides which are excellent adsorbents for arsenic [229]. Preliminary work, however, showed them to be prone to dissolution so that the prepared iron oxides 'leached' into treated water compromising both the treatment process and adsorbent lifespan. To address this challenge, iron oxides were deposited on CNF extracted from hemp fibers. Importantly, the CNF are thought to be porous [219]. CNF porosity may be the result of dissolution of lignin and hemicellulose between the lamellae by the chemical pulping process, or mechanical treatments e.g., blending and sonication applied to increase fibril separation. CNF with average pore sizes of 6 nm have recently been reported [219]. Manninen et al. [230] also reported a cumulative pore volume of 1.7 mL g −1 from pore sizes that ranged from\1 -3 nm in kraft pulp fibers, and from 3 nm to an undefined maximum between the CNF fibers.
The approach to adsorbent synthesis has, therefore, involved exploiting these pore spaces as embedment sites for contaminant adsorbents. In one study, a cellulose-ferrihydrite composite (Fig. 5a) was synthesized by the deposition of iron oxide onto TEMPO-oxidized fibrils at pH 10.5. Iron oxides, including poorly crystalline ones such as ferrihydrite, have a strong affinity for both As(III) and As(V). Adsorption occurred primarily by ligand exchange with surface OH 2 and / or OH − resulting in bidentate binuclear inner-sphere complexes [229,231]. This adsorbent displayed efficiencies of[99% for the removal of As in mine drainage contaminated water (Fig. 5b). Further, column experiments showed that 1 g of the adsorbent was needed to treat 1L of contaminated water. The reduction in absorbance intensity from the OH region (1000 cm −1 ) on FTIR spectra (Fig. 5c) suggests that adsorption of both arsenic species involved some loss of OH as suggested by Jain et al. [229].
A similar approach was employed in Sillanpää's group for the removal of Ni, Cd, Cu, as well as PO Exposure of the same adsorbent to Cr(VI) contaminated water also revealed similarly high efficiencies between pH 5 and 7: 94% of Cr(VI) ions were removed from solution in the first 5 min [233]. In a separate study, succinic anhydride was deposited onto CNF, and the composite investigated for adsorption of a range of metals (Fig. 5d)  A second approach has been through surface modification of cellulose e.g., via amination [216], thiolation [234], and cationization. This latter approach uses two quaternary ammonium salts: 3-chloro-2-hydroxypropyl trimethyl ammonium chloride (CHPTAC) and glycidyltrimethylammonium chloride (GTMAC), and has been explored ( Fig. 5e) and the resulting materials examined for Cr (VI) removal (Fig. 5f), and antibacterial activity. The results showed that at pH 4, 0.1 g of CHPTACmodified cellulose removed up to 47% of Cr (VI) ions, while 72% was adsorbed by GTMAC-cationized cellulose. GTMAC-cationized also displayed considerable antibacterial effects, reducing the viability of Escherichia coli by up to 45% after just 3 h of exposure. However, Cr(VI) uptake in contaminated water (pH 2.7) was diminished to 22% likely due to competition from sulfate and selenate ions [235] which are abundant in mine drainage. Nevertheless, together, b Removal efficiency of the CNF-Fe adsorbent when exposed to mine-drainage contaminated water c FTIR spectra of CNF-Fe before adsorption, and after uptake of As(III) and As(V). The box highlights changes in the surface -OH absorption region before and after adsorption. d Scanning electron micrograph of succinic anhydride-modified CNF [215]. e CP-MAS 13 C NMR spectra of unmodified and cationized cellulose with the additional peak at 52 ppm from (CH 3 ) 3 N ? -groups of the quaternary amines highlighted fr(VI) removal efficiency of cationized cellulose. (d) is reproduced from [215] with permission from Elsevier (Copyright Elsevier, 2013); e and f are reproduced under the terms of a CC-BY license from [218]. a-c are unpublished data.
these results suggest that cationized cellulose can be applied in the treatment of Cr(VI)-contaminated mine water particularly if pre-treatments to reduce concentrations of other anions are applied.
Supree Pinitsoontorn (Khon Kaen University, Thailand): carbonized bacterial cellulose aerogel as an efficient sorbent for oil-polluted water Bacterial cellulose (BC) is a class of nanocellulose with a unique structure and properties, such as a three-dimensional network of intertwined cellulose nanofibers, remarkable mechanical properties, high porosity, and low density [236,237]. In its original state, BC is a hydrogel with a significant water uptake volume ( 99%) [237]. Freeze-drying removes the water content but still preserves the BC 3D nanofibrous network and transforms a hydrogel into an aerogel. Moreover, pyrolysis of the BC aerogel in an inert gas atmosphere leads to the carbonization of the BC nanofibers, thereby forming a carbon nanofiber aerogel. The carbonized BC (c-BC) aerogel inherits the BC precursor's merits, so it still preserves the 3D continuous architecture, the interconnected nanofibrous network, and an extremely high porosity. The surface area of the carbon nanofibers can reach[400 m 2 g −1 , with a sizeable porous volume (ca. 3.00 cm 3g −1 ) and an ultra-light weight (4-6 mg cm −3 ) [238]. Furthermore, high-temperature pyrolysis can induce the hydrophobic/oleophilic properties of the c-BC aerogel, which makes it an ideal material for oil sorption.
Oil pollutants are one of the leading global environmental problems. Every year, the annual spillage of petroleum compounds to the marine environment is over 1.4 million tonnes which has caused catastrophic effects on ecological systems [239,240]. Therefore, a remedy for oil-spills is urgently needed. Although there are several approaches for treating oil-spills, the most effective approach is via physical adsorption, i.e., the use of oil sorbents, which has been proven to be energy-efficient, highly selective, environmentally friendly, fast, and recyclable [241,242]. Research over the last decade has reported several types of oil sorbent materials. Carbon aerogels have received considerable attention as one of the most effective materials for adsorption, separation, and recovery of spilled oil [243,244]. The carbon aerogels can be fabricated from various biomass-based products, such as cotton, bamboo, winter melon, or waste paper [241]. These aerogels have been applied for remedying oil-spills and have successfully demonstrated a large oil sorption capacity of up to 100 times of their weight. However, the preparation processes for such aerogels may involve severe mechanical and chemical pre-treatments, which are high in energy consumption, and subject to environmental concerns. Moreover, high-temperature pyrolysis may cause frangibility and brittleness to those biomass-based carbon aerogels [245], which hinders their practical use for oil recovery. On the other hand, carbon aerogels from carbonized BC (c-BC) have advantages over other biomass-derived materials. The fabrication of c-BC is simple and cheap without any use of harsh chemicals or complicated processes. Also, the natural 3D continuous nanofibrous architectures of BC make the c-BC material mechanically robust and flexible. The c-BC aerogel is typically obtained in a bulky macroscopic form, which is desirable for easy handling and recycling after oil sorption. Plus, the shapes and sizes of the c-BC aerogel are controllable via the bacteria cultivation process and scalable for industrial production.
The application of the c-BC aerogel for oil sorption was firstly reported by pioneering work done by Wu et al. [246]. They showed that pyrolyzing the BC aerogel under an argon atmosphere at 700-1300°C led to a c-BC aerogel with a density of only 4-6 mg cm −3 and a high porosity up to ca. 99.7%. The c-BC aerogel exhibits hydrophobicity and can adsorb a wide range of oils and organic solvents with excellent recyclability by direct combustion. The sorption capacity reached up to 310 times the weight of the c-BC aerogel. Moreover, it was highly flexible and mechanically robust. It could also be compressed to a more than 90% volume reduction and almost recover to its original shape after release, making a 'squeezing' process to recover oil possible [246]. In a recent study, a c-BC aerogel pyrolyzed at 1200°C was compressed to 99.5% strain, and it was able to be restored elastically to almost its original shape after release [247]. A detailed surface area and pore-size study showed that micropores and mesopores (2-100 nm) occupied a large portion of the pore-size distribution, providing huge spaces for oil sorption and leading to high oil sorption capacity. Furthermore, this c-BC aerogel was an excellent thermal insulator with extremely low thermal conductivity (0.025 W m −1 K −1 ) comparable to air [247].
Several efforts have been reported to increase the oil sorption capacity of c-BC aerogels by using a composite approach. c-BC was combined with carbon nanofibers derived from polyimide (PI) by freezing the mixture of BC and PI precursor suspension before imidization and pyrolysis [248]. The resultant carbon aerogel consisted of a 3D carbon skeleton with a cellular architecture from carbonized PI, decorated with 1D c-BC nanofibers. This hierarchical structure was advantageous for enhancing the compressive modulus leading to an improved shape-retention ability due to the effective crosslinking between the PI carbon skeleton and the c-BC nanofibers. The aerogel was so stiff that it supported a weight of 200 g without any noticeable deformation. Moreover, the combined 3D carbon skeleton and 1D c-BC nanofibers resulted in a reduced pore size and a narrow pore size distribution, which could be beneficial for oil sorption [248].
Wan et al. fabricated a c-BC aerogel nanocomposite with graphene [249]. To facilitate the uniform distribution of graphene in the BC network, an in situ biosynthesis route under agitated cultivation using a graphene-dispersed culture medium was employed. The spherical BC/graphene hydrogel was carbonized at 800°C to form a sphere-like c-BC/graphene aerogel. The nanocomposite aerogel exhibited an open honeycomb-like surface pattern consisting of numerous ridges and large cavities, which increased the aerogel's specific surface area and porosity. The unique nanostructure of the sphere-like c-BC/graphene aerogel is the key to enhance the sorption capacity (up to 457 times of its weight) for a wide range of oils and organic solvents [249].
Reduced graphene oxide (rGO) has also been composited in a c-BC aerogel [250]. This was done by freeze-casting and freeze-drying of the GO and BC mixed suspension. The GO nanosheets and BC nanofibers were uniformly assembled into a porous and interconnected 3D network. Subsequent pyrolysis transformed it into rGO/c-BC aerogel with the preserved 3D nanostructure where the c-BC nanofibers were still coated on the rGO sheets. The aerogel density was easily controlled by varying the concentration of the precursor in the suspension and the ratio of GO/BC. The lowest density was found for a GO/BC ratio of 1:1. The oil sorption capacities of the rGO/c-BC aerogel ranged from 300 to 1000 times of its weight, much higher than most carbon sorbents [250]. The ultrahigh sorption capacity was attributed to its low density and high porosity.
Luo et al. devised a new method of preparing c-BC aerogels containing rGO by a novel BC culturing process [251]. A thin BC pellicle was used as a substrate in the static culture. The solution containing a 2D few-layer rGO (FrGO) suspension was sprayed onto the BC substrate to form a thin layer, onto which BC was grown to consume the sprayed FrGO layer completely. The process was repeated to form a thick BC/FrGO layered structure, which was then freezedried and pyrolyzed. The c-BC/FrGO aerogel from the process exhibited an entangled nanostructure between the FrGO sheets and c-BC nanofibers, creating a multi-scaled porous structure and large specific area. As a result, the c-BC/FrGO aerogel exhibited mechanical robustness and an extremely high sorption capacity of 245-598 times its weight for a range of oils and organic solvents [251]. It also showed excellent reusability by both squeezing and combustion, with nearly the same sorption retention.
To assist the collection of a c-BC aerogel for reuse, regeneration, or recycling after oil sorption, functionalizing the aerogel with magnetic properties is a very useful and practical approach. A magnetic functionalized c-BC aerogel can be collected easily in large quantities with the aid of an applied magnetic field. Supree Pinitsoontorn and his team at the Institute of Nanomaterials Research and Innovation for Energy (IN-RIE), Khon Kaen University, have explored that concept by using in situ co-precipitation of magnetic Fe 3 O 4 nanoparticles in the BC pellicle before converting it into a magnetic c-BC aerogel [252]. Interestingly, by controlling the concentration of the initial Fe 3 O 4 precursors, the c-BC nanofibers were decorated with well-dispersed magnetic nanoparticles with a Fe/Fe 3 O 4 core-shell structure (Fig. 6a&b). The core-shell structure increased the magnetization of the NPs due to the large magnetization of the Fe core. This, in turn, improved the magnetic attraction ability when subjected to external magnetic forces. Although magnetic NPs were impregnated in the c-BC structure, the outstanding mechanical properties of the c-BC aerogel were still maintained. The magnetic c-BC aerogel was able to be compressed up to 90% strain and return to its original shape after release. This process was repeated up to 100 successive cycles, and the shape of the aerogel was nearly unchanged from the original state. Moreover, even with the addition of magnetic NPs, the magnetic c-BC aerogels had an ultralightweight property with a density of only 7.4 mg cm −3 , which is lighter than other magnetic carbon aerogels from several carbon sources [253,254]. The oil sorption capacity of the magnetic c-BC aerogel was still very high (37-87 times of its weight), which is comparable to other carbon aerogels from various sources [254][255][256]. Also, it can be rendered recyclable several times by dissolution. The highlight of the magnetic c-BC aerogel is its ability for magnetic retrieval from the liquid after sorption. As shown in Figs. 6c&d, the motion of the magnetic c-BC aerogel in a liquid can be controlled by an external magnet. The magnetic force can also lift it out of the liquid after use. This functionality is beneficial for manipulating the sorbents in a large area of polluted water, in a practical application of this technology.
The magnetic c-BC aerogel could not just be used for the remediation of an oil-spill but also be applied to other contaminant adsorption situations. Figure 7 shows a series of images demonstrating the dye (indigo carmine) sorption capability of the magnetic with magnetic manipulation and retraction of the c-BC aerogel. [252]. Image reproduced from [252] with permission from the ACS ( Copyright American Chemical Society, 2020).
c-BC aerogel [257]. The vivid blue color gradually faded, and the water became clear just like before the dye was added, indicating the complete dye removal, as confirmed by UV-Vis analysis. After the process, the aerogel was magnetically removed from water by a permanent magnet. The dye-adsorbed c-BC aerogel can be regenerated by dissolution followed by ovendrying before reuse. In addition, the magnetic c-BC aerogel can still be utilized in water treatment applications for other contaminants such as bisphenol A and Cr(VI) [258,259]. Some studies combine photocatalysis with other oxidation methods to further strengthen wastewater treatment. Rajagopal et al. [260] prepared microcellulose (MC) and TiO 2 composite materials, combined with hydrogen peroxide photocatalytic degradation (TiO 2 ?MC?H 2 O 2 ), to decolorized wastewater containing multiple dyes under sunlight. They showed that 99% of high-concentration methylene blue (MB) dye wastewater (200 mg/L) can be degraded within 150 min, and the removal efficiency of Chemical Oxygen Demand (COD) can reach 72%. It was highlighted that the synergy index of H 2 O 2 assisted photocatalytic degradation is 3.54, which shows that the above process coupling has a positive synergistic effect. Other research that is being developed by Cai et al. [261] takes advantage of a nanocomposite strategy, rendering stable cellulose-based hybrid materials with diverse functionalities for micropollutant removal. Through synergistic oxidation, e.g., Figure 7 A series of images demonstrating dye sorption using the magnetic c-BC aerogel, which was magnetically retracted afterward [257]. Image reproduced from [257] with permission from Springer-Nature ( Copyright Springer-Nature, 2020).
persulfate activation, photocatalytic water treatment has been proven to obtain practical value and can be further developed industrially [261].
As already discussed, heavy metal ions in water can be removed by adsorption. However, single physical adsorption can only enrich and transfer heavy metals but cannot completely remove them. Semiconductor photocatalysts have redox capabilities, which can change the chemical properties of heavy metal ions to reduce their toxicity. Taking TiO 2 as an example, the mechanism of photocatalysis to remove heavy metal ions is roughly as follows:1) nano-TiO 2 adsorbs heavy metal ions on its surface, 2) an ultraviolet lamp is excited to generate photogenerated electron-hole pairs, and the electrons transition to the conduction band and transfer to the TiO 2 surface, 3) photogenerated electrons reduce the adsorbed heavy metal ions to low valence states (such as chromium, mercury, lead) or elemental forms (such as silver), the metal ions in the lower valence state further generate compounds and precipitate (such as chromium) or further obtain electrons as elemental substances (such as lead, mercury) and deposit on the surface of TiO 2 particles. However, supporting TiO 2 (or other semiconductor photocatalyst) on the surface of cellulose can improve the removal efficiency of heavy metal ions. Although the adsorption of heavy metal ions by the hydroxyl groups in cellulose contributes to the removal of metal ions, the adsorption effect is weak. For this, researchers usually take advantage of chemical modification or graft copolymerization to introduce effective adsorption active sites on the surface of cellulose [262] such as carboxyl groups, amino groups, and sulfonic acid groups. These groups can selectively recognize and capture various heavy metal ions through electrostatic attraction or complexation and chelation coordination effects [263], thereby improving the photocatalytic removal efficiency.
Interfacial solar evaporation, which utilizes the abundant sunlight to evaporate saltwater, has gained significant attention as an environmentally benign and sustainable approach. Significant efforts have been made in realizing supporting substrates that provide optimal thermal management and unimpeded water transport to foster high-performance interfacial solar evaporation. The high degree of crystallinity of CNMs provides excellent mechanical stability, and their highly dense surface functional groups enable direct deposition or adsorption of various photothermal materials [264,265]. After subjecting MoS 2 /BC bilayered aerogels to vigorous mechanical agitation, no detachment of photothermal materials from a BC matrix was observed, and the bilayered structure remained intact [266]. Besides, CNMs can be easily processed into nanomicroporous structures, and this interconnected porous structure can enhance the light absorption of the photothermal materials loaded onto these structures because of the increased light reflection and scattering within the pores. Jiang et al. prepared carbon nanotube/cellulose composite aerogels as photothermal materials. Owing to the strong hydrogen bonding between the ample hydroxyl groups of CNFs and carboxylic groups of CNTs, these materials were found to be robust. The ultrahigh porosity of the CNF aerogel and high light absorption of CNTs led to a 97.5% light absorbance within the light range from 300 to 1200 nm [267]. In addition, owing to a low thermal conductivity they provide excellent thermal insulation. The thermal conductivity of a sophisticated solar evaporator designed by Li et al. was as low as 0.06 W m −1 K −1 . This solar evaporator achieved high evaporation efficiency, up to 85.6% under 1 sun illumination [267].

Introduction to polymer matrix biocomposites from well-dispersed cellulose nanofibers
Cellulose-based plant fibers can be readily used by themselves to form paper, packaging board materials and recently molded fibers [268]. Polymer matrix composites based on the same fibers offer possibilities to improve processing (improved geometrical complexity, rate of processing etc.), extend the range of physical properties, and improve the chemical stability, where the improved moisture stability is particularly important for engineering applications. The use of CNFs as a reinforcement is then of obvious importance: wherein the intrinsic 'fiber' properties should be better than for the plant fibers, specific nanoscale phenomena may occur such as structural improvements to the polymer matrix (crystallinity, orientation, reduced molecular mobility etc.) and it may become easier to fabricate small-scale geometries.
The first effort to make cellulose nanocomposites with a polymer matrix appears to have been the investigation by Boldizar et al. [269]. Poly(vinylacetate) nanocomposite films were prepared from hydrolyzed and homogenized cellulose pulp, with strongly improved mechanical properties. Also, hydrolyzed cellulose pulp was compounded with thermoplastics, injection molded into specimens for mechanical property measurements. The reinforcement effect was better than for pulp fiber reinforcement, possibly due to the higher aspect ratios obtained from the disintegration of the pulp fibers into nanofibers. The cellulose nanocomposites field, however, did not really take off until researchers in Grenoble investigated hydrolyzed CNCs (then called 'cellulose whiskers') as a reinforcement [270,271], followed by numerous studies, e.g., using nanocelluloses from parenchyma cells (potato tubers etc.), which are covered in a thorough review of the background to the whole field [272]. Yano and coworkers then investigated wood-based CNFs combined with poly(phenol formaldehyde) resins [273] and transparent nanocellulose films with an acrylate polymer matrix [274]. High optical transmittance and high mechanical performance is a highly interesting combination for biocomposites, which provides unique application opportunities, and will be reviewed in more detail in subsequent sections. Another important category of composites for engineering applications is 'cellulose biocomposites' based on thermosets. These are suitable for large-scale production, with nanopaper reinforcement in the form of prefabricated CNF mats. Epoxy and polyester resins typically used for glass fiber composites have been reinforced with a high content of wood CNFs [275,276], and mechanical properties were competitive with molded glass fiber composites [277]. A large collection of mechanical property data for cellulose nanocomposites have previously been analyzed [278]. The strongest and stiffest nanocomposites were those based on a high content of prefabricated nanopaper reinforcement, and properties scaled with nanopaper properties and volume fraction.
Key research and technical goals for the promotion of nanocellulose applications include scalable processing concepts and processing technologies for cellulose nanocomposites. Numerous technical studies are disappointing in that the mechanical performance does not meet requirements. Properties are simply not good enough to justify substitution of a petroleum-based polymer composite reinforced with glass fibers or mineral fillers. The most common reason for this first problem is that the CNFs are agglomerated in the composite, often into microscale aggregates. Although such a material is based on nanofibers, it is not nanostructured in a true sense. The modulus usually still shows improvement, although the reinforcement can be even lower than for comparable plant fiber composites. Strength is often reduced by the presence of aggregates, since they form stress concentrations which tend to initiate failure at low strain. A second problem is that cellulose nanocomposites are expected to contribute to sustainable development, but energy demand and carbon dioxide emissions related to CNF and nanocomposites fabrication tend to be high [279], which needs to be addressed. Holocellulose-based native CNF is one interesting possibility with low fibrillation energy, despite the lack of chemical modification [280]. Finally, the cost of the nanocomposite material needs to be competitive with alternative materials for a given application. The two last points (sustainability and cost) mean that the possibilities for chemical modification of nanofibers are not endless, contrary to many optimistic statements.
A common preparation route for cellulose nanocomposites is to dissolve the polymer in an organic solvent, mix in the CNF and then do solvent casting. In industry, this process could be used for some coatings, preparation of adhesive films etc., but it is not appropriate for semi-structural composites. The solvents used are frequently expensive and could even be toxic and difficult to recover. Casting from a hydrocolloidal dispersion is possible, though. The polymer matrix can be water-soluble, e.g., starch [281,282], although this limits the applications. The polymer can also be distributed in the form of waterdispersed nanoparticles, as was demonstrated for PLA [283]. The water-based processing route is again feasible for coatings and adhesives, and also for the preparation of some types of nanocomposite films and barrier layers. Layers of nanocomposite films of around 100 µm in thickness can be stacked, followed by hot-pressing into nanocomposite laminates. There are, however, significant industrial challenges in processing of the colloids. The concentration of nanofibers needs to be higher than the 1-2% typically used in basic research investigations, while preserving CNF dispersion and limiting the use of chemical modification to affordable and sustainable technologies. The removal of a large quantity of water is a technical difficulty, and a challenge in terms of cost and energy demands.
Melt processing of thermoplastic nanocomposites is of great technical potential. Injection molding of thermoplastic products is an enormous business and preferred in the automotive industry, due to the high processing rates. An important technical problem is the strong viscosity increase arising from the addition of CNF to thermoplastics, which can happen at a low cellulose content. The main reason for this increase is the large aspect ratio and small dimensions of the nanofibers. Yano and coworkers have addressed this by doing some of the pulp fiber disintegration in the compounding process itself [284]. Since mechanical properties scale with cellulose content, one limitation is that a typical reinforcement content is around 10 wt.%, which may be related to melt viscosity problems.
Thermoset nanocomposite processing is also challenging. Bulk mixing between liquid thermoset precursors and CNFs frequently results in their aggregation, simply because of the CNF-CNF affinity in such liquids. CNF reinforcement 'mats' will have nanoscale porosity, which means that good wetting is required for resin impregnation. Thermoset resin wetting of cellulose is, however, difficult since there will be water molecules at hydrophilic CNF surfaces under ambient conditions. Commercial production of glass fiber/epoxy prepreg utilizes organic solvent assisted epoxy impregnation, and this works well also for high content CNF/epoxy [277]. If the nanocellulose reinforcement is modified by acetylation [285] or by green chemistry approaches [286], bulk monomer impregnation is facilitated so that processing may be feasible and similar to industrial liquid molding of thermoset composites.
After more than 25 years of cellulose nanocomposites research, there is an urgent need to address challenges to realize large-scale use of nanostructured cellulose biocomposites (processing, moisture, performance, cost, sustainability, etc.) and intensify research and development on credible applications. In addition, the goals of materials substitution need to be defined: What are the specific reasons for substituting existing materials with polymer matrix cellulose nanocomposites? In the following sections, four cases of importance for the application potential of cellulose nanocomposites are covered: melt compounding of thermoplastic composites for mass markets will be addressed, optically transparent composites based on bacterial cellulose (BC) for impact protection, transparent wood, which is a polymer matrix nanocomposite for building and photonic applications based on nanostructured wood substrates, and finally functional CNM composites with structural color.

Kristiina Oksman (Luleå University of Technology, Sweden): large-scale melt compounding of cellulose nanocomposites
Melt blending is an important manufacturing process, and its development is important for the commercialization of cellulose nanocomposites and associated products such as automotive parts and packaging films. However, many challenges need to be overcome to enable large-scale production [287]. One challenge is the high material cost and complexity of the manufacturing process, which further increases the total cost of the material. In addition, the environmental impact of the manufacturing process is a vital consideration. Since significant amounts of energy are required to manufacture promising biobased cellulose nanocomposites, this has negative implications for product cost and commercial production. For this reason, the use of cellulosic nanomaterials in thermoplastic polymers must be extensively explored. The improvement in mechanical properties associated with the addition nanocellulose has been a research focus; however, the addition of nanocellulose can also result in other benefits. Regardless of the impact on polymer properties, nanomaterial additives must be optimally dispersed, distributed, and, in some cases, oriented within the polymer.
Generally, the manufacturing of nanocomposites is challenging since nanomaterials have a large surface energy and tend to agglomerate. CNMs are usually fabricated via top-down processes in the presence of water; although these materials initially disperse well in the water, once they have dried, redispersal is difficult owing to strong interactions that may include hydrogen bonding amongst other interactions, but ultimately leading to what is commonly termed 'hornification'. CNFs and CNCs form gels at low concentrations (1 wt.%) because of their tendency to form networks; however, gels derived from CNCs are less viscous owing their reduced length. Melt compounding is a high-temperature nanocomposite fabrication process in which polymers are heated above their melting temperature; this method restricts the use of CNMs as some are sensitive to high temperatures. These specific properties of nanocellulose materials make their use in melt compounding with thermoplastics even more challenging [287].
Many researchers have reviewed this topic; for example, Oksman et al., Wang et al., Zheng and Pilla, and Clemons and Sabo [288][289][290][291]. These reviews explore the developments in the extrusion processing of cellulose nanocomposites; Oksman et al. [288] focused on the processing and properties of cellulose nanocomposites, Wang et al. [289] focused on potential industrial processes, Zheng and Pilla [290] focused on melt processing with CNCs, and Clemons and Sabo focused on the wet compounding of cellulose nanocomposites [291].
In melt blending processes, various components, such as polymers, additives, and nanocellulose (reinforcing agents), are mixed in a compounding extruder where high temperatures and high shear forces melt the polymer; a specific screw configuration enables the mixing of the components. Two types of extruders are used in compounding processes: co-rotating and counter-rotating twin-screw extruders. Co-rotating extruders are preferred because they are more effective at mixing and dispersing the components; they also allow a flexible screw design, i.e., screws can be tailored to maximize dispersive, distributive mixing, or minimize shear forces to preserve fiber length. Co-rotating extruders are also more effective at removing moisture and volatiles, which is important if liquids are used as processing aids. The main challenges in the melt processing of cellulose nanocomposites are the controlled feeding of nanocellulose materials into the extruder and the dispersion and distribution of these nanomaterials in the polymer without degrading the cellulose or polymer [287,288]. Different approaches to the melt processing and associated dispersion of CNMs have been explored, namely, liquid-assisted extrusion or wet feeding, dry feeding, and single-or multi-step processing, including master-batch processing and solid-state processing.
In liquid-assisted extrusion processes, CNMs are suspended or dispersed in water with or without additives. This suspension is pumped into the extruder that vaporizes the liquid phase and removes it via a venting system [292,293,[295][296][297][298][299][300][301]. Karger-Kocsis et al. [292] listed several benefits of the liquidassisted feeding of nanomaterials into the extruder, including the absence of the necessity for the surface modification of the nanomaterials-thus avoiding the degradation associated with surface modifiers, reduced health risks owing to the suspension of the nanomaterials in liquid, and their improved dispersion because of 'blow-up' phenomena caused by the pressurized liquid evaporating from the melt, particularly in cases where water is used. Oksman et al. [287] also mentioned the economic and environmental benefits of the liquid-assisted process associated with the absence of supplementary treatment processes such as freeze-drying, which would increase its energy requirements, processing time, and risk to human health. CNCs are extremely small particles that are safer to handle in liquid owing to risks associated with inhalation.
However, if water-dispersed CNMs are directly subjected to high-temperature processing, rapid evaporation can lead to aggregation. Therefore, the use of processing or dispersion aids, that limit or prevent agglomeration during the evaporation of water or solvents, have been explored, such as polyethylene glycol (PEG) [293], triethyl citrate (TEC) [294,295], and prepolymers such as methyl methacrylate (MMA) [296], polyvinyl alcohol (PVOH) [297], and glycerol triacetate (GTA) [300]. Figure 8 shows a co-rotating extruder setup and screw design used for the fabrication of cellulose nanocomposites. The polymer is fed into the extruder, using a gravimetric feeding system, and melted before the introduction of the nanocellulose-containing liquid phase. Two atmospheric vents and a vacuum vent are used to remove the vaporized liquid phase. A typical temperature profile for the process with polylactic acid (PLA) as the polymer matrix is shown. The screws consist of feeding zones for the polymer and nanocellulose-containing liquid phase, a polymer melting zone, and dispersive and distributive mixing zones ahead of the vacuum vent. The total processing time is 30-50 s depending on the screw speed [297].
Liquid-assisted feeding presents many challenges such as the high viscosity of the nanocellulose suspensions-which hampers blending especially in the case of CNFs-and the high amounts of liquid (water or other solvent) that require removal. A suitable extruder, such as a co-rotating extruder, that effectively removes the liquid (solvent) should be used. The degradation of the polymer is also a common concern in liquid-assisted extrusion; however, provided no oxygen is available during processing, the polymer does not degrade. Herrera et al. [295] showed that the molecular weight of PLA is not affected by the liquidfeeding of CNFs; this result is consistent with that reported by Peng et al. [298], who investigated the reinforcing of PA6.
Geng et al. [296] polymerized MMA-latex onto CNCs (PMMA-CNC) and used this material to prepare a nanocomposite with PLA. A PMMA-CNCwater dispersion was pumped into a twin-screw extruder and mixed with PLA at a low CNC concentration. The nanocomposites were then oriented using a solid-state drawing process. The oriented nanocomposite exhibited an ultrahigh mechanical performance. Furthermore, this nanocomposite, containing well-dispersed CNCs, showed strain-responsive behavior, i.e., birefringence that changed with applied deformation, as shown in Fig. 9.
Peng et al. [298] used a water-assisted extrusion process to compound CNCs and PA6. They found that the addition of CNCs does not significantly affect the mechanical properties of PA6. However, they also found that CNCs act as a nucleation agent for the crystallization of PA6, increasing the cell density and reducing the cell size of the foam during microcellular injection molding. It was also demonstrated that water-assisted extrusion does not significantly affect the molecular weight of PA6. Herrera et al. [295,299,300] studied the effect of plasticizers or dispersing aids on the dispersion of CNMs in PLA using a liquid-assisted extrusion process. They found that the addition of plasticizers such as TEC and GTA effectively enhances the dispersion of CNFs and CNCs in PLA. Hietala et al. [301] prepared a liquid mixture of potato starch, plasticizer, and CNFs (up to 20 wt.%) and fabricated thermoplastic starch nanocomposites using a twin-screw extruder. The results were interesting in that the mechanical properties of the composite were similar to those of polyethylene; however, the added CNFs improved the moisture stability of the starch which is typically very moisture sensitive. Yasim-Anuar et al. [302] melt blended low-density polyethylene (LDPE), CNFs, and maleic anhydride-grafted polyethylene (MAPE) -a compatibilizer-using an internal mixer and a twin-screw extruder. They found that a twin-screw extruder better disperses CNFs, up to 3 wt.% fraction, than an internal mixer.

Dry feeding of nanocellulose during extrusion
Several studies have explored the use of dried nanocellulose powders in extrusion processing . , e.g., CNCs and CNFs [303][304][305][306][307]. The addition of dried nanocellulose particles is easier to feed, especially if high concentrations are of interest, but there are no studies showing that dry CNCs can be dispersed in melt compounding. Typically, the modulus of a composite prepared using dry nanocellulose increases slightly with increasing nanocellulose content, but its strength is inferior, or similar to that of the neat polymer. For example, Wang et al. [304] used spraydried CNFs, PP, and maleic anhydride-grafted polypropylene (MAPP)-a compatibilizer-to produce melt-extruded composites containing 3, 10, and 30 wt.% CNFs. The presence of CNFs in PP does not significantly affect its mechanical properties; the modulus and strength of composites containing 30 wt.% CNFs are higher and similar, respectively, to those of neat PP. Microscopy revealed the presence of large CNF agglomerates in the PP matrix. Venkatraman et al. [305] compared the performances of freezeand spay-dried CNCs in a process considered industrially scalable. A polymer (PA11) and dried CNCs were milled for 6 h before the powdered mixture was extruded or compression molded. Despite being time consuming and energy demanding, both processes produced composites with improved mechanical properties. Leao et al. [306] produced composites consisting of an acrylonitrile butadiene (ABS) matrix and 0.5, 1, and 1.5 wt.% CNCs with different lengths (150 and 220 nm) by twin-screw extrusion and injection molding. The CNCs were shown to increase the tensile modulus of ABS without increasing its strength, and the composite containing 0.5 wt.% CNCs (220 nm) exhibited the highest modulus. Sarul et al. [307] prepared a master batch consisting of spray-dried CNCs and a polymer mixture that was freeze-dried and used to form composites via direct mixing and twin-screw extrusion. The mechanical properties of the composites decreased with increasing CNC content, except in the case of the composites containing 5 wt.% CNCs, which exhibited a slightly increased modulus.

Master-batch processing
A master-batch process where a CNM is introduced in high concentrations into a polymer in solution is another approach for preparing cellulose-based nanocomposites. This mixture-master batch-containing high concentrations of CNMs is dried, crushed, and diluted in the extrusion process to the desired concentration [308][309][310][311][312]. This is a possible approach to the large-scale fabrication of cellulose nanocomposites; however, if solvent exchange is necessary, its environmental impact is significant. For example, it was found that the mixing of water-dispersed polyvinyl acetate (PVAc) latex with CNCs and CNFs at high concentrations (20 wt.%) promoted the dispersion of the CNCs and CNFs in the composite during the subsequent extrusion process [309,310]. Jonoobi et al. [311] prepared PLA-CNF composites by mixing an organic solvent-dissolved PLA (of different concentrations) with nanocellulose suspensions, drying and then extruding the mixtures using a twin-screw extruder. The resulting composites exhibit excellent mechanical properties and enhanced thermal stability. Jonoobi et al. subsequently investigated the reinforcing effects of acetylated (A C ) CNFs using a similar compounding process and concluded that the acetylation of CNFs does not further improve the nanocomposite properties [311]. Correa et al. Figure 9 Cellulose nanocomposite consisting of PLA and CNCs (0.2 wt%) with MMA-modified surfaces to aid dispersion, a ultrahigh mechanical properties and b strain-responsive birefringence, reproduced from Geng et al. [296] with permission from Elsevier (Copyright Elsevier, 2020).
[312] prepared a master batch of PA6 with a high concentration of CNCs by dissolving PA6 in formic acid. The aim was to coat the CNCs with PA6 to improve their thermal stability and enable their application in high-temperature processing. Nanocomposites consisting of PA6 and PA6-coated CNCs (1 wt.%) were prepared using a twin-screw extruder. It was found that, although well dispersed, the effect of the CNCs on the modulus and strength of PA6 was minimal and non-existent, respectively. Evidently, master-batch processing has advanced; however, the requisite solvent exchange and dissolution of the polymer matrix are not sustainable.
Researchers have extensively studied extruded cellulose nanocomposites, conducting interesting research on HDPE, PP, and PA6 nanocomposites, such as Sato et al. [313], Igarashi et al. [284], Suzuki et al. [314], and Semba et al. [315]. Sato et al. [313] compared the wet and dry processing of chemically modified cellulose, using multiple preprocessing steps, including mixing and kneading in an extruder at temperatures below the polymer melting temperature prior to the melt-compounding process. The cellulose and MAPP (compatibilizer) contents were kept constant (at 10 and 4.3 wt.%, respectively), while the degree of chemical modification of the cellulose was varied. They showed that the chemical modification and degree of substitution (DS) of the cellulose significantly affect the material properties and that chemically modified CNFs produce better composites than bead-milled CNFs. The treated cellulose was fibrillated during kneading but not to nanosized fibers.
Igarashi et al. [284] further explored the dry processing of cellulose nanocomposites; preparing dry mixtures of chemically pretreated cellulose pulp, HDPE, MAPP, and CaCO 3 using multiple preprocessing steps that were subsequently extruded to obtain composites. The results were similar to those reported by Sato et al. [313], who showed that the cellulose was fibrillated to smaller sizes during processing and that the mechanical properties of the composite were optimized at a DS of 0.43. The dispersion and distribution of the cellulose are shown in X-ray computed tomography images with increasing DS (Fig. 10a-f); Fig. 10d shows the composite with the highest mechanical properties. Although the preprocessing steps fibrillated the cellulose, improving the mechanical properties of the composites, they were time consuming, involving solvent exchange to an organic solvent and multiple washing steps with acetone, ethanol, distilled water, and isopropanol. Moreover, they did not reduce the cellulose to nanosized fibers. The preparation of the dry mixture required multiple steps and its feeding rate into the extruder was only 50 g/h, which is too slow for an industrial process.
Semba et al. [315] prepared nanocomposites consisting of refined A C cellulose and PA6. These components (PA6 and 10 wt.% A C cellulose) were mixed in propanol to form a slurry, which was dried and then melt compounded in a twin-screw extruder followed by injection molding. The authors reported a significant increase in the flexural properties of PA6, and the A C cellulose with DSs of 0.67 and 0.64 produced composites with the highest properties. Furthermore, the A C cellulose was fibrillated during the extrusion process and the thermal properties and heat deflection temperature of the composite were improved.

High-shear processes
High shear forces have been used in different ways, in wet and dry feeding processes, to improve dispersion, break cellulose agglomerates, and fibrillate cellulose [316][317][318][319][320]. Suzuki et al. [316] fabricated nanocomposites with high cellulose concentrations using a solid-state high-shear process. A wet mixture of cellulose (50 wt.%), powdered PP, and powdered MAPP was blended in a twin-screw extruder with a cooled extruder barrel (0°C). This process also fibrillated the cellulose to sub-micron sized fibers without melting the polymer. This mixture was then compounded by melt extrusion. The main outcome of this study was the fibrillation of cellulose pulp and the consequent enhanced mechanical properties of the associated composite, compared with those of neat PP.
Solid-state pulverization and melt processing is a multi-step fabrication process that has been employed by Iyer et al. [318] to prepare cellulose nanocomposites. A solid-state mixture of nanocellulose and PP was pulverized (and mixed and dispersed) at low temperature to avoid melting of the polymer before the pulverized homogeneous mixture was compounded through melt processing. They showed that the nanocellulose was well dispersed in PP but that it did not significantly affect its mechanical properties. Venkatraman et al. [319] used J Mater Sci (2022) 57:5697-5767 cryo-milling and planetary ball milling to prepare master batches consisting of PA6 and high concentrations of spray-dried CNCs. Master batches of PA6 and 75 wt.% CNCs and PA6 and 90 wt.% CNCs were prepared by ball milling. The powdered master batches were diluted during melt processing to produce composites with CNC contents of 5 and 10 wt.%. The properties of these composites were compared with those of corresponding samples produced by direct milling and compression molding. The results showed that the master-batch process produces a composite with a slightly higher modulus than neat PA6 but does not realize an improvement in strength. This study showed that spray-dried CNCs are difficult to finely disperse in PA6; moreover, the proposed process is very energy demanding as the milling time was 6 h. Olivera et al. [320] also explored the solid-state pulverization of CNFs. They mixed different cellulose nanofibers (10-30 wt.%), with 10 wt.% MAPP and PP, and extrusion mixed and compounded the mixtures at temperatures of 80-100 and 170-180°C, respectively. They reported improvements in the mechanical properties of the composites, especially for composites containing 30 wt.% cellulose. It has been shown that solid-state high-shear processing can effectively disperse cellulose. However, these processes are very energy demanding and may not be suitable for large-scale applications. Sridhara and Vilaseca [321] considered batch processing using a thermokinetic mixer to be suitable for scaling up to an industrial level. They fabricated nanocomposites by premixing CNF gel (containing 3 wt.% CNFs) with powdered PA6 to realize composites with CNF contents of 5, 15, and 25 wt.%; these mixtures were dried and melt compounded in a thermokinetic mixer before the resulting materials were milled or pelletized and compression molded. The mechanical properties of the composites increased with an increasing CNF content, and the best properties were found for composites containing 25 wt.% CNFs. This batch process is very fast as the material melts owing to high shear forces within 20 s.
Considering the latest research efforts on the melt processing of cellulose nanocomposites, interest in large-scale processing has increased; however, the expected breakthroughs have not yet materialized. Many processes are time consuming, have high energy and chemical requirements, and are expensive. It is very difficult to finely disperse nanocellulose, at a nanoscale, if composites with high nanocellulose contents (\1 wt.%) are desired. These studies were unable to determine whether cellulose at the nanoscale is a suitable reinforcing material because its dispersion at a nanoscale level is difficult in melt-extruded composites. To date, we have not realized cellulose nanocomposites with properties that cannot be achieved with micrometer naturalfiber reinforcement.
Techniques to assess the dispersion of CNFs inside thermoplastic composites, and the use of quantitative methods of the determination of dispersion and mixing are few and far between. Some work by Eichhorn and coworkers have shown that both photoluminescence [322], and Raman spectroscopy [323] are very useful in quantifying the spatial distribution and mixing of aggregates in CNF-reinforced thermoplastic composites. The relationship between the size of aggregates, and the mechanical properties of CNF-thermoplastic composites has also similarly been quantified using both photoluminescence, and Raman spectroscopy [324]. It was demonstrated that in spite of the use of a chemical dispersant, in this case tannic acid, sufficiently large aggregates of CNFs still persisted which reduced the fracture toughness of the composites [324]. It is possible that very low concentrations of well-dispersed nanomaterials can impart new functionalities to the polymer (as a nucleation agent etc.) and is expected to be the focus of future research on the melt processing of cellulose nanocomposites.

Koon-Yang Lee (Imperial College, UK): BC nanopaper-enhanced optically transparent composites for impact protection
Despite its high Young's modulus and tensile strength, one of the main challenges in the commercialization of nanocellulose as reinforcement for various composite applications is price. A recent market study reported that most producers will sell CNF gel at a price of US$100/kg [325]. Whilst the market price for BC is not widely reported, a recent technoeconomic analysis on the large-scale production of BC using an energy-efficient airlift reactor with modified Hestrin-Schramn medium estimated that the breakeven price for manufacturing BC was US $25/kg (a wet BC pellicle containing 99% water) [326]. Nanocellulose is therefore not cost-competitive in the high-volume composite market, especially when a high loading fraction of nanocellulose ([30 vol.-%) is required to achieve significant improvement in mechanical performance [278]. Cheaper sustainable reinforcing fillers, such as wood flour and natural fibers, are available for the composite industry [327][328][329][330]. It can be anticipated, however, that the high cost of nanocellulose could be offset by designing high value composites containing only a low loading fraction but still offering dramatically improved mechanical performance that conventional materials cannot achieve. One such area where nanocellulose could make a significant impact is their use to enhance the performance of transparent polymeric amour. The entry level for transparent amours is either monolithic acrylic or laminated polycarbonate/acrylic systems. The next choice up is glassclad polycarbonate, which offers a significantly higher level of impact protection but at the expense of added weight and cost. There are currently no lightweight polymeric transparent armor solutions that could bridge the gap between the two levels of impact protection.
The research group at Imperial College London is currently working on developing low loading fraction nanocellulose enhanced acrylic systems to close this property-performance gap. BC is an ideal candidate in this context due to its high single nanofiber tensile properties and its similarity to the refractive indices of acrylic resins [274], an essential requirement to achieve high level of optical transparency in a composite system [331]. Furthermore, BC is biosynthesized as pseudo-continuous cellulose nanofiber network [332][333][334] with a high specific surface area ([50 m 2 g −1 ) [335][336][337]. Therefore, the introduction of BC into an acrylic resin could create additional energy-dissipation mechanisms, including fiber-matrix and fiber-fiber debonding, as well as fiber re-orientation and fracture. To produce optically transparent BC-enhanced poly(methyl methacrylate) (PMMA) composites, Santmarti et al. [338] first pressdried BC pellicle (Fig. 11a) into a sheet of a BC nanopaper (Fig. 11b), followed by immersion and polymerization in a cell-cast mold containing a methyl methacrylate (MMA) syrup. Such a composite construct utilizes BC nanopaper as a two-dimensional reinforcement. While the starting BC nanopaper was not transparent, the resulting PMMA composite containing the BC nanopaper was optically transparent (Fig. 11c). The light transmittance of a 3 mm thick composite was found to be 73% at a wavelength of 550 nm. Such a high level of optical transparency was due to the low loading fraction of the BC nanopaper used (1 vol.-%), the filling up of the pores within the BC nanopaper structure with PMMA, the similarity between the refractive indices of BC and PMMA, as well as the small lateral size of the BC fibrils.
J Mater Sci (2022) 57:5697-5767 Significant improvement in the composite performance was achieved even though the BC loading was only 1 vol.-%. The tensile modulus of the resulting BC nanopaper-reinforced PMMA composite was measured to be 4.2 GPa, a 24% increase over the tensile modulus of neat PMMA. The BC nanopaper-reinforced PMMA composite was also found to possess significantly improved fracture resistance and flatwise Charpy impact strength. The initial critical stress intensity factor (K Ic ), a measure of resistance to fracture, of the BC nanopaper-reinforced PMMA composite was found to be 1.72 MPa m 0.5 ; a 20% increase over the K Ic of neat PMMA. A 20% increase in flatwise Charpy impact strength of the composite was also observed when compared to neat PMMA. Santmarti et al. [338] also investigated whether such improvements could also be achieved if BC was uniformly embedded within the PMMA matrix (i.e., BC as a three-dimensional reinforcement). The BC pellicle was first solvent exchanged from water though acetone into MMA, followed by polymerization in a cell-cast mold. While the resulting composite was also transparent, the mechanical performance of such a composite performed poorly when compared to neat PMMA. The K Ic and flatwise Charpy impact strength were found to be only 0.7 MPa m 0.5 and 4.7 kJ m −2 , respectively; a 50% decrease in K Ic and 25% decrease in flatwise Charpy impact strength compared to neat PMMA.
The mechanical performance of a composite is the volume-weighted average between the mechanical properties of the matrix and the reinforcement [339]. In theory, embedding BC uniformly within a polymer matrix should lead to a PMMA composite with improved performance as the high modulus and strength of a single BC nanofiber can be effectively utilized. However, fractographic analysis revealed the presence of significant matrix embrittlement when BC was embedded uniformly within PMMA (Fig. 12a). When BC was used as a two-dimensional reinforcement in the form of BC nanopaper, the effect of matrix embrittlement was minimized due to a reduced BC-PMMA interface. In addition to this, the reinforcing effect of such laminated composite Press-drying 120ºC, 1 t

(a)
Solvent exchanged into MMA, followed by polymerisation Direct MMA impregnation, followed by polymerisation BC as 3D reinforcement BC nanopaper as 2D reinforcement Figure 11 a BC pellicle with a water content of 99 wt%, b press-dried and wellconsolidated BC pellicle, followed by methyl methacrylate (MMA) impregnation and polymerization to produce 3-mm-thick BC nanopaperenhanced PMMA composites, and c 3 mm thick BC-PMMA composites produced from the solvent exchange of BC pellicle in water through acetone into MMA, followed by polymerization. The BC loading in all composites was 1 vol%. Reproduced from [338] with permission from the American Chemical Society (Copyright American Chemical Society, 2019).
construct stemmed from the mechanical properties of the BC nanopaper (Fig. 12b), which was found to possess a high tensile modulus and strength of 19.6 GPa and 188 MPa, respectively, as well as a high K Ic of 11.9 MPa m 0.5 [339], which is comparable to a single aramid fiber; reported to be 6.63 MPa m 0.5 [340]. The use of BC nanopaper also removed the need for complicated solvent exchange or drying steps, reducing the complexity of composite manufacturing. It should be noted that the long dewatering time of BC (or nanocellulose in general) to produce (bacterial) cellulose nanopaper, which is often cited as a bottleneck, could be addressed by reducing the grammage of the nanopaper or to induce flocculation of the nanocellulose by the addition of multivalent salts [341].
Since the loading fraction of BC required to achieve performance enhancement is low, the resulting BC nanopaper-PMMA laminated composite construct is expected to be cost-competitive, increasing its market uptake for advanced composite applications. However, there are still some outstanding issues in using BC nanopaper to enhance the properties of acrylic resins for impact protection that still need to be addressed, especially the effect of moisture on the long-term durability of the laminated construct as moisture is known to cause unpredictable delamination even in commercial transparent armor laminates under ambient service conditions [342]. Furthermore, the transparency of the BC nanopaper-acrylic laminated construct at elevated temperatures is another major issue requiring further research.

Lars Berglund and Yuanyan Li (KTH, Sweden): transparent wood as an application of nanostructured biocomposites
Cellulose biocomposites are often considered as environmentally friendly materials. The introduction of nanocelluloses is motivated by similar arguments, yet the energy associated with defibrillation of wood fibers into nanocellulose (process energy and energy for chemicals) is very high [343]. Wood substrates offer some advantages in this respect, provided the intrinsic cellulosic nanostructures in wood can be preserved and exploited. The category of woodpolymer composites, however, is not new; wood has been impregnated with formaldehyde-based thermosets and commercialized (Impreg®, Compreg®), and monomers for thermoplastics have also been impregnated and polymerized in wood substrates [344]. The new aspect of many recent studies is that the cellulose nanostructure in the cell wall of the wood substrate has been preserved and targeted for functionalization [345]. Research on transparent cellulose biocomposites started with studies where thin cellulose nanopaper films from nanofibers were impregnated by acrylic monomers and cured [274]. The concept of transparent wood was then suggested for engineering applications [346,347]. Wood substrates were delignified to remove the light-absorbing lignin component. In the next step, a monomer was impregnated into the substrate, followed by in situ polymerization to form a transparent wood biocomposite. Substrates other than wood have been used subsequently, for instance wood fibers [348] and bamboo [349]. From the point of view of cellulose biocomposites research, these materials are interesting because they combine structural properties (strength, stiffness) and the potential for making large structures with high optical transmittance. Optically transparent biocomposites opens a large field of materials research and opportunities for applications where semi-structural composites have photonic functions.
There are substantial challenges for transparent wood. Although processing is very similar to that of high-strength glass or carbon fiber composites (resin impregnation of a porous reinforcement), the added requirement of avoiding optical defects (voids, microcracks. cellulose agglomerates, etc.) makes it demanding. It is also difficult to achieve high optical transmittance in thick structures of high cellulose content, since light scattering inside the material is substantial. The research and materials development problems are illustrated in Fig. 13. When light reaches the transparent wood surface only a small fraction of it is reflected. Inside transparent wood, however, there is substantial light scattering, both forward-scattering and in other directions. Scattering takes place due to a mismatch in refractive index of phases, from optical defects (e.g., interfacial debonding) and possibly from Rayleigh scattering inside the cell wall. Very few ballistic photons can go through the material without scattering. Some light may also be absorbed by residual lignin. The detailed mechanisms of light propagation are still under investigation, and the results will support ongoing product development.
Yano and coworkers have prepared load-bearing biocomposites by impregnating delignified wood with polyphenol-formaldehyde thermoset precursors [351]. Li et al. [347] used NaClO 2 delignification for their transparent wood substrate to reduce light absorption from lignin whereas Zhu et al. used a kraft pulping approach [346]. Li et al. [352] further developed the possibilities by using a lignin-retaining method, where only lignin chromophores were removed.
Various polymer matrices have been used for transparent wood, where the best approaches involve impregnation of a monomer or thermoset precursor, including acrylic monomers [347,353], epoxies [346], thiol-enes [354], and polyimide [355]. Although it is well known that the refractive index of the polymer needs to match the refractive index (RI) of the substrate to reduce scattering, the details of the scattering mechanisms are not fully understood. This understanding will contribute to materials development efforts. The need to determine the refractive indices (in two directions) for the wood substrate has led to an experimental determination of their values [356]. This method is important, since chemical modification of the wood substrate will change its RI. Recently, the scattering in real wood microstructures from RI mismatches between the wood substrate and a polymer matrix has been modeled using numerical Figure 13 a Photograph of transparent wood [350] and b schematic of light transmission in transparent wood. Images reproduced from [350] under the terms of a CC-BY (Open Access) license.
methods and real microstructures [357]. This model can be used to predict haze, transmittance, etc., and also to analyze the effect of different wood substrate morphologies, wood content, and material thickness.
Transparent wood is of particular interest because of its contribution to sustainable development. For this reason, Montanari et al. [358] developed green procedures to modify wood substrates for improved polymer matrix compatibility and reduced moisture sensitivity. Subsequently, a new biobased acrylic polymer matrix was synthesized with a suitable RI, so that a fully biobased transparent wood biocomposite was obtained [359]. One of the challenges was to find chemical approaches which could be used in the chemically heterogeneous environment of a wood substrate.
For building materials applications, the mechanical behavior of transparent wood is important. Jungstedt et al. [360] measured a modulus of 19 GPa and a tensile strength of 263 MPa for birch/PMMA composites with 25 vol% wood substrate reinforcement. The optical transmittance of this composite was still as high as 70% at a thickness of 1.3 mm. Basic optical properties have also been investigated, including anisotropic scattering [361]. Vasileva et al. [362] reported on interesting polarization effects in transparent wood, where the effect depended on the wood species used for the reinforcing substrate. Haze was discussed by Li et al. [363], who pointed out favorable effects from high haze. Haze is defined as the proportion of forward scattered light to the total forwarded light. Broadband high haze ([80%) can create a uniform and consistent light distribution for comfortable living environments [363].
The thickness of transparent wood will have strong effects on optical transmittance, which is important for building applications. Chen et al. [364] investigated this effect since the well-known Beer-Lambert law in principle is not applicable to scattering materials. A model was developed to predict thickness effects, as well as a procedure to determine parameters in this model for specific materials. In a previous study, Li et al. [365] demonstrated how acetylated wood substrates showed improved transmittance so that thicker structures could be prepared. Acetylation will reduce the extent of optical defects, such as wood/polymer interface debonding and cracks. Acetylation also facilitates the amount of acrylic monomer diffusing into the wood cell wall and may influence the refractive index of the wood substrate.
It has been recently demonstrated how the cell wall in wood/PMMA composites prepared by solventassisted processing was impregnated by the MMA monomer so that PMMA becomes distributed at the nanoscale inside the wood cell wall [366].
The processing of very large structures is desirable and a requirement for many building applications. A key problem is that the permeability of liquid resins in wood substrates is often quite low. A practical consequence is therefore that thick structures are difficult to impregnate without the formation of significant optical defects. The main strategy suggested is to use lamination of veneer layers, as was demonstrated for plywood structures [367] and subsequently in several other studies [368]. Figure 14 suggests potential applications of transparent wood in smart buildings. High haze is a favorable feature, which allows uniform light distribution and can provide indoor privacy. Although high haze is beneficial for some applications, the possibility for low haze extends the range of product design possibilities. Li et al. [365] reported a smart window concept where indoor privacy was optically tuneable through tailoring of transmittance and haze in thick structures based on clear acetylated transparent wood. Transparent wood is an excellent base for further functionalization by the addition of functional particles and/or polymers. Montanari et al. [369] added phase change materials (PCM) to transparent wood, for heat storage purposes. Energy was adsorbed by the PCM melting during heating. Cooling resulted in heat release by PCM crystallization. Other examples include the addition of Cs x WO3 particles for heat-shielding since transmission in the near-IR range was reduced [370], and Fe 3 O 4 nanoparticles were used for magnetic transparent wood to provide electromagnetic interference shielding [371]. Recently, structural color has also been combined with optical transmittance. Hö glund et al. [372] precipitated metal nanoparticles inside the nanoporous wood substrate. This material was then impregnated with a refractive index-matched thiolene resin, followed by curing. The resulting composite showed structural color from the plasmonic nanoparticle effect, with only a minor reduction in the optical transmittance. Mechanical properties were well preserved. Another smart window concept is based on the integration of optoelectronic devices. Lang et al. [373] thus prepared electro-chromic windows by building conjugated-polymer-based devices on transparent wood substrates. The device demonstrated vibrant magenta-to-clear switching (ΔE*= 43.2) between -0.5 and 0.8 V, with a device contrast of 38 Δ%T at a 550 nm wavelength.
Another interesting modification possibility is to use stimuli-responsive particles. Li et al. [374] first reported transparent wood from luminescent particles in the form of quantum dots (QDs). Diffused luminescence was revealed under UV irradiation, where the luminescent color obviously depends on the QDs. Other stimuli-responsive properties have been developed, for instance by impregnation of thermo-and photochromic microcapsules [375], and photoluminescent properties through 1,3,3trimethylindolino-60-nitrobenzopyrylospira-based photoresponsive molecules [376]. One area of applications is the detection of changes in environmental conditions. Liu et al. [374] reported tunable roomtemperature phosphorescence through carbon dots doping for formaldehyde gas detection. The main research and development challenge is to obtain a uniform distribution of particles or active polymers despite the tendency for aggregation during processing. In situ particle synthesis [372] is one route to reduce the problems, compared with direct infiltration.
For solar cells, high haze is favorable due to increased length of the light path in the active layer, leading to improved efficiency. Therefore, transparent wood was used as light diffuser layer by direct attachment to a commercial solar cell [377]. Li et al. [378] instead fabricated a perovskite solar cell on a transparent wood substrate, for the purpose of energy positive buildings. Here, the main challenge was the introduction of the conductive layer since high optical transmittance, combined with good interlayer bonding, is required.
Durability and moisture stability are important for load-bearing building applications. One route is cell wall bulking of the wood substrate by acetylation [365] or other anhydrides [358]. This means that also the transparent wood biocomposites, based on such substrates [359,379], will be durable in a moist environment, although the high cost of polyimides is a limitation. For long-term durability, the yellowing problem known for paper needs to be avoided. Attempts to address this potential problem have been reported [380].
Transparent wood is also attractive for waveguiding photonics applications. Vasileva et al. [381] reported lasing from transparent wood impregnated by luminescent rhodamine 6G molecules. It was found that each fiber functioned as an optical resonator. The output signal is the collective contribution of the fiber-based resonators, which is broadly due to the fiber dimension variations and structural heterogeneity.
Optically transparent biocomposites based on cellulose or wood is a class of semi-structural materials which may combine load-bearing properties with eco-friendly characteristics, high optical transmittance, and photonics functionalities. The optical transmission criterion makes it necessary to improve micro-and nanostructural control during processing of cellulosic biocomposites. In addition, multifunctional biocomposites open new possibilities for applications of wood materials.

Silvia Vignolini and Bruno Frka-Petesic (University of Cambridge, UK): functional CNC composite materials with structural color
One of the most intriguing properties of nanocellulose is its intrinsic molecular chirality and how it can impact its behavior at much larger scales [382], which leads to the ability of CNCs to form functional materials with remarkable optical properties. The ability of CNCs to self-assemble into optically functional materials requires good dispersion in a solvent, usually water, and special care has to be taken to allow for the self-organization of the liquid crystalline order, followed by a removal of the solvent that preserves the acquired organization [383,384]. While these conditions are usually easily met when casting a pure dispersion in a dish, modifying the formulation or the assembly conditions to gain further functionality without compromising on the assembly abilities represents the main challenge. The formation of colored films using CNCs has been introduced by Revol et al. [385,386] and continues to be an active field of development, either to increase functionalities of such photonic films or to develop their potential as interference pigments by investigating scaling-up options. In this section, a few recent examples of optical functional materials are reviewed, leading to significant development in these two directions.

Optical functional materials
Plain photonic CNC films are usually brittle, and this has been considered as one of the major limitations to their practical use. Recently, several strategies have been proposed to mitigate their intrinsically poor mechanical resistance. Such strategies usually involve the co-assembly with a plasticizer that does not compromise the self-assembly of the CNCs, such as surfactants [387], a neutral or anionic polymer [388][389][390], globular proteins [391], resins or sol-gel precursors [392,393]. Two recent examples from Walters et al. [389] and from Saraiva et al. [390] showed that a common cellulose derivative, hydroxypropyl cellulose (HPC), can be successfully used as a plasticizer for CNC photonic films, significantly preventing crack growth upon bending with respect to plain CNC films (Fig. 15a-c). The tensile testing comparison between CNC/L-HPC 50/50 w/w with plain CNC films showed a clear reduction in stiffness (2.5 GPa vs 14.4 GPa) and maximum tensile strength (18.7 MPa vs 68.7 MPa) as well as a ten-fold increase in maximum strain (4.8% vs 0.5%). However, the addition of HPC, like most non-volatile co-solvents, caused a significant redshift [383,394], which usually comes with a trade-off on the range of accessible wavelengths. Another recent work involved a combination of bio-sourced and biocompatible silk fibroin (SF) and CNCs, showing improvement of uniformity and adhesion of the film, reminiscent of previous works using PEG to improve mechanical properties, uniformity, and adhesion on substrates ( Fig. 15d-g) [388,395]. The use of silk fibroin thus opens the possibilities of a fully biosourced, protein-based alternative to improve these properties.
Besides the co-assembly route to manufacture composite CNC-based materials, several post-treatments applied to plain CNC films have been also investigated to design original functionalities. Among them, one simple approach consists of immersing a CNC film in a prepolymer solution to produce a polymer-CNC composite. This strategy was used by Espinha et al. [396] to obtain a laminated structure with shape-memory properties. A more interesting and recent example was proposed by Boott et al. [397], whereby a CNC film was swollen and infiltrated with a prepolymer to form a mechanochromic elastomer (Fig. 16). Importantly, the infiltrated CNC films were initially prepared in the presence of 25%w/w glucose to facilitate later the infiltration of dimethylsulfoxide (DMSO), while specific slow assembly conditions were used to minimize the pitch to compensate for the red-shifting effect of the addition of the infiltrate [398]. After a first immersion in DMSO, two sequential soaking treatments were applied, first of 2,2′-azobis(2methylpropionitrile) (AIBN)/DMSO solution, and then with the monomer solution (containing ethyl acrylate, 2-hydroxyethyl acrylate and AIBN). The resulting composite elastomer presented clear mechano-chromic behavior upon stretching, visible in ambient light and solely caused by the compression of the pitch in the direction perpendicular to the applied stretch (and not simply relying on birefringence-induced Newton colors like in many other existing reports). A recent follow-up from the same authors proposed the fabrication of similar elastomers with additional shape-memory properties [399].

Upscaling strategies
The fabrication of CNC-based photonic materials usually involves laboratory scale processes (e.g., small and variable batches, slow processes, casting in Petri dishes) that do not easily translate to the industrial scale, limiting the applicability of CNCbased optical materials in end products.
The use of a roll-to-roll (R2R) approach to deposit CNCs into films has been proposed in the past but without any structural color [400,401]. The high throughput that is usually desirable makes any fast deposition incompatible with the formation of a cholesteric order; a reduced time suggests a high starting CNC concentration incompatible with a fast colloidal dynamic, and the resulting high shear applied near the slot-die aligns the CNCs in an achiral nematic film [400]. A significant step forward has been made recently by Droguet et al. [402] using R2R deposition to produce several meter-long CNCbased photonic films (Fig. 17a). In this work, a commercially available CNC source was employed (University of Maine Process Development Center) and simple pre-treatments were applied, involving tip sonication and fractionation. The starting suspension was diluted from ca 12%w/w down to 6%w/ w and sonicated, after which it was kept at rest until a macro-phase separation led to an upper isotropic phase and a lower cholesteric phase. The upper phase was discarded, and only the cholesteric phase, containing overall longer CNC particles, was deposited on a R2R to reduce the self-assembly time [398,403,404]. To reduce further the self-assembly time, the deposited CNC suspension was slowly translated through an in-line hot air dryer (T= 20 -60°C) at an effective speed of 0.2 mm s −1 , allowing the deposited film to be fully dry about a meter and a half from the suspension depositing slot-die, effectively reducing the self-assembly times from days to only a few hours.
The resulting film successfully demonstrated its relevance as a 100%-cellulose water-stable interference pigment after moderate heat treatment, grinding and size-sorting.
Other interesting strategies involve confinement in droplets, either in sessile drops or in water-in-oil (w/ o) emulsions. In the first case, selective dewetting was employed to deposit a CNC suspension onto hydrophilic spots with slow evaporation under an oil layer, leading to nearly perfect alignment, without any color distortion as usually caused by the coffee ring  (Fig. 17b) [405]. While the self-assembly then takes several days, the order is achieved within half an hour of the deposition, and a much faster assembly is achievable if the drying is completed in air. In the second case, w/o emulsions offer a substrate-free route to self-assembly, whereby the local alignment of the cholesteric is templated by the emulsion droplets themselves, and the need for casting on large surfaces when upscaling is suppressed. While this option was first explored using lab-made CNCs and gave too large a pitch to reflect in the visible [406], the recent optimization of the suspension formulation and exploitation of buckling upon further particle contraction led to a sufficient pitch compression to bring the photonic properties into the visible range and generate blue, green or red microparticles only made from cellulose (Fig. 17c) [407]. Interestingly, the blue and green hues were obtained by additional desiccation treatment, either via exposure to heat or to polar solvents (isopropanol for green, methanol for blue).
Overall, these alternative strategies open significant routes to industrial manufacturing of CNCbased photonic materials, each of them with specific advantages and disadvantages to adapt to the growing possibilities of co-assembly or post-treatments already available to develop the next generation of CNC-based optical materials.

Introduction to energy applications of nanocellulose
Solar cells have the capacity of converting inexhaustible and renewable solar energy into clean electricity and supply modern society with clean energy that promotes the carbon neutrality and sustainable development of our planet [408,409]. However, at present, only a small market share has been obtained for the electricity produced from solar cells. The widespread application of solar cells in our daily life is rooted in their cost reduction, environmentally friendliness, flexibility, and superior power conversion efficiency (PCE). Incorporation of cellulose paper into solar cells is an alternative solution to address the above issues [377,[410][411][412][413]. Solar cells have been demonstrated on common paper for more than fifteen years [414]. However, common cellulose paper has proven to be a poor substrate candidate due to its rough surface, porous structure, and impurities [415], which results in a complicated production procedure and a limited PCE (\1%).
Apart from energy conversion applications like solar cells, nanocellulose can also serve as the basis for the next-generation of sustainable energy storage technologies including supercapacitors and batteries [416]. Since many components of commercial energy storage devices like separators and carbon electrodes are fabricated from fossil fuels, their cost increases year by year with the overutilization of non-renewable raw materials [417]. However, the development of nanocellulose provides a potential but effective scheme to address this issue, where it can be used as the separator or the precursor material for carbon electrodes with enhanced electrochemical performance as well as sustainability [418]. In the near future, all-cellulose-based energy storage devices will be able to play an important role not only for energy conversion like solar cells, but also for energy storage like supercapacitors and batteries.  transparent to haze to highly scattering materials [419]. Since the first report of transparent nanocellulose films (also named 'cellulose nanopaper') [420], they have emerged as a promising flexible and green substrate candidate for solar cells. This is because of their better optical, mechanical, and barrier properties and surface smoothness compared with standard paper [421,422]. Hu et al. [423] initially demonstrated the direct fabrication of organic solar cells (OSCs) on a CNF film substrate that showed a PCE of 0.4% (Fig. 18a), and the corresponding I-V curve is displayed in Fig. 18b. Despite the poor device performance, this work first indicated that a CNF film was an attractive substrate for mechanically supporting OSCs. Meanwhile, the same group proposed an optical haze for a transparent CNF film (Fig. 18c & d) and predicted its potential application in solar cells as a functional light management layer.
Previous work had focused on the uses of nanocellulose films as a substrate material for solar cells. A variety of strategies were adopted to enhance the PCE of solar cells on nanocellulose film substrates. These have included the use of CNCs [424] and a mixture with other materials (such as acrylic resin and cellulose derivatives) [425], surface modifications of nanocellulose film [424], water-free manufacturing techniques [426], the use of novel active materials [427], and exquisite design of the cell configurations [427]. However, at present, the PCE of solar cells on nanocellulose film is still less than 5%, which is much lower than that of devices fabricated on plastic films; these indicate a PCE of over 21% on a laboratory scale [428].
To expedite the practical applications of solar cells on nanocellulose film, constant improvement in their performance should be given high priority. Therefore, future endeavor should focus on rationally engineering barrier properties, water resistance, weather durability, and surface roughness of the order of only a few nanometers over surface areas in the millimeter or even centimeter range. For instance, the mechanical isolation of CNFs from the cell wall of natural cellulose fibers produced microscale fibril bundles and debris in an aqueous suspension, thus increasing the surface roughness of the substrate, significantly deteriorating the device's performance, or even causing device failure.
Transparent nanocellulose film with high optical haze can go beyond the mechanically supporting substrate application and is becoming a functional  light management layer for solar cells. Generally, high transparency and high transmission haze (light scattering) are mutually exclusive. However, nanocellulose films not only exhibit a 90% transparency, but have also showed forward built-in light scattering behavior (Fig. 18a) [423]. These specific optical properties have allowed nanocellulose films to manipulate the propagation direction of transmitted light, which has extended applications toward solar cells as advanced light management layers to improve the efficiencies of light coupling into devices.
The interactions of visible light with nanocellulose plays a significant role in the optical properties of nanocellulose film. Theoretically, visible light can easily pass through nanocellulose films with negligible light scattering because the fibril diameters (3.5 30 nm) are much smaller than the wavelength (400-900 nm) of visible light. Therefore, nanocellulose films can display a transparent and clear appearance. However, optical haze is observed in most reported nanocellulose films (Fig. 18c&d). This abnormal phenomenon is primarily due to the existence of scattering particles (e.g., microscale fibril bundles and debris) derived from the incomplete mechanical homogenization of wood fibers.
The presence of micro-sized scattering elements gives rise to refractive index inhomogeneities and increased surface roughness of the as-prepared nanocellulose films, thereby increasing the scattering efficiency and finally indicating an improvement of optical haze. According to Mie theory, when the scattering particles have a diameter equal to or larger than the wavelength of the incident light, most of the transmitted light will be scattered along the incident direction.
Transparent nanocellulose films show forward scattering of light (Fig. 19a). As we can see from Fig. 19b, c and d, the original CNF suspension (rightmost image in Fig. 19b) has a hazy appearance due to the occurrence of micro-sized fibril bundles and debris (Fig. 19d) serving as scattering particles to enhance the transmission haze (Fig. 19c). However, after removing these scattering materials by a centrifugation procedure, the nanocellulose suspension indicated a clear appearance and the green laser could propagate through the suspension with a narrow-angle scattering (middle vials shown in Fig. 19b and c) [429]. Purified nanocellulose showed a uniform distribution of fibril diameters at a nanoscale level, which contributed to the reduced inhomogeneities and surface roughness of the nanocellulose film that suppressed wide-angle light scattering.
Taking inspiration from previous works, Fang et al. [430] reported a highly transparent nanostructured paper with a high haze using micro-/nanocellulose, exhibiting a transparency of[90% and a transmission haze of 60% at 550 nm (Fig. 20a). Through 2,2,6,6-  tetramethylpiperidine-1-oxyl (TEMPO) oxidization of wood fibers, micro-/nanocellulose was obtained, where the microscale cellulose fibers functioned as a light scattering source to increase the optical haze while the nanocellulose worked as a filling matrix to improve the transparency. They also demonstrated the use of this nanostructured paper in organic solar cells as a light management layer and a 10% increase of PCE was achieved (Fig. 20b). Since then, microscale cellulose fibers have been the main raw material to prepare highly transparent cellulose film with high haze by bottom-up methods (e.g., filtration [431], impregnation [432,433], and surface dissolution of fibers) due to their strong light scattering behavior [434].
In conclusion, nanocellulose obtained from mechanical isolation inevitably produces scattering particles. A transparent and hazy film with microscale surface roughness was obtained using the original nanocellulose as a raw material, which can serve as a functional layer for the improvement in the PCE of solar cells. Nanocellulose film without microscale particles (e.g., fibril bundles and debris) is a desired transparent and clear substrate with nanoscale surface roughness for solar cells. Despite much work being devoted to improving the properties of nanocellulose films, their use in solar cells is still in its early stages and is not ready for practical applications. In addition, in comparison with nanocellulose, microscale cellulose fibers are a much more suitable starting material to prepare highly transparent films with high transmission haze for solar cells as a light management layer due to their stronger light scattering behavior. Rechargeable batteries and supercapacitors are the two most popular types of electrochemical energy storage devices, and they have attracted increasing attention and enjoyed great success both in academic research and commercialization over the past few decades [416,[435][436][437]. With a high aspect ratio, and abundant surface functional groups to interact with ions, nanocellulose is an ideal starting material for energy storage devices based on ion transport [438,439]. It has been widely investigated as various important functional components in batteries and supercapacitors, such as current collectors, binders, electrolytes/separators, and electrodes [439][440][441][442]. The diverse structural tunability of nanocellulose (e.g., pores, fibril orientation, fibril diameter/length, surface functional groups, surface charge, surface energy, and surface wettability, degree of crystallinity, crystal phase structure, etc.) is particularly attractive for the design of high-performance functional components of batteries and supercapacitors. A key design principle is to meet the specific parameters depending on the functions of each component of the device. For example, improving the electrical conductivity of nanocellulose is critical for current collector applications, while separators must be electrically insulating. The combination of a wide potential window, high ionic conductivity and good mechanical properties are critical for developing Figure 20 a Light scattering behavior of highly transparent nanostructured paper with high haze. b The IV curves of the solar cells with/without nanostructured paper [430]. Reproduced from [430] with permission from The American Chemical Society (Copyright 2014, American Chemical Society).
high-performance nanocellulose-based electrolytes and separators.
As a common component in batteries and supercapacitors, the properties of binders directly affect the properties of electrodes, particularly their mechanical and electrochemical performances [441][442][443]. Polyvinylidene fluoride (PVDF) has been one of the most widely used binders in the most recent batteries and supercapacitors. This material, however, requires the use of volatile and toxic/hazardous solvents for processing. Alternatively, biomass-based binders such as nanocellulose, and its derivatives, are 'greener' and safer to use, and therefore are attracting increasing interest from the academic and industrial communities [444]. For example, carboxymethyl cellulose (CMC), as a derivative of the linear polymeric cellulose with the substitution of hydroxyl groups for anionic carboxymethyl groups, is widely used as a water-soluble binder in both the cathodes and anodes of batteries and supercapacitors [441].
Nanocellulose itself can also be used as binder for both cathode and anode electrodes. The high aspect ratio and tunable rheological properties of nanocellulose are desired for electrode manufacturing. Electroactive electrode particles can be wrapped by the high-aspect-ratio nanocellulose fibrils, forming a mechanically robust network. Moreover, at a finer scale, the abundant hydroxyl groups of nanocellulose fibrils can interact with the electroactive electrode particles, providing an excellent binding effect. Kuang et al. [445] demonstrated the excellent binding effect of nanocellulose (or cellulose nanofiber, CNF) in their recent work (Fig. 21). They first prepared a conductive CNF by mixing with carbon black (CB), which was then used as a binder to 'glue' the electroactive electrode (lithium iron phosphate, LFP) particles, forming a 3D interconnected porous foam (Fig. 21a). This 3D interconnected porous foam was further pressed to increase the density of the nanopaper electrode (Fig. 21b). A TEM image of the nanopaper showed that the LFP particles were glued by the CNF/CB network, which provided pathways for ion and electron transport (Fig. 21c). The nanopaper electrode demonstrated not only improved areal capacity over a conventional LFP electrode with similar thickness, but also excellent flexibility ( Fig. 21d and e). Alternatively, nanocellulose can also be used as a viscosifier to prepare printable inks for printed electrode fabrication owing to its unique characteristics such as rich hydroxyl groups, negative zeta potential, one-dimensional (1D) fibrous structure, and chemical functionalities [440,446,447]. In these printed electrodes, nanocellulose not only acts as a viscosifier during the printing process, but also as a binder to 'glue' the electroactive electrode particles in the dried state.
Nanocellulose-based separators have been widely studied and developed. For instance, mesoporous CNC membranes with high surface areas were prepared by Gonçalves et al. [448] as the separator for environmentally safer lithium-ion batteries (Fig. 22). The obtained nanocellulose-based separators exhibited outstanding wettability when soaked with conventional ester-based and ionic liquid-based electrolytes. Meanwhile, the three-dimensional porous structures formed by nanocellulose can shorten the ion diffusion pathway, thus obtaining a competitive ionic conductivity of 2.7 mS cm −1 . Good compatibility of the interface between the electrode and electrolyte can also be obtained, in comparison with commercial separators like glass fiber or polypropylene (PP)/polyethylene (PE) membranes. Coupled with the LiFePO 4 cathode, the assembled lithium-ion batteries can deliver a specific capacity of 122 mAh g -1 at a 0.5 C and an excellent rate capacity of 85 mAh g -1 at 2.0 C, indicating the important role of the threedimensional porous structures formed by nanocellulose. Meanwhile, the mechanical properties of nanocellulose-based separators are also superior to those of commercial separators according to the literature. In addition, there have been some companies like the Nippon Kodoshi corporation who have fabricated battery separators made from 100% cellulose which exhibit good heat resistance, exceptional porosity, and high liquid retention rates [449]. Therefore, by replacing conventional materials with nanocellulose-based separators, the development of real eco-friendly energy storage devices could be achieved in the near future.
As well as separators, nanocellulose can also be used as flexible substrates. Combined with a coating of conductive nanomaterials (carbon nanotubes, graphene, etc.) on the nanocellulose, free-standing electrodes can be obtained for energy storage devices such as supercapacitors [450].
When employed as the precursors, nanocellulose can also be converted into sustainable carbon materials for next-generation energy storage devices as well. For example, Li et al. [451] prepared hollow hard carbon microtubes derived from renewable cotton as high-performance anode materials for the next-generation of sodium-ion batteries, where the resulting hard carbon anodes at 1300°C delivered an outstanding sodium-ion capacity of 315 mAh g −1 at 0.1 C and a good rate capability as a result of their unique turbostratic structures. In addition, Xu et al. [452] developed sustainable hard carbons from CNCs at a lower carbonization temperature of 1000°C for use in sodium-ion batteries. The low-cost hard carbon anodes derived from nanocellulose displayed a reversible specific capacity of 256.9 mAh g −1 at a current density of 0.1 A g −1 , which was superior to the hard carbons obtained from cellulose microfibers at the same carbonization temperature because the layer-by-layer self-assembly of CNCs during the drying process formed a relatively low surface area to suppress the formation of a solid electrolyte interphase (SEI). The sodium-ion capacity of the hard carbons can compete with commercial graphite anodes for lithium-ion storage by consuming less energy. Sodium-ion batteries have also widely aroused interest from the battery community as promising successors to lithium-ion batteries for grid energy storage based on abundant sodium resources [453]. Therefore, nanocellulose-derived sustainable hard carbons perfectly meet the environmental and electrochemical requirements of eco-friendly sodiumion batteries.
Meanwhile, all-cellulose-based devices have also been achieved inspired by the hierarchical building blocks of natural cellulose (Fig. 23) [452]. By coupling with the cellulose microfiber-derived porous carbon cathodes using the hydrothermal process and subsequent chemical activation, an all-cellulose-based sodium-ion hybrid capacitor has been assembled [452]. This uses a modified cellulose-based gel electrolyte instead of using oil-based feedstocks, and its capacity reached 58.2 mAh g −1 at 0.2 A g −1 [452]. The energy density of this all-cellulose-based sodium-ion capacitor reached 139 Wh kg −1 at a power density of 478 W kg −1 , which can bridge the gap between highenergy batteries and high-power supercapacitors [452].
As we can see, great efforts have been made to develop the next generation of sustainable energy storage devices based on nanocellulose. There are still some challenges that remain to be resolved. For the commercialization of energy storage devices based on nanocellulose, a large-scale production platform is desirable to be established with decreased energy consumption [454]. In the meantime, how to rationally control the chemical and physical properties of nanocellulose to further enhance the electrochemical performance of devices needs to be considered at a scaled-up level [454]. In addition, most of the development of nanocellulose research in batteries and supercapacitors in the past few decades has been limited to sizes no smaller than the elementary fibril level, while the fundamental science and technologies at the molecular level of nanocellulose deserve further exploration [455]. All in all, the emerging applications of nanocellulose are of obvious benefits to energy storage toward carbon neutrality, but more efforts are needed to overcome existing shortcomings of technologies enabled by nanocellulose.

Conclusions
A review of the use of nanocellulose fibers in a variety of potential applications has been presented, targeting specifically health, water purification, composites and energy. The world needs to move to more sustainable choices for its materials, but not only that it needs to embed sustainability across all sectors of the economy. Nanocellulose has the potential to contribute to sustainability, but there needs to be careful consideration about the ways in which it is used in everyday applications, and there are hurdles to overcome in the embedded energy costs for its production. This review has demonstrated the use of CNMs for the directed growth of tissues, showing that both chemical and topological cues from these materials could be used to judiciously design bespoke tissue for implantation. One of the most common forms of materials for solid dosage pharmaceuticals is cellulose, in the form of microcrystalline cellulose. However, it has been demonstrated that nanocellulose is not just useful for controlled release in a solid dosage format, but also for transdermal and implantable systems. There is potential for nanocellulose to make inroads into commercial products since the barriers of cost are not so restrictive in such applications.
Water purification, and the supply of clean water, is critical to much of the world's population. It has been shown that nanocellulose can be very useful in this respect, and with some chemical modification can help to clean contaminated water. The development of 'smart' aerogel materials that can both clean oil-contaminated water, but also provide a means to remove the materials after use using magnetic properties has great potential for real-world applications. We have also shown that nanocellulose also has the potential to address other forms of water contamination, such as desalination, removal of dyes, and heavy metals.
Polymer composites remain a topic of great interest with respect to nanocellulose. Perhaps one of the most exciting developments in recent times has been the ability to use the inherent nanostructure of wood, combined with suitable resins, to make transparent materials that could have tremendous benefits for housing and other buildings. These types of application, where carbon can effectively be buried in a building, while also providing a critical function to enhance living, have enormous potential to contribute to sustainability. In terms of the industrial exploitation of composites, it has been made clear that for any advances to be made there are critical issues around dispersion and mixing that need to be addressed. Several in situ methods for the fibrillation of cellulose have been introduced, but much work is required to address mechanical performance. One way to overcome the need to mix the nanocellulose in a polymer is to use a pre-form. Bacterial cellulose networks have been demonstrated to be useful in this respect, although work needs to be done to control the basis weight of the networks, thereby their porosity and the achievable resin penetration. The barrier of cost of the nanocelluloses, over other materials such as talc and conventional fibers, might not be such an issue if the target application is carefully chosen. In niche applications, such as structural colored films, the added functional benefit and the ability to now apply roll-to-roll methods for production have real potential, particularly for packaging applications.
Finally, nanocellulose has real potential for energy storage devices: both as supercapacitors and batteries. Here cost is key, but the potential to make small devices with a high-power delivery, competing with conventional materials, is very much a real prospect. In this way nanocellulose could truly contribute to the development of new sustainable battery systems, and also with new developments such as the use of sodium-ion-based systems. The use of the material for the anode and separators in the latter has been demonstrated. Large-scale production of such devices is still a barrier for entry into industrial products, but we are close to solutions to these issues, perhaps drawing on the roll-to-roll production methods used for other applications. When it comes to solar cell applications, the efficiency of the devices made with nanocellulose is still not competitive. Nevertheless, the forward scattering provided by nanocellulose fiber networks makes these interesting materials which could be implemented in low-cost devices.
In conclusion then, much has developed in the field of nanocellulose in the last 10 years since the previous review was published. There are now several largescale manufacturers of materials (both fibrils and nanocrystals), and so the scaled-up production is no longer an issue. What is required now are for applications to be realized, and this can only be aided by a continued use of basic research to underpin these developments.

Open Access
This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licen ses/by/4.0/.