Cellulosic surfaces endowed with chemical reactivity by physical adsorption of functionalized polysaccharides

A strategy to functionalize cellulosic surfaces through physical adsorption of xyloglucan (XG) and carboxymethyl cellulose (CMC) derivatives bearing allyl or alkyne groups is reported. A set of functional polymer derivatives with degrees of substitution -DS- ranging from 0.10 up to 0.44 are first prepared through the opening of the epoxide ring of allyl glycidyl ether or propargyl glycidyl ether under mild basic aqueous medium. Contrary to alkyne-functionalized polymers, the radical copolymerization of allyl-XG and -CMC derivatives with acrylamide/acrylic acid leads to the formation of hydrogels, confirming their reactivity. The quantitative analysis of the deposition of these functionalized polysaccharides onto Whatman paper and wood pine fibers (spraying of aqueous solutions, drying and desorption step in water) shows that the physisorption of the polymer chains is not altered neither by the extent of the modification nor by the nature of the substituents. QCM-D experiments highlight a high affinity of allyl-XG for cellulosic substrates. The topochemical mapping by confocal Raman microscopy of cellulosic substrates on which alkyne polysaccharide derivatives have been deposited underpins that the surface coverage is rather uniform and that the diffusion of the polymer chains into the substrate reaches 40 μm. This aqueous functionalization/spraying procedure appears as a promising approach to confer novel adjustable surface properties to various cellulosic substrates, in a sustainable manner.


Introduction
The development of biobased materials has become central to answer to the depletion of fossil resources and to the necessity to develop recycling processes limiting the release of hazardous waste, e.g. endocrine disruptors and micro-and nanoplastics into the environment (Chemat et al. 2021;Tardy et al. 2022). In this context, cellulose and its derivatives are key substrate for the present and future emergence of high-performance green materials (Ferreira et al. 2021). Cellulosic substrates can be found under different shapes and compositions depending on their origin, their refining level and also on the specific applied treatments such as partial oxidation. Cellulosic substrates are commonly used in a wide range of applications from paper (De Assis et al. 2018) to textile (Felgueiras et al. 2021) but also in the conception of advanced materials such as green nanocomposites (Kargarzadeh et al. 2017), conductive materials (Hassan 2018) or medical devices with nanocellulose fibers (Jorfi et al. 2015). A crucial point with cellulosic materials lies in the tunability of their chemical and physical surface features, through covalent grafting of (macro)molecules (Klemm et al. 2005) by "grafting from" or "grafting onto" strategies (Roy et al. 2009). Covalent surface modification of cellulosic substrates offers many advantages, due to the diversity of species amenable to be grafted and to the strength of the covalent bond enhancing the durability of the resulting materials. Sometimes, conditions of derivatization alter the attributes of cellulose in terms of crystallinity and mechanical properties (Roy et al. 2009;Onwukamike et al. 2019). Besides, grafting approaches are not so universal and must be selected in regards with the substituent to be anchored. Alternatively, coating strategies have long been used in the industry to confer desirable properties to cellulosic materials (Chang et al. 2010). As an example, wood panels are generally sealed with acrylate coated polymerized under UV irradiation. However, the poor chemical affinity at the interface between the acrylic layer and the cellulosic substrate usually limits the durability of the material performances (Maurin et al. 2012).
In order to alleviate these concerns, another strategy to tailor the surface properties of cellulosic substrates consists in the physical adsorption of polymers exhibiting good chemical affinities for cellulose (Brumer et al. 2004;Xu et al. 2012a, b;Arumughan et al. 2021;Littunen et al. 2015;Zhou et al. 2007;Filpponen et al. 2012). The main advantage of this mild approach refers to the possibility to modulate surface properties of cellulosic substrates without alterating its native appealing properties. This simple process is particularly attractive from the sustainability point of view, since the deposition process can be carried out from aqueous solutions and relies on the formation of multiples non-covalent interactions at the interface and open promising perspectives in terms of reversibility and recyclability.
As most of cellulosic substrates are negatively charged due to the presence of hemicelluloses and/ or carboxylic groups onto cellulose backbone, cationic polymers are ideal candidates for physisorption thanks to the gain of entropy stemming from counter-anion release. Studies on cationic polymers adsorption onto cellulosic substrates are numerous and range from starch (Niegelhell et al. 2018) to xylans (Deutschle et al. 2014) used in paper manufacture to chitosan (Nordgren et al. 2009) and cationized polyacrylate latex (Pan et al. 2013) or even more complex functional materials such as positively charged polymeric nanoparticles generated by polymerization induced self-assembly (Alexakis et al. 2021).The use of polyanions, such as carboxymethylcellulose (CMC) in combination with anionic substrates has also been reported. It was shown that the adsorption is then governed by several parameters such as: the pH value which controls the degree of ionisation, the degree of carboxymethylation of CMC (Blomstedt et al. 2007) or the nature of the electrolyte and its concentration. Filpponen et al. (2012) reported on the adsorption of azide-or alkyne-modified CMC onto amorphous cellulose and subsequently functionalized the resulting materials with bovine serum albumin BSA or poly(ethylene glycol) through click chemistry. Similarly, Xu et al. (2012a, b) developed a platform of clickable cellulose surface using alkyne functionalized galactosylated hetero-polysaccharides. The adsorption phenomenon can also be modulated by playing with the hydrophilic-hydrophobic balance of the polyelectrolyte. This strategy has been described by Grigoray et al. (2017) who designed water-soluble photo-responsive cellulosic derivatives capable to adsorb onto cellulosic fibers. The adsorption was then tailored by adjusting the balance between the χ polymer-solvent interaction parameter and the χ s polymer-surface interaction parameter . Recently, our team demonstrated that CMC derivatives partially hydrophobized synthetized by Passerini three-component reaction can be greatly adsorbed onto cellulose surfaces (Pettignano et al. 2019). Interestingly, Konttury et al. (2017) have also highlighted the possibility to modify cellulose nanopapers by directly adsorbing hydrophobic polymers chains (polystyrene, polytrifluoroethylene) from solution in aprotic solvents. According to the authors, the mechanism adsorption, for most of the deposited polymers, follows a Langmuir adsorption isotherm.
In aqueous conditions, non-ionic polymers and in particular neutral polysaccharides from the lignocellulosic biomass can also be adsorbed onto cellulose surfaces, as described for xylans (Grantham 2017), arabinoxylans (Kôhnke et al. 2008), or even mannan (Berglund et al. 2020). The adsorption depends not only on the tendency of the polymer to develop hydrogen bonding interactions at the interface but also on the conformation adopted by the chains, which is controlled by the overall hydrophilicity of the polymer. Thus, the composition of polysaccharides, the nature of the substituents and the extent of derivatization are of great importance. Xyloglucan (XG), mainly found in the cell wall of plants, is another a water-soluble polysaccharide of interest whose high affinity for lignocellulosic surfaces is well-documented (Doineau et al. 2020;Dammak et al. 2015;Stiernstedt et al. 2006;Benselfelt et al. 2016). Its adsorption results from an endothermic process dominated by the entropy of released interfacial water molecules when the adsorption occurs near room temperature (Kishani et al. 2021). (Kishani et al. 2021). Whereas the composition of the xyloglucan has little effect on its capability to interact with cellulosic surfaces (Zhao et al. 2014), the molecular weight closely dictates the conformation of the XG chains deposited onto the substrate and highly impacts the yield of physisorption (Lopez et al. 2010;Villares et al. 2017). Owing to their peculiar affinity for cellulosic materials, chemically modified XG derivatives have been extensively deposited onto cellulose surfaces, with the aim to impart cellulose with varied chemical functionalities, from chromophore to cationic species (Zhou et al. 2007).
In view of developing new eco-designed cellulose materials, we propose herein to endow cellulosic surfaces with chemical reactivity through the physical adsorption of xyloglucan (XG) and carboxymethyl cellulose (CMC) derivatives bearing allyl or alkyne functions. We aim at generating an intermediate layer strongly interacting with cellulosic surfaces and capable to chemically react with a top-coat (scheme 1). This original synthetic approach combining efficiency and sustainability, should allow for post-reactions of allyl functions through various chemical routes such as thiol-ene (Lowe 2010;Meng and Edgar 2016), hydrosilylation (Dobrynin et al. 2020), radical reactions (Iio et al. 2007) where alkyne groups can be engaged in click chemistry coupling reaction (Filpponen et al. 2012;Sun et al. 2019;Deng et al. 2016) or used to map functionalization within the cellulosic material through confocal Raman microscopy (Mangiante et al. 2013).
Consistent with the principles of sustainability, allyl and alkyne functions were introduced onto polysaccharide backbones through simple and low environmental impact procedure involving an epoxide opening of allyl glycidyl ether (AGE) or propargyl glycidyl ether (PGE) in mild basic aqueous medium conditions. The functionalized polysaccharides (XG and CMC) were carefully characterized and their reactivity was evaluated in radical copolymerization in water medium. Finally, we examined the deposition of AGE and PGE-functionalized polymers onto cellulosic surfaces and investigated how the nature and characteristics of the polysaccharides and their degree of substitution impact their physisorption. Scheme 1 Cellulosic substrates (paper, wood fibers) covered by an adsorbed layer made of chemically modified XG or CMC, able to further react with a reactive formulation

Materials
Carboxymethyl cellulose sodium salt (NaCMC: CMC in the following text), having a degree of substitution of 0.7 (DS = average number of carboxymethyl groups per anhydroglucose unit) and a Mw = 250,000 g mol −1 , was provided by Sigma Aldrich. Xyloglucan was obtained from Megazyme with 95 wt% of purity, a molar mass Mw of 795,000 g mol −1 and the composition, namely the proportions of each unit given by the supplier are: Glucose 49 wt%, Xylose 31 wt%, Galactose 17 wt% and 3 wt% of other sugars. Cellulose substrates were cellulose Whatman and lignocellulosic fibers from maritime pine provided by the FCBA technology institute. The other chemicals were purchased from Sigma Aldrich (France) and used without further purification: propargyl glycidyl ether (PGE), allyl glycidyl ether (AGE), sodium hydroxide (≥ 98%), (+)-sodium (L)-ascorbate (≥ 99%), copper (II) sulfate (≥ 99%), acrylamide, sodium acrylate and sodium persulfate. The solvents were supplied by Carlo Erba.

Xyloglucan and carboxymethyl cellulose functionalization with AGE or PGE
The functionalization of XG or CMC with AGE or PGE was performed in basic aqueous medium (through opening of the epoxy rings of AGE or PGE by the hydroxyl groups of the sugar). XG or CMC (200 mg) was first solubilized in 20 mL of H 2 O or H 2 O/IPA (Isopropanol) solution (80/20 w/w, V = 20 mL) at room temperature. After the complete dissolution of the polymer, some NaOH pellets were added (from 0.33 to 3 eq/OH). The solution was then stirred for 30 min or until the total dissolution of NaOH. AGE or PGE (from 1 to 9 eq/OH) was then added to the solution and the vial was sealed. The reaction was carried for 7 h at 60 °C under magnetic stirring. Then, the solution was slowly transferred to a cold IPA or acetone solution, in order to precipitate the polymer. The precipitate was filtered using a Büchner (pore size 4) and washed several times with a solution of 20-30 mL of IPA or acetone. The polymer was dissolved in water and re-precipitated under the same conditions to remove any trace of starting reagents. The precipitate was dissolved one last time in distilled water and freeze-dried to get a white fluffy powder.
DS of allyl-functionalized derivatives were calculated by 1 H NMR in D 2 O (see below). Direct estimation of the DS is not possible for alkyne-derivatives and requires further modification of the polymers by "click" chemistry. In this view, 100 mg of product was dissolved in 20 mL of distilled water overnight. Sodium L-ascorbate and an azide molecule (11-Azido-3,6,9-trioxaundecan-1-amine, M = 218 g/ mol) were added to the solution. The solution was homogenized for 10 min under magnetic stirring and CuSO 4 .5H 2 O was introduced in solution. The quantities of each reagent are given by the following molar ratio: alkyne/azide/Cu 2+ /ascorbate = 1/1/0.05/1, the number of alkyne functions being calculated with respect to the theoretical maximum DS (DS = 3). The solution was left under magnetic stirring at room temperature for 72 h. After reaction, the solution was dialyzed against water using a regenerated cellulose membrane (cut-off: 10 kDa) for 3 days to remove the catalyst and the excess of reagent. Finally, the dialyzed solution was freeze-dried and the DS alcyne calculated from NMR analysis (see below).

Hydrogels preparation by radical copolymerization
The allyl-or alkyne-functionalized polysaccharide (m = 100 mg) were dissolved in 2.5 mL of distilled water at 70 °C. Then 0.5 mL of an aqueous solution containing 50 mg of acrylamide and 50 mg of sodium acrylate was added. The total mass of polysaccharide derivatives and monomers corresponds to m 0 . The solution was degassed by bubbling argon for 30 min. Potassium persulfate (2 wt%) was finally added to the medium to trigger the polymerization (and the formation of the hydrogel). After polymerization, the solution was maintained at 70 °C until gelation was macroscopically observed (usually after 15 min). The resulting hydrogel was immersed in 200 mL of distilled water for 24 h and the water was renewed several times to remove unreacted products and free chains. The hydrogel was weighted for determining the mass m s of the swollen hydrogel at equilibrium in water at 25 °C. The hydrogel was finally freeze-dried in order to determine the mass m d of the dry network corresponding to the mass of the reactants that have reacted. Extractable rate (T ex .%) and swelling ratio (SR) were achieved from Eq. 1: T ex % = (m 0 − m d )/ m 0 × 100 and Eq. 2: SR = (m s − m d )/m d respectively.

Raman analysis
Raman experiments were performed on a DXR Raman Microscope (ThermoFisher Scientific) equipped with an excitation wavelength at 532 nm and a 10 mW beam power. The spectra were collected using a 50 µm pinhole aperture and a polynomial fitting (at a polynomial order of 4) was used for fluorescence background correction. The Raman micrographs (1300 × 800 μm) were obtained from spectral data using point-by-point scanning with a 100 μm step size.
To study the epoxide ring stability in aqueous solution by 1 H-NMR spectroscopy, the glycidyl ether, (AGE or PGE, C = 20-30 g.L −1 ) was first dissolved in 1 mL of D 2 O/IPA (80/20 w/w). The onset of kinetics was marked upon the addition of NaOH (previously dissolved in D 2 O/IPA) to reach a concentration of 0.05 mol/L. The solution was rapidly homogenized under magnetic stirring and transferred into an NMR tube (5 mm). The kinetics were then followed by 1 H-NMR at 60 °C and analyses were performed at defined times with 64 scans for each analysis. To investigate the kinetics of epoxide hydrolysis, proton integrals of epoxide rings (at 2.78 and 2.97 ppm) were compared with those of the allyl (CH 2 =CH-) or alkyne (-CH 2 -C≡CH) functions for AGE or PGE (at 5.30 and 4.18 ppm respectively) which remain constant over time.
DS values were estimated from 1 H NMR spectra. Concerning allyl-functionalized derivatives, DS were calculated from the integral of the anomeric protons I anomer between 4.45 and 4.9 ppm and the one of one allyl proton I allyl between 5.9 and 6.1 ppm, according the following equation: DS allyl = (I allyl )/(I anomer ). In the case of alkyne-functionalized derivatives, DS were calculated from the "clicked" derivatives as following DS alkyne = (I triazole )/(I anomer ).

Deposition of polysaccharide derivatives and adsorption investigation
Coating of cellulosic substrates by spraying (Scheme 2): an aqueous solution of polymer (5 g L −1 , pH 6) was sprayed onto a cellulosic substrate (cellulose Whatman and wood fibers) and dried in an oven at 100 °C for 30 min in order to remove water and to favor adsorption of the polymer chains onto the substrate. The coated substrate was further immersed into distilled water solution to remove unbound chains (200 mL, 30 min). Through titration measurements, analyses of washing water allowed to determine the amount of physisorbed polymers onto the substrate according this equation: m adsorbed = m 0 -(C aq x V aq ) with m 0 the mass of polymer sprayed on the substrate, C aq the mass concentration of polymer in the washing water, measured by UV/Vis spectrometry according the Albalasmeh method (SI.1) and V aq the volume of washing water. The adsorption yields were given by m adsorbed /m 0 × 100. Scheme 2 Polymer adsorption onto Whatman paper by solution spraying/drying procedure Analysis with Quartz Crystal Microbalance with Dissipation Monitoring (QCM-D) Adsorption Measurements were performed with a QCM-D E4 apparatus (Q-Sense, Sweden) equipped with 4 measuring cells (or modules). The QCM-D quartz crystals were coated with gold electrode on each side (QSX 301, Q-Sense, Sweden) and had a fundamental resonance frequency (f 0 ) of 5 MHz. The temperature of the QCM-D chamber was controlled at 20 °C. Different cellulosic substrates were prepared from a coating onto commercially available gold quartz substrates (QSX 301, Q-Sense) previously cleaned in a Piranha bath (H 2 SO 4 :H 2 O 2 , 7:3) rinsed thoroughly and dried under nitrogen flow before activation by a plasma treatment for 20 min. The first category of cellulosic substrate was obtained by firstly spin-coating of an aqueous solution of cationic poly(allyl amine) (PAH) (C = 1 g L −1 ) onto gold sensor, which acts as an anchoring polymer to improve adsorption of cellulose onto the electrode surface (60 s of adsorption before rotation: 60 s, 3600 rpm, acceleration of 180 rpm s −1 ). Then, a second layer composed of cellulose nanocrystals (CNC) was spin-coated from a dispersion at 3 g. L −1 , before rotating the substrate (3000 rpm) for 60 s. CNC are derived from the acid hydrolysis of cotton linters. The conditions of hydrolysis are 64% H 2 SO 4 , 35 min at 65 °C. The second category of cellulosic surfaces was obtained after the spin-coating of PAH (C = 1 g L −1 ) onto the gold sensor, and a second layer of amorphous cellulose was spin-coated (no adsorption before rotation: 180 s, 3600 rpm, acceleration of 180 rpm s −1 ) from a 5 g L −1 solution in DMAc/ LiCl (Pettignano et al. 2019). After every layer was fabricated, the films were rinsed with water, by using the same spin-coating procedure employed for the two polymers. A final washing step was performed by immersing the surface in water, overnight, in order to ensure the complete removal of DMAc/LiCl. So, the obtained surfaces were dried under nitrogen flow, followed by drying in an oven at 40 °C for 20 min, and stored in a desiccator until use. The thickness of the amorphous cellulose layer, measured by ellipsometry, was approximately 27 nm.
In the QCM-D cell, the PAH-NCC surfaces were treated by a continuous flow of water (100 μL min −1 ), allowing for reaching equilibrium at 20 °C (i.e., when stable baselines for normalized frequency, Δf n /n, and dissipation, ΔD n , signals were obtained). The flow was then left at 50 μL min −1 for 10 min. The solutions of polymers (15 µg mL −1 in pure water at pH 7, filtered on a 5 µm cellulose nitrate filter) were then flowed for approximately 40 min. Finally, a rinsing step was performed by flushing the chamber with pure water solution, in order to remove the macromolecules that were loosely bound to the surface and verify the reversibility of the adsorption.

Results and discussion
Allyl and alkyne functionalization of xyloglucan and carboxymethylcellulose The introduction of allyl and alkyne functions onto polysaccharides relies on epoxide-alcohol reaction between the hydroxyl groups of XG and CMC, and allyl glycidyl ether (AGE) or propargyl glycidyl ether (PGE), respectively (Scheme 3). These coupling reactions involving allyl glycidyl ether have been relatively well-documented in the case of starch (Huijbrechts et al. 2007), xylan (Nurmi et al. 2015), cellulose (Qi et al. 2012) and even CMC (Lawal et al. 2011) but the use of propargyl glycidyl ether has scarcely been described (Nielsen et al. 2010). Herein, derivatization of the polysaccharides was performed in aqueous basic medium following a procedure established by Bigand et al. (2011). As the epoxy rings are subjected to a non-desired hydrolysis competing with the etherification, the stability of the glycidyl derivatives was monitored by 1 H-NMR analysis at 60 °C in basic hydro-alcoholic conditions (see details in experimental part).
The comparison of the 1 H-NMR spectra of AGE in D 2 O/IPA at 25 °C (pH 6) ( Fig. 1) shows a progressive decrease of the 1 H peaks corresponding to the protons of the epoxide ring at 2.78, 2.97 ppm (a) and 3.35 ppm (b) over time. Concomitantly, multiple 1 H peaks gradually appear between 3.5 and 3.7 ppm, related to the formation of by-products such as glycerol 1-allyl ether and the dimer resulting from the coupling between AGE and glycerol 1-allyl ether ( Figure SI.2). From Eq. (1), it is possible to quantify the rate of residual epoxide group with time (Fig. 1). It appears that the half-life (time at which 50% of the initial epoxide functions are consumed) of the epoxide ring of AGE is about 1h30, and the quasi complete opening of the epoxide ring occurs after 8 h (90% of the epoxide functions consumed). A similar hydrolysis profile was emphasized in the case of PGE ( Figure SI.3.2.), clearly anticipating that the chemical modification of polysaccharides (CMC and XG) with the glycidyl derivatives should not exceed 7 or 8 h under these conditions. From this statement, the reaction of XG with AGE was first carried out in aqueous basic medium at 60 °C using a stoichiometric amount of epoxide and hydroxyl functions, a AGE/NaOH ratio = 3 and a concentration of AGE = 20 g L −1 . Raman analysis of the purified modified XG derivative evidences the presence of the absorption band of alkene bond at 1640 cm −1 (Fig. 2a). In order to confirm the efficiency of the purification step, a mixture of polysaccharide (CMC or XG) with pre-hydrolysed AGE was prepared and submitted to similar washing procedure to be analyzed by Raman analysis (spectra 2 Fig. 2a and 5 Fig. 2b). No signal at 1640 cm −1 was identified on these Raman responses. These results clearly demonstrate that the coupling reaction with AGE is effective and that allyl groups are covalently bound to the polymer chains.
1 H-NMR analyses further confirm the anchorage of the allyl group onto the polysaccharide chains as 1 H signals at 5.3 and 6 ppm, corresponding to the allyl protons (Ha, and Hb) are clearly identified onto the spectra of modified XG (Fig. 3a). The evolution (1) % residual epoxide = 100 * I a I f of DS over time was monitored from the integration of allyl (Ha, and Hb) and anomer protons of the different sugar units (Arruda et al. 2015), emerging at respectively 5.13, 4.94 and 4.55 ppm. As shown in Fig. 3b, regarding the allyl functionalization of XG, the DS reaches a plateau at 0.14 after 7 h of reaction. Different experiments were performed with various amounts of AGE and NaOH in order to study their impact on the grafting efficiency. The results given in Table 1, indicate that DS values can be tuned either by adjusting the AGE/NaOH ratio while keeping the NaOH/OH ratio constant (0.33 eq/OH; entries 1-4) or by varying both ratios (entries 5-7). Substantial DS values up to 0.54 (entry 7) was obtained with the highest concentrations of AGE and NaOH at 60 °C. However, xyloglucans with DS superior to 0.4 exhibit relatively poor water solubility, even at 60°. Interestingly, the functionalization of the polysaccharide can also be carried out at low temperature (30 °C, entry 8) or in pure water (entry 9), but at the expense of the grafting efficiency (DS = 0.05 and 0.11 for entries 8 and 9, respectively vs. 0.14 for entry 1).
The same chemical strategy was then successfully transposed to CMC (see Table 1, entries 10-18, Figure SI.4.2), as evidenced by Raman and NMR spectroscopy (Fig. 2b and SI.4). The DS of the modified CMC derivatives are slightly lower than the ones resulting from the XG modification with AGE (DS = 0.34 for CMC vs 0.54 for XG in the same conditions of reaction, see entries 7 and 15, Table 1). This lower reactivity may reasonably Fig. 2 Raman spectra of a XG and b CMC (unmodified polysaccharide (spectra 1 and 4), after mixture with hydrolyzed AGE followed by washing step (spectra 2 and 5), and after reaction with glycidyl allyl ether, 1 eq/OH at 60 °C, 7 h) followed by washing steps (spectra 3 and 6)) be ascribed to the steric hindrance caused by the pendent carboxymethyl side groups and to the lower concentration of OH groups in CMC (as compared with XG). To evaluate the general applicability of this derivatization, a second functional glycidyl derivative i.e. propargyl glycidyl ether (PGE), was engaged in grafting procedures with XG and CMC under similar experimental conditions (Table 2).  Again, the ligation of PGE alkyne groups was unequivocally underpinned by the appearance of the characteristic signal of alkyne bond at 2115 cm −1 in the Raman spectra of derivatized polymers ( Figure  SI.5). Unfortunately, no alkyne proton could be detected by 1 H NMR in D 2 O (owing to its acidic character) making direct estimation of the DS impossible. To evaluate the extent of the modification, the alkyne-functionalized polymers were then engaged in copper alkyne-azide cycloaddition click reactions (CuAAC) with a large excess of azido-PEG (11-azido-3,6,9-trioxaundecane-1-amine) in water (Mangiante et al. 2013). 1 H NMR analyses confirmed the formation of the triazole ring at 8.1 ppm (Figure SI.6) and Raman analysis collected after the CuAAc coupling, confirmed the full consumption of the alkyne groups. Making assumption that the CuAAC coupling is quantitative, DS ranging from 0.09 to 0.24 and from 0.05 to 0.20 were obtained for alkyne-functionalized XG and CMC derivatives ( Table 2). Comparison of DS values suggests that the reactivity of AGE and PGE are roughly similar (see entry 5 Table 1 vs. entry 2  Table 2 or entry 13 Table 1 vs. Entry 4 Table 2).
Aiming at evaluating their reactivity, the allylfunctionalized XG and CMC derivatives were subjected to radical (co)polymerization in aqueous medium with acrylamide and sodium acrylate which are widely used in industry and will be eventually biosourced (Sobus et al. 2022). Due to the high functionality in allyl groups of XG, and CMC, we anticipated that their participation to a copolymerization process in water would give rise to the formation of chemically cross-linked hydrogels. Contrary to acrylamides or acrylates (Cabaness et al. 1971), allyl and especially propargyl functions exhibit poor reactivity in radical reactions and to our knowledge the reactivity ratios of AGE and PGE in radical copolymerization systems have not been reported in the literature. However, for comparison's sake, the reactivity ratios have been estimated for allyl monomers (A) such as allyl acetate or allyl chloride together with n-butyl acrylate (B) (Heatley et al. 1993;Brandrup et al. 1999). The reactivity ratios reported in these studies: r A ≪ 1 and r B≫ 1 (5.8 and 11.7 respectively), indicate that allyl monomers are not suitable for homopolymerization (via free radical polymerization) but can be involved in radical copolymerization/crosslinking procedures with acrylate/acrylamide monomers (Duanmu et al. 2007).
On this basis, the copolymerization of acrylamide, sodium acrylate and allyl-functionalized polysaccharide was investigated at 70 °C in water. Examples of hydrogels are shown in Fig. 4.
Contrary to alkyne-functionalized polysaccharides that failed at forming hydrogels whatever the DS and the conditions of reaction, radical copolymerization in the presence of the allyl-functionalized XG or CMC derivatives resulted in self-standing hydrogels which maintained their integrity after swelling in water (Table 3 and Fig. 4). Tuning the DS of the allyl-functionalized polymers allowed to modulate the swelling behavior so that hydrogels with a swelling ratio (SR) ranging from 180 to 390 were conveniently prepared by using a DS = 0.38 and 0.10, respectively. This trend is coherent because the crosslinking density is expected to increase with the number of reactive chemical functions, as reported in literature (Bencherif et al. 2008;Reis et al. 2003). Another key parameter is the chemical nature of the derivatized polysaccharide employed as multi-functional cross-linking reagent. As shown in Table 3 (entry 1 vs. entry 5), for a given DS in allyl groups, hydrogels exhibiting the higher swelling ratios were prepared from allyl-functionalized CMC (SR value of 1000, entry 5, Table 3). Surprisingly, the extractable rates from the xyloglucan-based hydrogels remained constant at around 60% whatever the DS of the XG derivatives, suggesting that a significant proportion of the grafted allyl functions do not participate to the copolymerization radical process. Conversely, the DS value has a great impact on the features of the hydrogels when allyl-functionalized CMC are used. Whereas the derivatized CMC with the lowest DS (0.06) fails at generating a hydrogel (Table 3 entry 4), the extractable rate is 57% using a CMC with a DS = 0.13 (Table 3 entry 5) and decreased down to 17% using a CMC with a DS = 0.24 (Table 3 entry 6). These contrasting properties of XG and CMC based hydrogels suggest that the reactivity of allyl groups is not equivalent on CMC and XG chains resulting in disparate conversion (of allyl groups). Such behavior can be reasonably explained by (i) a lower overall mobility of the XG chains due to their significantly higher molecular weight (Mw of 795,000 g mol −1 for XG vs. 250,000 g.mol −1 for CMC), and to the ramified nature of the XG polysaccharide structure and/or (ii) a different chain conformation in aqueous basic medium leading to a different behavior in solution. Indeed, the presence of carboxylate groups on CMC at this pH enhances its solubility, promotes extended conformation of the polymer chains owing to intra and interchain electrostatic repulsions, which might improve the accessibility of the allyl side-groups for radical copolymerization. Thus, the generation of hydrogels from the use of allyl-functionalized XG and CMC as cross-linking reagent confirms the ability of the grafted allyl functions to react upon radical conditions in aqueous medium.

Adsorption of functionalized xyloglucan and carboxymethylcellulose onto cellulosic substrates
As mentioned in the introduction, the adsorption of chemically modified polysaccharides appears as a powerful tool for the valorization of cellulosic substrates. This sustainable and simple approach relying on the physisorption of modified polysaccharides in aqueous medium, is particularly adapted to tailor the surface properties of cellulosic materials without jeopardizing its attractive bulk properties. The interest herein is to impart cellulose surfaces with new reactivity thanks to the presence of allyl and alkyne functions on the physisorbed polymer chains. In order to evaluate the propensity of derivatized XG and CMC to be adsorbed onto cellulosic substrates, aqueous solutions of the native and allyl-functionalized XG and CMC (pH ≈ 6) with DS ranging from 0.07 to 0.54 were first sprayed onto two cellulosic substrates namely a model Whatman paper and wood fibers (Muller et al. 2013) at a concentration of 5 g L −1 (semi-diluted regime since C>>C*) and the resulting materials were subsequently dried to favor interactions between the substrate and the deposited polysaccharide. This spraying deposition process was preferred over more conventional techniques such as dip-coating because of its facile transposition at the industrial scale (large substrates with various geometries and shapes can be covered). Moreover, this process allows to preserve the dimensional integrity of the cellulose substrate, and the drying step promotes interactions at the interface between the substrate and Table 3 Main characteristics of the hydrogels obtained through free radical polymerization of sodium acrylate, acrylamide and allyl-functionalized XG and CMC derivatives (the weight ratio between polysaccharide/monomers is 50/50)

Entry
Polysaccharide DS T ex (%) SR (g water /g gel ) the deposited polysaccharide. Adsorption yield that is to say the weight percentage of physisorbed chains after a washing step are given Fig. 5. As shown in Fig. 5 (right), the good adsorption of XG chains on Whatman paper is not altered by the grafting of allyl-functions as satisfactory adsorption yields around 50% are obtained, whatever the value of DS (from 0.13 up to 0.54). Regarding wood fibers, adsorption of native XG chains seems to be less favorable (adsorption yield ~ 30%). The insertion of rather apolar allyl residues onto XG chains has a positive impact on the adsorption yields (66% for a DS = 0.40). This spectacular enhancement is consistent with the chemical composition of wood fibers which contain rather higher content of lignin in addition to cellulose. In this context, allyl-functionalized XG chains are capable to establish hydrogen bonds at the interface but also hydrophobic and Van der Waals interactions with cellulose (Benselfelt et al. 2016) and lignin (Barakat et al. 2007). Such interactions lead to a gain of entropy that derives from the release of water trapped on the surface during adsorption.
Contrary to XG, the adsorption of unmodified CMC is noticeably more pronounced on wood fibers (~ 20 vs. 60% yields, see Fig. 5 left) probably because the electrostatic repulsions are higher with Whatman paper which is slightly oxidized (Henniges and Potthast 2009).
Allyl functionalization of CMC has a negligible effect on the adsorption yields onto both substrates at moderate DS values (0.07 and 0.19). In contrast, adsorption yields are significantly increased (up to 80% onto Whatman paper and up to 84% onto wood fibers) using CMC with higher DS (0.34).
Note that contrary to the other aqueous CMC solutions engaged in this adsorption study, dissolution of CMC chains with the highest allyl content in water afforded slightly turbid solution. This is consistent with the presence of chains aggregates which may resulted in a more efficient packing of the adsorbed macromolecules onto the cellulosic surfaces. Moreover, this difference in adsorption ability could be reasonably explained by the formation or reinforcement of other types of favorable driving forces (Van der Waals, hydrophobic interactions) at the interface with the cellulose substrate caused by the covalent tether of allyl residues. Relatively similar behavior has been previously reported for dodecanyl esters of CMC (DS dodecane = 0.012) which form aggregates in aqueous medium at pH 4-6 and efficiently adsorbed onto polystyrene surfaces for which unmodified CMC has no chemical affinity (Stuart et al. 1998). In this particular example, as mentioned by the authors, the presence of dodecane residues favors the establishment of hydrophobic interactions with polystyrene. The adsorption of alkyne-functionalized XG (DS = 0.09) and CMC (DS = 0.04) was additionally examined (SI.7). The behaviors of unmodified XG and alkyne-XG toward Whatman are quite similar. These results are in concordance with data reported by Xu et al. (2012a, b) who mentioned that an alkyne modified xyloglucan (Mw = 17,000 g/mol) (regioselective modification of galactose units by enzymatic oxidation followed by reductive amination with propargylamine) is adsorbed onto Whatman paper in a manner similar that of unmodified xyloglucan. It is important to note that for a given DS, the adsorption ability of alkyne-XG chains is roughly the same as allyl-XG chains, which is consistent with the similarities of the two derivatives in terms of chemical composition (relatively low DS). The propensity of alkyne-modified XG to interact with cellulose surfaces will be advantageously exploited to tackle a precise 3D topochemical mapping allowing to image the distribution of the modified polysaccharides at the surface and within the cellulosic substrates, through Raman confocal microscopy by collecting the alkyne response used as a specific tag (vide supra).
In order to gain insights into the ability of modified polysaccharide derivatives to be adsorbed onto cellulosic substrate and to complete the data achieved by spraying deposition (Fig. 5), in-situ QCM-D analysis of adsorption were undertaken. With that aim, a gold sensor was first covered by a spin-coated poly(allylamine) hydrochloride layer to provide hydrophilicity and further coated with a thin layer of CNC. Solutions of XG and allyl-functionalized XG at 15 μg mL −1 in pure water at pH 7 were injected allowing for depositing the polymers onto the cellulose surface for about 40 min. The evolution of Δf n /n and ΔD n was monitored for native XG and modified XG having different DS (DS = 0.10, 0.21 and 0.29) as a function of time (Fig. 6). For sake of clarity, the variations of frequency and energy dissipation for a single overtone (n = 3) are given to compare the adsorption features of the different samples.
The noticeable change of the frequency of resonance after each polymer injection proves that the polymer chains are adsorbed onto the cellulose surface (Fig. 6). More specifically, Δf n /n value (Table 4, Entries 1-4) is equal to − 17 Hz with unmodified XG and vary from − 10.6 to − 6.3 Hz with allyl-XG derivatives, highlighting the effect of the XG derivatization on the adsorption. As underpinned by the absence of noticeable shift of the signal upon addition of pure water in the chamber, the rinsing step do not Fig. 6 Normalized frequency (Δf n /n, white symbols) and dissipation (ΔD n , black symbols) changes for the overtone number n = 3 of the cellulose surfaces, exposed to unmodified XG (○ and •), and allyl-modified XG with DS = 0.10 (◊ and ♦), 0.21 (▲ and Δ), and 0.29 (□ and ■) solutions (15 μg/mL, at pH 7, in pure water), as a function of time at 25 °C. The arrows indicate the initial baseline (cellulosic substrate in water), the injection of the polymer solutions and the rinsing step (pure water) Allyl-XG (DS = 0.29) − 6.3 ± 1.1 0.9 ± 0.3 1.1 ± 0.2 trigger desorption of the polymer chains indicating that the adsorption process is irreversible under these conditions of analysis. In parallel, the injection of XG is accompanied by relatively small energy dissipation changes (Table 4, Entries 1-4. ΔD n ≈ 1 × 10 -6 ), which reflects the formation of a relatively rigid and non-viscoelastic layer. In these conditions, the mass of adsorbed polymers can be estimated through the Sauerbrey equation (Table 4, Δm). The calculated adsorbed mass, Δm of 3 mg m −2 (entry 1) is closed to the one reported by Eronen et al. (2011) and Villares et al. (2015) for XG of high molecular weight adsorbed onto cellulose nanofibrils and cellulose nanocrystals, respectively. Comparison of the frequency profiles for unmodified XG and derivatized XG clearly highlights the impact of the chemical modification on the adsorption of the chains. Adsorption seems to be less prominent in the case of the derivatized XG (see Δm values, Table 4), and this trend is being accentuated as the number of allyl residues along the XG backbone increases (entry 2, 3 and 4). All together, these changes in frequency of resonance demonstrate that the presence of allyl side groups affects the adsorption mode of the polysaccharide chains onto the cellulose surface. The presence of allyl moieties might cause steric hindrance along the XG backbone and the substitution of the hydroxyl groups by allyl groups probably impedes and partially impairs the formation of hydrogen bonds at the interface with the cellulose surface. Note that, even if the impact of the allyl functionalization is clearly visible, the adsorption of allyl-XG derivatives remains completely acceptable whatever the value of DS. It is also worth mentioning that these results should be compared with caution with the adsorption yields values resulting from the titration experiments as the concentrations of polysaccharide solutions used for QCM-D analysis are well below the critical overlap concentration C*. Villares et al. (2015) have shown that at a concentration > 6 mg L −1 , the XG chain adsorption is fast enough to hamper chains rearrangements thus favoring the saturation of the cellulose surface with the formation of polymer loops. The lower values of energy dissipation resulting from the deposition of allyl-XG, compared to unmodified XG, indicate a less hydrated character of the deposited layers, caused by the presence of apolar allyl residues which leads to the formation of hydrophobic zones onto the surface and tends to expulse water molecules from the layer. Therefore, the decrease of Δm values observed for the allyl-functionalized polymers probably stems not only from a modification of the interfacial interactions but also from a water molecules exclusion during the adsorption process.
In the case of CMC and allyl-CMC derivatives, QCM-D analyses revealed that the adsorption onto CNC layers was unfavored and that partial desorption occurred with allyl derivatives during the rinsing step (SI.8.a). This can be ascribed by the formation of electrostatic repulsions between the carboxylate groups of CMC and the negatively charged surface of the substrate composed of nanocrystals of cellulose (Kargl et al. 2012). As reported in the literature, the addition of CaCl 2 electrolyte is required to promote CMC adsorption on this type of cellulosic substrates (Liu et al. 2011). Keeping this in mind, we turned to the preparation of a neutral amorphous cellulose surface resulting from the spin-coating of cotton linters dissolved in dimethylacetamide/LiCl. In such experimental conditions, the deposited mass Δm after injection of unmodified CMC and allyl derivative (DS = 0.07) is high, between 2.5 and 2.7 mg m −2 and no desorption occurs during the rinsing stage (SI.8.b). The dissipation values (ΔD n ) is higher than 3.10 -6 , signature of a higher hydration level of the layer, compared to the XG-based analogous layers. This feature can be linked to (i) the hygroscopic character of CMC, enhancing the water uptake and the hydration degree, and (ii) the presence of negatively charged carboxylate groups (without electrolyte) providing additional intra and interchain repulsive electrostatic interactions which exalt its ability to swell in aqueous medium. Consistent with the titration data on Whatman papers given in Fig. 5, no noticeable difference in the adsorption response is observed between unmodified CMC and allyl-CMC owing to the low DS values of the derivative CMC. These encouraging results emphasize the possibility to adsorb CMC and its reactive derivatives onto lignocellulosic substrates.
Taking advantage of the intense Raman response of alkyne functions (at 2115 cm −1 ) (Sun et al. 2019), we finally examine the spatial distribution of the alkyne-functionalized CMC and XG chains sprayed onto Whatman paper by Raman confocal microscopy. Raman mappings were carried out on a surface of several hundred μm 2 (380 × 380 μm 2 ) at various depth levels with step sizes of 10 μm. As shown in Fig. 7a, c, the Raman images reveal the presence of the expected alkyne signals on the whole surface of the substrate, confirming that the paper sheet is completely covered with adsorbed CMC/XG chains. As no signal related to alkyne functions is observed between fibers, we believe that the polymer chains are adsorbed at the paper surface or inside the cellulose fibers since microscopy images did not show any organic matter between the fibers (Fig. 7c). The Raman surface mapping of the coated Whatman paper suggests some disparities and fluctuations in terms of alkyne concentration (see regions in red and yellow) emerge. However, these intensity gradients probably stem from the Raman detection itself, which can be interfered by the surface geometric features of the sample impacted by the non-planarity of the substrate or from a heterogeneous deposition of the polymer chains in the course of the spray process. Moreover, the analyzed samples were subjected to a drying step, which can generate a roughness able to disrupt the Raman signal, and thus the final detected intensity. To gain insight into the penetration of the polymer chains inside the cellulosic substrate, Raman mapping was further carried out in the depth. As evidenced in Fig. 7b, d, the response of the alkyne bonds remains detectable over a thickness ~ 40 μm, i.e. the diameter of one cellulose fiber, whatever the nature of the adsorbed alkyne-functionalized polymer. This diffusion of polysaccharide chains inside the substrate is an advantageous feature to promote the adhesion of the coated layers.

Conclusion
Herein, xyloglucan (XG) and carboxymethyl cellulose (CMC) have been chemically modified by means of ring opening reaction with propargyl glycidyl ether and allyl glycidyl ether in basic aqueous medium at low temperature. A deep scrutiny of the reaction parameters showed that the hydrolysis side reaction of the epoxy functions does not hamper the desired ligation and that a modulation of the DS can be reached by playing with the molar ratio of NaOH and reagent. XG is slightly more reactive than CMC and allyl and propargyl glycidyl ethers have roughly similar reactivity. The radical copolymerization in water of multifunctional allyl-derivatives with acrylamide and sodium acrylate led to the formation of hydrogels, confirming the reactivity of such polysaccharide derivatives. The depositions of XG and CMC onto Whatman paper and wood pine fibers through spray-drying of aqueous solutions of polysaccharides was confirmed by the significant rise of the adsorbed mass. The adsorption yields of allyland alkyne-derivatives were similar or higher than the ones observed for the XG and precursors proving that neither the nature of grafting moieties nor the DS alter the deposition of the polymer chains. Additional data from QCM-D confirmed that CMC and its derivatives had less affinity with nanocellulose surface because of the electrostatic repulsion whereas the affinity of XG for cellulosic surfaces was slightly reduced upon incorporation of apolar allyl residues along the chains. Finally, a topochemical mapping with alkyne derivatives highlighted a rather heterogeneous deposition of the polymer chains very likely related to the use of a lab spray process. Interestingly the polysaccharide chains were able to diffuse inside the cellulosic substrate. This new strategy allows for modifying cellulosic surface features in a simple, robust and ecofriendly way and paves the way for the design of new cellulosic materials.