One-pot preparation of bi-functional cellulose nanofibrils

Successful carboxymethylation and preserved carboxylation after functionalisation was confirmed by conductometric titration. The targeted total charge was 600 μeq g⁻¹, and as seen in Table 1, the functionalised samples, CM-ene-fibres and CM-yne-fibres, maintained their total charge density after the functionalisation step. X-ray diffraction and CP/MAS ¹³C-NMR spectroscopy were utilised to evaluate any changes in crystallinity between the different samples (Table 1 and Figure S3). There is a discrepancy between the estimates of the degree of crystallinity obtained by CP/MAS ¹³C-NMR and XRD (Park et al. 2010), both in absolute values and in trends. The degree of crystallinity by XRD is systematically higher than corresponding results obtained by NMR. Both techniques indicate that there is no significant change in the degree of crystallinity as a result of the chemical modifications. Differences in the absolute values of crystallinity assessments obtained by XRD and NMR results is not uncommon, however the absence of common trends in XRD and NMR results deserves some attention. The chemical modification performed in this study primarily attach substituents to polymers that constitute the cellulose fibril surface, the cellulose particle phase boundary. NMR estimates of the degree of crystallinity are calculated based on the surface-to-volume ratio of the cellulose fibrils (Nocanda et al. 2007; Wickholm et al. 1998). The surface-to-volume ratio is sensitive to any changes in the signal intensity originating from polymers located at the cellulose fibril surfaces, why it is inherently sensitive to chemical changes occurring at the cellulose fibril surfaces. It can be argued that the fibril surface polymers, located at the cellulose particle phase boundary, are not part of a complete unit cell, why they could be considered a contribution to the amorphous (non-crystalline) signal in XRD, a situation that would most likely prevail also after a partial chemical modification of such polymers. This could explain the lack of consistent trends in the NMR and XRD data on crystallinity. Nonetheless, both NMR and XRD results indicate that no remarkable change in crystallinity is seen due to the functionalisation.

In the case of alkynation, successful functionalisation was confirmed by FT-IR in transmission mode utilizing KBr-pellets, by detecting the sharp but weak peak around 2120 cm⁻¹, ascribed to the carbon–carbon alkyne bond stretch vibration (Socrates 2001). Although moderately sharp, the peak was apparent in the CM-yne-fibre sample (Fig. 1a), confirming a successful functionalisation.

Azide-alkyne Husigen cycloaddition employing azide-functionalised disperse red 13 (DR13-N₃) was also conducted to further establish alkynation (Fig. 1b, c) and the strong red colour of the alkynated sample (Fig. 1c) and the absence of colour in the neat sample (Fig. 1b) affirm a successful alkynation.

Attempts to directly verify allylation were ineffective since the characteristic alkene bond stretch or bending vibrations (3010–3100 cm⁻¹, 1620–1680 cm⁻¹ or 675–1000 cm⁻¹) (Socrates 2001) were all hidden in the characteristic peaks attained from the carboxymethylated cellulose fibres (Fig. 2). However, when looking in detail at the region 1800–1500 cm⁻¹ (Fig. 2) it is evident that a new peak has appeared at 1739 cm⁻¹, anticipated to be attributed to a carbonyl bond stretch of an ester (Socrates 2001). The appearance of this new peak is possibly a result of the formation of an allyl ester at the carboxylate group. This may occur via an S_N1-reaction due to the dissociation of the bromide from the allyl bromide, allowing the carboxylate group to react with the allyl (Sykes 1986). However, the formation of allyl esters during the etherification is a side-reaction, and is not considered to be a competing reaction since no significant difference was noted in the total charge density between CM-fibres and CN-ene-fibres (Table 1), which would have been the inevitable result of a competing reaction.

Alternately, a thiol-ene coupling reaction using methyl 3-mercaptopropionate was employed to determine the presence of alkenes since a new peak attributed the carbonyl bond stretch vibration for the ester should appear approximately around 1730 cm⁻¹ (Socrates 2001) after successful coupling, distinguishable from the carboxylate group bond stretch vibration of the carboxymethylated cellulose fibres at 1580 cm⁻¹. It was, however, anticipated that the ester would also physically adsorb to the reference sample (CM-fibre) and not only covalently couple with the allylated samples. Hence, a blank reaction with only methyl-3-mercaptopropionate was conducted to estimate the amount of adsorption (Fig. 3a). The results show that 3-mercaptopropionate is adsorbed to both CM-fibres and CM-ene-fibres in small amount and that the adsorption is seemingly equivalent in both samples. In Fig. 3b, it is observed that a new peak has appeared at 1725 cm⁻¹ in both samples but the ratio between the ester carbonyl (1725 cm⁻¹) and the cellulose carbonyl (1580 cm⁻¹) is greater for the CM-ene-fibre sample indicating that this sample comprise of both adsorbed and covalently coupled ester.

The samples were analysed by XPS to determine the degree of substitution (DS) by analysing the samples after coupling with methyl 3-mercaptopropionate or DR13-N₃ and calculating the content of coupled sulphur onto the CM-ene-fibres (Figure S5) and the nitrogen content, originating from DR13-N₃, onto the CM-yne-fibres (Figure S6), respectively. The data obtained were given as atomic percentage and hence, these were converted to weight percentage. Using Eq. (2), the DS of allyl and alkyne-functionality were calculated to be 0.016 and 0.0018, respectively. These degrees of substitution may seem low but the mild conditions need to be taken into account when evaluating the effect of the etherification. As previously studied and discussed (Mangiante et al. 2012), the reaction parameters are governed by the reaction medium, the NaOH weight ratio, the ratio between the functional reagent and the anhydroglucose units (F/AGU) and the reaction time. However, a prerequisite for our study was to employ identical reaction conditions as commonly used for carboxymethylation of cellulose fibres targeting a fibre charge of 600 µeq g⁻¹, i.e. a purely alcoholic medium, approximately 0.7 wt% NaOH and not more than 60 min for the subsequent reaction in order to not implicate the targeted charge. Hence, the discussion is hinged on these postulated conditions.

The rationale behind the reaction time was the incomplete understanding how the reaction time would influence the targeted charge, hence a subsequent reaction time of 60 min was added for the functionalisation and studied. In this regard, our choice of reaction time proved to be reasonable since the apparent charge was not affected by the extended reaction time. Further, the carboxymethylation protocol is conducted in a purely alcoholic medium. On the contrary to the previous work (Mangiante et al. 2012) the absence of water as a co-solvent in our system did not give rise to de-mixing issues for NaOH, which in turn was reported as a probable cause for the major alterations of the crystallinity when a high alcohol/water ratio was used. Instead, our analyses of the changes in crystallinity showed no major alterations of the crystallinity. The F/AGU ratio was chosen with a sensible environmental approach, and a higher DS would reasonably have been achieved with a higher F/AGU ratio (Mangiante et al. 2012). A DS of approximately 0.02 for the CM-F-fibres should be considered as a good value considering the restrictions in reaction parameters.

The neat and functionalised fibres were further treated mechanically to disintegrate the fibres into fibrils, followed by sonication and centrifugation to obtain stable aqueous dispersions. The surface charge of the fibrils was determined by polyelectrolyte titration (Table 2). The surface charges are lower compared to the total charge density of the fibres (Table 1), however, this is expected since it is assumed that although stable dispersions are obtained not all fibrils are completely disintegrated and small aggregates remaining after centrifugation will lower the surface charge value compared to the apparent total charge density value.

AFM was used to analyse the morphology of the nanofibrillar samples (Figure S2), and the height images do not indicate any noticeable changes of the morphology when comparing the neat and functionalised nanofibrils.

Films from the cellulose nanofibrillar dispersions were made in order to evaluate if the hydrophobic functionalities would increase the contact angle (CA) compared to the neat CNF film. As seen in Table 3, the initial CA measured (10 s) for the functionalised samples had increased with roughly 20%, indicating a more hydrophobic surface compared to the neat sample. Moreover, the water droplet prevailed after 30 and 60 s for both the functionalised samples, although with a decreasing angle, while the water droplet on the neat sample was adsorbed in less than 30 s.

To confirm maintained functionalisation after mechanical treatment aerogels of CM-yne-CNF were made and then subjected to the azide-alkyne Husigen cycloaddition employing DR13-N₃. As seen in Fig. 4, the reference sample (CM-CNF) has not been stained with the dye, whilst the CM-yne-CNF sample has been coloured. The colour is not as uniform and solid as for the fibre sample but this is considered more a consequence by the difficulty to homogeneously soak the dry aerogel with solvents and reagents and not due to less functionality sites compared to the fibres.

The maintained presence of alkenes in CM-ene-CNF aerogels was indirectly confirmed by employing the same thiol-ene reaction used for the CM-ene-fibre samples, and blank reactions were also conducted to display the non-covalent adsorption of the methyl-3-mercaptopropionate (Fig. 5a). Similarly, as with the fibre coupling the bond stretch vibration attributed to the carbonyl ester at 1725 cm⁻¹ is present in both samples, displaying yet again the physical adsorption of the ester but as with the fibre sample the ratio between the ester carbonyl (1725 cm⁻¹) and the CM-cellulose carbonyl (1580 cm⁻¹) is greater for CM-ene-fibril, which indirectly demonstrates the presence of alkenes (Fig. 5b).

The stability of the CM-F-CNF compared to CM-CNF was further evaluated by DLS since the mobility of the dispersed fibrils can be expressed by the decay rate of the normalized intensity function (Tables S7 and S2). It is observed that the added functionality had not significantly changed the mobility, i.e. dispersion stability, nor the size of the fibrils, ensuring that stable dispersions are achievable with an added functionality.