Cooxidant-free TEMPO-mediated oxidation of highly crystalline nanocellulose in water

Abstract

Selective oxidation of C6 hydroxyls to carboxyls through 2,2,6,6,-tetramethylpiperidine-1-oxyl (TEMPO)-mediated oxidation, where the oxidizing species (TEMPO⁺) is generated by cooxidants, such as NaBrO, NaClO or NaClO₂, has become a popular way to modify the surfaces of nanocellulose fibrils in aqueous solutions. Employing highly crystalline nanocellulose from Cladophora sp. algae we demonstrate that the same degree of oxidation (D.O.) can be achieved within approximately the same time by replacing the cooxidants with electrogeneration of TEMPO⁺ in a bulk electrolysis setup. The D.O. is controlled by the oxidation time and the maximum D.O. achieved (D.O. 9.8%, 0.60 mmol g⁻¹ of carboxylic acids and 0 mmol g⁻¹ aldehydes) corresponds to complete oxidation of the surface-confined C6. This shows that TEMPO⁺ is not sterically hindered from completely oxidizing the fibril surface of Cladophora nanocellulose, in contrast to earlier hypotheses that were based on results with wood-derived nanocellulose. The oxidation does not significantly affect the morphology, the specific surface area (>115 m² g⁻¹) or the pore characteristics of the water-insoluble fibrous particles that were obtained after drying, but depolymerization corresponding to ∼20% was observed. For extensive oxidation times, the product recovery of water-insoluble fibrils decreased significantly while significant amounts of charge passed through the system. This could indicate that the oxidation proceeds beyond the fibril surface, in contrast to the current view that TEMPO-mediated oxidation is confined only to the surface.

---

Introduction

Cellulose is a linear polymer composed of β-1-4-linked D-anhydroglucose units and is almost inexhaustible as a raw material.^1,2 Nanocellulose has attracted much interest during the last decade due to its mechanical properties, biodegradability, wide functionalization possibilities and nanoscale features, such as large specific surface areas.³ The most commonly investigated forms of nanocellulose are the short (<300 nm) and highly crystalline cellulose whiskers and the longer (micrometer lengths) but less crystalline nanofibrillated cellulose (NFC), which both are extracted from wood pulp. There are other sources of nanocellulose as well, e.g. certain algae and cellulose-producing bacteria, and a number of excellent reviews cover the production and properties of nanocelluloses extracted from different native cellulose sources.^1,2,4–9 For the production of NFC from wood pulp, surface charges are nowadays introduced on the cellulose fibrils, in order to decrease the energy consumption that is required to disintegrate the cellulose fibers into nanofibers, or fibrils. Most commonly carboxyl groups are introduced by either carboxymethylation or 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) mediated oxidation. The same methods may also be used to functionalize any type of nanocellulose.^1,10 This is interesting as the introduced carboxyl groups can be utilized for further functionalization through e.g. covalent attachments^11–13 or electrostatic interactions^14,15 to produce novel cellulose-based functional nanomaterials.

An interesting form of nanocellulose can be extracted from Cladophora sp. green algae, which pollutes shores around the world. This nanocellulose is highly crystalline, in contrast to NFC, and has high aspect ratio (micrometer lengths), in contrast to the shorter rod-like cellulose whiskers from wood.^9,16,17 Our group has dedicated a great deal of research into finding new applications and developing functional materials based on this type of nanocellulose, including easily produced and functionalized cellulose beads,¹⁸ virus-removal filters,¹⁹ pharmaceutical excipients,^20,21 rheology enhancers,²² and electroactive and conducting composites²³ for environmentally friendly energy-storage devices²⁴ or hemodialysis.²⁵ We have further demonstrated that the nanocellulose is non-toxic, and that TEMPO-oxidized Cladophora nanocellulose promotes fibroblast adhesion and proliferation, making it interesting for tissue engineering applications.²⁶

The usefulness of nitroxyl radicals, e.g. TEMPO, for oxidation of alcohols has long been known.²⁷ These radicals are stable, non-toxic as well as non-mutagenic and can be used for mild and selective oxidation of primary alcohols as the oxidation of secondary alcohols is significantly slower.^27,28 Other examples include e.g. 4-acetamido-TEMPO,²⁹ 2-azaadamantane N-oxyl (AZADO),³⁰ or phtalimide-N-oxyl (PINO).³¹ The oxidation is carried out by the corresponding oxoammonium ion, e.g. TEMPO⁺, which then is reduced to its hydroxylamine form (TEMPOH). TEMPOH and TEMPO⁺ subsequently form TEMPO through comproportionation. This cycle is illustrated in Fig. 1, where a typical chemical procedure of TEMPO-mediated oxidation of cellulose is shown to the left; TEMPO⁺ is continuously generated in a catalytic cycle by cooxidants in aqueous solutions at pH 10. In the TEMPO/NaBr/NaClO oxidation system, the cooxidant NaClO oxidizes NaBr to the cooxidant NaBrO, which presumably oxidizes TEMPO to TEMPO⁺.

Fig. 1 TEMPO-mediated oxidation of cellulose C6 hydroxyl groups at pH 10 with the TEMPO/NaBr/NaClO system (left) or through electrogeneration (right) in a bulk electrolysis setup.

In the case of cellulose the hydroxyls at the C6 position are selectively oxidized to carboxyls via intermediate aldehydes (C6 highlighted in Fig. 1). In contrast to TEMPO-mediated oxidation of generated³² or mercerized³³ cellulose, the oxidation of native cellulose has been observed to generate mostly (>80%) water-insoluble products, which has been ascribed to the higher crystallinity of the latter substrate, and thereby a lower accessibility of reagents to the crystalline bulk cellulose.³⁴ Recently, it was found that the water-insoluble fraction of TEMPO-oxidized wood pulp could easily be disintegrated into individual fibrils of 3–5 nm widths.^35,36 Since then, TEMPO-mediated oxidation has become one of the most common pretreatments in the preparation of NFC from wood.⁵ Compared to alternative ways of introducing carboxyl groups, e.g. carboxymethylation, TEMPO-mediated oxidation has been shown to have lower environmental impact.³⁷ The TEMPO/NaBr/NaClO system has apart from wood pulp, and regenerated or mercerized pulp also been applied to nanocellulose from tunicate,¹⁰ bacteria,³⁸ and Cladophora algae.^20,26,35

Although the TEMPO/NaBr/NaClO system is widely applied, it has drawbacks. For example, cellulose depolymerization occurs as the intermediate aldehydes make the glycosidic bonds susceptible to β-elimination under the employed alkaline conditions.⁵ The TEMPO/NaClO/NaClO₂ system (at pH 6.8, 60 °C) reduced the extent of depolymerization but also resulted in a significantly lower number of carboxyl groups (∼0.8 vs. 1.5–1.7 mmol g⁻¹) with softwood pulp as the substrate,^39,40 showing that the oxidation system parameters significantly affect the oxidation rate and the degree of cellulose oxidation that is achieved. Another drawback is the stoichiometric amounts of halogen-containing oxidants that are used, which are particularly unwanted in industrial scale processes. Furthermore, the reaction medium can be reused after the cellulose has been removed, which is appealing from both economic and environmental perspectives, but additional NaClO has to be added and the resulting build-up of NaCl in the medium reduces the cellulose oxidation rate for each use.⁴¹

A greener route is to generate TEMPO⁺ electrochemically (Fig. 1), where the use of cooxidants is completely eliminated. Such an oxidation system could also be anticipated to better facilitate the reuse of the oxidation medium without loss in oxidation rates, as there is no build-up of NaCl. TEMPO-mediated oxidation of carbohydrates through electrogeneration of TEMPO⁺ has, however, received only limited attention. Those reports show a complete or nearly complete oxidation of water-soluble mono-, di- and polysaccharides, as well as of complete oxidation of mercerized cellulose into water-soluble products.^42,43 Regenerated cellulose fibers have, on the other hand, been oxidized through electrogeneration of 4-acetamido-TEMPO⁺ without loss of the fiber morphology.⁴⁴

There are two reports on the oxidation of native cellulose using electrogeneration.^29,45 In the most recent report the authors demonstrate an electrochemical method for comparing oxidation rates of different mediators and reaction conditions, involving cotton fabric in close contact with the working electrode.⁴⁵ Oxidation of the cellulose was confirmed through infrared spectroscopy but no data concerning the degree of oxidation (D.O.) was presented. In the second report, native softwood pulp was oxidized and resulted in oxidized fibrils with lower carboxyl content (∼1.0 mmol g⁻¹ vs. 1.7 mmol g⁻¹) and higher aldehyde content (∼0.2 mmol g⁻¹ vs. 0.07 mmol g⁻¹) as compared to the corresponding fibrils prepared using the TEMPO/NaBr/NaClO system.²⁹ This suggests that there may be limitations regarding the ability of TEMPO⁺ to oxidize native cellulose fibrils by themselves. Specifically, it has been hypothesized that TEMPO⁺ is sterically hindered by inter- and intramolecular hemiacetals formed between intermediate aldehydes and cellulose hydroxyls, thereby preventing the complete oxidation of the aldehydes to carboxyls.^46,47 In line with this it has been described that either NaBrO or NaClO, as well as NaClO₂, are, at least partly, responsible for completing the oxidation of intermediate aldehydes to carboxyls (illustrated in Fig. 1).^5,29

To be a viable option to the conventional TEMPO-processes involving cooxidants, it is important that the cellulose can be oxidized to the same degree through electrogeneration, and preferably within a similar time frame. In the present investigation we have systematically explored this possibility for the oxidation of highly crystalline native Cladophora nanocellulose and show that the method indeed fulfills both criteria. Highly crystalline cellulose was chosen in order to minimize uncertainties due to the excessive reactivity of disordered domains, thereby allowing comparison of our data with literature values.³⁵ We also demonstrate that significant depolymerization occurs during the process and that significant amounts of water-soluble products are formed if the oxidation is carried out for extensive periods of time, suggesting that TEMPO⁺ may in fact penetrate into the crystalline domains of the cellulose. Apart from the depolymerization, the oxidation had no large impact on the water-insoluble products in terms of morphology, specific surface areas or pore size distributions.

Experimental

Materials

Cladophora sp. algae nanocellulose was provided by FMC Biopolymers (USA) as a spray-dried powder (hemicellulose content <1%) and had a crystallinity index of 92%, as determined from X-ray diffraction data in earlier work.²⁶ TEMPO and other chemicals used were of reagent or analytical grade and used as received from commercial suppliers. Deionized water was used throughout all experiments.

TEMPO-mediated oxidation in a bulk electrolysis setup

TEMPO (150 mg, 1 mmol) was dissolved in 175 mL aqueous 0.1 M carbonate buffer (adjusted to pH 10.3) and 750 mg (4.6 mmol of glucose units) of Cladophora nanocellulose was added. A graphite working electrode (∼12 cm²) was placed in the main compartment of the electrochemical cell, while a counter electrode (coiled Pt-wire, ∼3 cm²) and reference electrode (3 M NaCl Ag/AgCl) were placed in separate compartments (sintered glass filters, 4–8 μm pores) containing only carbonate buffer. Electrolyses were carried out for 30 min to 72 hours under vigorous stirring with the potential controlled at 0.7 V vs. Ag/AgCl by an Autolab potentiostat (EcoChemie, The Netherlands). Ethanol was immediately added after the set oxidation times for quenching purposes, and the products were washed by centrifugation (at 2600g) with ethanol several times and then dialyzed in water for 3 days. The products were then collected in 0.01 M HCl and centrifuged, after which the solvent was changed to ethanol and then ether, and finally dried from ether. The products were white powders. The product recovery was obtained gravimetrically from the quantities of starting material and recovered product. For reference, the starting material (non-oxidized Cladophora nanocellulose) was mixed for 30 min in the carbonate buffer and then subjected to the same washing and drying treatments as the oxidized samples.

FTIR

FTIR spectra were collected on a Bruker Tensor 27 (Germany) with KBr pellets (1 wt% sample). The resolution was 4 cm⁻¹ and 100 scans were averaged. Two analyses were made with each sample. A rubberband background was subtracted from all spectra using the instrument software (Opus 7.0, Bruker, Germany) and normalized with respect to the absorption at 2897 cm⁻¹, corresponding to a C–H stretching vibration.⁴⁸

Conductometric titration for determination of carboxylic acid content

The carboxylic acid content in samples subjected to TEMPO-mediated oxidation and samples subsequently subjected to chlorite oxidation were determined through conductometric titrations. ∼100 mg of sample was dispersed in 60 mL 0.010 M NaCl (aq) through high-energy ultrasonication (VibraCell, USA) and the pH was adjusted to <3 through addition of concentrated HCl. The dispersion was equilibrated at 25 °C with nitrogen purging for 30 min prior to titrations. The dispersions were titrated under nitrogen atmosphere with addition of 0.020 mL 0.050 M NaOH (aq) every 30 seconds until pH ∼11. The carboxylic acid content was determined by making linear fits to the linear regions (corresponding to strong acid and base) and the plateau region (corresponding to weak acid, i.e. carboxylic acid) of the conductivity vs. volume titration curves. The amount of carboxylic acid was calculated from the intercepts of the fitted curves. Each sample was analyzed three times.

Oxime formation and determination of the combined aldehyde and ketone content by CHN elemental analysis

Samples were subjected to Schiff base reactions with hydroxylamine, converting any aldehydes (C6 position) and ketones (C2 and C3 position) to oximes, according to a previously described procedure.⁴⁹ Briefly, 100 mg of sample was added to 40 mL 0.01 M acetate buffer (pH 4.5) and mixed, after which 1.65 mL aqueous hydroxylamine solution (50 wt%) was added and the mixture was stirred at room temperature for 24 h. The product was collected and washed through centrifugation (4700g) with 0.01 M HCl and ethanol and finally dried. The product recovery was >95 wt% for all samples. CHN analysis was performed by Analytische Laboratorien (Lindlar, Germany).

Chlorite oxidation and determination of aldehyde content by conductometric titration

Aldehydes were oxidized to carboxylic acids according to a previously published protocol,⁴⁹ with minor modifications. In short, 500 mg of sample and 3.6 mg sodium chlorite was added to 100 mL 0.01 M acetate buffer (pH 4.5) and the reaction was allowed to proceed under stirring for 20 h at 40 °C. The product was collected and washed through centrifugation (4700g) in ethanol, then dialyzed for three days in water. The product was then collected in 0.01 M HCl through centrifugation after which the solvent was changed to ethanol and then ether, and finally dried from ether. The products were white powders and the product recovery was >93 wt% for all samples. The samples were titrated conductometrically according to the above protocol, and the aldehyde content was obtained by subtracting the amount of carboxylic acid obtained before chlorite oxidation from the amount obtained after chlorite oxidation.

Determination of ketone content

The ketone content was taken as the difference between the combined aldehyde and ketone content (from CHN analysis of oximes) and the aldehyde content (from titration of carboxylic acids after chlorite oxidation).

Faradaic yield

The faradaic yield was calculated according to eqn (1).where q_n and q_n−1 represent the charge required for the number of carboxyl and aldehyde groups of sample n and n−1, respectively. Q_n and Q_n−1 represents the amount of charge passed through the electrochemical system for sample n and n−1, respectively.

Intrinsic viscosity

The intrinsic viscosity was assessed for samples subjected to TEMPO-mediated oxidation as well as for samples subjected to chlorite oxidation following the TEMPO-mediated oxidation. The ASTM standard method D1795-96 (ref. 50) for determining the intrinsic viscosity of cellulose was followed, using an Ubbelohde viscometer (SI Analytics GmbH, Germany). Each sample was measured five times. The degree of polymerization was calculated as described in the standard method.

SEM

Scanning electron micrographs were recorded on a LEO1550 field emission SEM instrument (Zeiss, Germany). Samples were mounted on aluminum stubs with double-sided adhesive carbon tape and sputtered with Au/Pd to reduce charging effects.

Nitrogen sorption isotherms

Nitrogen sorption isotherms were acquired using an ASAP 2020 instrument (Micromeritics, USA). The BET specific surface area was assessed from the adsorption branch of the isotherm using the instrument software,⁵¹ while the BJH pore size distributions and total pore volumes were assessed from the desorption branch.⁵²

Results and discussion

Cellulose oxidation in a bulk electrolysis setup

Electrogeneration of TEMPO⁺ was carried out for different time periods (ranging from 30 minutes up to 72 hours) in an aqueous carbonate buffer (pH 10) under vigorous stirring in order to facilitate fast transport of TEMPO to the working electrode, and TEMPO⁺ from the electrode. A representative chronoamperogram for 72 hours of electrolysis in the presence of Cladophora nanocellulose is shown in Fig. 2. For reference, a chronoamperogram recorded in the absence of cellulose is also shown, as well as the corresponding charges. The current decreased rapidly during the first hours when cellulose was not present. This was contrasted by a nearly constant current at ∼19 mA during the first four hours with cellulose present, followed by a slower current decrease. The total charge over 72 hours was five times larger when cellulose was present compared to without cellulose. Thus, with cellulose present, TEMPO⁺ was generated to a greater extent, implying higher TEMPO⁺ consumption and cellulose oxidation.

Fig. 2 Representative chronoamperograms (filled symbols) and the corresponding chronocoulograms (open symbols) for the generation of TEMPO+ during 72 hours with and without cellulose present. Data acquisition was continuous and is represented by the solid lines. The symbols are only visual aids.

Degree of oxidation

The FTIR spectra of the water-insoluble products were characteristic of cellulose (full spectra are shown in Fig. S1 in ESI†), except in the region corresponding to a carbonyl stretching vibration (1732 cm⁻¹), shown in Fig. 3. In this region a gradual increase in the absorbance could be observed during the first four hours of electrolysis, after which the intensity remained constant. This shows that the fibrils were indeed oxidized. It also indicates that the extent of oxidation was controlled by the oxidation time, until a maximum D.O. was reached. The products were converted from the carboxylate to the carboxylic acid form during washing. Therefore the carboxylate content was low, as confirmed by the lack of an absorbance band at ∼1600 cm⁻¹.⁵³ The absorbance peak at 1640 cm⁻¹ corresponds to the O–H stretching vibration of adsorbed water.⁵³

Fig. 3 FTIR spectra in the region 1500–2100 cm⁻¹ of the recovered water-insoluble products after normalization with respect to the C–H stretching vibration at 2897 cm⁻¹.

The carboxylic acid content of the water-insoluble products was determined using conductometric titrations (Fig. 4 and Table 1) and was found to increase linearly to the maximum value of 0.59–0.60 mmol g⁻¹ during the first four hours of electrolysis, after which it became constant. This corresponds to a D.O. just below 10%.

Fig. 4 The amount of carboxyls and aldehydes determined from conductometric titrations and the combined amounts of aldehydes and ketones determined from CHN elemental analysis. Data points represent mean values and error bars correspond to the standard deviation (n = 3). Lines are only guide to the eye.

Table 1 Carboxylic acid, combined aldehyde and ketone, aldehyde, and ketone contents and the corresponding degrees of oxidation of the water-insoluble products

Oxidation time (h)	Carboxylic acid^a (μmol g^−1)	Combined aldehyde and ketone^b (μmol g^−1)	Aldehyde^c (μmol g^−1)	Ketone^d (μmol g^−1)	Carboxylic acid^e (mol%)	Aldehyde^e (mol%)	D.O.^f (mol%)
a From conductometric titrations. Values represent the mean ± 1 standard deviation (n = 3).b From CHN analysis of oximes.c The difference in carboxylic acid content before and after chlorite oxidation. Values represent the mean ± 1 standard deviation (n = 3).d The difference between the combined aldehyde and ketone content and the mean aldehyde content.e Based on the mean values.f The sum of the mol% carboxylic acid and aldehyde.
0	37 ± 2	<71	32 ± 12	<39	0.6	0.5	1.1
0.5	79 ± 7	107	95 ± 15	12	1.3	1.5	2.8
1	143 ± 13	107	110 ± 14	0	2.3	1.8	4.1
2	301 ± 9	107	93 ± 20	14	4.9	1.5	6.4
3	461 ± 10	114	88 ± 19	26	7.5	1.5	9.0
4	591 ± 10	<71	3 ± 10	<68	9.7	0	9.7
8	602 ± 7	<71	0	<71	9.8	0	9.8
24	595 ± 11	<71	0	<71	9.7	0	9.7
72	599 ± 21	<71	0	<71	9.8	0	9.8

---

The combined aldehyde and ketone content was determined by subjecting the samples to a Schiff base coupling reaction with hydroxylamine, where oximes were formed in place of any aldehydes and ketones. This allowed CHN elemental analysis to be used to quantify the combined aldehyde and ketone content. The detection limit of the analysis was 0.071 mmol g⁻¹ and the combined aldehyde and ketone contents of the starting material as well as samples subjected to oxidation for four hours or more were below this value (Fig. 4 and Table 1). The samples oxidized for 0.5 h to 3 h, on the other hand, contained ∼0.11 mmol g⁻¹ of aldehydes and ketones.

To determine purely the aldehyde content, aldehydes were oxidized to carboxylic acids using chlorite and titrated conductometrically. The aldehyde content was taken as the difference between the carboxylic acid content before and after chlorite oxidation. As can be seen in Fig. 4 and Table 1, significant amounts of aldehydes were only present in samples subjected to oxidation for 0.5 h to 3 h. For times equal to or longer than four hours no significant amounts of aldehydes could be detected.

The ketone content of samples oxidized for 30 minutes to three hours was found to be essentially 0 mmol g⁻¹ (Table 1). For the remaining samples the ketone content was found to be <0.071 mmol g⁻¹, but these assessments are limited by the fact that the detection limit of the CHN analysis was 0.071 mmol g⁻¹ and that no aldehydes were present in these samples. Still, this shows that ketones are only present in low amounts, if not completely absent, in all samples as would be expected due to the specificity of TEMPO⁺.

Cladophora nanocellulose with 0.52 mmol g⁻¹ of carboxyl groups and 0 mmol g⁻¹ of aldehydes were obtained using the TEMPO/NaBr/NaClO system in earlier work.³⁵ In the same work this was shown to correspond to the complete oxidation of the fibril surface. The previously reported values are close to the maximum values obtained in this work, viz. 0.6 mmol g⁻¹ of carboxyl groups and 0 mmol g⁻¹ of aldehydes. Thus, the current results show that nanocellulose fibrils can be oxidized to the same extent using electrogeneration as with the TEMPO/NaBr/NaClO system. Furthermore, the results show that all aldehydes present on the fibrils can be oxidized to carboxyls by TEMPO⁺. This suggests that no significant steric hindrance for TEMPO⁺ exists in terms of its ability to oxidize the aldehydes on Cladophora nanocellulose, contrary to what has been hypothesized earlier for wood-derived nanocellulose.^46,47

In agreement with the FTIR data (Fig. 3), the results also show that the D.O. is controlled by the oxidation time, which means that the D.O. can be amply controlled in future work when a certain D.O. may be desired. While complete surface oxidation of 750 mg cellulose was reached within 4 hours using the currently investigated unoptimized cooxidant-free setup, 3 hours were required for complete oxidation of 1 g of Cladophora sp. cellulose using the TEMPO/NaBr/NaClO system.³⁵ Thus, in terms of the time needed to reach full surface oxidation, the current electrochemical setup requires slightly longer time. It is expected that the oxidation time can be reduced by optimizing the electrochemical setup, possibly by using a similar methodology as recently presented.⁴⁵ This is however outside the scope of the current work.

The faradaic yield relates the charge required for the obtained D.O. of the samples to the total charge passed through the electrochemical system (see eqn (1)). As can be seen in Fig. 5, the faradaic yield during the first three hours is 60–70%, after which it decreases to 35% after four hours and then to 0% for longer oxidation times. This shows that a significant amount of charge is consumed in reactions resulting in other products than water-insoluble oxidized nanocellulose fibrils. One such side reaction is likely the oxidation of cellulose or cellulose derivatives that are, or become, water-soluble, and therefore are not detected.

Fig. 5 Faradaic yield as a function of oxidation time.

Depolymerization during oxidation and formation of water-soluble products

To evaluate if any depolymerization occurred during oxidation, the intrinsic viscosity was determined for the water-insoluble products and the results are shown in Fig 6. The average intrinsic viscosity of the water-insoluble products decreased by ∼20% over the initial 3–4 hours of electrolysis, corresponding to a degree of polymerization (D.P.) decrease from ∼740 to ∼570.⁵⁰ This decrease was coinciding with the D.O. increase up to the maximum D.O. The depolymerization may be attributed to β-elimination, occurring during oxidation because of the alkalinity of the solution (pH 10) and the presence of aldehydes.⁴⁰ As a control experiment to ensure that the depolymerization in fact occurred during the oxidation and not during the course of dissolution and/or viscosity measurements in the alkaline cupriethylenediamine hydroxide solution, the products were subjected to chlorite oxidation in order to convert all aldehydes to carboxylic acids. Thereby the possibility of β-elimination was eliminated. The viscosity values obtained from these samples were nearly identical to the original values of the samples only subjected to TEMPO-mediated oxidation (Fig. 6), showing that the depolymerization indeed took place during the TEMPO-mediated oxidation.

Fig. 6 Mean intrinsic viscosities determined after TEMPO-mediated oxidation and after subsequent oxidation with chlorite (n = 5, error bars = 1 s.d. hidden behind the symbols, left y-axis), as well as the product recovery after TEMPO-mediated oxidation (right y-axis). Lines are only guide to the eye.

An unexpected progressive increase in intrinsic viscosity was observed for samples oxidized for more than four hours (Fig. 6), i.e. for the samples that had reached the maximum D.O. This may be understood by observing the product recovery of the water-insoluble products after drying, also shown in Fig. 6. During the first 4 h of oxidation the product recovery was 81–86%, whereas it gradually decreased to 51% after 72 hours, inferring that the amount of water-soluble products increased during the same time period. Since the intrinsic viscosity reflects some average value of the cellulose chain length distribution, the observed increase indicates that the distribution of the water-insoluble products shifts towards longer chains as the fraction of water-soluble cellulose increases, as is illustrated in Fig. 7. It should be emphasized that the actual distributions, such as kurtosis, skewness, and multi-modality, of the recovered products are unknown at this point and will require further studies.

Fig. 7 Hypothetical illustration of the shifts of the chain length distribution and the progressively decreasing amounts of water-insoluble oxidized fibrils following TEMPO-mediated oxidation.

Cellulose oxidation by TEMPO⁺ is considered to be surface-confined and unable to penetrate into the crystalline domains of the fibrils.^5,35 In the present work, approximately 1650 C of charge passed through the system during the last 68 hours of a 72 hour oxidation when cellulose was present, i.e. after full oxidation of the fibril surface. Even when subtracting the charge passed for only oxidation of TEMPO to TEMPO⁺ without any cellulose present during 72 hours (∼300 C), this would correspond to oxidation of approximately 3.5 mmol of hydroxyl groups to carboxyl groups. This implies that the oxidation proceeds beyond the surface of the fibrils and generates water-soluble products, which is reflected in the product recovery in Fig. 6. Given that the presently investigated cellulose is highly crystalline, the results could suggest that TEMPO⁺ in fact is capable of oxidizing crystalline cellulose if the oxidation is carried out for sufficiently long times.

Morphology, specific surface area, pore size distributions and pore volumes

SEM was used to investigate if the morphology of the nanocellulose fibrils had been significantly affected by the oxidation and/or the alkaline solution in which the oxidations were carried out. Micrographs at different magnifications (3, 50, and 500kX) for the starting material and for a sample subjected to four hours of oxidation (D.O. 9.7%) are shown in Fig. 8. The oxidized sample serves as an example for all oxidized samples as no significant differences were observed between these samples. Apart from the different particle sizes of the unmodified (Fig. 8a) and oxidized (Fig. 8b) nanocellulose samples, no significant differences were observed. In Fig. 8d it can be seen that the powder particles of the oxidized samples were fibrous and porous, much like the starting material (Fig. 8c). The fibril widths, observable in Fig. 8e and f, were approximately 35–40 nm in both the starting material and the oxidized samples. The fibril width of Cladophora nanocellulose has previously been found to be 25–35 nm through atomic force microscopy⁵⁴ and estimated to be 29–38 nm from density and surface area data.^54,55 The slightly larger fibril widths observed in the present investigation were most likely due to the Au/Pd coating required to avoid charging effects.

Fig. 8 Scanning electron micrographs of the starting material (a, c and e) and a sample oxidized for 4 hours (b, d and f) at 3 (a and b), 50 (c and d), and 500 (e and f) kX magnification.

Typically, the fibril lengths of nanocellulose from Cladophora and other algae are described to be >1 μm, and to our knowledge there are no reports of more precise values.² Therefore we were interested in trying to obtain a more accurate estimate of the fibril lengths, but it is difficult to evaluate the lengths of individual fibrils from the SEM micrographs shown in Fig. 8c and d because the high degree of fibril entanglement. Still, parts of fibrils corresponding to 1.5–2.0 μm in length are clearly visible. The fibril lengths may also be evaluated from the D.P. values presented above, with the assumption that the relationship between the length-weighted average fibril length and D.P. of NFC established by Shinoda et al.⁵⁶ is valid also for Cladophora nanocellulose. In this case, comparable fibril lengths are obtained, viz. ∼2.4 and ∼1.7 μm for D.P. values of 740 and 570, respectively. Furthermore, Kim et al.⁴⁹ have published TEM-micrographs of individualized Cladophora fibrils, prepared in a similar way as the starting material used in the present work. Although not described by the authors, the micrographs showed fibrils ranging in length from ∼0.8 μm to ∼2.5 μm where the longest fibrils spanned over the entire micrograph, making the upper limit uncertain. Still, this indicates that the average fibril length estimates in the present work are reasonable.

The BJH pore size distributions and total pore volumes as well as BET specific surface areas of the powders were evaluated from nitrogen sorption isotherms (isotherms are presented in Fig. S2 in ESI†). The BJH pore size distributions of selected powder samples are shown in Fig. 9, and the pore size characteristics and specific surface areas for all samples are tabulated in Table 2. All samples display wide pore size distributions, spanning the entire mesopore region (2–50 nm) and well into the macropore region (>50 nm). Compared to the starting material the pore size distributions are shifted towards slightly larger pores following oxidation (Fig. 9), irrespective of D.O. or oxidation time. The total BJH pore volume generally decreased slightly with increased oxidation but was not related to the oxidation duration. However, none of the isotherms (see Fig. S2 in ESI†) displayed limiting adsorption at high relative pressures, indicating that macropores larger than ∼100 nm in diameter cannot be excluded.⁵⁷ The SEM micrographs (Fig. 8) not only reveal larger macropores but also show that the structure, at least on the surface, is dominated by meso- and macropores <100 nm in width. Additionally, the large surface area of the starting material was retained following oxidation, irrespective of the D.O. or oxidation duration.

Fig. 9 BJH pore size distributions for samples with different D.O.

Table 2 BJH pore size characteristics and BET specific surface area

Oxidation time (h)	D.O. (mol%)	BJH pore size distribution mode (nm)	BJH total pore volume^a (cm^3 g^−1)	BET specific surface area (m^2 g^−1)
a For pores less than 108 nm at p/p0 = 0.98.
0	1.1	42	1.01	116
0.5	2.8	49	1.03	126
1	4.1	46	0.91	126
2	6.4	49	0.90	129
3	9.0	45	0.87	126
4	9.7	46	0.79	130
8	9.8	52	0.85	132
24	9.7	52	0.77	126
72	9.8	57	0.78	122

---

Conclusions

TEMPO-mediated oxidation of highly crystalline nanocellulose from Cladophora sp. algae was successfully carried out without cooxidants in a water-based system using electrochemical generation of TEMPO⁺. Complete fibril surface oxidation, corresponding to 0.60 mmol g⁻¹ of carboxylic acids and 0 mmol g⁻¹ aldehydes, was achieved within four hours of oxidation. Both the maximum D.O. and the required oxidation time are comparable to values achieved with conventional TEMPO-process involving cooxidants, showing that NaBrO, NaClO and NaClO₂ are unessential for the Cladophora nanocellulose oxidation process and that TEMPO⁺ is not sterically hindered from completely oxidizing the fibrils. While the fibril morphology, specific surface area and pore characteristics were not significantly affected by the oxidation, depolymerization corresponding to ∼20% occurred during the initial four hours of oxidation. For longer oxidation times, the product recovery decreased due to formation of water-soluble oxidation products while large amounts of charge passed through the electrochemical system without any further detectable oxidation of the recovered water-insoluble products. This could indicate that TEMPO-mediated oxidation is not restricted to merely the fibril surface.

Acknowledgements

The Bo Rydin Foundation, The Carl Trygger Foundation, The Swedish Foundation for Strategic Research (SSF) and The Knut and Alice Wallenberg Foundation are gratefully acknowledged for the financial support of this work. One of the authors (AM) is Wallenberg Academy Fellow and thanks the Knut and Alice Wallenberg Foundation for their support.

References

1. Y. Habibi, L. A. Lucia and O. J. Rojas, Chem. Rev., 2010, 110, 3479–3500.
2. R. J. Moon, A. Martini, J. Nairn, J. Simonsen and J. Youngblood, Chem. Soc. Rev., 2011, 40, 3941–3994.
3. D. H. Milanez, R. M. Do Amaral, L. I. L. De Faria and J. A. R. Gregolin, Mater. Res., 2013, 16, 635–641.
4. S. J. Eichhorn, A. Dufresne, M. Aranguren, N. E. Marcovich, J. R. Capadona, S. J. Rowan, C. Weder, W. Thielemans, M. Roman, S. Renneckar, W. Gindl, S. Veigel, J. Keckes, H. Yano, K. Abe, M. Nogi, A. N. Nakagaito, A. Mangalam, J. Simonsen, A. S. Benight, A. Bismarck, L. A. Berglund and T. Peijs, J. Mater. Sci., 2010, 45, 1–33.
5. A. Isogai, T. Saito and H. Fukuzumi, Nanoscale, 2011, 3, 71–85.
6. D. Klemm, F. Kramer, S. Moritz, T. Lindström, M. Ankerfors, D. Gray and A. Dorris, Angew. Chem., Int. Ed., 2011, 50, 5438–5466.
7. N. Lavoine, I. Desloges, A. Dufresne and J. Bras, Carbohydr. Polym., 2012, 90, 735–764.
8. I. Siró and D. Plackett, Cellulose, 2010, 17, 459–494.
9. A. Mihranyan, J. Appl. Polym. Sci., 2011, 119, 2449–2460.
10. Y. Habibi, H. Chanzy and M. Vignon, Cellulose, 2006, 13, 679–687.
11. R. K. Johnson, A. Zink-Sharp and W. G. Glasser, Cellulose, 2011, 18, 1599–1609.
12. J. Araki, M. Wada and S. Kuga, Langmuir, 2000, 17, 21–27.
13. A. P. Mangalam, J. Simonsen and A. S. Benight, Biomacromolecules, 2009, 10, 497–504.
14. P. Sarrazin, L. Valecce, D. Beneventi, D. Chaussy, L. Vurth and O. Stephan, Adv. Mater., 2007, 19, 3291–3294.
15. L. Wågberg, G. Decher, M. Norgren, T. Lindström, M. Ankerfors and K. Axnäs, Langmuir, 2008, 24, 784–795.
16. A. Mihranyan, A. P. Llagostera, R. Karmhag, M. Strømme and R. Ek, Int. J. Pharm., 2004, 269, 433–442.
17. M. Strømme, A. Mihranyan and R. Ek, Mater. Lett., 2002, 57, 569–572.
18. J. Lindh, D. O. Carlsson, M. Strømme and A. Mihranyan, Biomacromolecules, 2014, 15, 1928–1932.
19. G. Metreveli, L. Wågberg, E. Emmoth, S. Belák, M. Strømme and A. Mihranyan, Adv. Healthcare Mater., 2014, 3, 1546–1550.
20. D. O. Carlsson, K. Hua, J. Forsgren and A. Mihranyan, Int. J. Pharm., 2014, 461, 74–81.
21. A. Mihranyan, M. Strømme and R. Ek, Eur. J. Pharm. Sci., 2006, 27, 220–225.
22. A. Mihranyan, K. Edsman and M. Strømme, Food Hydrocolloids, 2007, 21, 267–272.
23. D. O. Carlsson, M. Sjödin, L. Nyholm and M. Strømme, J. Phys. Chem. B, 2013, 117, 3900–3910.
24. G. Nyström, A. Razaq, M. Strømme, L. Nyholm and A. Mihranyan, Nano Lett., 2009, 9, 3635–3639.
25. N. Ferraz, D. O. Carlsson, J. Hong, R. Larsson, B. Fellström, L. Nyholm, M. Strømme and A. Mihranyan, J. R. Soc., Interface, 2012, 9, 1943–1955.
26. K. Hua, D. O. Carlsson, E. Alander, T. Lindstrom, M. Strømme, A. Mihranyan and N. Ferraz, RSC Adv., 2014, 4, 2892–2903.
27. A. E. J. De Nooy, A. C. Besemer and H. Van Bekkum, Synthesis, 1996, 1153–1174.
28. G. Sosnovsky, J. Pharm. Sci., 1992, 81, 496–499.
29. T. Isogai, T. Saito and A. Isogai, Cellulose, 2011, 18, 421–431.
30. S. Takaichi and A. Isogai, Cellulose, 2013, 20, 1979–1988.
31. G. Biliuta, L. Fras, M. Drobota, Z. Persin, T. Kreze, K. Stana-Kleinschek, V. Ribitsch, V. Harabagiu and S. Coseri, Carbohydr. Polym., 2013, 91, 502–507.
32. C. Tahiri and M. R. Vignon, Cellulose, 2000, 7, 177–188.
33. A. Isogai and Y. Kato, Cellulose, 1998, 5, 153–164.
34. T. Saito, M. Yanagisawa and A. Isogai, Cellulose, 2005, 12, 305–315.
35. Y. Okita, T. Saito and A. Isogai, Biomacromolecules, 2010, 11, 1696–1700.
36. T. Saito, Y. Nishiyama, J. L. Putaux, M. Vignon and A. Isogai, Biomacromolecules, 2006, 7, 1687–1691.
37. Q. Li, S. McGinnis, C. Sydnor, A. Wong and S. Renneckar, ACS Sustainable Chem. Eng., 2013, 1, 919–928.
38. C. Lai, L. Sheng, S. Liao, T. Xi and Z. Zhang, Surf. Interface Anal., 2013, 45, 1673–1679.
39. T. Saito, M. Hirota, N. Tamura, S. Kimura, H. Fukuzumi, L. Heux and A. Isogai, Biomacromolecules, 2009, 10, 1992–1996.
40. A. Potthast, T. Rosenau and P. Kosma, in Polysaccharides II, ed. D. Klemm, Springer, Berlin Heidelberg, 2006, vol. 205, ch. 99, pp. 1–48.
41. L. Mao, P. Ma, K. Law, C. Daneault and F. Brouillette, Ind. Eng. Chem. Res., 2010, 49, 113–116.
42. P. Parpot, K. Servat, A. P. Bettencourt, H. Huser and K. B. Kokoh, Cellulose, 2010, 17, 815–824.
43. K. Schnatbaum and H. J. Schäfer, Synthesis, 1999, 864–872.
44. T. Isogai, T. Saito and A. Isogai, Biomacromolecules, 2010, 11, 1593–1599.
45. Y. Jin, K. J. Edler, F. Marken and J. L. Scott, Green Chem., 2014, 16, 3322–3327.
46. T. Saito and A. Isogai, Biomacromolecules, 2004, 5, 1983–1989.
47. T. Saito and A. Isogai, Colloids Surf., A, 2006, 289, 219–225.
48. M. Åkerholm, B. Hinterstoisser and L. Salmén, Carbohydr. Res., 2004, 339, 569–578.
49. U. J. Kim, S. Kuga, M. Wada, T. Okano and T. Kondo, Biomacromolecules, 2000, 1, 488–492.
50. ASTM, Standard D1795–96(2007)e1: Standard Test Method For Intrinsic Viscosity of Cellulose, ASTM International, 1996.
51. S. Brunauer, P. H. Emmett and E. Teller, J. Am. Chem. Soc., 1938, 60, 309–319.
52. E. P. Barrett, L. G. Joyner and P. P. Halenda, J. Am. Chem. Soc., 1951, 73, 373–380.
53. C. Eyholzer, N. Bordeanu, F. Lopez-Suevos, D. Rentsch, T. Zimmermann and K. Oksman, Cellulose, 2010, 17, 19–30.
54. R. Ek, C. Gustafsson, A. Nutt, T. Iversen and C. Nyström, J. Mol. Recognit., 1998, 11, 263–265.
55. C. Gustafsson, H. Lennholm, T. Iversen and C. Nyström, Drug Dev. Ind. Pharm., 2003, 29, 1095–1107.
56. R. Shinoda, T. Saito, Y. Okita and A. Isogai, Biomacromolecules, 2012, 13, 842–849.
57. R. Pierotti and J. Rouquerol, Pure Appl. Chem., 1985, 57, 603–619.

---

Footnote

† Electronic supplementary information (ESI) available: FTIR spectra and N₂ sorption isotherms. See DOI: 10.1039/c4ra11182f

---