Cellulose from the green macroalgae Ulva lactuca: isolation, characterization, optotracing, and production of cellulose nanofibrils

Hydrogen peroxide (H₂O₂, 30%, CAS nr: 7722-84-1), hydrochloric acid (37%, CAS nr: 7647.01-0), sodium acetate (99%, CAS nr: 127-09-3), sodium hydroxide (NaOH, 99%, CAS nr: 1310-73-2), lithium chloride (99%, CAS-nr: 7447-41-8), dimethylacetamide (99%, CAS-nr: 127-19-5), α-cellulose, from Spruce wood (99%, CAS-nr: 9004-34-6), L-(+)-arabinose (99%, CAS nr: 5328-37-0), D-(+)-glucose (99%, CAS nr: 50-99-7), D-(+)-mannose (99%, CAS nr: 3458-28-4), D-(+)-galactose (99%, CAS nr: 59-23-4), D-(+)-xylose (99%, CAS nr: 58-86-6), L-(+) rhamnose (99%, CAS nr: 10030-85-0, D-(+)-glucoronic acid (98%, CAS nr: 6556-12-3) and D-(+) galacturonic acid monohydrate (98%, CAS-nr: 91510-62-2) were all purchased from Sigma Aldrich. Ethanol (99%, CAS nr: 64-17-5) was purchased from VWR Chemicals. Iduronic acid (99%, CAS nr: 2073-35-0,61199-83-5) was purchased from CarboSynth.

Ulva lactuca were collected from Inre Vattenholmen (58° 52′ 37.4″ N 11° 6′ 52.1″ E) and brought back to the lab within 2 h of collection. Collected Ulva lactuca were rinsed several times in natural seawater to remove grazers and loose epiphytes. The seaweeds were placed into cultivation tanks (90 L) under a neutral light cycle (16 h daylight, 8 h darkness) at a light intensity of 140 µE m⁻² s⁻¹. The light source was an INDY66 LED 60 W 4000 K 6000 lm. The seaweeds continuously received filtered seawater that was passed through 1 µm filters. No additional medium or chemicals were added to the water. The natural seawater used in the flow-through system was pumped in from the bay outside the Tjärnö Marine Laboratory (58° 52′ 36.4″ N 11° 6′ 42.84″ E). Thus, the salinity and temperature fluctuated depending on the prevailing weather and seasonal conditions. During the cultivation period (March–May 2018), the salinity ranged from 11 to 25 PSU and the temperature increased from 4 to 19 °C.

The extraction of the cellulose was performed using the method suggested by Siddhanta et al. (2009) with some modifications to avoid the use of sodium hypochlorite and methanol. The extraction procedure is schematically shown in Fig. 1. Freeze-dried Ulva lactuca (50 g) was ground to a fine powder using a Bosch MKM6003 coffee miller (Clas Ohlson, Sweden) and subjected to Soxhlet extraction in EtOH (85%) for 24 h at 120 °C to remove pigments and fatty acids. The liquid phase was discarded and the insoluble fraction was washed several times with EtOH (99%) and dried in an oven at 37 °C for 16 h. After drying, the sample was further treated to remove residual green pigments and other colored impurities by suspending the sample in 400 mL of 4% H₂O₂ followed by heating of the mixture to 80 °C for 16 h. After cooling to room temperature, the mixture was centrifuged at 5000 rpm for 15 min and the supernatant was discarded. The insoluble fraction was washed several times with water until the pH of the washing water was 7 and was then suspended in 400 mL of 0.5 M NaOH. The mixture was kept at 60 °C for 16 h. After cooling to room temperature, the mixture was centrifuged at 5000 rpm for 15 min and the supernatant was discarded. The insoluble fraction was washed several times with water until the pH of the washing water was 7 and the insoluble fraction was then suspended in 200 mL of 5% (v/v) HCl. The mixture was heated to boiling and then cooled to 30 °C and kept at this temperature for 16 h. The solution was centrifuged at 5000 rpm for 15 min and the supernatant was discarded. The insoluble fraction containing cellulose was washed several times with water until the pH of the washing water was 7, collected and freeze-dried.

Freeze-dried extracted cellulose (0.5 g) was dissolved in 100 mL dimethylacetamide (DMAc) containing 8% (w/v) LiCl and stirred at 80 °C for 16 h until the complete dissolution of the cellulose. The solution was cooled to room temperature and the dissolved cellulose was regenerated by the dropwise addition of LiCl/DMAc solution to 100 mL of water. The mixture was centrifuged at 5000 rpm for 15 min and the supernatant was discarded. The insoluble fraction containing the regenerated cellulose was washed with water, collected and freeze-dried.

Cellulose was extracted as described above, but instead of freeze drying, the extracted cellulose was kept as a suspension in water and diluted to 250 mL with distilled water. The resulting slurry was mechanically disintegrated in a homogenizer (M-110EH, Microfluidics Corp, United States). The slurry was first passed through two large chambers in series (400 and 200 μm, respectively) at 925 bar. Thereafter, the samples were passed through two smaller chambers in series (200 and 100 μm, respectively) at 1600 bar four times in total. The remaining slurry after the process was freeze-dried for further characterization.

To obtain a cellulose fraction with higher crystallinity, a subsample (0.5 g) of the extracted cellulose was treated with 100 mL 3 M HCl at 90 °C for 90 min. After cooling to room temperature, the mixture was centrifuged at 5000 rpm for 15 min and the supernatant was discarded. The insoluble fraction was washed several times with water until the pH of the washing water was 7 and the insoluble fraction was collected and freeze-dried.

The thermal stability of the cellulose fractions, char and ash contents of Ulva lactuca before extraction and the cellulose fractions, were estimated using a Mettler Toledo TGA/DSC. Approximately 4 mg of each freeze-dried sample was heated in alumina cups from 40 to 800 °C at a heating rate of 10 °C/min under an N₂ atmosphere. After being heated to 800 °C, O₂ was introduced instead of N₂ and the temperature was kept at 800 °C for 15 min. The flow rate was set to 50 mL/min. The char content was calculated as the weight that remained after the samples were heated to 800 °C in N₂ and the ash content was calculated as the mass remaining after finalizing the heating program. The data were processed and analyzed using STARe software.

The morphology of the CNF was observed by ultra-high-resolution field emission scanning electron microscopy (FE-SEM) using a Hitachi S-4800 operating at 5 kV. The CNF was freeze-dried overnight and then attached to the sample supports using double-sided adhesive carbon tape and sputter-coated with a 7 nm Pt/Pd layer using a Cressington 208HR under an inert atmosphere.

The crystal structure and the crystallinity index of the cellulose samples were measured at room temperature using an Empyrean PAN-analytical XRD operating at 45 kV and 45 mA using monochromatic CuKα radiation (λ = 0.1541 nm) in the range 2θ = 10°–40° with a scan rate of 1.0 min⁻¹. The crystallinity index Crl(%) was calculated using Eq. 1 (Segal et al. 1959):where I₂₀₀ is the intensity of the crystalline peak at 2θ = 22° corresponding to the (200) plane and I_am is the intensity at the local minimum of the curve at 2θ = 18° corresponding to the amorphous regions of the cellulose.

The morphology of the CNF in the dry state was analyzed using AFM (Multimode V, Bruker, Santa Barbara, CA). Silicon wafers (1 × 1 cm) were used as substrates for the adsorption. The wafers were washed with absolute EtOH and MilliQ water. After washing, the wafers were placed in an air plasma cleaner (model PCD 002, Harrick Scientific Corporation, NY) for 3 min at 30 W. The wafers were immersed in 0.1 M NaOH for 2 min after which it was immersed in a 0.020 g/L solution of polyethyleneimine (PEI) for 2 min. After washing with water, the samples were immersed in a 0.06 mg/mL suspension of CNF in water for 2 min. The samples were dried using a flow of nitrogen gas.

The silicon wafers were attached to AFM specimen disc. The images were taken in the scanning-mode in height, amplitude and phase modes by using an RTSP silica cantilever (Bruker) having a spring constant of 5 Nm, a tip radius of 8 nm and a resonance frequency of 150–200 kHz. Nanoscope Software (version 1.4, Bruker Corporation) was used for data processing.

¹H → ¹³C cross-polarization magic-angle-spinning (CPMAS) NMR spectra were collected on a Bruker Avance-III NMR spectrometer at a magnetic field strength of 14.1 T (Larmor frequencies of 600.1 and 150.9 MHz for ¹H and ¹³C, respectively) using 4.0 mm zirconia rotors at a MAS rate of 14.00 kHz. Acquisitions involved proton 90 degree excitation pulse of 4 μs and matched spin-lock fields that obeyed a modified Hartmann-Hahn condition υ_H = υ_C + υ_r. Contact time of 500 us was used and SPINAL-64 proton decoupling at 60 kHz. From 16,384 to 24,576 signal transients with 4 s relaxation delays were collected for each sample. Chemical shifts were referenced with respect to neat tetramethylsilane (TMS).

q-FTAA, p-FTAA and h-FTAA was synthesized as reported previously (Åslund et al. 2009; Klingstedt et al. 2011) and HS-310 was synthesized as outlined below. Chemicals and solvents were obtained from commercial sources and used as received (Scheme 1).

Microwave reaction was performed using Biotage Initiator microwave system. TLC was carried out on Merck precoated 60 F254 plates using UV-light (λ = 254 nm and 366 nm) and charring with ethanol/sulfuric acid/p-anisaldehyde/acetic acid 90:3:2:1 for visualization. NMR spectra were recorded on a Varian 300 instrument (Varian Inc., Santa Clara, CA, USA) operating at 300 MHz for 1H and 75.4 MHz for 13C, using the residual solvent signal as reference. Analytical liquid chromatography/mass spectrometry (LC/MS) was performed on a Waters system equipped with a Waters 1525 gradient pump, 2998 Photodiode Array Detector, 2424 Evaporative Light Scattering Detector, SQD 2 Mass Detector and an Xbridge^® C18 column (4.6 × 50 mm, 3.5 μm).

A mixture of dimer 1 (115 mg, 0.272 mmol), 5,5′-dibromo-2-2′-bithiophene (2) (40 mg, 0.123 mmol), K₂CO₃ (86 mg, 0.618 mmol) and PEPPS-IPr (4 mol%) in 1,4-dioxane/methanol (8: 2, 3 mL/mmol, degassed) were added to a reaction vial, back filled with nitrogen and the mixture was heated at 100 °C in microwave system for 30 min. After cooling to room temperature, the reaction was quenched by glacial acetic acid and water (20 mL) added. The residue was extracted with DCM (3 × 40 mL/mmol), washed with water (2 × 30 mL), brine (30 mL) and the combined organic phase was dried over Na₂SO₄, filtrated and the solvent was evaporated. The crude product was treated with warm MeOH and filtrated to obtain hexamer as red solid, yield: 0.35 g (38%).

The hexamer was dissolved in 1,4-dioxane/water (9:1) and hydrolyzed with NaOH (1 M, aqueous, 1.5 equiv./ester) at 80 °C for 16 h. The solution was lyophilized to render HS-310 as a red solid with quantitative yield.

¹H NMR (300 MHz, D₂O) δ 7.31 (d, J = 3.9 Hz, 1H), 6.95 (s, 2H), 6.89 (d, J = 3.9 Hz, 1H), 6.86 (s, 1H), 3.50 (s, 2H). 13C NMR (126 MHz, D2O) δ 179.0, 169.5, 139.9, 139.7, 135.6, 135.1, 134.9, 134.7, 131.3, 130.9, 128.0, 125.7, 124.8, 124.6, 38.3. LCMS (ESI): m/z calcd for C₃₀H₁₈O₈S₆ (M + H) 698.9 found: 699.7.

Stock solutions (1 mg/mL) of h-FTAA, HS-310, p-FTAA and q-FTAA in deionized water were stored at 4 °C. Microcrystalline cellulose fibres (M. cellulose, 100 μm) from cotton liners (CAS no. 9004-34-6) were from Sigma-Aldrich, Stockholm, Sweden, and used as reference samples to compare with the aglae fraction produced in this work. Xyloglucan and xylose were obtained from Sigma-Aldrich. Polysaccharides were maintained as 2 mg/mL stock solutions at 4 °C.

Aliquots (100 μL) of each carbohydrate stock solution were pipetted into individual wells of a white 96-well round bottom microtiter plate (Corning, Stockholm, Sweden). Optotracer stock solutions were diluted in PBS, pH 7.4, (Sigma-Aldrich, Stockholm, Sweden) to a final concentration of 1.33 μg/mL, from which 100 μL was added to each well. Samples were sealed with adhesive PCR plate seals (Termo-Fischer, Stockholm, Sweden) and incubated at 4 °C on a rocking shaker for 30 min. The plate was then positioned in a Synergy MX platereader (Biotek, Stockholm, Sweden), the excitation spectra (emission at 545 nm) were collected between 300 and 525 nm and the emission spectrum (excitation at 450 nm) was collected between 475 and 700 nm. Spectra were collected with the instruments’ ‘top-down’ setting with 1 nm steps. Data from experiments performed in triplicates were processed and analyzed using Prism 6 (Graphpad, USA).

About 1.1 g of extracted cellulose was obtained in the extraction starting from 50.0 g freeze-dried biomass, giving an overall yield of 2.2%(w/w) on a dry weight basis. This value is very similar to the value reported from extraction of Ulva lactuca collected in France (Lahaye et al. 1994) but lower than the yield reported from the extraction of cellulose from Ulva lactuca collected in India (Siddhanta et al. 2009). The chemical composition of Ulva lactuca is strongly dependent on the local growing conditions, so Ulva lactuca collected along the west coast of Sweden cannot be directly compared to Ulva lactuca from other localities. The age of algae individuals is another factor that strongly influences the chemical composition (Starko et al. 2018). Ulva lactuca individuals used in this study were harvested at an early stage of maturity, which likely caused an overall lower content of cellulose.

After the dissolution of 0.5 g of the extracted cellulose fraction in LiCl/DMAc followed by precipitation and freeze-drying, 0.35 g of regenerated cellulose was obtained which corresponds to 70% (w/w) of the dissolved cellulose. This means that 30% (w/w) of the dissolved cellulose remains in the supernatant after precipitation. A possible explanation is that dissolution and heating of cellulose in LiCl/DMAc lead to partial degradation of the cellulose over time due to random cleavage of the glycosidic bonds (Potthast et al. 2002, 2003) The cleavage of the glycosidic bonds leads to a decrease in the average molecular weight and formation of low molecular weight fragments. These fragments might be soluble in the DMAc/water mixture and will not precipitate upon addition of water. Another possibility is that the extracted cellulose fraction contains smaller amounts of other Ulva lactuca cell-wall polysaccharides such as xyloglucan or ulvan which may have a higher solubility than cellulose in the DMAc/water mixture and thus not fully precipitate upon the addition of water.

After the acidic treatment of 0.5 g of the extracted cellulose, 0.34 g cellulose remained as an insoluble fraction, which means that the cellulose was partially degraded during the acidic treatment. The degradation process is caused by acid hydrolysis which causes cleavage of the glycosidic bonds.

Extracted cellulose was mechanically disintegrated in a homogenizer to produce CNF, as further discussed below.

HPAEC-PAD analysis on the cellulose fractions was carried out to determine the carbohydrate composition, targeting, in particular, the glucose purity indicating if the cellulose fractions contain other cell-wall carbohydrates such as ulvan, xylan and xyloglucan. The chromatograms with assigned peaks from the HPAEC-PAD analysis are shown in the supporting information (Figs. S1–S4).

The cellulose fractions extracted from Ulva lactuca contain mostly glucose (Table 1), but also a smaller amount of xylose, which sustains that indeed a glucose-rich polysaccharide was isolated in the extraction. This is in agreement with previous studies, which showed that an insoluble fraction consisting of mostly glucose and smaller amounts of xylose is obtained if Ulva lactuca is bleached and treated with alkali (Lahaye et al. 1994; Bobin-Dubigeon et al. 1997; Cronshaw et al. 1958; Ray and Lahaye 1995). However, the origin of the xylose moieties is still not fully established. Lahaye et al. (1994) used enzymatic depolymerization of the extracted cellulose followed by HPLC analysis in an attempt to identify the origin of the xylose. Their analysis suggests that the extracted cellulose is likely a mixture of cellulose and smaller amounts of xyloglucan rather than the cellulose chain itself carrying the xylose residues. As shown in our study, neither pigment removal nor an additional post-production acidic treatment removed the xylose from the sample. Even after treatment with 3 M HCl, the sample still contains xylose. This is a possible indication that xylose and glucose units are, at least to some extent, covalently linked and that the cellulose is indeed a mixture of pure cellulose and a xylose-glucose polymer such as xyloglucan. Another possibility is that the insoluble fraction consists only of a glucose-xylose polysaccharide and not true cellulose I or II at all. Rhamnose, glucuronic acid, and iduronic acid were not found in the samples verifying that there is no ulvan in the samples. The overall carbohydrate content in the regenerated cellulose fraction was lower than in the extracted cellulose and the CNF. This can be explained by the fact that cellulose partially degrades upon dissolution in LiCl/DMAc at higher temperatures. One of the suggested degradation mechanisms is the peeling and degradation of monomer units at the reducing end of the cellulose chain (Potthast et al. 2002, 2003). The degradation of the monomers will lead to a lower monosaccharide content in the regenerated sample and therefore a lower carbohydrate content. The acid-treated cellulose had the highest carbohydrate content with an overall carbohydrate content of 89% (w/w), probably because the acid-treatment removed impurities that were present in the extracted cellulose. Another interesting observation is that the regenerated cellulose had a lower relative amount of xylose (10%) compared with the extracted cellulose and CNF (15%). This is another indication that the extracted cellulose fraction is a mixture of cellulose and xyloglucan. The xyloglucan is likely to have a higher solubility in the DMAc/water mixture and will not precipitate completely. Consequently, the regenerated cellulose will be more enriched in pure cellulose than the extracted cellulose before dissolution. The acid-treated cellulose also contains a lower relative amount of xylose (9%) than the extracted cellulose and CNF (15%) further sustaining that the xyloglucan was partly removed during the acid-treatment process. In conclusion, the HPAEC-PAD analysis verifies the absence of ulvan and indicates that the extracted fractions contain cellulose mixed with a xylose-glucose polysaccharide, but cannot convincingly verify that the fractions contain cellulose I or II. Further characterization is necessary.

The cellulose fractions were further analyzed by FTIR (Fig. 2). Lists of all assigned peaks are included in the Supporting information (Tables S1–S4). The FTIR spectra of the cellulose fractions show many of the characteristic peaks for polysaccharides. The broad bands at 3500–3200 cm⁻¹ correspond to the O–H stretching for the hydroxyl groups and the bands in the range 3000–2800 cm⁻¹ correspond to C–H stretching. The peaks at 1420 cm⁻¹ and 1315 cm⁻¹ correspond to H–C–H scissor vibration and H–C–H tip vibration, respectively. The sharp peaks around 1020–1050 cm⁻¹ correspond to symmetric stretching of C–O–C linkages. The small peak around 895 cm⁻¹ is characteristic of the β-glycosidic bond, which is a strong indication of the presence of polymeric polysaccharides such as cellulose (Ververis et al. 2004). Importantly, the spectra show no traces of strong peaks around 1220 cm⁻¹ and 840–845 cm⁻¹. These peaks correspond to S=O stretching and C–O–S stretching and are characteristic of the polysaccharide ulvan, a major component of Ulva lactuca. The absence of such peaks hence indicates that ulvan was successfully removed from the cellulose fraction using the proposed fractionation protocol, which is in agreement with the results from the HPAEC-PAD analysis. Furthermore, no peak around 1550 cm⁻¹ corresponding to N–H bending in proteins was observed indicating that the cellulose samples were free from proteins. A small peak is observed around 1710 cm⁻¹, possibly corresponding to C=O stretching in oxycellulose (Zeronian and Inglesby 1995) indicating that oxidation of cellulose into oxycellulose took place to a minor extent during the extraction, probably during the bleaching step with hydrogen peroxide. Oxidation of xyloglucan probably also occurred to some extent. No significant differences were observed between the extracted, regenerated, and the CNF samples and all the samples show all the peaks that are expected for cellulose.

Ulva lactuca before fractionation, the extracted cellulose fractions, and the produced CNF were further analyzed by TGA. The ash and char contents are given in Table 2 and the TG and DTG curves are shown in Figs. 3 and 4. The mass of the cellulose fractions decreases by 4% (w/w) at temperatures up to 100 °C. This is caused by the evaporation of bound water. The largest decrease in mass starts at around 260 °C, which is caused by the thermal degradation of cellulose. This indicates that the cellulose in Ulva lactuca is thermally stable up to this temperature. Previous studies have shown that the thermal stability of cellulose from seaweed varies between species (Lakshmi et al. 2017; Chen et al. 2016; Singh et al. 2017) and is usually in the range 250–300 °C. The thermal stability increases with increasing crystallinity index of the cellulose. By comparing the TGA curves for the extracted cellulose and the regenerated cellulose, respectively, it is clear that the regenerated cellulose starts to decompose at a slightly lower temperature than the extracted cellulose. Studies on wood cellulose has shown that regenerated cellulose from DMAc/LiCl has a slightly lower thermal stability than cellulose before dissolution (Mahadeva et al. 2013) and our data suggest that cellulose from Ulva lactuca shows the same behavior. This is probably caused by the fact that regenerated cellulose has fewer hydrogen bonds between the cellulose chains and also a lower crystallinity than cellulose before dissolution. The acid-treated cellulose had a slightly higher thermal stability than the extracted cellulose, probably because the acid-treated cellulose has higher crystallinity than the extracted cellulose. The DTG curves for the cellulose fractions show one peak corresponding to the degradation of cellulose. The peak values are as follows: T(extracted cellulose) = 318 °C, T(regenerated cellulose) = 305 °C, T(CNF) = 342 °C, and T(acid treated cellulose) = 328 °C.

The overall ash contents in the extracted and regenerated cellulose samples (Table 2) were significantly lower than the ash content of Ulva lactuca before extraction, probably because typical water-insoluble inorganic substances such as metal oxides, carbonates, and/or sulfates were converted to water-soluble metal chlorides during the HCl treatment (the final step in the cellulose extraction). By making these impurities water-soluble they can be removed from the cellulose since the liquid phase containing the impurities is discarded after the extraction. After the homogenization process, the ash content of the extracted cellulose decreased from 5.7% (w/w) to 0% (w/w) which means that all inorganic impurities were removed from the cellulose during the homogenization process. The regenerated cellulose fraction shows a lower ash content than the extracted cellulose before dissolution, probably because the regenerated sample was subjected to additional washing steps after the precipitation, which likely removed additional impurities.

CNF produced from the extracted cellulose was assessed with FE-SEM and AFM to verify that the homogenization process was successful in converting the cellulose into individual fibrils. The SEM images (Fig. 5) show individual fibrils with a thickness and length in the nanometer range, and regions with a sheet-like structure. The sheet-like structure can possibly be non-defibrillated regions indicating the homogenization process was not able to defibrillate the cellulose completely. It is also possible that the sheet-like structures represent aggregated cellulose fibrils caused by the freeze-drying process that was performed prior to the FE-SEM analysis. The sheet-like regions can also be because the cellulose contains a xylose-glucose polysaccharide such as xyloglucan, as discussed previously and indicated by the HPAEC-PAD analysis. Formation of these nanofibrils requires close packing of the molecules. Linear polysaccharides like cellulose are able to form nanofibrils due to their ability to pack densely while branched polysaccharides like xyloglucan typically do not form nanofibrils. The mixture of nanofibrils and sheet-like regions is another indication that the CNF is a mixture of cellulose and xyloglucan where the nanofibers consist of cellulose and the sheet-like structures are xyloglucan.

The crystal structure and the crystallinity index of the cellulose fractions were estimated by XRD (Fig. 6). The extracted cellulose, acid-treated cellulose, and CNF gave rise to three peaks at 2θ = 16°, 22°, and 34°, corresponding to the crystal planes (1 1 0), (2 0 0) and (0 0 4), respectively, sustaining the presence of crystalline regions (Sarko and Muggli 1974; Park et al. 2010; Wada and Okano 2001; French and Cintrón 2013; French 2014). These are typical peaks for the cellulose I allomorph. The XRD diffractogram of regenerated cellulose (Fig. 6e), looks different and has a broad peak at 2θ = 21° which is characteristic for the cellulose II allomorph (Chanzy et al. 1986; Stipanovic and Sarko 1976; Kolpak and Blackwell 1976). The main difference between cellulose I and II still remains a question of intense debate. One theory suggests that the main difference between the two allomorphs is the orientation of the individual cellulose chains: in cellulose I, the cellulose chains are parallel, while in cellulose II, the chains are antiparallel (Chanzy et al. 1986; Stipanovic and Sarko 1976; Kolpak and Blackwell 1976). Also, the hydrogen bond pattern between the cellulose chains is different for cellulose I and II. The conversion from cellulose I to cellulose II is an irreversible process. The crystallinity index (Crl %) was calculated from the XRD diffractograms using Eq. 1 (Segal et al. 1959). The crystallinity index for the regenerated cellulose fraction could not be calculated since the value of I_am could not be derived due to the broad peak at 21°. However, the broad peak suggests that this sample is essentially amorphous. This is a possible explanation to why the regenerated cellulose showed lower thermal stability than the extracted cellulose in the TGA analysis.

The extracted cellulose and CNF had a crystallinity index of 48% indicating that the homogenization process did not change the crystallinity to any significant extent (Table 3). However, the crystallinity index increased to 63% after the acidic treatment, probably because the acidic treatment partially removed the amorphous regions of the cellulose. The low crystallinity of the extracted cellulose and the CNF can be explained by the samples being a mixture of cellulose and xyloglucan. The xyloglucan residue might possibly interrupt the formation of cellulose crystals leading to a low crystallinity. The HPAEC-PAD analysis showed that xylose content decreased as an effect of acid-treatment, which indicates a partial removal of the xylose-glucose polysaccharide mixed with cellulose in the fractions. This would explain why the acid-treated cellulose had a higher crystallinity index than the other samples. Overall, the degree of crystallinity of the Ulva lactuca cellulose seems low. The estimated values of the crystallinity however show the crystallinity of a mixture between cellulose and xyloglucan rather than the crystallinity of the pure cellulose. Hence, even if the estimated values of the crystallinity is low it does not necessarily mean that the crystallinity of pure cellulose from Ulva lactuca is low. Studies of various algae have shown that the crystallinity of cell wall cellulose varies substantially between different classes and orders of algae (Mihranyan 2011). In some algae, cellulose is the major component of the cell wall and the cellulose tends to be highly crystalline while other algae are characterized by lower contents and crystallinity of cell wall cellulose. In addition, the crystal structure varies between algal species (Koyama et al. 1997). Acid treatment of cellulose recovered from Cladophora green algae produced a cellulose with very high degree of crystallinity (Wada and Okano 2001), while the acid-treated cellulose fraction produced in this study did not reach a crystallinity index higher than 63%. In general, the degree of crystallinity of CNF is dependent of the biomass source and on the preparation method. Gopakumar et al. (2016) report that ‘crystallinity ranging from 40 to 80% has been reported for the CNFs obtained from softwoods and hardwoods.’

Further structural investigations of the cellulose extracted from Ulva lactuca and the produced CNF were performed with solid-state NMR (Fig. 7). The reference cellulose sample shows all the characteristic signals for cellulose. Signals at 65 and 63 ppm correspond to the C6 carbon in crystalline and amorphous regions in cellulose, respectively. The strong signals at 76 and 73 ppm correspond to the C2, C3, and C5 carbons. The weaker signals at 89 and 85 ppm correspond to the C4 carbon in crystalline and amorphous regions in cellulose, respectively. The signal at 105 ppm is attributed to the C1 carbon (Wada et al. 2004). The spectra for the cellulose samples from Ulva lactuca look similar, but some important differences can be observed. For the extracted cellulose, acid-treated cellulose and CNF, the intensity of the signal at 89 ppm corresponding to crystalline domains is lower than for the reference samples indicating that the crystallinity of these samples is lower than the crystallinity for the reference samples, which is in agreement with the XRD data (Table 3). The regenerated cellulose shows only a very faint signal at 89 ppm indicating that this sample is largely amorphous. The same behavior can be observed for the signals from the C6 carbon. In the reference sample, the signals corresponding to the crystalline regions have the same intensity as the signal from the amorphous regions. However, for the extracted, acid-treated and CNF, the amorphous signal has a higher intensity than the crystalline signal. For the regenerated cellulose, we only see the amorphous signals at 63 ppm and 85 ppm, which is another indication that this sample is amorphous. As previously discussed, if the cellulose samples are indeed mixtures of cellulose and xyloglucan, we would expect individual signals from each monosaccharide (glucose and xylose). However, the chemical shifts for the C1–C5 carbons in xylose are very close to the chemical shifts for C1–C6 in glucose (Lahaye et al. 1994) and since the peaks in the NMR spectra are very broad, the peaks for glucose and xylose are likely to overlap. The results from the solid-state NMR in summary sustain that the extracted cellulose fractions are mixtures of cellulose with a xylose-glucose polysaccharide. Still, just as in the case of HPAEC-PAD, the results cannot convincingly verify that the fractions contain cellulose I or II.

Optotracing was developed as a non-disruptive method to optically detect and visualize cellulose in biological tissues in their native states. The method utilizes optotracers, a class of molecular probes called oligothiophenes, which bind selectively to glucans producing a unique spectral pattern that identifies the bound target. Applying heptameric (Choong et al. 2018) and pentameric (Choong et al. 2019) optotracers, cellulose can be clearly distinguished from a range of glucans and glucose-containing heteropolysaccharides based on differences in size and stereochemistry. Here, we applied optotracing as a complementary tool to overcome the limitation of HPAEC-PAD and solid-state NMR to verify the polymeric origin of the detected xylose and glucose building blocks in the extracted cellulose. To determine if optotracing could differentiate cellulose from xyloglucan, a panel of 4 optotracers ranging from heptameric to tetrameric was screened against pure referenced samples of microcrystalline cellulose (M. cellulose), xyloglucan and xylose. For easy comparison of spectral patterns and wavelengths representing the peak amplitudes, the fluorescence emitted by optotracers in each interaction was normalized by calculating the percentage of the emitted fluorescence at each excitation wavelength relative to the highest and lowest values in the spectrum (N. RFU). The N. RFU was then plotted in Normalized Spec-Plots (Fig. 8). The wavelengths of maximum excitation (Ex. λmax) and emission (Em. λmax) are summarized in Table 4.

Analysis of the normalized Spec-Plots of h-FTAA showed visibly different spectral patterns when in interaction with cellulose and xyloglucan, which was reflected in the 5 nm and 11 nm differences in Ex. λmax and Em. λmax respectively. This indicated that optotracing with h-FTAA could differentiate between cellulose and xyloglucan.

Comparison of the normalized Spec-Plots of xylose and xyloglucan indicated that the spectral pattern collected of xyloglucan was not due to interactions with monosaccharide, but more likely due to interactions with the xyloglucan polysaccharide. To determine if the detection of xyloglucan, like cellulose, was due to binding of the β-1,4-glucan segments, 1 mg/mL of each polysaccharide were screened against hexameric (HS-310), pentameric (p-FTAA) and tetrameric (q-FTAA) optotracers. Analysis of each normalized spec. plot showed unique spectral patterns in all instances, whereby the spectral pattern of cellulose and xyloglucan were different and distinct. This indicates that the interaction between optotracers and xyloglucan was along segments of the polysaccharide substituted with 1,6-linked xylose side chains.

Observing the enhanced resolution of optotracing by the comparative analysis of multiple optracers, we applied the panel of heptameric to tetrameric optotracers to CNF produced from Ulva lactuca. M. cellulose was screened alongside as a cellulose reference of high purity (Fig. 9, Table 5). Comparison of the normalized spec plot of CNF and M. cellulose revealed highly similar spectral patterns, Ex. λmax and Em. λmax of bound optotracers. p-FTAA was the only exception, wherein M. cellulose binding produced an additional Ex. λmax at 446 nm, which may be a reflection of the highly dense crystal morphology of M. cellulose. Analysis by the optotracer panel indicates that glucose detected in CNF (Table 1) originates from cellulose. The absence of other spectral patterns reminiscent of xyloglucan as described in Fig. 8, suggests that the xylose fraction detected during monosaccharide analysis is likely due to the presence of another glucose-xylose copolymer. This copolymer could possibly be the linear β-(1,4) xyloglucan reported in previous studies (Lahaye et al. 1994; Lahaye and Robic 2007) which would then differ from the branched β-(1,6) xyloglucan used as the reference.