Isolation and purification of high-molecular weight hemicelluloses from radiata pine wood chips prior to thermo-mechanical pulp (TMP) production

2.1 Preparation of compression screw pressate (CSP)

Fresh debarked radiata pine (Pinus radiata D. Don) slabwood chips from a sawmill based in Rotorua, New Zealand were steamed at 80 °C for 5 min and then transferred via a compression screw to the preheater, where they were held at 140–170 °C for 20 or 30 min. Preheated chips were then fed through a second compression screw which produced the pressate (CSP) for the study (stored in the dark at 4 °C until needed) (Lloyd and Murton 2016).

2.2 Filtration of CSP

All CSP samples were vacuum-filtered through wet-strength Whatman 113 filter paper to remove suspended solids.

2.3 Washing of XAD resins

Amberlite^® resins XAD4, XAD7 and XAD16 were purchased from Sigma-Aldrich. Before use, resins (200–1000 g) were stirred in several litres of Milli-Q water for 30 min to remove the preservative salts and recovered by vacuum filtration through wet-strength Whatman 113 filter paper. The washed resin was then oven-dried at 50 °C.

2.4 Treatment of CSP with XAD resins

Treatments of prefiltered CSP with washed XAD resins were carried out at resin-to-liquid ratios of 0.02–0.08, calculated on washed resin weight. CSP was placed in a 0.5 L or 1 L Schott bottle and the required amount of washed resin was added. The bottle was then shaken at 180 rpm in an orbital shaker for the scheduled length of time (1–3 h) at ambient temperature (ca. 24 °C) or 40 °C, as indicated. After treatment the resin particles were removed by vacuum filtration through wet-strength Whatman 113 filter paper and the filtrate was stored at 4 °C in the dark until needed.

2.5 Regeneration of XAD7 resin

Regeneration of XAD7 resin was performed by desorption of lignin and other adsorbed material with 0.1 M NaOH. Each batch of used XAD7 was stirred magnetically in 1–2 L of 0.1 M NaOH after which the resin particles were removed by vacuum filtration and washed with several litres of distilled water. This process was performed three or four times for each batch until the washings were only lightly yellow coloured.

2.6 Light absorbance measurements

The light absorbance of CSP samples was measured in the ultraviolet (UV) region at 280 nm and/or in the visual region at 457 nm in a quartz cell with a path length of 1 cm using a Shimazdu UV-1800 UV–Vis spectrophotometer. Before measurement samples were diluted with Milli-Q water until the absorbance readings were in the range 0.2–1.0. The results were expressed as the values obtained by multiplying these readings with the corresponding dilution factors.

2.7 Determination of total dissolved solids (TDS)

Total dissolved solids (TDS) in filtered CSP samples were determined gravimetrically after freeze-drying.

2.8 Determination of lignin content

Acid-insoluble (Klason) lignin in freeze-dried CSP was determined in triplicate by a modified method based on TAPPI Standard Methods, T 222 om-88 and acid-soluble lignin in triplicate by a modified TAPPI Useful Method UM 250.

2.9 Determination of extractive content

Quantification of wood extractives in CSP was performed using an earlier published method (McLean et al. 2005; Örsa and Holmbom 1994) but with a partly different selection of internal standards. A 5.0 mL aliquot of each sample, adjusted to pH 3.5 using 1.0 mol L⁻¹ HNO3, and 100 µL of internal standard (0.55 mg mL⁻¹ solution of heptadecanoic acid, betulin, cholesteryl heptadecanoate and 1,3-dipalmitoyl-2-oleoylglycerol (Sigma) in toluene (Merck)) were transferred to a screw cap test tube and extracted into 1.0 mL of methyl tert-butyl ether (MTBE (BDH)). After extraction the MTBE layer was isolated by centrifugation, transferred into gas chromatography (GC) vials and blown to dryness under a nitrogen atmosphere. Samples were silylated with 100 µL of BSA (N,O-bis(trimethylsilyl)-acetamide, Alltech) and 100 µL of pyridine (Merck) at 60 °C for 20 min before cooling to room temperature and making up to 1.0 mL with toluene.

Silylated samples were analysed on an Agilent 6890 N GC FID (flame ionisation detector) instrument with Agilent 7683B autosampler. Samples (1 µL) were injected directly onto a Phenomenex Zebron (ZB-1, 15 m × 0.53 mm × 0.15 µm FT) capillary GC column under a constant pressure (0.21 bar) flow of helium. The injector temperature was held at 90 °C for 30 s after injection before increasing to 325 °C at 100 °C min⁻¹. The column was held at 90 °C for the first 90 s before increasing to 320 °C at 20 °C min⁻¹. The FID was held at 360 °C for a total run time of 60 min. Using a calibration standard of 0.55 mg mL⁻¹ dehydroabietic acid (Helix Biotech), palmitic acid (Aldrich) and triolein (Sigma) in toluene, relative response factors were calculated and used to determine the level of resin acids, fatty acids, sterols, lignans, steryl esters and triglycerides. Samples were analysed in triplicate and the mean and standard deviation reported.

Total dichloromethane (DCM) soluble extractives were determined gravimetrically on freeze dried solids by Soxtec-extraction for 1 h and evaporation to dryness under reduced pressure.

2.10 Determination of mono- and oligo-/polysaccharides

Monosaccharides in liquid samples were determined by ion chromatography before and after acid hydrolysis at Celignis Ltd, University of Limerick, Ireland.

2.11 Determination of MW distributions

Gel permeation chromatography (GPC) was used to determine the MW distribution of freeze-dried CSP samples. GPC was carried out on a Polymer laboratories PL-GPC-50 integrated GPC system with a Knauer K-301 refractive index (RI) detector and a Knauer K-2501 UV detector (280 nm), using two PolarGel-L columns (Varian, 300 × 7.5 mm) connected in series and protected by a guard column (50 × 7.5 mm) of the same material. The system was run at 50 °C with analytical grade DMSO + 0.05 M LiCl as eluent. Flow rate was maintained at 1 mL min⁻¹. The system was calibrated using pullulan saccharide standards (PSS, 180–47,000 MW). Samples were dissolved overnight on a shaker at a concentration of 3 mg mL⁻¹. No filtering was necessary. Injection volume was 100 μL.

2.12 Ultrafiltration

Ultrafiltration (UF) of CSP samples was carried out at the highest feed pressure allowed by the UF system (1.0–1.2 kPa) using a Masterflex ultrafiltration unit (motor model 7553-75; Cole-Parmer) that was equipped with a 10,000 Da polyethersulphone membrane (cylindrical, surface area 0.29 m2) purchased from Smart Membrane Solutions, Rolleston, NZ. Each run was continued until the permeate flow ceased. At this point the run was terminated or distilled water was added to the retentate to continue the ultrafiltration and further concentrate the high-MW fraction. In total, three successive UF runs were performed on the purified CSP 160/20-XAD7.

3.1 Optimal TMP conditions for dissolving high-MW hemicellulose

The yield and MW distribution of hemicelluloses in CSP depend on the severity factor which is a function of the chip pre-heating temperature and time (Table 1) (Lloyd and Murton 2016). The amount (yield) of freeze-dried total dissolved solids (TDS) in CSP samples originating from pulping at different severities is shown in Figure 1. It is seen that as severity increased, TDS increased in a logarithmic fashion while the proportion of material of 10,000 Da or larger was at its highest at the middle of the severity range investigated.

Figure 1:

Effect of preheating severity on total dissolved solids (TDS) content of compression screw pressate (CSP, squares) and the percentage of >10,000 Da material in CSP (circles).

The approximate composition of CSP TDS is shown in Table 1. The ratio of carbohydrates to lignin increased with increasing severity as the rate of hemicellulose hydrolysis increased at a higher rate than that of lignin depolymerisation and solubilisation. This has been reported before for the water soluble portion of TMP produced from Nordic softwoods (Widsten et al. 2001).

The MW distribution of TDS samples is shown in Table 2. Two different GPC detectors were used. The refractive index (RI) detector detects both carbohydrates and lignin while the UV detector detects lignin and any aromatic hemicellulose dehydration products such as furfural and hydroxymethylfurfural (HMF) or pseudolignin. The weight-average MW (M_w) initially increased and then started to decline with an increase in severity, presumably because the rate of secondary depolymerisation reactions was faster than primary hydrolysis reactions at higher severities. Combining the TDS data from Table 1 and the RI detector data from Table 2 shows that the CSP produced by pre-heating at 160 °C for 20 min (hereafter called ‘160/20’) had the highest content of material with MW of 10,000 Da or higher (1.2% of CSP, or 19.2% of its TDS) and was therefore selected for further work.

3.2 Purification of CSP before ultrafiltration

3.2.2 Removal of lignin and other UV-absorbing material

Larger reductions were achieved at greater resin-to-liquid ratios and at the lower (ambient) treatment temperature (Figure 2). Extending the treatment time to 2 h caused a slight further reduction in UV absorption but increasing it to 3 h gave results similar to 2 h (Figure 3a). In general, the ability of the three resins to remove UV-absorbing material decreased in the order XAD16 > XAD7 > XAD4 although the differences were small in relation to the UV absorbance of untreated CSP. The portion of UV-absorbing material that was not removed by the XAD treatments may have been bound to carbohydrates in the form of lignin-carbohydrate complexes (Lawoko et al. 2005) that are not sorbed onto the hydrophobic XAD resins (Westerberg et al. 2012).

Figure 2:

Effect of XAD resin type, dose and treatment temperature on UV absorption (280 nm) of pre-hydrolysis TMP water 160/20 (treatment time 1 h).

Figure 3:

Effect of XAD resin treatment time on absorption of pre-hydrolysis thermos-mechanical pulping (TMP) water 160/20 in the UV (A) and visible (B) regions. Treatments were done at 24 °C with a resin-to-liquid ratio of 0.08.

All three types of XAD resin were effective at removing UV-absorbing material during 1 h batch treatment of CSP.

For industrial application it is important to be able to regenerate the adsorbent resins used by desorption of the adsorbed material. To investigate the regenerability of XAD7 part of it used in the 3 h treatments was regenerated by stirring it in 0.1 M NaOH (which became dark brown), followed by washing with Milli-Q water and air-drying. Although the dried resin retained a yellowish colour it performed similarly in 3 h treatments (76% of absorbance removed at 280 nm) as the white unused XAD7 resin (78% of absorbance removed at 280 nm) in terms of removing UV-absorbing material from CSP. Regeneration of XAD resin with organic solvents can also be effective (Schwartz and Lawoko 2010).

3.2.3 Removal of colour

In terms of colour, all three XAD resins were able to reduce absorbance of CSP 160/20 in the visible region (457 nm) (Figure 3b). Extending the treatment time from 1 h to 2 h reduced absorbance substantially while a further extension to 3 h had little additional effect. Figure 4a shows the effect on colour of two successive 1 h treatments using fresh XAD resins. Samples treated once with XAD16 were darker still than the untreated sample, indicating that the strongly chromophoric material initially became more concentrated in the sample treated with XAD16 before some of it was removed in the second treatment. Most of the dark colour can probably be attributed to lignin oxidation products such as quinonoids (Paulsson and Parkås 2012) that may also absorb light in the UV region and pseudolignin originating from degradation of polysaccharides (Shinde et al. 2018). The visual appearance of the samples was not always in line with the measured light absorbances in the visible region, as is clearly seen by comparing Figures 3b and 4a.

Figure 4:

Unpurified and purified compression screw pressates (CSPs) before and after drying. (A) Compression screw pressate (CSP) 160/20 before (left) and after treatment (XAD samples). In each pair of XAD samples the sample on the left is from a single treatment and the sample on the right from two successive treatments with fresh resin (1 h; resin-to-liquid ratio 0.08; 24 °C). (B) Freeze-dried CSP samples: 160/20 before XAD7 treatment (left), after XAD7 treatment for 3 h at 24 °C (middle) and the high-MW fraction (retentate 2 of Table 3) isolated by ultrafiltration (right).

3.2.4 Removal of extractives

The presence of dissolved and colloidal extractives in CSP is undesirable as they contribute to membrane fouling in ultrafiltration (Puro et al. 2011). The concentration of extractives in CSP 160/20 was reduced by ca. 40% upon passing CSP through filter paper (Figure 5). This reduction in extractives was attributed to the removal of fines material, which is known to contain considerable levels of bound resin (Johnsen et al. 2006). When the filtered CSP was treated with XAD resins the total extractive contents were reduced by 76–89%. Typically, the remaining extractives were mostly triglycerides, with XAD4 also retaining a significant level of resin acids. The removal efficiencies of the three types of resin fell in the same order as seen for UV-absorbing material (XAD16 > XAD7 > XAD4) but the differences were small relative to the amount of extractives found in the original or filtered CSP.

Figure 5:

Dissolved and colloidal extractives in compression screw pressate (CSP) 160/20 before and after XAD resin treatments (0.08 resin-to-liquid ratio; 24 °C; 2 h). The filtered (solids-free) CSP 160/20 was used for the XAD treatments. Error bars show standard deviation.

3.2.5 Retention of high-MW carbohydrates and lignin

During treatment of CSP with XAD resins it was desirable to preserve high-MW carbohydrates to the highest extent possible while removing lignin and extractive fractions that are detrimental to ultrafiltration. In another investigation (Strand et al. 2015) lignin ranging from 500 to 5,700 Da in size was found to have a significant negative effect on UF flow rate when the membrane used had a cut-off of 5,000 Da, whereas smaller molecules and those larger than 7,400 Da had little negative impact. Determination of MW distribution by GPC with an RI detector showed that the CSP samples treated with XAD resins retained 83–94% of the carbohydrates and lignin of high MW (>10,000 Da) present in untreated CSP (Table 3). In general, the proportion of high-MW carbohydrates was slightly reduced and that of high-MW lignin increased. The resin treatments removed a much larger percentage of the UV-absorbing material of low MW than material with an MW greater than 10,000 Da (Figure 6, UV detector); the largest reductions were seen for the XAD7-treated sample (Figure 6). The fact that XAD7 was the most effective resin in decolouration of CSP and gave the lowest high-MW lignin content (as estimated by GPC with UV detection) suggests that a lot of the dark colour may reside in the high-MW lignin fraction. Thus, although high-MW lignin can potentially be a useful raw material for coating and films if it can be etherified in the same way as hemicellulose, it may be best to avoid accumulating it in the high-MW fraction. From this point of view XAD7 resin is arguably the best choice for purifying CSP intended for coating and film applications.

Table 3:

Percentage of high-MW material (>10,000 Da) in CSP 160/20 before and after XAD resin treatment (0.08 resin-to-liquid ratio; 24 °C; 3 h).

Sample	Mainly carbohydrates and lignin (RI detector; %)		Mainly lignin (UV detector; %)
	% Of TDS	% In water	% Of TDS	% In water
Untreated	18.2	1.07	0.24	4.1
XAD4	23.2	1.09	0.74	15.7
XAD7	18.7	0.99	0.38	7.3
XAD16	22.2	0.95	0.60	13.8

Figure 6:

GPC chromatographs of untreated and resin-treated freeze-dried compression screw pressate (CSP) 160/20 samples.

3.3 Ultrafiltration of CSP 160/20 treated with XAD7 resin

When the XAD resin-treated CSP samples were ultrafiltered using a membrane cut-off of 10,000 Da, the permeates were less coloured than the retentates. CSP 160/20 treated with XAD7 was selected for UF runs due to the superior colour reduction and lower content of high-MW UV-absorbing material (lignin) compared to the CSP samples purified with the other two XAD resins. However, most of its carbohydrates were of low MW (<10,000 Da; Figure 6). To render the material suitable for coatings and films its Mw had to be substantially increased by ultrafiltration. This was possible at a reasonable UF flow rate to isolate high-MW hemicellulose (Table 4). During UF some of the remaining coloured material was also removed as can be seen from the freeze-dried samples in Figure 4b.

The monosaccharide distribution of post-hydrolysis 160/20 (Figure 7) reflects the reported composition of non-cellulosic polysaccharides in radiata pine wood, dominated by GGM and xylan (Smelstorius 1974), except for the large amount of arabinose that probably originates from arabinoxylans and arabinogalactans (Smelstorius 1974). 13% of pre-hydrolysis 160/20 solids were monosaccharides (mostly arabinose), increasing to 91% after acid hydrolysis of oligo- and polysaccharides (this includes the mass of water added on hydrolysis). Regarding the UF of 160/20-XAD7 (Figure 7), the monosaccharides (almost all of them arabinose) were almost completely removed already during the first run. The second and third UF runs did not change the composition of the hemicelluloses but increased the molar mass of the retentate solids that includes UV-absorbing material such as lignin (possibly as LCCs, lignin-carbohydrate complexes) or pseudolignin in addition to the high-MW hemicelluloses (Table 4). The hemicellulose content of the final retentate (R3) at 89% of total solids was similar to that of 160/20 (91%) and 160/20-XAD7 (86%), indicating that they were well retained in the UF process (Figure 7).

Figure 7:

Monosaccharide content of 160/20, 160/20-XAD7 and the retentates and permeates from ultrafiltration of resin-purified compression screw pressate (CSP) 160/20-XAD7 before and after acid hydrolysis (dry matter basis). Error bars show standard deviation. R1-R3 denote retentates from successive UF runs and P1 the permeate from the first run.