Study on the residual lignin in Eucalyptus globulus sulphite pulp

Molecular weight of residual lignin (L_R)

Lignin from unbleached sulphite eucalypt pulp of κ 18.2 was isolated by a modified acidolytic technique previously described for kraft pulps (Pinto et al. 2003a). The molecular weight (Mw) of residual lignin (L_R sample) was lower than that reported for dioxane lignin (DL sample) isolated from eucalypt wood (E. globulus) by the same technique (Table 1). This fact may be due to partial degradation of lignin remained in pulp during acidic sulphite pulping. The determination of Mw of lignin dissolved in sulphite spent liquor (L_SSL sample) is difficult without its previous purification. Due to the presence of significant amount of sugars, inorganic salts and lignin degradation products with low molecular weight in SSL, the Mw of LS was unreasonably low (ca. 950 Da). Accordingly, LS was purified by dialysis with a membrane of 5000 NMWCO. About 40% of SSL dry matter was removed during SSL dialysis. Marques et al. (2009b) found that besides sugars and inorganics, the highly sulphonated monomeric/dimeric structures are removed during dialysis with this membrane and retentate becomes enriched in oligomeric LS. The Mw of purified LS was 2350 Da, which is almost triple of that of purified lignin isolated from kraft black liquor (Pinto et al. 2003a). It was confirmed that L_SSL possesses significantly higher molecular weight compared to kraft lignin. However, the Mw of L_R in unbleached sulphite is very similar to that in kraft pulps (κ 14–18) (Pinto et al. 2002b) (2200 and 2170 Da, respectively).

Wet chemistry analyses

The results of chemical analyses of L_SSL and L_R are presented in Table 1. These are compared with previously reported data of DL isolated from E. globulus wood (Evtuguin et al. 2001). It may be concluded that L_SSL and L_R have less OMe groups compared to DL and are partially sulphonated. The amounts of sulphonic groups, however, were almost triple in L_SSL compared to that in L_R, which was water-insoluble and incompletely soluble in DMSO. L_R contains almost 50% more phenolic groups than L_SSL, indicating a more ramified structure of the former.

The information about different “condensed” and “non-condensed” structures in L_SSL and L_R was assessed based on degradation products arising from KMnO₄ oxidation (PO) (Table 2). Lignins were ethylated prior to oxidation to protect phenolic units against aromatic ring degradation and to be able to distinguish PO products with free phenolic groups. Nearly 96% of PO products were identified. Products 1, 2 and 3 were assigned to non-condensed lignin structures derived from the p-hydroxyphenyl (H), guaiacyl (G) and syringyl (S) structural units, respectively. The isohemipinic acid methyl ester (4) is originated from degraded phenylcoumaran (β-5′) structures and the products 5 and 6 are assigned to G and S units, respectively, condensed at the C-6 position. Other condensed structures of biphenyl and diaryl ether types are presented by products 7 and 8/9, respectively (Table 2). The product 10 may be assigned both to demethoxylated G units and to condensed tannins of catechin type (Marques et al. 2009a).

Table 2

Results on permanganate oxidation analysis of L_SSL, L_R and DL.

Oxidation product	(mol%)	Oxidation product	(mol%)
	LSSL: 2		LSSL: 8
	LR: 1		LR: 11
	DL: 3		DL: 6
	LSSL: 24		LSSL: <1
	LR: 14		LR: <1
	DL: 18		DL: 2
	LSSL: 54		LSSL: <1
	LR: 67		LR: <1
	DL: 58		DL: 1
	LSSL: 3		LSSL: 2
	LR: 1		LR: 4
	DL: 3		DL: 7
	LSSL: 5		LSSL: 2
	LR: 3		LR: 1
	DL: 3		DL: 0

The comparison of PO degradation products 1–10 from L_R and L_SSL is very conclusive. L_R contains less amounts of phenolic non-condensed G units (product 2) compared to L_SSL and DL, and higher amounts of S units (product 3). The higher abundance of S structures in L_R than in L_SSL is obvious. At the same time, unlike RL in kraft pulp (Pinto et al. 2003a), no increase of biphenyl and diaryl ether structures has been detected in L_R in comparison to DL isolated from wood. However, L_R demonstrates an impressive increase of S-type structures condensed at C-6 of an aromatic ring; more precisely, the abundance of structures 6 was almost double those in wood and almost 50% higher than in L_SSL. Hence, syringyl structures in L_R of sulphite pulp are highly condensed.

Unlike DL and L_SSL, L_R contained rather significant amounts of carbohydrates (mainly xylan and glucan) that could not be removed upon purification and which are probably chemically linked to lignin (Table 2). A similar observation was made previously based on NMR studies of eucalypt LS isolated from SSL (Marques et al. 2009b). Apparently, strong association with carbohydrates, condensation reactions and limited lignin sulphonation are the main reasons for the retention of LR in pulp.

Analyses by NMR

The main structural features of L_SSL and L_R were evaluated by quantitative ¹³C NMR (Figures 2 and 3). L_SSL was dissolved in D₂O and acetylated L_R in DMSO-d₆. Acetylation was necessary to improve solubility in DMSO, because quantitative ¹³C NMR spectroscopy needs high lignin concentration in solution (ca. 30%). The assignments of signals in ¹³C NMR spectra are based on the analysis of proton-carbon correlation (HSQC) 2D NMR spectra of corresponding lignins (Figures 4 and 5). Moreover, the published database of sulphonated eucalypt lignins (Luthaes et al. 2008; Marques et al. 2009b) and native eucalypt lignins (Evtuguin et al. 2001) served for interpretation. The assignments were additionally confirmed by the single-bond proton-proton correlation spectra (COSY, spectra are not shown) and HSQC spectrum of non-acetylated L_R (Figure 6). The signal assignments of acetylated L_R relies on databases published for model compounds (Ralph et al. 2004) and hardwood/softwood lignins (Kilpeläeinen et al. 1994; Ämmälahti et al. 1998). The signals belonging to concomitant carbohydrates were assigned according to known data on non-acetylated and acetylated xylan (Evtuguin et al. 2003; Miyagawa et al. 2013) and glucan (Bock et al. 1984; Lisboa et al. 2005).

Figure 2

Quantitative ¹³C NMR spectrum of L_SSL (D₂O, 50°C). Structures A′, B′ and D′ designate sulphonated β-O-4′, β-5′ and β-1′ structures, respectively.

Figure 3

Quantitative ¹³C NMR spectrum of acetylated L_R (DMSO-d₆, 45°C). The impurities are marked by asterisks and were not considered for integrations. The expanded region at 165–175 ppm shows acetylated phenolic (OH_ph) and aliphatic primary (OH_prim) and secondary (OH_sec) hydroxyl groups. Structures A, B, C and D designate β-O-4′, β-5′, β-β′ and β-1′ structures, respectively.

Figure 4

HSQC spectrum of L_SSL (D₂O, 50°C). Designations for the main lignin structures are presented in Figures 1 and 2. S-n and G-n designates the signals from syringyl and guaiacyl units, respectively, where n is the number of proton/carbon atom (sulphonated units designated with the index sulph). Xn, Gn and GlcAn designates β-d-xylopyranose, β-d-glucopyranose and 4-O-methyl-α-d-glucuronopyranose structural units, respectively, where n is the number of proton/carbon atom; the reducing terminal units are designated by letter t, and α- and β-isomers are denoted by α and β, respectively (e.g., αX1t means anomeric proton/carbon in α-isomer of reducing β-d-xylopyranose residue). XnAcmGlcA means the proton/carbon n in β-d-xylopyranose residue possessing the acetyl group (Ac) at carbon m and substituted at O-2 by 4-O-methyl-α-d-glucuronopyranose residue.

Figure 5

HSQC spectrum of acetylated L_R (DMSO-d₆, 45°C). Designations for the main lignin structures are the same as in Figure 3. Xn, Gn and GlcAn designates acetylated β-d-xylopyranose, β-d-glucopyranose and 4-O-methyl-α-d-glucuronopyranose structural units, respectively, where n-number of proton/carbon atom. The reducing terminal units are designated by letter t, and α- and β-isomers are denoted by α and β, respectively (e.g., αX1t means anomeric proton/carbon in α-isomer of reducing acetylated β-d-xylopyranose residue). XnGlcA means the proton/carbon n in acetylated β-d-xylopyranose residue substituted at O-2 by 4-O-methyl-α-d-glucuronopyranose residue. LSα and LC6 are cross-signals assigned to benzylic and H6/C6 in glucopyranose, respectively, involved in α-O-6 benzyl ether linkage (depicted by the dashed red line).

Figure 6

HSQC spectrum of L_R (DMSO-d₆, 45°C). Designations for the main lignin structures are presented in Figure 3. LSα is the cross-signal of benzylic proton/carbon in α-O-6 benzyl ether LC linkage (depicted by the dashed red line).

The analysis of the main lignin structures in L_SSL and L_R is summarized in Table 3. The calculations have been carried out per one aromatic ring according to previously established methodology (Evtuguin et al. 2001; Marques et al. 2009b). The aromatic region equivalent to six carbon atoms was integrated at 103–156 ppm. The abundance of lignin structures linked by β-O-4′ bonds was calculated based on signals integrals at 61.0–63.0 ppm in acetylated L_R or at 60.0–62.0 ppm in L_SSL (C_γ resonance in β-O-4′ structures). Similarly, the abundance of phenyl coumaran (β-5′) and pino/syringaresinol (β-β′) structures was assessed based on C_β resonances in corresponding structures (Figures 4 and 5).

The abundance of β-O-4′ structures in L_SSL was much lower than that in L_R, which contained in turn about 20% less amount of these structures than wood lignin (Table 3). Interestingly, the relative occurrence of β-O-4′ structures found in L_SSL and L_R was comparable to those detected previously in purified lignin from kraft black liquor and in RL of kraft pulp with κ ca. 14 (Pinto et al. 2003b). This means that β-O-4′ bonds were cleaved to a similar extent in kraft and acidic sulphite cooking of eucalypt wood. The principal difference between β-O-4′ structures of L_SSL and L_R consisted in the amount of sulphonic groups at the benzylic carbon (BnC). According to rough estimation made by the volume of C_β peaks in the HSQC spectra of lignins (Figures 4 and 5), more than 95% of β-O-4′ structures in L_SSL are sulphonated at the Bn position (structures A′, Figure 2), whereas about 90% of β-O-4′ structures in L_R were not sulphonated (structures A, Figure 3). Other structures of L_R containing sulphonic groups are β-5′ units that are almost completely sulphonated at the Bn position. This conclusion could be drawn based on the absence of the cross-signal of C_α in corresponding B structures (Figure 3) of acetylated L_R at ca. 5.4–5.5/87–88 ppm and the characteristic cross-peak at ca. 4.65/68.8 ppm assigned to sulphonated BnC in B′ structures (Figure 2). The sulphonation at the Bn position of a part of S units in L_R is also evidenced by the characteristic down field chemical shift of C2,6 in corresponding structures (Figure 5).

The analysis of quantitative ¹³C NMR spectra (Figures 2 and 3) also shows a noticeable increase in the relative abundance of S structures in L_R compared to L_SSL (Table 3). This observation can be explained, at least partly, by the removal of highly sulphonated monomeric S-type structures in the purification step of L_SSL by dialysis (Marques et al. 2009b). However, the proportion of S units in L_R was even higher than that in dioxane lignin (DL) isolated from wood. This is in agreement with data of KMnO₄ oxidation of lignins (Table 2) and may be attributed to the topochemical peculiarities of the acidic sulphite process in eucalypt wood. It was highlighted previously that the delignification of vessels, rays and middle lamellae (rich in G units) of hardwood is faster than the secondary wall of libriforms enriched in S structural units (Glennie 1971; Terashima 1990). Despite sometimes contradicting opinions on the topochemistry of sulphite and sulphate pulping, it is recognised that at the same level of retained lignin, its content in the S2 of sulphite pulp is higher than in kraft pulp (Wood and Goring 1973; Koch et al. 2003). In view of the prevalence of S structures in L_R, it can be concluded that RL in sulphite pulp is localised essentially in the S2. The opposite is true for kraft E. globulus pulp, where the RL has lower proportion of S units than wood lignin (Pinto et al. 2003a,b). The relatively high amount of S units in the RL contributes to the easy bleachability of eucalypt sulphite pulp.

The quaternary carbons in aromatic rings as detected by quantitative ¹³C NMR, which are for structures with alkyl-aryl and aryl-aryl linkages, are in agreement with results of PO (Table 2) as they also indicate more condensed structures in L_R than in DL (Table 3). At the same time, the relative abundance of quaternary aromatic carbons in L_SSL is even higher than that in L_R. Accordingly, the degree of lignin condensation itself does not explain the lignin retention in pulp upon cooking. Most probably, the accessibility of lignin in the cell wall towards pulping acid and eventual lignin-carbohydrate (LC) interactions plays an essential role for the retention of RL in pulp fibres.

Glucan and glucuronoxylan are associated with RL according to sugars analysis in L_R (Table 2) and the data of the HSQC spectrum (Figure 5). These are supposed to be chemically linked to RL. The last hypothesis about LC bonds in L_R is confirmed by characteristic cross-signals in the HSQC spectrum at 3.23/61.2 ppm and at 4.56/79.2 ppm assigned to H6/C6 in glucopyranose unit and H_α/C_α in phenylpropane unit, respectively, involved in α-O-6 Bn ether linkages (Taneda et al. 1987). The signal of BnC in benzyl ether-type LC structures was shifted to 4.60/81.5 ppm in the HSQC spectrum of non-acetylated L_R (Figure 6), which is in agreement with previously reported assignments in the corresponding structures of lignin-carbohydrate complexes isolated from birch (Betula pendula) wood (Balakshin et al. 2011) and eucalypt (E. globulus) wood (Miyagawa et al. 2013). Such LC linkages are present originally in the cell wall, though their formation during acidic sulphite pulping cannot be completely excluded.

It is noteworthy that two expected cross-peak signals in non-acetylated L_R were not detected, namely Bn proton/Bn carbon involved in ether linkage with xylopyranose (5.10–5.15/81.5–82.5 ppm) and in ester linkage with glucuronopyranose residues (5.5–6.0/75–76 ppm) (Figure 6). This fact, however, does not mean that lignin and glucuronoxylan are not chemically associated. Evtuguin et al. (2003) found that at least a part of eucalypt glucans is linked with xylan by glucosidic (1→2) linkages with terminal 4-O-methyl-α-d-glucuronosil residues. The structural association between RL in sulphite pulp and hemicelluloses needs further study.