Practical high-yield production of vanillins from kraft or soda lignin using highly alkaline hydrogen peroxide treatment

2.1 Materials

Semiconductor grades (99.99+ %) of sodium hydroxide (NaOH), iron (III) chloride (FeCl₃), and copper (II) chloride (CuCl₂) were purchased from Sigma-Aldrich Japan LLC (Tokyo, Japan). A 30 % H₂O₂ solution containing no stabilizer was purchased from FUJIFILM Wako Pure Chemical Corp. (Osaka, Japan). All other chemicals were purchased from either of these or Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). V, VA, and AV were purified by recrystallization from ethanol before use.

Wood chips (about 20 × 10 × 3 mm) prepared from Japanese cedar (Cryptomeria japonica) were gifted from Nippon Paper Industries Co., Ltd. (Tokyo, Japan). Ultrapure water (H₂O) was prepared by a generator, Puric Z II (Organo Corporation, Tokyo, Japan), and used in all runs of HA–H₂O₂–Tr conducted in this study.

2.2 Preparation of iKL and iSL

The Japanese cedar wood chips (100 g, oven dry basis) were pulped in a stainless-steel autoclave (1.0 L), TEM-V1000M (Taiatsu Techno Corp. Tokyo, Japan), under the following kraft or soda cooking conditions: active alkali charge of 38.5 % as sodium oxide (Na₂O); sulfidity of 28.0 or 0 %, respectively; liquor to wood ratio of 7.0 L/kg; total volume of 700 mL. Common grade of NaOH was used in cooking, and sodium sulfide (Na₂S) was titrated with iodine in a sulfuric acid (H₂SO₄) solution to quantify the actual sulfide content before use. The temperature was raised to 100 °C for 10 min, and then immediately to 165 or 170 °C for 50 min in the kraft or soda cooking, respectively. The maximum temperature was maintained for 120 min (total: 180 min). The kraft or soda pulp was filtered by a Büchner funnel and washed well with deionized H₂O. The combined effluent, called “black liquor”, was a total volume of 1,250 or 1,460 mL after the kraft or soda cooking, respectively. The yields of the obtained kraft and soda pulps were calculated from the weights after drying in an oven at 105 °C for one day (Table 1). The kappa numbers of the obtained kraft and soda pulps were measured, following the TAPPI standard method, T236 om-13 (Table 1).

A portion of the kraft or soda black liquor (400 or 420 mL, respectively) was acidified with a H₂SO₄ solution to adjust the pH to 2.0. The precipitate generated was collected by several centrifugations with washing, followed by lyophilization and successively final drying in vacuo. The obtained brown to black powder was iKL or iSL, respectively.

2.3 Determination of lignin contents in original wood, iKL, and iSL and constituent sugar contents in iKL and iSL

Wood meal was prepared from the Japanese cedar wood chips using a Wiley mill. The Japanese cedar wood meal, iKL, or iSL was subjected to the following method to quantify the lignin content: two successive acidic treatments of the Japanese cedar wood meal (1.00 g), iKL (100 mg), or iSL (100 mg) in a 72 % H₂SO₄ solution (10, 1.0, or 1.0 mL, respectively) at room temperature for 4 h and in a 4 % solution prepared by diluting the 72 % solution at 120 °C for 1 h in an autoclave, BS-305 (Tomy Seiko Co., Ltd., Tokyo, Japan). The obtained residue after complete drying was weighed as the Klason (acid-insoluble) lignin content (Table 2). The acid effluent was analyzed by an ultraviolet and visible spectrophotometer, UV-240 (Shimadzu Co., Kyoto, Japan), to measure the acid-soluble lignin content, using a quartz cell with an optical path length of 1.0 cm, absorptive coefficient of 110 L g⁻¹ cm⁻¹, and absorbance at 205 nm, following the TAPPI standard method, T222 om-2 (Table 2).

Constituent sugars of iKL or iSL were analyzed by utilizing the above-described acid effluent and alditol acetate method, following the TAPPI test method (T249 cm-21). Quantification was done with an internal standard compound, myo-inositol, using a gas-chromatograph (GC) with a flame-ionization detector (FID), GC-2010 plus (Shimadzu Co.). A capillary GC column, TC-17 (length: 30 m, inner diameter: 0.25 mm, film thickness: 0.25 μm; GL Sciences Inc., Tokyo, Japan), was used with helium as the carrier gas. A constant temperature of 210, 220, or 230 °C was employed for the oven, injector, or detector, respectively. A calibration curve for each sugar was separately created from four samples. These four samples were 4 % H₂SO₄ solutions containing four different amounts of the sugar and a specific amount of myo-inositol, and subjected to the alditol acetate method in a manner identical to that in the constituent sugar analysis of iKL or iSL.

2.4 Alkaline nitrobenzene oxidation (ANO) of iKL and iSL

iKL or iSL was subjected to traditional ANO, following a method described in the literature (Chen 1992). V and VA produced were quantified by another GC (FID), GC-2014 (Shimadzu Co.), using ethylvanillin (3-ethoxy-4-hydroxybenzaldehyde) as the internal standard compound, after trimethylsilylation with N,O-bis(trimethylsilyl)acetamide (BSA). A capillary GC column, InertCap 1 (length: 30 m, inner diameter: 0.25 mm, film thickness: 0.40 μm; GL Sciences Inc.), was used with helium as the carrier gas. The oven temperature was initially maintained at 150 °C for 25 min, followed by two successive rises to 190 and 280 °C at rates of 3 and 10 °C/min, respectively, while 215, 230, and 280 °C were maintained for 45, 20, and 10 min, respectively, with an injection or detector temperature of 280 °C (total time: 122.3 min). A calibration curve was separately created for V or VA by preparing four samples containing four different amounts of each and a specific amount of ethylvanillin. In this curve creation, an alkaline solution of V or VA containing ethylvanillin was primarily prepared and subjected to a work-up procedure identical to the above analysis for iKL or iSL.

2.5 HA–H₂O₂–Tr of iKL or iSL

A 30-mL ultrapure H₂O solution was prepared to contain 18 mg (600 mg/L) of iKL or iSL, 0 or 97.2 μmol (3.24 mmol/L) of a chelating agent (d-mannitol, 3,4-dihydroxybenzoic acid (protocatechuic acid, PCA), ethylenediaminetetraacetic acid (EDTA), or hydroxyethylethylenediaminetriacetic acid (HEDTA)), 32.4 μmol (1.08 mmol/L) of FeCl₃ or CuCl₂ (semiconductor grade), and 90 mmol (3.0 mol/L) of NaOH (semiconductor grade) in a volumetric flask made of Teflon that was thoroughly washed with ultrapure H₂O in advance. These were added in this order. This solution was transferred to a Teflon flask (50 mL) with a Teflon cap and heated to 90 °C with magnetic stirring in a brine bath. Then, a 0.245-mol/L H₂O₂ solution prepared by diluting the 30 % solution with ultrapure H₂O was continuously added at a specific rate for a specific period, using a syringe pump, YSP-202 (YMC Co., Ltd., Kyoto, Japan), and a rubber cap instead of the Teflon one. The Teflon flask was soaked in an ice H₂O bath 20 min after the completion of adding the H₂O₂ solution to terminate the reaction. Table 3 lists all reaction systems and conditions employed in this study.

V was used as another starting compound instead of iKL or iSL under the same conditions as in system 6 or 19 listed in Table 3. Another 30-mL ultrapure H₂O solution was separately prepared to contain only iSL (600 mg/L) and semiconductor grade NaOH (3.0 mol/L), and then immediately subjected to the quantification of V, VA, and AV described below.

2.6 Quantification

A NaOH solution (1.0 mL) containing ethylvanillin, the internal standard compound, was added to the cooled Teflon flask, followed by immediate additions of 200 μL of acetic acid (AcOH) and 1.0 mL of methanol (MeOH) and successive thorough shaking. A portion of the resultant mixture was passed through a membrane filter, and then a further portion of the filtered mixture was injected by a syringe into a high-performance liquid chromatograph (HPLC) consisting of a system controller (SBM-20A, Shimadzu Co.), online degasser (DGU-12A, Shimadzu Co.), solvent delivery unit (LC-10AD & LC-20AD, Shimadzu Co.), column oven (CTO-10Avp, Shimadzu Co.), and photodiode array detector (SPD-M 10Avp, Shimadzu Co.). Quantification was based on absorbance at 280 nm. An HPLC column, Luna 5 μm C18(2) 100 Å (length: 150 mm, inner diameter: 4.6 mm, particle size: 5.0 μm; Phenomenex, Inc., Torrance, CA, USA), was used at a solvent flow rate of 1.0 mL/min and an oven temperature of 40 °C. The solvent employed was a mixture of MeOH and aqueous 1.0 wt% AcOH. A solvent ratio of 19/81 (v/v) was initially maintained for 30 min, gradated to 44/56 for 5 min, and maintained for 10 min (total: 45 min). A calibration curve was separately created for V, VA, or AV by preparing four samples containing different amounts of each and a specific amount of ethylvanillin.

3.1 Theory on viewing HA–H₂O₂–Tr as the most appropriate system

As described in chapter 1 (Introduction), the authors’ group has studied and reported the mechanism of chemical reactions in pulp bleaching processes using oxygen and oxygen-related species (Imai et al. 2007, 2008; Ohmura et al. 2012, 2013; Posoknistakul et al. 2016, 2017a,b; Sakai et al. 2022; Yokoyama et al. 1996, 1998, 1999a,b, 2002, 2005, 2007). The major findings were: i) the degradation of carbohydrates, a serious problem that should urgently be solved in these bleaching processes, is caused by active oxygen species, among which the hydroxyl radical (HO⦁) and oxyl anion radical (O⦁⁻, the conjugate base of HO⦁) are the most responsible; ii) these radicals are generated by transition metal-catalyzed self-decomposition of organic peroxides as well as H₂O₂; iii) HO⦁ reacts with aromatic portions in lignin slightly faster than with aliphatic counterparts that constitute carbohydrates and side chains of lignin, whereas O⦁⁻ attacks the aliphatics much more preferably owing to the repulsive force between negatively charged O⦁⁻ and the electron-rich aromatic portions (Posoknistakul et al. 2016, 2017a; Yokoyama et al. 2005). Additionally, O⦁⁻ is the only responsible species in a system applying a pH higher than around 12.5, because the acid-dissociation equilibrium between HO⦁ and O⦁⁻ largely shifts to the latter at that pH (pK_a of HO⦁: 11.9). On this basis, it has been considered that aromatic nuclei present in iKL or iSL survive HA–H₂O₂–Tr while the side chains are degraded and cleaved to efficiently generate V and VA as well as AV despite employing structurally unknown iKL or iSL as the raw material (Figure 3). A key point is whether the unimolecular self-decomposition of hydroperoxide anion (HO₂⁻, the conjugate base of H₂O₂) can be promoted (probably not the self-decomposition of H₂O₂ due to the mostly complete dissociation of H₂O₂, pK_a of H₂O₂: 11.6) to efficiently generate O⦁⁻ (Figure 3). Reaction conditions appropriate for this were examined in model experiments using VeG in previous studies by the authors, as introduced in Section 3.2 (Posoknistakul et al. 2017b; Sakai et al. 2022).

Figure 3:

Rough schematic description of HA–H₂O₂–Tr production.

Alkaline H₂O₂ oxidation (not highly alkaline) is practically applied to current multistage ECF (elemental chlorine free) bleaching processes as the most common final stage or one before the final stage in pulp and paper mills worldwide. This practical alkaline H₂O₂ oxidation targets the degradation of chromophoric groups in chemical pulp to whiten it without causing extensive delignification, and thus applies such conditions as those where neither HO⦁ nor O⦁⁻ but HO₂⁻ is the most contributing species. HO₂⁻ actively reacts with carbonyl and carbonyl-related groups and degrades them without extensive delignification. As only a few reaction conditions differ between practical alkaline H₂O₂ oxidation and HA–H₂O₂–Tr, HA–H₂O₂–Tr should be applicable using already-established mill equipment for practical alkaline H₂O₂ oxidation after the setting of reaction conditions. HA–H₂O₂–Tr is thus ready for practical installation, a fact absent from most previous studies, as described in chapter 1, and so this is an important advantage of HA–H₂O₂–Tr.

3.2 Results obtained in the previous model studies by the authors’ group

Posoknistakul et al. (2017b) showed that HA–H₂O₂–Tr of VeG under specific conditions efficiently produces Ve and VeA, appropriate analogues of V and VA, respectively. The total yield of Ve and VeA was 77 % based on the amount of degraded VeG, although only 22 % of VeG used was degraded. It was thus only 17 % based on the total amount of VeG used. These results suggest that abundant H₂O₂ (actually HO₂⁻) added to the reaction system bimolecularly self-decomposed into O₂ and H₂O (actually the hydroxide anion (HO⁻)), because H₂O₂ can be considered to have not accumulated but completely decomposed on the basis of the results of another authors’ previous study (Yokoyama et al. 2002). This is a waste of H₂O₂ from the viewpoint of producing Ve and VeA. Sakai et al. (2022) showed that modifications of reaction conditions increased the total yield from 17 to 44 %, accompanying an improvement in the amount of degraded VeG from 22 to 61 %, but with just a slight decrease in the total yield based on the amount of degraded VeG from 77 to 72 %. Thus, the previous model studies by the authors’ group were able to suggest that HA–H₂O₂–Tr has the potential to efficiently produce V and VA from an actual lignin sample. In this study, iKL or iSL was treated in HA–H₂O₂–Tr, and reaction conditions were set to be appropriate for the production of V and VA as well as AV with yields as high as possible.

Incidentally, a preliminary trial of HA–H₂O₂–Tr using the phenolic analogue of VeG, guaiacylglycerol-β-guaiacyl ether (1-(4-hydroxy-3-methoxyphenyl)-2-(2-methoxyphenoxy)propane-1,3-diol), yielded V, which corresponds to Ve in HA–H₂O₂–Tr of VeG, but only a small amount, although the degradation of this phenolic analogue was greater than that of VeG. Thus, a phenolic nucleus cannot effectively survive HA–H₂O₂–Tr, although it exists as phenolate that has a clear negative charge and electric repulsive force with O⦁⁻.

3.3 Evaluation of kraft and soda cookings, iKL, iSL, and ANO of iKL and iSL

The Klason (acid-insoluble) and acid-soluble lignin contents of the employed Japanese cedar were 35.5 and 0.8 %, respectively, and the total lignin content was thus 36.3 % (Table 2). These contents are common as those of a softwood sample.

A high active alkali charge (38.5 %) was employed in the kraft or soda cooking, because a high liquor to wood ratio (7 L/kg) was inevitably used due to the application of a batch system. The kraft or soda cooking was able to successfully pulp the Japanese cedar wood chips, although the kappa number of the soda pulp was high. The yield and kappa number of the obtained kraft pulp were 43.9 % and 37.7, respectively, indicating that the kraft cooking was effective (Table 1). Those of the obtained soda pulp were 44.6 % and 92.8, respectively (Table 1). The simultaneously obtained kraft or soda black liquor was acidified to precipitate iKL or iSL, respectively. Each whole black liquor would have generated 13.3 g of iKL or iSL after the whole work-up procedure, although actually a portion of the kraft or soda black liquor (400 mL of 1,250 mL or 420 mL of 1,460 mL, respectively) was acidified to precipitate 4.25 g of iKL or 3.82 g of iSL. The yield of iKL or iSL was thus 13.3 % based on the original wood and 36.6 % based on native lignin (13.3/36.3 × 100) in the original wood (Table 4). This yield was converted to another, 39.3 % for iKL or 44.2 % for iSL, based on all organic materials present in each whole black liquor before acidification (Table 4). In this conversion, the residual lignin content in each pulp was calculated by multiplying the kappa number by 0.15, and kraft or soda lignin was supposed to exist as the exclusive organic material in each black liquor.

The Klason, acid-soluble, and total lignin contents of iKL or iSL were determined, accompanying its constituent sugar analysis. The contents of iKL were 78.9, 5.3, and 84.2 %, respectively (Table 4). Those of iSL were 77.7, 4.7, and 82.4 %, respectively (Table 4). The constituent sugar analysis confirmed that polysaccharides occupy about 10 % of iKL or iSL (Table 4). Xylan was the major contaminant, with a much smaller amount of arabinan. This is in accordance with the fact that a major component of softwood hemicellulose is arabino-4-O-methylglucuronoxylan, whose side chains are L-arabinose and 4-O-methyl-d-glucuronic acid residues. The latter side chain residues are acidic and hence not detected in the constituent sugar analysis, although they were actually converted to the hexenuronic acid residues, which are highly labile to the acidic constituent sugar analysis to be small fragments, during the cookings. The contamination of iKL or iSL with xylan was caused by either or both of the following two possibilities: i) kraft or soda lignin and xylan were connected by LC (lignin-carbohydrate) covalent bonds in the black liquor; ii) polymeric or oligomeric xylan present in the black liquor was physically incorporated into iKL or iSL during the acidification and not removed during the subsequent work-up procedure. The latter may be more frequent, because xylan present in black liquor is known to readsorb onto co-existing solid (pulp) surfaces and hence has a tendency to physically adsorb. The other major component of softwood hemicellulose, O-acetyl-galactoglucomannan (despite deacetylation during the cooking processes), contaminated iKL or iSL only slightly, although it is generally more abundant than xylan in softwood cell walls. This slight contamination suggests that glucomannan was degraded into small oligomers, monomers, or small fragments derived from the monomers more readily than xylan in the black liquor and/or remained inside pulp fibers unless it was degraded into those. This is because the physical incorporation of glucomannan into iKL or iSL during acidification should not have differed from that of xylan. Because of the followings: i) galactan originated from d-galactose residues as side chains in galactoglucomannan; ii) the amount of d-galactose side chains is much smaller than total amounts of d-glucose and d-mannose intermediate residues in the galactoglucomannan backbone, the contaminating galactan may be considered a lot relative to the contaminating glucan and mannan. Thus, a portion of the glucomannan frequently carrying d-galactose side chains may have preferentially been incorporated. Because xylan, in contrast to glucomannan, does not significantly undergo the peeling reaction due to the presence of d-glucuronic acid (possibly as well as L-arabinose) residues at the C2-positions, the above discussion may be rational. Although the amounts of polysaccharides in iKL (as well as iSL) were much larger than those in isolated kraft lignin in a previous report, probably due to the acidification conditions (Hu et al. 2016), the contaminating xylan may also have been larger than the contaminating glucomannan in this study.

ANO of a lignin-containing sample under the conditions described in the literature is the most traditional and general method for producing V as well as VA (Chen 1992). The yield of V (or total yield of V and VA) in ANO of a lignin-containing sample is always higher than those in any other methods, and so it is often employed as the potentially maximum yield from the sample. Despite this, unfortunately, nitrobenzene cannot practically be applied due to its carcinogenicity. iKL or iSL was primarily subjected to ANO to examine the potentially maximum yield of V as well as VA from iKL or iSL. The yields of V and VA from iKL were 11.9 and 1.4 %, respectively, and those from iSL were 13.4 and 2.0 %, respectively. The yield of V from softwood, i.e., softwood native lignin, in ANO commonly reaches around 30–40 % based on the native lignin. V as well as VA is generally considered to mostly originate from the β-O-4-type of substructures (Figure 4). Because the cleavage of the β-O-4-type of interunit linkages is targeted to achieve delignification in cooking processes, kraft or soda lignin present in the black liquor commonly contains small amounts of substructures with the β-O-4 bond. In fact, the progress of this cleavage in the cooking processes is suggested by the yield of V as well as VA from iKL or iSL being much lower than that from softwood native lignin. The yield of V from iKL or iSL in ANO, 11.9 or 13.4 %, respectively, and total yield of V and VA, 13.3 or 15.4 %, respectively, are employed as potential maxima in discussions on the yield of V and total yield of V and VA (and AV) in HA–H₂O₂–Tr of iKL or iSL in the following contents.

Figure 4:

Major substructures present in softwood native lignin. The β-O-4-type is considerably more abundant than any others, commonly at 50 % or more.

3.4 Basic HA–H₂O₂–Tr of iKL and iSL

Primarily, iSL was dissolved in a NaOH solution (3.0 mol/L) without any other additive. The prepared solution was immediately subjected to quantification of V, VA, and AV to confirm whether these compounds originally contaminated iSL. The yield of V was 1.5 %, while VA and AV were hardly detected. This suggests the following: i) V originally existed in the soda cooking black liquor; ii) some molecules of the originally existing V were incorporated into the precipitates generated by acidification; iii) the incorporated molecules of V were not completely removed during the filtration, centrifugation with washing, and lyophilization, and drying in vacuo, and so a small amount of V still contaminated iSL. Thus, some V formed in the soda and kraft cookings of the Japanese cedar and contaminated iSL and iKL in this study. In fact, Ve, the analogue of V, was detected as a major reaction product in soda cooking of VeG in a previous study by the authors (Kato et al. 2019). Some V must have similarly contaminated the soda or kraft lignins isolated from the black liquors and used as raw materials in many previous studies (Araújo et al. 2009, 2010; Bjelic et al. 2018; Borges da Silva et al. 2009; Fargues et al. 1996; Gomes and Rodrigues 2020a,b; Jeon et al. 2020; Khwanjaisakun et al. 2020; Liu et al. 2020; Mathias and Rodrigues 1995; Pinto et al. 2011; Singh and Ghatak 2017; Sutradhar et al. 2024; Vu et al. 2023; Zhang et al. 2020; Zirbes et al. 2020, 2023), although these studies did not discuss this original contamination.

In the model study of HA–H₂O₂–Tr by Sakai et al. (2022), the highest total yield of Ve and VeA, appropriate analogues of V and VA from VeG, respectively, was obtained under the following conditions: 3.92 mmol H₂O₂ was added stepwisely 800 times with an interval of 15 s (total reaction time of 220 min, including 20 min after the completion of H₂O₂ addition) to a 30-mL reaction solution containing 1.0 mmol/L of VeG, 1.08 mmol/L of FeCl₃, and 3.0 mol/L of NaOH at 90 °C. In this study, these optimal conditions were primarily applied to HA–H₂O₂–Tr of iKL or iSL (system 1 or 14, respectively, in Table 1). Because V can be converted to VA in HA–H₂O₂–Tr, the total yield of V, VA, and AV is discussed more than each yield in the following contents. Table 3 lists all yields of V, VA, and AV and their total yields observed in this study.

V, VA, and AV were produced from iKL with yields of 3.0, 0.5, and 0.4 %, respectively, to give a total yield of 3.9 % in system 1. Their yields from iSL were 3.9, 0.4, and 0.5 %, respectively, to give that of 4.8 % in system 14. Thus, these compounds were produced from iKL less than from iSL, which was also observed in any comparison between other iKL and iSL systems that employed specifically constant conditions. This tendency can rationally be explained by the above-described results on ANO of iKL and iSL: V and VA were produced from iKL less than from iSL. This smaller production is due to cleavage of the β-O-4-type of interunit linkages in the kraft cooking being more extensive than in the soda cooking; therefore, substructures that are origins of V and VA existed in iKL less than in iSL, although their origins in ANO may not be identical to those in HA–H₂O₂–Tr. The smaller production may also be attributed to the possibility that V originally contaminated iKL less than iSL.

It was not quantified how many iKL or iSL reacted or were degraded in HA–H₂O₂–Tr, because weight measurement of the solid residue obtained by acidification after the reaction is not enough for quantification. The solid residue contains lots of substructures with structural changes, and so cannot correspond to “residual iKL or iSL”.

H₂O₂ must have completely self-decomposed at the end of any reaction system, according to the stability of H₂O₂ observed in an authors’ previous report (Yokoyama et al. 2002).

3.5 Effect of chelating agents in HA–H₂O₂–Tr of iKL and iSL

The authors’ previous model study showed that the addition of d-mannitol (3.24 mmol/L) as a chelating agent is effective to increase yields of Ve and VeA, analogues of V and VA from VeG, respectively, when H₂O₂ was added with half the amount of this study (Sakai et al. 2022). d-Mannitol has been suggested to directly interact with iron ions to form complexes, although it is generally known as a radical scavenger (Gutteridge 1984; Magara and Ikeda 2013). Thus, 3.24 mmol/L of d-mannitol was also added in system 2 or 15 for iKL or iSL, respectively. d-Mannitol or another chelating agent (PCA, EDTA, or HEDTA) had to be added with FeCl₃ (or CuCl₂) before adding NaOH in the preparation of reaction solutions; otherwise, hydroxides and/or oxides of Fe (or Cu) precipitated, and consequently, the addition was not effective. Yields of V, VA, and AV were 4.8, 0.4, and 1.0 %, respectively, to give a total yield of 6.2 % in system 2. Those were 6.1, 0.3, and 0.9 %, respectively, to give that of 7.3 % in system 15. The total yield was 1.5 times or more of that without the addition of d-mannitol in HA–H₂O₂–Tr of iKL or iSL (comparison between systems 1 and 2 or between systems 14 and 15, respectively). Thus, the addition of d-mannitol was effective in HA–H₂O₂–Tr of iKL or iSL.

The addition of PCA or EDTA, another chelating agent, was less effective than that of d-mannitol in HA–H₂O₂–Tr of iKL (comparison between systems 2 and 8 or between systems 2 and 9, respectively) or iSL (that between systems 15 and 22 or between systems 15 and 23, respectively). However, the addition was slightly more or ineffective for iKL or iSL, respectively, than no addition of any chelating agent in HA–H₂O₂–Tr (comparison between systems 1 and 8 or 9 or between systems 14 and 22 or 23, respectively). The addition of HEDTA had an effect similar to that of d-mannitol in HA–H₂O₂–Tr of iKL or iSL (comparison between systems 2 and 10 or between systems 15 and 24, respectively).

All positive effects of adding these chelating agents can be explained as follows: i) many Fe hydroxides and oxides precipitated and did not catalyze the uni- but bimolecular self-decomposition of HO₂⁻ to generate O₂ and HO⁻ in the absence of any chelating agent; ii) the chelating agents showing the positive effects formed the soluble complexes with some Fe³⁺ and Fe²⁺ (generated by reduction of Fe³⁺), leaving these cations their catalytic activities in the self-decomposition of HO₂⁻ and generation of O⦁⁻.

Because HA–H₂O₂–Tr was conceived as an environmentally benign system, d-mannitol rather than HEDTA was employed in systems described in the following text.

3.6 Effect of forms of H₂O₂ addition in HA–H₂O₂–Tr

Various forms of H₂O₂ addition were applied to examine the effects on the yields of V, VA, and AV in HA–H₂O₂–Tr of iKL or iSL in the presence of d-mannitol (systems 2–7 for iKL and 15–21 for iSL). The total amount of H₂O₂ added was 3.92 mmol in systems 2, 3, and 4 for iKL and systems 15 and 16 for iSL, but the addition rates were different between these systems. The total as well as each yield decreased with either accelerating or decelerating the addition from 80 μL/min (19.6 μmol/min), employed in the authors’ previous study (Sakai et al. 2022). This rate was therefore the best and used as the standard in this study. Subsequently, the total amount of H₂O₂ added was changed while maintaining the standard rate in systems 2, 5, 6, and 7 for iKL and systems 15, 17, 18, 19, 20, and 21 for iSL. Because iSL showed higher values for the total as well as each yield than iKL, HA–H₂O₂–Tr of iSL was examined in more detail. The total as well as most yields in HA–H₂O₂–Tr of iSL increased with decreases in the total amount added from the highest, 5.88, to 1.96 mmol (comparison between systems 15, 17, 18, and 19) but reversely decreased with a further decrease to the lowest, 0.98 mmol (comparison between systems 19, 20, and 21). HA–H₂O₂–Tr of iKL showed an identical tendency, and the addition of 1.96 mmol also afforded the maximum total yield. The maximum total yield obtained was thus 7.5 or 8.5 % in HA–H₂O₂–Tr of iKL or iSL, respectively. This maximum was no lower than those from isolated kraft or soda lignin achieved in most previous studies (Araújo et al. 2009, 2010; Bjelic et al. 2018; Borges da Silva et al. 2009; Fargues et al. 1996; Gomes and Rodrigues 2020a,b; Jeon et al. 2020; Khwanjaisakun et al. 2020; Liu et al. 2020; Mathias and Rodrigues 1995; Pinto et al. 2011; Singh and Ghatak 2017; Sutradhar et al. 2024; Vu et al. 2023; Zhang et al. 2020; Zirbes et al. 2020, 2023), as summarized in a previous report (Xu et al. 2023). Because iKL or iSL contained a large amount of polysaccharides (about 10 %), the maximum yield may be higher than 7.5 or 8.5 % when it is considered that V, VA, and AV originated from only organic materials derived from lignin. These facts clearly show the superiority of HA–H₂O₂–Tr over the methods employed in previous studies when HA–H₂O₂–Tr is considered not only ready for practical installation but also its high efficiency for the formation of V, VA, and AV.

The maximum total yield of V, VA, and AV in HA–H₂O₂–Tr of iKL reached 56.4 % of the total yield of V and VA obtained in ANO of iKL, which is the potential maximum of the total yield. This level was 55.2 % for iSL. An authors’ previous report suggested the following novel possibility: formation of the α-β-type of interunit linkage is a major mode of lignin condensation in alkaline cooking processes and progresses much more frequently than that of the α-5- or α-1-type, which has generally been considered to be the most common in condensation (Komatsu and Yokoyama 2022). The α-β-type of substructures present in iKL or iSL can probably be the origins of V and VA in ANO more efficiently than in HA–H₂O₂–Tr, which at least partly results in the maximum total yield in HA–H₂O₂–Tr being lower than in ANO.

Because the total yield of V, VA, and AV decreased with increases in the total amount of H₂O₂ added from 1.96 mmol, some produced V was suggested to be converted to something other than VA, the most plausible oxidation product of V. V was thus treated as the starting compound in HA–H₂O₂–Tr to examine whether VA was exclusively produced. The recovery yield of V and yield of VA were 78.0 and 3.9 %, respectively, when V was treated in HA–H₂O₂–Tr under the conditions of system 6 or 19. V was thus confirmed to be moderately degraded in HA–H₂O₂–Tr and converted to unknown compounds as well as VA, but only less than 20 % of the disappearing V was converted to VA. This is probably because the major active species in HA–H₂O₂–Tr is not HO₂⁻ but O⦁⁻, and these species have different reaction modes, as described in Section 3.1. Formation of the unknown compounds from V probably accompanied degradation of the aromatic nucleus by the attack of O⦁⁻, because the preliminary experiment described at the bottom of Section 3.2 suggested that the aromatic nucleus of a phenolic compound is degraded by O⦁⁻ despite the presence of an electrostatic repulsive force between the aromatic nucleus and O⦁⁻.

3.7 Effect of addition of different metal salts in HA–H₂O₂–Tr

CuCl₂ was added to HA–H₂O₂–Tr of iKL or iSL instead of FeCl₃ (systems 11–13 or systems 25–28, respectively), although its addition to HA–H₂O₂–Tr of VeG was not effective for the production of Ve and VeA, the analogues of V and VA, respectively, in the model study by the authors (Sakai et al. 2022). The addition was not effective for the production of V, VA, and AV in any system either. The total yield was always lower than that in HA–H₂O₂–Tr with the addition of FeCl₃ under the same conditions (comparison between systems 11 and 1, 12 and 2, 13 and 9, 25 and 14, 26 and 15, or 28 and 23). The total yield was higher with the addition of EDTA than in that of d-mannitol when CuCl₂ was the metal salt added, which is the opposite to the trend noted in systems with the addition of FeCl₃ (comparison between systems 12 and 13 or between 26, 27, and 28). Cu²⁺ (and Cu⁺ possibly generated) as well as precipitates of Cu hydroxides and/or oxides is considered to only slightly catalyze the unimolecular self-decomposition of HO₂⁻ to generate O⦁⁻ but markedly catalyze the bimolecular self-decomposition to afford O₂ and HO⁻, wasting H₂O₂ (actually HO₂⁻) from the viewpoint of producing V, VA, and AV. Because the yields of V in many systems were lower than 1.5 %, which is the V yield obtained when iSL was just dissolved in alkaline solution, HA–H₂O₂–Tr with the addition of CuCl₂ degraded V more markedly than that with the addition of FeCl₃.