Revisiting the condensation reaction of lignin in alkaline pulping with quantitativity part I: the simplest condensation between vanillyl alcohol and creosol under soda cooking conditions

The condensation reaction of lignin is believed to interfere with delignification in alkaline pulping processes, without any clear evidence, which has motivated us to quantitatively revisit it. This paper is the first of a series, and hence we employed the simplest model system using 4-hydroxymethyl-2-methoxyphenol (vanillyl alcohol, Va) and 2-methoxy4-methylphenol (creosol, Cr) under soda cooking conditions. The α-5-type condensation product between these compounds [VaCr, 2-(4-hydroxy-3-methoxybenzyl)-6-methoxy-4-methylphenol] was identified and quantified as exclusive. VaCr was yielded with a mole amount of 24%, 46%, 62%, or 72% based on that of disappearing Va at a reaction time of 120 min when the ratio of the initial concentration of Cr to that of Va was 1.0, 2.5, 5.0, or 7.5, respectively. These yields and an HPLC analysis of the reaction solution obtained by a treatment of Va as the sole compound under the same soda cooking conditions suggested the formation of self-condensation products of Va even in the treatments containing Cr. The obtained results comprehensively suggested that the self-condensation of Va progresses more readily than the condensation between Va and Cr. The factors behind this will be the topic of our next paper.


Introduction
Condensation is a major reaction mode of phenolic substructures of lignin in alkaline pulping processes. Figure 1 illustrates these condensation reactions of lignin on the basis of previous reports by Gierer et al. [1][2][3][4], although they often employed compounds that are not appropriate as lignin models. First of all, condensation progresses only between phenolic substructures. The primary reaction is the liberation of the hydroxide anion (HO − ) from the benzyl position (α-position) of a dissociated phenolic (phenoxide) substructure to afford the important intermediate, a quinone methide structure (QM, Fig. 1). The aromatic nucleus of another phenoxide substructure nucleophilically attacks the α-carbon of QM to afford an α-5-or α-1-type condensation product. The side-chain portion is released as an aldehyde fragment in the formation of the latter type. Formaldehyde (HCHO) is often liberated from a γ-hydroxymethyl group during the processes, and reacts with the aromatic nucleus of a phenoxide substructure to introduce a hydroxymethyl group on the empty ortho-position. This ortho-hydroxymethylated substructure similarly liberates the HO − to convert to the ortho-type QM (o-QM, Fig. 1), which is also nucleophilically attacked by the aromatic nucleus of another phenoxide substructure to afford a 5-CH 2 -5-type product.
It was thus confirmed that the α-5-, α-1-, and 5-CH 2 -5-type condensations actually progress and that the condensation products are stable under alkaline conditions. As the progress of condensation is accompanied by polymerization of lignin, it has been considered to be undesirable for delignification in alkaline pulping processes. This undesirability is further reinforced by the fact that residual lignin in pulp or lignin-originating organic compounds in black liquor, whose structures are not well understood, affords only small amounts of vanillin and its analogues in the alkaline nitrobenzene oxidation method, although low yields of these compounds do not show abundance of condensed substructures in the lignin samples but only absence of the uncondensed substructure.
However, there are several issues with unquestioningly concluding that condensation is absolutely undesirable for delignification. The above-described knowledge is not quantitative but qualitative. Gierer and Ljunggren showed that a trimeric lignin model compound of α-5type condensation product carrying the β-O-4 bond undergoes the β-O-4 bond cleavage much more readily than a common dimeric β-O-4-type lignin model compound, owing to the ready neighboring group participation of the phenoxide of the condensed aromatic nucleus (Fig. 2) [14]. Although this finding originates from model experiments and hence cannot be directly applied to an actual alkaline pulping process, it suggests that the β-O-4 bond is more readily cleaved when a common phenolic β-O-4-type substructure condenses at its aromatic C-5  [1][2][3][4] position to convert to an α-5-type condensed substructure with the β-O-4 bond. In other reports, α-5-type condensation substructures are primarily generated by an acidic pretreatment together with p-cresol to enhance the neighboring β-O-4 bond cleavage in a following alkaline cooking process [15][16][17][18][19][20]. It was also shown that an analogous trimeric model compound of α-1-type condensation product undergoes β-O-4 bond cleavage, although the rate is lower than that of a common dimeric β-O-4-type lignin model compound. Therefore, the α-1-type condensation does not strongly interfere with the cleavage of the neighboring β-O-4 bond (Fig. 2) [14]. It is thus unclear whether the progress of condensation reactions invariably interferes with delignification in alkaline pulping processes, and we were motivated us to quantitatively revisit the condensation reactions in these processes.
Because this paper is the first in a series, the simplest system was employed. Two guaiacyl-type phenolic lignin model compounds, vanillyl alcohol (Va, Fig. 3) and creosol (Cr, Fig. 3), were treated together under soda cooking conditions. The disappearances of Va and Cr were examined in detail, with identification and quantification of the condensation products.

Materials
All chemicals were purchased from Fujifilm Wako Pure Chemical Industries Co. (Osaka, Japan), Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan), or Sigma-Aldrich Japan K. K. (Tokyo, Japan), and used without further purification except for Va. Va was recrystallized from methanol (MeOH) before use. High purity of Va and Cr were confirmed by 1 H-NMR (JNM-A500, 500 MHz, JEOL Ltd., Tokyo, Japan) before use. Deionized water (H 2 O) was used.

Soda cooking for isolation and identification of condensation product
A solution (100 mL) was prepared containing sodium hydroxide (NaOH, 1.06 mol/L), Va (10 mmol/L), and Cr (50 mmol/L), to simulate a soda cooking process. A portion of the solution (5.0 mL) was transferred into a stainless-steel autoclave (10 mL volume, Taiatsu Techno ® Co., Tokyo, Japan), and the air present in the head space was replaced with nitrogen gas. A further 19 autoclaves with exactly the same contents were simultaneously prepared. All 20 autoclaves were immersed together in an oil bath at 150 °C and left soaking with shaking.
After 240 min, all the autoclaves were taken out, immediately immersed in an ice/water-bath, and then the content of each was neutralized with acetic acid (AcOH, 0.6 mL). The combined neutralized solution was extracted with dichloromethane three times and then once with ethyl acetate (EtOAc). The combined organic layer was washed with H 2 O and brine, dried over anhydrous sodium sulfate, and concentrated by an evaporator under reduced pressure to afford a brown syrup (776 mg). The syrup was fractionated by preparative thin layer chromatography (PTLC) using n-hexane/ EtOAc mixture (2/1) as the eluent. A crystal (102 mg) was obtained from an area of the PTLC surface, where a condensation product could exist, by extraction of the  area with EtOAc and successive evaporation. The crystal was recrystallized from 50% aqueous MeOH, which afforded a white needle crystal (87 mg).

Soda cooking for quantitative analysis
A reaction solution (5.0 mL) was prepared containing NaOH, Va (10 mmol/L), and Cr using degassed H 2 O, and transferred into the stainless-steel autoclave. Table 1 lists the initial concentrations of the contents and systems employed in this work. The air present in the head space was replaced with nitrogen gas. A further six autoclaves with exactly the same content were simultaneously prepared. All 7 autoclaves were immersed together in an oil bath with shaking at 130 °C, 150 °C, or 170 °C and left soaking for 0, 10, 20, 30, 45, 60, or 120 min with shaking. All reactions were conducted three times to confirm reproducibility.

Quantification
One autoclave was taken out from the oil bath at each specified reaction time and immediately immersed in an ice water bath. AcOH (0.6 mL) was added for neutralization followed by addition of a MeOH solution (5.0 mL) containing the internal standard compound (IS), 4-hydroxybenzaldehyde. After thoroughly shaking the autoclave, a portion of the content was filtered with a membrane filter, and the filtrate was analyzed by HPLC (LC-2010C HT , Shimadzu Co.) equipped with an UV-Vis detector for quantification (280 nm). Conditions for the HPLC analysis were the same as described above.

Building the reaction system
The simplest system was desirable, because this is the first paper of the series. As shown in Fig. 1, QM primarily forms accompanying the liberation of HO − from the α-position of a phenoxide substructure in an alkaline pulping process. Va is the simplest lignin model compound that can convert to a QM, because it has the phenolic hydroxy group, the shortest side-chain with the α-hydroxy group, and the guaiacyl nucleus that is a characteristic aromatic nucleus of lignin. QM is nucleophilically attacked by the aromatic nucleus of another phenoxide substructure to mainly afford an α-5-or α-1type condensed substructure. An aldehyde fragment is released in the formation of the latter type. A prerequisite for the simplest lignin model compound as this nucleophile is to have a guaiacyl nucleus and a methyl group as the side-chain that does not have any hydroxy group to quench the α-1-type condensation with the QM derived from Va (QM Va , Fig. 3). High stability under alkaline conditions is another prerequisite. Cr is thus the simplest lignin model compound as this type of nucleophile.
Soda cooking conditions were employed, since this is the simplest alkaline pulping process. Because the condensation between two molecules of Va (Va and QM Va , self-condensation) should be suppressed as much as possible to simplify the reaction, the initial concentration of Cr should be higher than that of Va. The initial concentrations of Va and Cr were thus set to 10 and 50 mmol/L, respectively, in the isolation and identification of condensation products. In a reaction using these compounds with these initial concentrations, 60 mmol/L of NaOH is always consumed by these compounds before the soda cooking reaction progresses. Thus, the concentration of NaOH was set to be the same as the total of the initial concentrations of Va and Cr plus 1.0 mol/L in all reactions, to maintain a concentration of 1.0 mol/L after the consumption of NaOH by the initially added Va and Cr. Table 1 lists the reaction systems employed not for the isolation and identification but for the quantitative analysis. Several different initial concentrations of Cr and temperatures were employed. The initial concentration of Va was always 10 mmol/L. Thus, the α-5-type condensation product between Va and Cr, 2-(4-hydroxy-3-methoxybenzyl)-6-methoxy-4-methylphenol (VaCr, Fig. 3), was expected to form as the exclusive major product, at least in systems Cr50 and Cr75.

Identification of condensation product
Va and Cr were treated together to isolate and identify condensation products under the same conditions as system Cr50 for the quantitative analysis, except that the scale was larger than that of this system. The white needle crystal of the isolated condensation product was obtained in yields of 38 and 32 mol% before and after recrystallization, respectively, based on the initial mole amount of Va. In accordance with our expectation, this isolated compound was identified as VaCr, the α-5-type condensation product between Va and Cr, on the basis of all the observed NMR and MS spectra. Figure 4a, b, c, d shows the 1 H-NMR spectrum, its aromatic region, 13 C-NMR spectrum, and MS spectrum of VaCr, respectively. The 13 C-NMR DEPT135 (Additional file 1: Fig.  S1), 2D-NMR 1 H-1 H COSY (Additional file 1: Fig. S2), 2D-NMR 1 H-13 C HSQC (Additional file 1: Figs. S3-S6), and 2D-NMR 1 H-13 C HMBC spectra (Additional file 1: Figs. S7-S11) are shown in the Additional file. The peaks appearing in the 1 H-NMR, 13 C-NMR, and MS spectra were assigned as below. In the following assignments, the letter 'Va-' or 'Cr-' indicates from which compound, Va or Cr, the targeted sites originated.  Figure 5d shows the HPLC chromatogram of the reaction solution in system Cr50 that was performed not for isolation and identification of VaCr but for the quantitative analysis. All the conditions employed in system Cr50 were the same as those in the reaction for isolation and identification of VaCr except that the reaction was terminated at a reaction time of not 240 but 120 min and IS was added. Therefore, Fig. 5d can be substituted for the chromatogram drawn in the analysis of the reaction for isolation and identification of VaCr. The peak appearing at a retention time of 23.6 min corresponds to VaCr. Dimers and molecules larger than dimers appeared after a retention time of about 20 min. VaCr was thus the exclusive major condensation product, which is reasonable given that the absorptivities of the condensation products at 280 nm are not largely different from one another. VaCr was quantified by preparing a calibration line using this isolated and identified VaCr together with IS in the following contents.

Overview of the condensation in systems Cr0-Cr75
Va and Cr were treated under the soda cooking conditions in systems Cr0-75, where the initial concentrations of Cr were different, to examine how the condensation reaction differed in these systems. Figure 5a, b, c, d, e shows the HPLC chromatograms of the reaction solutions obtained at a reaction time of 120 min in systems Cr0, Cr10, Cr25, Cr50, and Cr75, respectively, after adding IS and filtration. The vertical scales were standardized by the peak heights of IS. Many peaks in Fig. 5b-e appear at the same retention times as those in Fig. 5a. Thus, self-condensation products of Va and disproportionation products described below must have formed even in system Cr75, where the initial mole amount of Cr was 7.5 times that of Va, although the formation of these products requires further confirmation by other methods. The peaks of Va are smaller with increasing initial concentration of Cr, indicating that Va condensed more rapidly with the increase. The peaks of VaCr are larger with increasing initial concentration of Cr, which indicates that VaCr became the more exclusive condensation product with the increase.
The peaks at retention times of 12.2 min and 13.7 min showed vanillin (V, 4-hydroxy-3-methoxybenzaldehyde) and guaiacol (G, 2-methoxyphenol), respectively. V was isolated from the reaction solution and identified by the retention time and structural analyses by 1 H-NMR and LC/MS (see Additional file 1). The authentic compound of G appeared at the same retention time and showed the same UV-Vis spectrum as the peak at 13.7 min when the reaction solution and authentic compound were analyzed by a photodiode array (PDA) detector (SPD-M10A, Shimadzu Co.) in the HPLC apparatus, although complete identification should be additionally conducted by other methods. V is presumed to have formed via a reaction route whereby Va is oxidized by QM Va , namely, the disproportionation of Va (Fig. 3). Several previous reports indicate that phenolic compounds can be oxidized by QMs [21][22][23][24][25][26]. The yields of V with standard deviations from three duplicate runs (shown in parentheses) were 4.5% (± 0.4), 3.0% (± 0.4), 2.5% (± 0.1), 1.2% (± 0.1), and 1.3% (± 0.4), based on the initial mole amount of Va at a reaction time of 120 min in systems Cr0, Cr10, Cr25, Cr50, and Cr75, respectively, when 70%, 85%, 91%, 97%, and 99% of Va disappeared, respectively. These yields became smaller with increasing initial concentration of Cr, which suggests that the increase suppressed the reaction between Va and QM Va and hence supports the above-described formation route of V. The formation of Cr would accompany this disproportionation, although neither Cr nor VaCr was detected in system Cr0 (Fig. 5a). G is presumed to have formed when the α-1-type self-condensation followed a route whereby the σ-complex intermediate was generated and then the π-electron system of the aromatic nucleus originating from QM Va abstracted the proton from the hydroxymethyl group originating from Va (Fig. 3). The reformation of QM Va would accompany this route. The yields of G were 5.5% (± 0.2), 4.4% (± 0.1), 3.9% (± 0.0), 2.7% (± 0.1), and 2.4% (± 0.2), based on the initial mole amount of Va in the same reactions as described above for V. Those yields were also smaller with increasing initial concentration of Cr, which suggests that the increase suppressed the reaction between Va and QM Va and hence supports the above-described formation route of G. Figure 6a, b, c, d, e shows the time courses of the changes in the concentrations of Va, Cr, and VaCr in systems Cr0, Cr10, Cr25, Cr50, and Cr75, respectively. Figure 6f, g, h shows those in the recovery yields of Va, Cr, and in the yields of VaCr, respectively, in these systems based on the initial amounts of Va, Cr, and Va, respectively.

Detailed analysis of the disappearance of Va and formation of VaCr
Cr hardly disappeared when it was treated as the sole compound under the same conditions as the systems employed in this work, which shows its high stability under the employed conditions and also the complete absence of oxygen from the system judging from the lability of Cr to oxygen oxidation under alkaline conditions [25]. In spite of this, the mole amounts of disappearing Cr were roughly about 1.5 times larger than those of the generated VaCr, the only exclusive condensation product identified in this work, at a reaction time of 120 min in most runs. This phenomenon cannot currently be explained. Cr may have condensed with some compounds to afford products other than VaCr, although such condensations do not seem to have afforded any major products. These minor products will be identified in our forthcoming paper.
The yields of VaCr monotonically increased in all the systems and seem to have approached ceiling levels, indicating that VaCr does not frequently condense further as a nucleophile and is rather stable. The ceiling levels seemed to be 20-25%, 40-45%, 55-60%, and 65-70% in systems Cr10, Cr25, Cr50, and Cr75, respectively, although Va still remained with yields of 14.7% (± 0.8) and 5.7% (± 0.0) in the former two systems, respectively, at a reaction time of 120 min. These levels and the ratios of the initial amounts between Va and Cr suggest that the self-condensation of Va (as well as the disproportionation) progresses more readily than the condensation between Va and Cr. It should be noted that there are several routes for self-condensation, and not a major but many minor products form. Although the self-condensation of Va consumes more Va than the condensation between Va and Cr, the yield of VaCr at a reaction time of 120 min in system Cr10 was much lower than the half amount of consumed Va based on the initial amount of Va (43%). The ceiling levels in the other systems also seem to indicate more labile progress of the self-condensation of Va.
There must be some factors that promote self-condensation, because the electron density of the aromatic nucleus, which is a factor commonly used to judge the reactivity of an aromatic nucleus as a nucleophile, is not largely different between Va and Cr, and hence the aromatic nuclei of Va and Cr should have similar reactivities as nucleophiles. Two possible factors are as follows: (i) Va can additionally condense at its aromatic C-1 position owing to the presence of the benzyl hydroxy group; and (ii) Dimers produced by the self-condensation of Va can further condense to be larger self-condensation products, resulting in acceleration of the disappearance of Va. It is one of our research topics to examine what factors actually promote the self-condensation of Va. Kinetic analyses are required to quantitatively compare the reactivities between Va and Cr as nucleophiles toward QM Va .
Although self-condensation products larger than dimers were produced, and the disproportionation would reproduce QM Va in system Cr0, it was noteworthy that the disappearance rate of Va approximated to a common second-order reaction rate equation: concentration of Va at a reaction time of t, k Cr0 : second-order reaction rate constant). Because the reaction period before 20 min involves a temperature increase [26][27][28] and the remaining yield of Va at a reaction time of 120 min could be too low for inclusion in the approximation, the data points at reaction times of 20, 30, 45 and 60 min were employed. The k Cr0 value obtained by three duplicate runs was 1.77 ± 0.10 L mol −1 min −1 , with three squared correlation coefficients (R 2 ) in each run: 0.992, 0.997, and 0.997. The major reaction in system Cr0 can thus be the selfcondensation between two molecules of Va.
Not only all reactions occurring in system Cr0 but also the condensation between Va and Cr to afford VaCr could progress in systems Cr10, Cr25, Cr50, and Cr75, in addition to other minor reactions. Therefore, the disappearance rates of Va could have been expressed well by a second-order reaction rate equation concentration of Cr at a reaction time of t, k: second-order reaction rate constant). However, the approximations employing the data points at reaction times of 20, 30, 45 and 60 min were poor in all the systems. When the k value was individually calculated at each of these four reaction times, it increased with the progress of the reaction (0.601-1.27, 0.460-0.925, 0.479-0.850, or 0.203-0.823 L mol −1 min −1 in system Cr10, Cr25, Cr50, or Cr75, respectively) and showed the maximum value at a reaction time of 45 min. Although the term, k[Va] t [Cr] t , in the above equation is expressed only by the concentrations of Va and Cr, it actually connotes the rate of the disappearance of Va caused by not only the condensation between Va and Cr but also all reactions that did not occur in system Cr0 but progressed in the other systems owing to the co-presence of Cr. It is thus suggested that the latter kind of reaction becomes more favorable with the progress of the reactions in these systems. The k values became smaller with increasing initial concentration of Cr, which suggests that the latter kind of reaction becomes favorable with a degree smaller than that expected from the increase in the initial concentration. As even the maximum values were smaller than the k Cr0 , the rate constant of the former term in the right side of the equation, it can be safely concluded that the self-condensation of Va progresses more readily than the condensation between Va and Cr even in the systems containing Cr.
Because VaCr did not clearly decrease after its formation, as shown in Fig. 6h, the formation rates of VaCr could have been expressed well by a second-order reaction rate equation: d[VaCr] t /dt = k′[Va] t [Cr] t (k′: second-order reaction rate constant). The approximations employing the data points at reaction times of 20, 30, 45 and 60 min were not good either, although the deviations were smaller than those in the previous approximations. When the k′ value was individually calculated at each of these four reaction times, the k′ values at these times were in ranges of: 0.515-0.678, 0.512-0.567 (0.147 at 60 min), 0.452-0.544, or 0.418-0.588 (0.231 at 60 min) in system Cr10, Cr25, Cr50, or Cr75, respectively. The k′ values seem to be smaller with increasing initial concentration of Cr. Because the formation rate of VaCr would be the proportional connection to initial concentration of Cr when the k′ values are constant regardless of the initial concentration, the formation of VaCr is accelerated by increasing the initial concentration to a degree smaller than the proportional connection. The ceiling levels of the formation of VaCr estimated from Fig. 6h also seem to show this phenomenon. Because the k′ values of the ranges were about 1/4-1/3 of the k Cr0 value, the formation of VaCr was slower than the self-condensation of Va. This is partly because self-condensation consists of not the only route affording the α-5-type product but also routes generating the α-1-type, other dimers, and products larger than dimers, as described above. The k′ value contrarily corresponds to the only reaction route affording the α-5-type condensation product between Va and Cr, VaCr. Identification and quantification of the selfcondensation products of Va are required to compare the formation rates between possible condensation reactions. Figure 7a, b, c shows the time courses of the changes in the concentrations of Va, Cr, and VaCr in systems 130, 150, and 170, respectively. System 150 is identical to system Cr50. Figure 7d, e, f shows those in the recovery yields of Va, in those of Cr, and in the yields of VaCr in these systems, respectively, based on the initial mole amounts of Va, Cr, and Va, respectively.

Effect of temperature on the condensation between Va and Cr
The disappearances of Va and Cr and formation of VaCr were naturally more rapid with rising temperature. The ceiling levels of formation of VaCr at 150 °C and 170 °C were about 60%, based on the initial mole amount of Va. Because the ratio of the formation amount of VaCr to the disappearing amount of Va seems to be also about 60% during the whole reaction at 130 °C, the ceiling level can also be about 60% at 130 °C. The consumption of Cr was thus greater than the formation amount of VaCr at any temperature.
When a pseudo-first-order reaction can be applied to the approximation of the disappearance of Va, the change in the concentration of Cr during the reaction should be negligible or much smaller than that of Va. Although the concentrations of Cr clearly decreased at all temperatures, as shown in Fig. 7e, a pseudo-first-order reaction rate equation: − d[Va] t /dt = k obs [Va] t (k obs : pseudo-firstorder reaction rate constant) was applied to the approximation of the disappearance rates of Va on trial. Table 2 lists the values of k obs and R 2 in three duplicate runs. As a result, all the approximations were good. The Arrhenius activation energy (E a ) and frequency factor (A) were   [26], the values were 130 kJ/mol and 2.94 × 10 14 min −1 , respectively, for the erythro isomer, and 133 kJ/mol and 1.66 × 10 14 min −1 , respectively, for the threo isomer. The k obs values obtained in this work do not correspond to a specific condensation reaction but comprehensively to the whole set of reactions occurring in the systems, while those in our previous report were specifically for the β-O-4 bond cleavage reaction. Nevertheless, a comparison of the E a and A values between this and our previous reports should have some meaning. The E a value obtained in this work is smaller than that in our previous report, which indicates that the k obs value varied with temperature less significantly in this work than in our previous report. The A value obtained in this work is smaller than that in our previous report, which indicates that the k obs value in this work is smaller than that in our previous report in a high-temperature region. These findings suggest that condensation can progress more rapidly than β-O-4 bond cleavage in a relatively low-temperature region, although no concrete temperature region can yet be described. Temperature may have to be raised to the maximum as quickly as possible in a soda cooking process to minimize condensation.

Conclusions
The condensation reaction of lignin was quantitatively examined in the simplest model systems: phenolic lignin model compounds, Va and Cr, were treated together under soda cooking conditions with various different initial concentrations of Cr. The α-5-type condensation product between these compounds, VaCr, was isolated and identified in accordance with previous reports. The quantitative analyses in this work showed that VaCr is the exclusive condensation product and rather stable under these conditions. The yields were 24%, 46%, 62%, or 72% based on the mole amount of disappearing Va at a reaction time of 120 min when the ratio of initial mole amount of Cr to that of Va was 1.0, 2.5, 5.0, or 7.5, respectively. These yields are not high enough to say that the condensation between these compounds is the exclusive major reaction in the systems. This fact and the comparison with the treatment of Va as the sole compound suggested that the self-condensation of Va progresses more readily than the condensation between Va and Cr, even in the system containing the largest amount of Cr, affording various minor condensation products.
The same treatments at three different temperatures, 130 °C, 150 °C, or 170 °C, may suggest that the condensation reaction progresses more rapidly than β-O-4 bond cleavage in a relatively low-temperature region in the soda cooking process, and hence the temperature should be raised to the maximum as soon as possible.