Using guaiacol as a capping agent in the hydrothermal depolymerisation of kraft lignin

2.1 Materials

Softwood kraft lignin isolated from black liquor by the LignoBoost process at the Bäckhammar Mill in Sweden from Pinus sylvestris and Picea abies, with a dry content of 82.9 % ± 1.2 % (Sartorius MA30 moisture analyser; Sartorius, Göttingen, Germany) and a sulphur content of 1.97 wt% ± 0.2 wt% (Ahlbom et al. 2021), was depolymerised in a mixture of guaiacol (≥99 %, Sigma-Aldrich, Steinheim, Germany), anhydrous sodium carbonate (≥99.9 %, VWR Chemicals, Leuven, Belgium), sodium hydroxide (≥99 %, Merck, Darmstadt, Germany) and deionised water.

The following were all used as received for the analytical measurements: LiBr (>99 %, Sigma-Aldrich, Steinheim, Germany), dimethyl sulphoxide (DMSO, >99.7 %, Sigma-Aldrich, Steinheim, Germany), pullulan standards (PL2090-0100, Varian, Church Stretton, UK), sodium standard (UltraScientific, North Kingstown, USA), diethyl ether (DEE, >99.0 % with 1 ppm BHT as the inhibitor, Sigma-Aldrich, Steinheim, Germany), nitric acid (65 %, Merck Suprapur, Darmstadt, Germany), syringol (99 %, Sigma-Aldrich, Steinheim, Germany), DMSO-d₆ (99.5 atom% D, 0.03 (v/v) TMS, Sigma-Aldrich, Sigma-Aldrich, Steinheim, Germany), 1,2-dihydroxybenzene (>99 %, Sigma-Aldrich, Steinheim, Germany), guaiacol (Sigma-Aldrich, Steinheim, Germany), 1,2-dimethoxybenzene (99 %, Sigma-Aldrich, Steinheim, Germany), 4-methyl guaiacol (≥98 %, Sigma-Aldrich, Steinheim, Germany), 4-ethylguaiacol (≥98 %, Sigma-Aldrich, Steinheim, Germany), vanillin (99 % Sigma-Aldrich, Steinheim, Germany), acetovanillone (≥98 %, Sigma-Aldrich, Steinheim, Germany), homovanillic acid (Sigma-Aldrich, Steinheim, Germany) and HCl (1 M, Honeywell Fluka, Seelze, Germany).

2.2 Methods

Softwood kraft lignin was depolymerised at a mass fraction of 5 wt% in a mixture of water, sodium carbonate (1.6 wt%), sodium hydroxide (1.0 wt%) and guaiacol (0–2 wt%). The reactor employed was a custom-made 99 ml batch reactor that allows a charge (containing lignin, base, water and guaiacol mixed for 15 min at 15,000 rpm with an UltraThurrax [IKA T25; IKA-Werke GmbH & Co. KG., Staufen, Germany]) to be injected into a preheated pre-charge. Non-reactive species in the precharge, such as water and salts, can thereby be heated to a temperature greater than that of the reaction. The injection is thus made into a mixture of vapour and liquid, where the condensation of the vapour heats the injection charge. The reaction charge quickly reaches the reaction temperature desired which, together with a rapid cooling down after the reaction, is beneficial to the hydrothermal treatment of biomass (Brand et al. 2014). A more in-depth description of the reactor is given by Arturi et al. (2017).

The injection causes the temperature in the reactor to fall, after which the control system of the reactor aims at re-establishing the set point temperature; this is also accompanied by small variations in pressure but these can be controlled by minor injections and ejections (<1 ml) of reaction material. In this work, the residence time of the reaction mixture was defined as the time between the end of the injection and the start of the discharge. Since the temperature and pressure vary during the residence time, an average reaction temperature and an average reaction pressure were defined as

where y is either pressure or temperature, and t₁ and t₂ are the times of the end of the injection and the start of the discharge, respectively.

At the end of the residence time in the reactor, the product was discharged into a flask containing 200 g water placed in an ice-bath to quench the reacting mixture and halt the reactions. No quantification, nor identification, of gases formed was made. A sample was taken to measure the guaiacol content in the product fraction by gas chromatography using a flame ionisation detector (GC-FID). The product was then fractioned into three parts: char, precipitated solids (PS) and acid soluble organics (ASO). This fractionation is outlined in Figure 1, along with the analytical procedures. The reactor product was divided in two product fractions and each was filtered over a #5 glass filter, with a nominal cut-off of 1.0–1.6 µm. The resulting filter cake was washed with a water amount equivalent to half the filtered product liquor to remove residual product liquor containing guaiacol, and was dried at 40 °C for four days to give the char fraction. Data presented of the char yield is reported as the average and standard deviation of the two filtrations.

Figure 1:

Fractionation flow chart of the reactor product.

The filtrate was acidified with 1 M HCl to pH 1.5 to precipitate the solids, which were filtered over a #5 glass filter. The resulting filter cake was washed with an equal amount of water as the acid filtrate sample, again to dispel potential residual guaiacol. The filter cake was then dried at 40 °C for five days, thereby forming the PS fraction.

The remaining acid filtrate was dried at ambient conditions for three days, followed by drying at 40 °C for a day, to give the ASO fraction. It is assumed that residual guaiacol from the reaction mixture will be found in this ASO phase because guaiacol is soluble in the reaction mixture. Salt crystals, most likely NaCl resulting from acidification of the Na₂CO₃ and NaOH with HCl, were found in the ASO fraction as the water was being evaporated. In order to quantify this amount of NaCl, the sodium content of the acidified filtrate was measured by inductively-coupled plasma – optical emission spectrometry (ICP-OES) and, assuming all of the sodium was present as NaCl, this content of salt was deducted from the ASO yield.

Yields of char, PS, ASO and monoaromatic compounds were calculated based on the dry lignin loaded to the reactor accordingly:

with Y _i being the yield of fraction i, m _i the mass of fraction i and m_{dry lig} the mass of dry weight lignin loaded to the reactor.

2.3 Analytical procedures

2.4 Test plan

The test plan included 10 experimental runs, with residence times in the range 1–12 min, for four nominal levels of guaiacol mass fractions in the range 0–2 wt%, see Table 1. The injection of the charge containing the lignin and guaiacol aimed to reach a pressure in the reactor of 250 bar after heating to the reaction temperature. The resulting mass fraction of lignin was in the range 5.1–5.4 wt%, while the mass fractions of guaiacol at the four levels were 0, 0.2, 1.1 and 2.1–2.2 wt%, as reported in the table. A control sample without lignin was run to investigate the stability of guaiacol under the reaction conditions.

It was estimated from the experimental data of a previous work (Ahlbom et al. 2022) that the mass fraction of guaiacol in the product mixture could be around 0.2 wt%. This was chosen as the lowest mass fraction level to evaluate the effect of the capping agent as opposed to no guaiacol loading. Furthermore, previous work (Belkheiri et al. 2018) also showed that increasing the mass fraction above 2 wt% of the capping agent, in that case phenol, did not improve the yield of either bio-oil or water-soluble organics significantly. Consequently, the highest mass fraction of guaiacol added in this study was chosen as 2 wt%. Moreover, 2 wt% guaiacol in the feed was also close to the limit of mass fraction of guaiacol in the injection charge at which it was possible to mix and pump the charge in the injection system: at higher mass fractions, the separation of a highly viscous guaiacol-rich phase was observed.

There are two duplicate sets of experiments, with a nominal mass fraction of 0 and 2 wt% guaiacol in the reactor feed, respectively, at 4 min residence time. The yield data and content of guaiacol after the reaction are presented as averages of the two experiments, along with their standard deviations.

3.1 Reactions of guaiacol

In Figure S1, the content of guaiacol remaining in the product is presented as fractions of the mass fraction of guaiacol in the reactor feed. The control experiment, without lignin but with guaiacol, was performed in order to investigate the reactions of the guaiacol. Neither char nor PS were then formed, but Figure S1 shows that about 45 % of the guaiacol in the feed was consumed in the control experiment, thereby confirming that guaiacol reacts during the treatment. The products of the guaiacol reaction, however, remain dissolved both before and after acidification to pH 1.5 and end up in the ASO fraction. The oxygen content of guaiacol could possibly result in the products of the reacted guaiacol having oxygen-containing functional groups that cause it to remain dissolved, or be present in some colloidal form, thus avoiding precipitation in the aqueous product phase.

The molecular weight distribution of the ASO fraction of the control experiment is shown in Figure S2, together with samples run at the same reaction conditions but also including lignin. As can be seen, the ASO fraction without lignin shows components of high molecular weight. Indeed, oligomeric and polymeric fractions are formed. These fractions show a weight average molecular weight (M _w ) of 8.0 kDa, see Table S1, which indicates a degree of polymerisation of guaiacol of about 65. Polymerisation of guaiacol has been reported previously in supercritical water and in pyrolysis; in these cases, the polymerised guaiacol formed a precipitating char fraction (Lawson and Klein 1985; Wahyudiono et al. 2011, 2007). However, at sub-critical conditions, 250 °C, no char was formed but crosslinking of guaiacol derived phenolic structures from was noted (Wahyudiono et al. 2011).

An investigation into the behaviour of guaiacol at high pH at ambient temperature was also carried out to verify whether high pH alone causes polymerisation. No formation of high M _w products occurred, though the mixture turned from clear via blue to dark brown in 2 h, similar to the observations of Taurog et al. (1992). It was thus confirmed that polymerisation of guaiacol did not occur spontaneously in the reaction mixture but required elevated temperatures and pressures.

Identification of monoaromatic compounds with GC-MS in the product mixture of the control experiment, with guaiacol but without lignin added, showed that guaiacol, aside from polymerising, also formed 1,2-dimethoxybenzene. At higher temperatures, and without any alkaline components added, guaiacol was instead reported to form catechol, phenol, o-cresol and methanol (Lawson and Klein 1985; Wahyudiono et al. 2011, 2007).

The reactions of guaiacol, with lignin in the reaction mixture, are presented in the following sections. Figure 2 shows the mass of guaiacol in the product mixtures at varying residence times (A) and additions of guaiacol (B). From the experiments without any guaiacol in the feed, it can be seen that the formation of guaiacol starts within the first minute of reaction, and that the net rate of formation slows down between 1 and 4 min, see Figure 2A. Furthermore, when starting with 2 wt% guaiacol in the feed, the rate of consumption of guaiacol is higher in the first minute and decreases as the reaction time increases, see Figure 2A. Formation of guaiacol from lignin could include retro-aldol reactions cleaving of side chains from the guaiacyl unit similar to reactions during kraft pulping (Crestini et al. 2017; Giummarella et al. 2020). Guaiacol is therefore both formed and consumed during hydrothermal treatment. In the absence of guaiacol in the feed, there is a net production of guaiacol from lignin depolymerisation whereas, in the presence of 1 or 2 wt% guaiacol in the feed, the net rate of consumption of guaiacol in reactions with the lignin-derived fragments is higher than the net rate of formation of guaiacol from lignin depolymerisation. Strangely enough, at a low loading of guaiacol (0.2 wt%), the guaiacol was formed at a slightly higher rate than it was consumed: the final content of guaiacol was 114 % of the amount added, see Figure S1B. It is interesting to note that the fraction of guaiacol remaining after the reaction was lower when lignin was added, see Figure S1B. Although the exact reactions are unknown, it is likely that there are several reaction paths, particularly for the consumption of guaiacol because it can react with different lignin fragments. It is possible that lignin increases the consumption of guaiacol by reacting with it.

Figure 2:

Mass of guaiacol in the reaction mixture versus residence time (A) and mass fraction of guaiacol in the reactor feed (B) at average reactor temperatures and pressures of 288–301 °C and 214–251 bar, respectively. Data for 0 min of residence time in (A) is the average amount of guaiacol injected into the reactor.

The high M _w products from guaiacol polymerisation are not observed in the samples with lignin, see Figure S2. This indicates that guaiacol reacts faster with fragments generated by depolymerisation of lignin than with itself, else there should be traces of guaiacol oligomers in the molecular weight distributions of the ASO fractions with lignin added. An alternative possibility is that the higher pH when guaiacol is run without lignin, seen in the higher product phase pH in Table 1, could promote polymerisation of guaiacol.

3.2 Yields of product fractions

The yields of the three product fractions char, PS and ASO versus the dry lignin loaded are presented in Figure 3. The dominating product fraction is clearly the PS. It is likely that most of the non-reacted guaiacol ended up in the ASO fraction, since guaiacol is soluble in the reaction mixture; the yields for this fraction are presented with this guaiacol included.

Figure 3:

Yields of product fractions run at average reactor temperatures and pressures of 288–301 °C and 214–251 bar at varying mass fractions of guaiacol in the feed at 1 min (A) and 4 min (B) residence times, respectively, and at varying residence times with 0 wt% (C) and 2 wt% (D) guaiacol in the feed, respectively.

The amount of char formed increases with increasing guaiacol charge and residence time, see Figure 3. The filter cake forming the char was washed with water and only showed minor amounts of guaiacol in the molecular weight distribution, see Figure S3. This indicates that the increase in yield of char when guaiacol is added is an effect of guaiacol promoting char formation, rather than non-reacted residual guaiacol ending up in the char phase.

In Figure 3A it can be seen that, at a short residence time of 1 min, the addition of 2 wt% guaiacol in the reaction mixture results in a char yield of 12 % ± 4 %, which should be compared to a yield of 0 % when no guaiacol was added. Guaiacol thus reacts rapidly (within 1 min) with the depolymerised lignin molecules and forms new molecules that, to some extent, end up in the char; since they precipitate, it is likely that these new molecules are non-polar (Ahlbom et al. 2022). At a longer residence time of 4 min, char is formed without any guaiacol in the feed, see Figure 3B and C. Consequently, while the reaction conditions are thus sufficiently harsh to yield char at 4 min without guaiacol, adding guaiacol causes its formation to occur sooner, at 1 min already. On the other hand, the PS fraction is largely unaffected by the guaiacol addition, see Figure 3A and B, but the yield decreases with increasing residence time, see Figure 3C and D. The yield of the ASO fraction is seen to increase somewhat with increased guaiacol feed and residence times with guaiacol added, see Figure 3B and D.

3.3 Molecular weights

The weight average molecular weights, M _w , of the product fractions and the original lignin are presented in Tables 2 and 3. It is clear that hydrothermal treatment reduces the M _w of the products (char, PS and ASO fractions) compared to the original lignin. The PS fraction has the highest M _w , the char fraction shows a lower M _w and the ASO exhibits the lowest M _w . The molecular weight distributions from which the M _w in Table 2 are calculated are given in Figures 4 and S4.

Figure 4:

Molecular weight distributions of the char (1), PS (2) and ASO (3) fractions at 4 min residence time and varying mass fractions of guaiacol in the feed (A), and with 2 wt% guaiacol in the feed at different residence times (B). The unit is Da.

Table 2 reports the M _w of the product fractions at 1 and 4 min of residence time. It is obvious that increasing the guaiacol feed from 0 to 2 wt% decreases the M _w of the char at 4 min and of PS fractions at both 1 and 4 min of residence time.

Guaiacol had the strongest influence on the M _w of the char fraction, in which it decreased as the addition of guaiacol increased. The M _w of the PS, on the other hand, seemed to be affected when some guaiacol, 0.2 wt%, was added but further additions in the range 0.2–2 wt% did not have any significant effect. This could also be found in the molecular weight distributions presented in Figures 4A and S4A: it is from these distributions the M _w in Table 2 are calculated. More pronounced shifts in the molecular weight distributions with higher guaiacol feeds are seen in the char fraction, see Figure 4A1, than for the PS and ASO fractions, see Figure 4A2 and A3, respectively. This is in accordance with the greater change in M _w of the char fraction compared to the PS and ASO fractions. Adding 0.2 wt% guaiacol to the reaction mixture does not change the molecular weight distribution of the char fraction as much as in the PS fraction. Instead, increasing the amount of guaiacol is necessary, from 0.2 to 1 wt% guaiacol, to change the char molecular weight distribution significantly, see Figure 4A1.

In Figure 4A it can also be found that, for both the char and the PS fractions, the content of high-molecular weight material decreases, and the content of low-molecular weight material increases with increasing guaiacol charge, thereby showing the capping effect of guaiacol. For the ASO fraction, it seems that the content of high molecular weight material increases slightly with increasing guaiacol charge, see Figure 4A3: this may be due to the fact that guaiacol reacts with some low molecular weight compounds of lignin, forming structures with a higher molecular weight.

The M _w of the product fractions at different residence times with 0 and 2 wt% added guaiacol are shown in Table 3. Here, it can be seen that, after initial depolymerisation of the lignin within the first minute, the M _w increases with increasing residence time as the products show a net repolymerisation, which begins at different times for the various product fractions. Figure 4B and S4B present the molecular weight distributions from which the M _w in Table 3 is calculated. For the char fraction in particular, it is obvious that the repolymerisation that occurs with increasing residence time results in shifts to the right, to higher molecular weights, see Figure 4B1. The same is seen for the PS and ASO fractions, but to a lesser degree. This indicates that a fast initial depolymerisation is followed by reaction steps where product molecules, oligomers and polymers react further, forming larger molecules and even polymers. These reactions are nevertheless slower than the initial depolymerisation. Thus, in order to minimise the molecular weight of the products, the residence time should be short and at least some guaiacol should be added.

3.4 Changes in molecular structure

The inter-unit region of the NMR spectrum shows cleavage of the inter-unit ether linkages in both the char and the PS fractions, see Figure S5. 1 min of residence time in the reactor causes the peaks of the inter-unit ether linkages, β-O-4′, β-β′ and β-5′, to disappear, i.e. these inter-unit ether linkages are lost in the char and PS fractions, as shown in previous work by Ahlbom et al. (2022). The breaking of the inter-unit ether linkages was also found at the longer residence times of 4 and 12 min, as well as without guaiacol. It thus seems that the hydrothermal conditions in themselves cleave the inter-unit ether linkages rapidly, regardless of whether guaiacol is added or not. The original lignin depolymerises, which agrees not only with the reduction in M _w given in Tables 2 and 3 but also the loss in high molecular weight compounds shown in Figure 4B.

Signals for residual guaiacol should show up in the aromatic region (Giummarella et al. 2020). These signals are located under the aromatic peaks for the char and PS fraction and therefore hard to distinguish. In Figure 5 the aromatic regions of the lignin (A), the PS fraction at 1 min of residence time without guaiacol in the feed (B) and with 2 wt% guaiacol in the feed (C), are presented. With the addition of guaiacol, new peaks emerge in the PS fraction, see Figure 5C. The peaks that emerge are highlighted with a black oval; the very same peaks also appear in the aromatic region of the char spectra when guaiacol is added after 4 min of reaction, see Figure S6. The intensity of these new peaks also increases at longer residence times for both char and PS. Giummarella et al. (2020) obtained similar peaks in an HSQC spectrum for guaiacol oligomers polymerised, by radical coupling, with hydrogen peroxide and horseradish peroxidase. Thus, while these new peaks could be the result of guaiacol reacting with lignin fragments, they could also be traces of the polymerisation of guaiacol, similar to the polymerisation of guaiacol in the control sample, see Figure S2. Potentially, the guaiacol could be coupled in a radical reaction in this case too, just as in the work by Giummarella et al. (2020).

Figure 5:

Aromatic region of the HSQC spectrum of lignin (A), the PS fraction run at 1 min of residence time without guaiacol in the feed (B) and with 2 wt% guaiacol in the feed (C). N.B. The new peaks that emerge when guaiacol is added are circled in black.

3.5 Yields of monoaromatic compounds

Monoaromatic compounds in the ASO fractions were identified, and semi-quantified, by GC-MS. Most were found to be guaiacol derivatives, even when guaiacol was not added to the reaction mixture. Since the lignin is sourced from softwood, and thus comprised of guaiacyl propane units, this is in line with expectations. Control measurements made on the diethyl ether extract of dissolved, unreacted lignin showed no monoaromatic compounds, suggesting that hydrothermal treatment is indeed required if the depolymerisation of lignin is to form monoaromatic components.

Table 4 reports the yields of monoaromatic components versus dry lignin at 1 and 4 min of residence time. Guaiacol is excluded because it dominates the monoaromatic compounds identified, even when it is not added to the feed. With increasing amounts of guaiacol added, the yields of monoaromatic compounds identified by GC-MS increase. For instance, the yield of 1,2-dimethoxybenzene, earlier mentioned as a product of guaiacol reacting without lignin, increases with the addition of guaiacol to the reaction mixture.

Figure 6 shows the yields of monoaromatic components with 0 wt% guaiacol addition for 1 and 4 min of residence time. These yields increase with residence time, with the exception of vanillin. Guaiacol was also identified by GC-MS: with a yield of 2–2.5 % on dry lignin, it dominated the yield of monoaromatic components and is therefore not given in Figure 6. The yield of guaiacol versus dry lignin increased with residence time at 0 wt% guaiacol in the feed, thereby agreeing with the result shown in Figure 2A.

Figure 6:

Yields of monoaromatic compounds identified and semi-quantified by GC-MS, with standard deviations for duplicate reactor runs, versus dry lignin at 0 wt% guaiacol feed at different residence times. The yield of guaiacol is excluded.

Figure 7 shows the yields of monoaromatic components obtained at different residence times with the addition of 2 wt% guaiacol. Since guaiacol was added to the feed, the product is dominated by guaiacol and it is therefore excluded. However, like in Figure 2A, the guaiacol yield level, semi-quantified with GC-MS, was seen to drop with increasing residence time. It is notable that the yields of monoaromatic compounds are higher with guaiacol added to the feed than without the addition of guaiacol, cf. Figures 6 and 7. Catechol was not separated reliably from 4-methyl guaiacol; at 12 min of residence time, the catechol value therefore contains an unknown fraction of 4-methyl guaiacol. With increasing residence time, the 1,2-dimethoxybenzene, 4-methyl guaiacol and catechol appear to increase somewhat with time, although they remain at low levels. It could be hypothesised that alkyl guaiacols form either from the depolymerisation of lignin or alkylation of the guaiacol, in a fashion similar to that observed in previous works in which phenol was used as a capping agent (Arturi et al. 2017; Barbier et al. 2012; Saisu et al. 2003). The loss of guaiacol is nevertheless greater than the gain of alkyl guaiacols, which can be explained at least partly by the polymerisation of guaiacol.

Figure 7:

Yields of monoaromatic compounds identified and semi-quantified by GC-MS, with standard deviations for duplicate reactor runs, versus dry lignin at 2 wt% guaiacol feed for different residence times. N.B. The yield of catechol could include an unknown amount of 4-methyl guaiacol. The yield of guaiacol is excluded.