Adapting the kraft cooking process in glycerol media. Studies of impregnation kinetics

Materials

Spruce wood chips for the glycerol experiments were kind gifts from SCA Munksund. Sodium hydroxide of Emsure grade was procured from (Sigma-Aldrich, Darmstadt, Germany). ≥ 60 % Sodium hydrysulphide hydrate was procured from (Sigma-Aldrich, St. Louis, MO, USA). Glycerol of GPR Rectapur grade was procured from (VWR International, Leuven, Belgium). All other chemicals were of analytical grade. Temperature was monitored using an Amarell Precision Thermometer (Amarell, Kreuzwertheim, Germany).

Glycerol cooking system

Kraft cooking was replicated in glycerol as follows: sodium hydroxide and sodium hydrogen sulphide was added neat into a round bottom flask according to desired effective alkali and sulphidity, after which an amount of GPR Rectapur grade glycerol, appropriate to the desired liquid to wood ratio, was added. Temperature was increased (incidentally removing any lingering impurities of water), stirring initiated, and a flow of nitrogen added to the reaction vessel. Once all chemicals were dissolved, starting aliquots of the cooking liquor were collected to measure initial NaOH and NaHS concentrations. Finally, wood chips were added to the reaction vessel and the resulting reaction monitored.

For reference, a control experiment was also conducted, following the same steps as per the above description but obviating the addition of chemicals and nitrogen gas.

Alkaline glycerol consumption experiment

Experiments were conducted as described above on 18 g of spruce wood chips in 130 ml of glycerol liquor using starting chemical concentrations of 1 M NaOH and 0.1 M NaHS (corresponding to an effective alkali of 28 % and a sulphidity of 18 % at a liquid to wood ratio of 7.2 to 1). Aliquots of 5 ml were collected at 0 min, 15 min, 30 min, 60 min, 90 min and 120 min and were used to determine hydroxide concentration over time. The experiment was replicated using increasing temperatures of 120 °C, 130 °C, 140 °C, and 150 °C.

Atmospheric cooking experiment

Full experiments of cooking in glycerol media were conducted as described above on 18 g of spruce wood chips in 130 ml of glycerol using starting chemical concentrations of 1 M NaOH and 0.3 M NaHS (corresponding to an effective alkali of 28 % and a sulphidity of 46 % at a liquid to wood ratio of 7.2 to 1). Impregnation was performed at 140 °C for 3 h after which the temperature was ramped to 160 °C and cooking was performed over 4 h. After the reaction, the reaction vessel was cooled, the black liquor extracted and the (delignified) wood chips washed overnight. After washing, the wood chips were defibrated using a NAF defibrator operating at 2 bar of water pressure. The resulting pulp was centrifuged, and its yield determined gravimetrically. Similarly, the reject (material retained in a metal mesh basket attachment of the NAF defibrator with 1.5 mm perforations) was collected and gravimetrically compared to the starting material.

Alkaline water consumption experiment

For reference, an experiment of alkaline consumption was also done in water media. In this experiment, batches of 200 g of spruce wood chips were added to a set of rotative steel autoclaves, which were allowed to react in a polyethylene glycol bath at temperatures of 90 °C, 120 °C, 130 °C, 140 °C and 150 °C using starting chemical concentrations of 0.6 M NaOH and 0.2 M NaHS and using a liquid to wood ratio of 5.5 to 1 (thereby corresponding to an effective alkali of 12 % and a sulphidity of 53 %). For each temperature, five autoclaves were added and taken out after 5 min, 10 min, 20 min, 30 min and 60 min, respectively. The autoclaves were rapidly cooled, and the liquor extracted for residual alkali and sulphur analysis.

Lignin precipitation, isolation and sample preparation

Lignin was precipitated from the glycerol black liquor as follows: 10 ml of glycerol black liquor was diluted with distilled water (in order to lower the viscosity of the solution) after which alkali was added in order to redissolve the lignin. The solution was then passed through a filter in order to remove any remaining pulp in the liquor.

Once the solution was thus filtered, it was acidified to a pH of 3, precipitating the lignin. The acidified solution was sequentially centrifugated in a Hettich Centrifuge Rotina 420 (Andreas Hettich GmbH & Co. KG, Tuttlingen, Germany) at 4800 rpm for 20 minutes at a pH 2.5 a total of seven times, after which a final centrifugation at pH 3.5 was applied and the precipitated lignin thus obtained was transferred to a Vacucell vacuum drying oven (BMT Medical Technology s. r. o., Zábrdovice, Czech Republic), where it was left to dry overnight at a temperature of 40 °C.

After being dried thus, 92.3 mg of sample was dissolved in 0.6 ml of DMSO-d6, after which it was subjected to inverse NMR experiments as described below. A separate sample of 27.8 mg was prepared according to the method of Argyropoulos (1995) after which it was subjected to ³¹P NMR analysis as detailed below.

Analysis methods

Residual alkali measurements were done on the black liquor aliquots according to SCAN-N 33:94 and residual hydrogen sulphide ion measurements were done according to SCAN-N 31:94, while Kappa number measurements were done on the produced pulp in accordance with ISO 302:2015 and viscosity measurements were done on the produced pulp in accordance with ISO 5351.

The precipitated lignin was subjected to a nuclear magnetic resonance (NMR) spectroscopy analysis including a heteronuclear single quantum coherence (HSQC) analysis, a complementary edited heteronuclear single quantum coherence (E-HSQC) analysis, a heteronuclear multiple bond correlation (HMQC) analysis and a quantitative phosphorous ³¹P NMR analysis.

The HSQC experiment was done using the hsqcetgpsi pulse sequence with increments of 1024 × 256, 72 scans per increment, an acquisition time of 86 ms, a relaxation delay of 2 s and a receiver gain of 203 with a spectral width of 15 ppm in the proton dimension and 165 ppm in the carbon dimension. The E-HSQC experiment was done using the hsqcedetgp pulse sequence with increments of 1024 × 256, 40 scans per increment, an acquisition time of 83 ms, a relaxation delay of 2 s and a receiver gain of 203 with a spectral width of 15 ppm in the proton dimension and 165 ppm in the carbon dimension. The HMBC analysis was done using the hmbcqpndqf pulse sequence with increments of 4096 × 128, 224 scans per increment, an acquisition time of 0.33 s, a relaxation delay of 2 s and a receiver gain of 203 with a spectral width of 15 ppm in the proton dimension and 222 ppm in the carbon dimension. The ³¹P NMR analysis was done using the zgig30 pulse sequence using 64k data points, 5120 scans, an acquisition time of 1.68 s, a relaxation delay time of 6 s and a receiver gain of 203 with a spectral width of 120 ppm.

All experiments were done on a Bruker Avance III HD 400 MHz instrument (Bruker Corporation, Billerica, MA, USA) equipped with a BBFO probe (Bruker Corporation, Billerica, MA, USA), including a Z-gradient coil. Processing was done using the software MestReNova v14.2.1-27684 (Mestrelab Research, Santiago de Compostela, Galicia, Spain), including apodisation (90°-shifted square sine bell for the HSQC spectra, 0°-shifted sine bell for the HMBC spectra, and 1 Hz exponential multiplication for the ³¹P NMR experiment), manual phase adjustment, and automatic baseline correction using a third order Bernstein Polynomial Fit for the ³¹P NMR experiment.

Alkaline consumption in water

An initial study of alkaline consumption in water media was done by subjecting five autoclaves filled with 200 g of spruce wood chips as well as white liquor at chemical concentrations of 0.6 M NaOH and 0.2 M NaHS to temperatures of 90 °C, 120 °C, 130 °C, 140 °C and 150 °C for varying amounts of time, after which residual alkali was measured, and the alkali consumption calculated. The result of these experiments can be seen in Figure 2.

Figure 2

Alkaline consumption of spruce at different temperatures during kraft pulping impregnation conditions in water based white liquor.

In this experiment, it can be seen that most of the alkaline consumption is achieved in the first 15 minutes, similarly to what has been observed in previous works (Määttänen and Tikka 2012) after which alkaline consumption is seen to follow a similar trend to that observed by Karlström (2009), who reported consumptions of 100 kg ton⁻¹ wood after 90 min of reaction at 120 °C, 90 kg ton⁻¹ wood after 90 min of reaction at 110 °C and 70 kg ton⁻¹ wood after 90 min of reaction at 100 °C, all cases using initial concentrations of 0.6 M NaOH and 0.2 M NaHS.

An Arrhenius-analysis of the data was done by fitting the data to a first order kinetic expression. In principle, assuming an elementary reaction, the neutralisation of acetyl groups by alkali would be described by a second order kinetic expression proportional to the concentration of acetyl groups as well as the concentration of alkali. However, for a system of wood undergoing such neutralisation, the concentration of alkali can be assumed to be in excess of the accessible acetyl groups, resulting in a pseudo-first order expression dependant on the concentration of acetyl groups. Similarly, the molar consumption of acetyl groups, which is equal to the molar consumption of hydroxyl groups and proportional to the alkaline consumption expressed as g kg⁻¹ wood, will also follow pseudo-first order kinetics as a result. By treating the alkaline consumption as a pseudo-first order phenomenon following a A × ( 1 − e − k t ) expression, rate values for each temperature were obtained through regression, giving values as presented in Table 1 as well as an Arrhenius plot which is provided in Supplementary Figure 1 in the supplementary information. From the Arrhenius plot, an apparent activation energy was calculated, found to be 12 kJ mol⁻¹. This activation energy is well below 30 kJ mol⁻¹, thereby suggesting a primarily diffusion driven phenomenon (Germgård 2017).

Alkaline consumption in glycerol

One of the hypotheses of the present work was that glycerol, being a polar solvent able to swell wood in a similar fashion to water (Demirbaş 1998, Küçük 2005, Novo et al. 2011) while seemingly being able to also transport solutes in a similar fashion due the high mobility of residual water in the matrix, (Glycerine Producers’ Association 1963) could be used for investigations of alkaline consumption similar to the more classic experiment above. Doing so would have the benefit of allowing the experiment to proceed under atmospheric conditions as well as lowering the barrier of entry to similar research in the future. As such, a similar set of experiments were conducted on spruce wood chips in glycerol. The glycerol experiments were conducted in a smaller scale, using 18 g of wood chips and volumes of 130 ml of GPR Rectapur grade glycerol using initial chemical concentrations of 1 M NaOH and 0.1 M NaHS. In this case, aliquots of 5 ml were taken out of the same batch at intervals of 15 min, 30 min, 60 min, 90 min and 120 min respectively. Residual alkali and alkaline consumption were measured and calculated analogously to the water-based experiments above, and the results are shown in Figure 3.

Figure 3

Alkaline consumption of spruce at different temperatures during kraft pulping impregnation conditions in glycerol based white liquor.

For temperatures below 150 °C, consumptions of alkali are seen to be significantly lower than their counterparts in the water-based experiment although interestingly, the difference between the two systems is significantly mitigated at higher temperatures, particularly at 150 °C. One possible reason for this effect is the exponential lowering of glycerol viscosity at high temperatures, which is halved from 7.797 centipoise at 120 °C to 3.823 centipoise at 150 °C for solutions of glycerol (Glycerine Producers’ Association 1963).

The above profiles together with their respective temperatures were used to model the experiments in accordance with the same assumptions as for the water experiments conducted above with the resulting analysis giving rate values as seen in Table 1, an Arrhenius plot depicted in Supplementary Figure 2 in the supplementary information and an apparent activation energy of 22 kJ mol⁻¹. While higher than the apparent activation energy of 12 kJ mol⁻¹ observed in the water-based experiment, this activation energy is still below 30 kJ mol⁻¹ and therefore suggestive of a primarily diffusion-driven phenomenon (Germgård 2017).

Atmospheric cooking in glycerol

With the impregnation behaviour in glycerol elucidated, and with the initial goal of demonstrating glycerol to be a viable setup for continued experiments under laboratory conditions concluded, the cooking system was modified to a full pulping process. Although impregnation proved most efficient at 150 °C, a lower temperature of 140 °C was selected due to the high risk of initiating cooking reactions at the higher temperature. To compensate, impregnation was extended to three hours. Cooking was conducted at 160 °C for four hours and starting chemical concentrations were selected to 1 M NaOH and 0.3 M NaHS, corresponding to an effective alkali of 28 % and a sulphidity of 46 % at a liquid to wood ratio of 7.2 to 1.

Using these conditions in the same set-up as for the consumption experiments above, spruce wood chips were delignified. They were then washed overnight and defibrated using a NAF defibrator operating at 2 bar of water pressure. By these means, a pulp of kappa number 28 and a pulp viscosity of 1113 ml g⁻¹ was produced with a screened yield of 37 % and a reject content of 15 %. While the kappa number and total yield of the experiment shows promise for utilising glycerol for atmospheric kraft cooking, the reject content is unacceptably high.

To control for the nucleophilic properties of the pure glycerol, a control experiment was performed on spruce wood chips in glycerol without the addition of chemicals. Although some extraction was seen to take place, the thus treated wood chips retained 94 % of their dry weight and could not be defibrated into pulp when subjected to the NAF defibrator.

The high amount of reject produced in the atmospheric cooking despite the investigation into impregnation could be a result of insufficient penetration (given that glycerol has previously been found to be an excellent solvent for hydroxide in the work of Demirbaş (1998) and has been found able to swell wood as in the work of Demirbaş and Celik (2005), it seems unlikely to the authors that either diffusion or swelling would be at fault) of the glycerol into the wood matrix. Penetration of cooking liquor into wood chips through cavities concerns the flow of liquid into wood chips by means of a pressure gradient (Määttänen and Tikka 2012). Given that liquid flow is inversely proportional to the viscosity of the liquid in question, this process would naturally be lower for the glycerol system than in a system were water is used as solvent. The impregnation of wood in glycerol could also have been partially hindered by the lower dissolution of oxygen in glycerol compared to water, which makes glycerol less able to remove air bubbles, thereby reducing penetration (Malkov 2002).

If penetration is indeed to blame for the high amounts of reject (particularly if the cause is due to issues with the dissolution of oxygen), it is possible that the situation could be remedied by pre-steaming the wood chips (Gustafson et al. 1989), although this would require experimental verification. Alternatively, insufficient penetration could be avoided by using thinner wood chips, as seen in the work of Jiménez et al. (1990), who found wood thickness to be a significant variable for the penetration of cooking liquor into Douglas fir chips or in the work of Demirbaş and Celik (2005), who achieved significantly better delignification using wood chips at a thickness of 0.3 mm compared to 2 mm when cooking poplar and spruce wood chips in alkaline glycerol.

As a final set of experiments for elucidating the behaviour of glycerol cooking, lignin was precipitated from black glycerol liquor, lyophilised, and analysed using nuclear magnetic resonance (NMR) spectroscopy. Specifically, 92.3 mg of lyophilised lignin was redissolved in 0.6 ml of DMSO-d6 and analysed using a heteronuclear single quantum coherence (HSQC) experiment, a multiplicity edited heteronuclear single quantum coherence (E-HSQC) experiment and a heteronuclear multiple bond correlation (HMBC) experiment, while 27.8 mg of lyophilised lignin was phosphorylated and analysed using quantitative ³¹P NMR. The HSQC and ³¹P NMR protocols represent established methods of semi-quantifying characteristic bonds in lignin, and quantifying hydroxyl functionalities, respectively.

The HSQC spectrum, presented in Figure 4, was semi-quantified according to the method of Sette et al. (2011). The HSQC spectrum is seen to contain functionalities typical of kraft lignin, with a β-O-4 content of 2.03 mol%, a β-β content of 1.97 mol% and a β-5 content of 1.09 mol%, normalised on the C2 carbon in lignin. Although these values are in the range typically seen in other kraft-lignins (for instance, see Giummarella et al. (2020), reporting values of 6.7 mol% β-O-4 and 1.9 mol% β-5 for an industrial softwood kraft lignin), the β-O-4 content in particular is lower than would be expected of regular kraft lignin, indicating a higher amount of degradation than is normal at this kappa number. This is likely a consequence of the liquor penetration issues discussed above, resulting in excessive degradation of the dissolved lignin. If this penetration issue could be solved or mitigated, it is therefore possible that this degradation could also be mitigated in turn.

Figure 4

HSQC spectra of lignin extracted from the glycerol black liquor, with HMBC data of the same lignin superimposed in green. Also superimposed in green is a window around the glycerol signal at δH 3.5, δC 62-72, showing that there are no long-range correlations between glycerol and lignin.

In addition to the regular functionalities normally seen in kraft lignins, signals related to the glycerol solvent can be seen at shifts (δH 3.5, δC 62-72). This is particularly noticeable in a complementary E-HSQC spectrum that was acquired on the same material, which can be found in Supplementary Figure 3 in the supplementary material. To investigate whether any long-range correlations could be seen for the glycerol, a complementary HMBC analysis was made, which is shown alongside the HSQC spectra in Figure 4. While glycerol peaks can again be identified at shifts (δH 3.5, δC 62-72), no long-range correlations to lignin can be seen in the HMBC spectrum. While one should be careful when making conclusions based on the absence – rather than the presence – of NMR peaks, the large prevalence of glycerol by virtue of being used as solvent together with the slow t2 relaxation of glycerol makes this absence suspicious and corroborates the hypothesis that the system has simply reacted according to regular kraft mechanics.

In order to quantify the hydroxyl functionality of the procured lignin, a ³¹P NMR spectra was taken and quantified according to the method of Granata and Argyropoulos (1995). In this spectrum, which is presented in Supplementary Figure 4 in the supplementary material, the lignin is seen to contain 1.83 mmol g⁻¹ of aliphatic hydroxyls, 4.16 mmol g⁻¹ of aromatic hydroxyls and 0.51 mmol g⁻¹ of carboxylic hydroxyls with a total functionality of 6.50 mmol g⁻¹. These values are, again, in the range of values typically seen for kraft lignin, with Giummarella et al. (2020) reporting values of 2.30 mmol g⁻¹ of aliphatic hydroxyls, 3.84 mmol g⁻¹ of aromatic hydroxyls and 0.37 mmol g⁻¹ of carboxylic hydroxyls for a sample of industrial softwood kraft lignin.