Fractionation of beech wood pyrolysis liquids with supercritical and gaseous CO₂

2.1 Materials

The beech wood SPL (Figure 2) used in setup 1 was supplied by proFagus GmbH (Bodenfelde, Germany). The extract used in setup 2 was obtained from the same SPL and produced as described in Möck et al. (2023) (chapter 3.4) with a scCO₂ density of 724 kg/m³. There, a single extract was collected during the extraction process. The extraction conditions were as follows: 20 MPa, 333.15 K, flow = 5 g scCO₂/min, 25 g SPL, 4 h.

Figure 2:

Beech wood SPL used for scCO₂ extraction in setup 1.

2.2 Setup 1: supercritical CO₂ extraction with different off-gas temperatures

Off-gas is a mixture of gaseous CO₂ and the extract formed during depressurization after scCO₂ extraction. The influence of off-gas temperature on extract composition and loss was investigated. For this, a scCO₂ extraction system from HDT Sigmar Mothes GmbH (Wildau, Germany) (setup 1) was used. The system is illustrated schematically in Figure 3. Liquid CO₂ was compressed by using a high-pressure pump (Figure 3; 2) and heating (Figure 3; 5) brought it into the supercritical state. Then the scCO₂ flowed through the extraction vessel (Figure 3; 6), which was filled with liquid SPL, and was depressurized via the heated back pressure regulator valve (Figure 3; 7). The extraction vessel had a volume of 185 cm³. A second heated valve (Figure 3; 8) distributed the cooling during decompression, allowing the off-gas to be heated up to 333.15 K. The off-gas flowed into an insulated vial (60 cm³) (Figure 3; 11) filled with 2 mm glass beads to collect the extract. A thermocouple located in the vial (Figure 3; 10) controlled the heating of the second valve, which allowed the temperature of the off-gas to be adjusted. The extraction plant had been used in previous studies with modified configurations (Feng and Meier 2015, 2016, 2017; Möck et al. 2023, 2024). A more detailed description can be found in Möck et al. (2023).

Figure 3:

Flow diagram of the scCO₂ extraction system used for SPL extraction in setup 1 (extraction parameters: 12.5 MPa, 313.15 K, ρscCO₂ = 731 kg/m³, 4 h, flow = 5 g scCO₂/min, 25 g SPL).

Extractions were carried out at three different off-gas temperatures (333.15 K, 313.15 K, 293.15 K). The extraction parameters were as follows: 12 MPa, 313.15 K, ρ_scCO2 = 731 kg/m³, 25 g SPL, flow = 5 g scCO₂/min, 4 h. Extractions were carried out in duplicate and the system’s leak-tightness was verified using a Gas Alert Micro 5 IR (Honeywell International Inc., Charlotte, USA).

2.3 Setup 2: fractionation of SPL extract with hot gaseous CO₂

Setup 2 used five tempered vials (353.15 K, 333.15 K, 313.15 K, 293.15 K, 283.15 K) for fractionating the SPL extract with gaseous CO₂. Figure 4 shows a scheme and Figure 5 an image of setup 2. The SPL extract, produced by scCO₂ extraction, was pumped at a rate of 0.025 g/min into a heated (353.15 K) gaseous CO₂ flow of 2.5 g/min. A syringe pump (Teledyne ISCO, Model 500 D, Lincoln, USA) and two connected 24 ml syringes were used for pumping (Figure 5, C). The syringe pump filled the first syringe with ethanol, which pressed the SPL extract out of the second syringe. The SPL extract was not pumped directly by the pump because the use of syringes made it possible to easily weigh the amount of SPL extract pumped. In order to maximize the contact surface area between the CO₂ and the SPL extract, the SPL extract was pumped into a tube filled with cotton. Preheated CO₂ (353.15 K) flowed through the cotton-filled tube and then through five vials containing 2 mm glass beads, gradually reducing the temperature of the mixture of CO₂ and SPL extract from 353.15 K to 283.15 K. The vials were immersed in either a heated or cooled water bath, and thermocouples measured the temperature of the water and the CO₂ inside the vials. The first three vials were heated using magnetic stirrers with hotplates. The fourth vial was cooled with water, and the fifth vial with crushed ice and water. The tubes between the vials were isolated with aluminum foil. Prior to pumping the SPL extract, the vials and CO₂ flow were heated at the target temperature for 30 min. Four hours were allotted for pumping SPL extract into the CO₂ flow, during which 6 g of extract were pumped into 600 g of CO₂ (ratio 1:100). Then, the vials and the syringe filled with SPL extract were weighed using a VWR LPW-2103i (d = 1 mg, max 2,100 g) (Avantor Inc., Radnor, USA) scale. Fractionation was carried out in duplicate, and the leak-tightness of setup 2 was verified using a Gas Alert Micro 5 IR (Honeywell International Inc., Charlotte, USA).

Figure 4:

Flow diagram of the gaseous fractionation system in setup 2 used for the fractionation of the SPL scCO₂ extract at different temperatures (fractionation parameters: 353.15 K, 333.15 K, 313.15 K, 293.15 K, 283.15 K, 4 h, flow = 2.5 g CO₂/min and 0.025 g extract/min; T= thermocouple, MFC = mass flow controller).

Figure 5:

Images of the gaseous fractionation system (setup 2) used for fractionation of the SPL scCO₂ extract at different temperatures (fractionation parameters: 353.15 K, 333.15 K, 313.15 K, 293.15 K, 283.15 K, 4 h, flow = 2.5 g CO₂/min and 0.025 g extract/min: (A) assembled setup 2; (B) vial with glass beads and thermocouple; (C) syringes for extract pumping).

2.4 GC-MS/FID

The SPL and the extracts were analyzed by GC-MS/FID. The setup consisted of an Agilent HP6890 gas chromatograph (Agilent Technologies. Inc., California, USA) equipped with a flame ionization detector (FID) and an Agilent HP5972 mass spectrometer (MS). Helium was used as the carrier gas at a flow rate of 2 ml/min. A silica-fused column (Varian VF-1701 ms, 60 m x 0.25 mm x 0.25 µm) was used for separation. The temperature program was as follows: 318 K constant for 4 min, heating to 553 K (3 K/min), constant for 20 min at 553 K. Fluoranthene in acetone (ß=203 μg/ml) was used as the internal standard because it is not present in the SPL. Additionally, the retention time of fluoranthene does not overlap with those of other compounds contained in the SPL, thus preventing peak interference. The proportion of the sample to acetone with internal standard solution was 50 mg of sample per ml of internal standard solution. To achieve this, 150 mg of the sample was weighed into a 7 ml vial by using a Gilson Microman pipette with M100 tip (Gilson Inc., Middleton, USA) and an XSE205 Dualrange balance (Mettler-Toledo International Inc., Columbus, USA). Then, 3 ml of the internal standard solution were pipetted with a Microman pipette with M1000 tip (Gilson Inc., Middleton, USA). The sample was dissolved by shaking. One milliliter of the solution was withdrawn with a disposable syringe and transferred through a 0.45 µm regenerated cellulose filter into a 2 ml GC vial. One microliter was injected with a split ratio of 1:15 and an injection temperature of 523.15 K. Based on the MS signal the substances were qualified with Mass Finder 4 and the FID signal was used for quantification. In the SPL, 117 substances were detected, some in small amounts. Therefore, the chromatogram was evaluated only for the 44 substances with the highest proportions. These 44 substances represent approximately 90 % of the GC-detectable portion of the SPL, and 33 of these substances were calibrated against standards. The remaining substances were evaluated using an estimated response factor based on chemically similar substances. This method has been used in two previous studies (Möck et al. 2023, 2024). There was a strong overlap of peak areas for the substances syringol, 4-allyl-, and syringol, 4-propyl-. Since these substances have the same response factor, they were reported together. In cases of minor overlap of peak areas, the course of the peak was estimated for integration. The chromatograms of the SPL and the scCO₂ extract are provided in the Supplementary Material. Additional information on the standards used including concentration range, calibration points, R-squared values, producer, purity, and CAS number is provided based on Möck et al. (2023).

2.5 Water content

The water content of the SPL and the scCO₂ extract used was measured by Karl Fischer titration as described in Möck et al. (2024). For this, a Titroline Alpha titrator (Schott AG, Mainz, Germany) was used. Forty milligrams of the sample were dissolved under stirring in anhydrous Hydranal Methanol Rapid (Honeywell Fluka™, Seelze, Germany). Then, Hydranal Composite 2 (Honeywell Fluka™, Seelze, Germany) was automatically dosed until all the water present in the sample was consumed. Each ml of water required 0.5 ml of Hydranal Composite 2 (Honeywell Fluka™, Seelze, Germany). The amount of water in the sample was calculated based on the titrated volume of Hydranal Composite 2. The titration endpoint was determined using a double platinum electrode.

2.6 Kinematic viscosity

The kinematic viscosity of the SPL and the scCO₂ extract used was determined at 298.15 K using an automatic lead time measuring system from Schott Geräte GmbH (CT62, AVS 350, AVS/S) (Mainz, Germany) in accordance with DIN 51 562 (Deutsches Institut für Normung e.V. 1999). The Ubbelohde viscometer MII/1050741 was used to measure the SPL, and the Ubbelohde viscometer MIc/1033959 was used to measure the scCO₂ extract.

2.7 Density

The density of the SPL and scCO₂ extract was measured using a 10 ml Isolab Boro 3.3 pycnometer (Isolab Laborgeräte GmbH, Eschau, Germany) and a METTLER AT261 DeltaRange analytical scale (Mettler Toledo, Columbus, USA) at 293.15 K. To determine the density of the sample, the exact volume of the pycnometer was measured using water, and its empty mass was weighed. Then the pycnometer was filled with the sample and weighed again. The density was calculated by dividing the mass of the sample by the volume of the pycnometer.

2.8 Calculation of the vapor phase loading

Vapor phase loading (VPL) indicates the number of grams of the extract dissolved in one kilogram of CO₂. The VPL was calculated for gaseous and supercritical CO₂ using the following equation:

3.1 Characterization of the SPL and scCO₂ extract used for setup 2

The SPL used in setup 1 was dark black, had a high kinematic viscosity of 1,101 mm²/s at 293.15 K and a water content of 4.09 %. Its density was 1,204 kg/m³ at 293.15 K. The SPL scCO₂ extract used in setup 2 had a kinematic viscosity of 21.6 mm²/s at 293.15 K and a density of 1,117 kg/m³ at 293.15 K. The water content of the scCO₂ extract was only 1.67 %. The colour of the scCO₂ extract was dark red. Table 1 shows the GC-detectable composition of the SPL and the scCO₂ extract used in setup 2.

With GC-MS/FID it was possible to characterize 37.2 wt. % of the SPL and 56.6 wt. % of the scCO₂ extract. Since most high molecular weight substances are poorly soluble in scCO₂ (Mukhopadhyay 2000), the majority of these were separated during the extraction, thereby enriching the low molecular weight substances in the scCO₂ extract. For instance, the proportion of acids increased from 5.9 wt. % to 10.0 wt. %, non-aromatic ketones from 4.0 wt. % to 9.0 wt. % and syringols from 10.5 wt. % to 19.4 wt. %, respectively. Since scCO₂ is a non-polar solvent, the sugar content decreased from 10.1 wt. % in SPL to 2.4 wt. % in the scCO₂ extract.

3.2 Extraction with different CO₂ off-gas temperatures (experimental setup 1)

In order to investigate how different off-gas temperatures influence the losses and compositions of the extracts during decompression to atmospheric pressure, SPL (25 g) was extracted with supercritical CO₂ (flow = 5 g/min, 1,200 g) at three off-gas temperatures (293.15 K, 313.15 K, 333.15 K). A mass loss was observed when comparing the initial mass to the mass of extracts and residues. The losses consist of small amounts of extract remaining in the valves and tubing of the extraction system, and of substances that are not separated in the separator and are carried out with the gaseous CO₂. Figure 6 shows the mass balances as the mean of the duplicates, with the error bars indicating the range.

Figure 6:

Influence of off-gas temperature on losses during scCO₂ extraction of SPL in setup 1 (extraction parameters: 12.5 MPa, 313.15 K, ρscCO₂ = 731 kg/m³, 4 h, flow = 5 g scCO₂/min, 25 g SPL).

The loss at 293.15 K was 9.5 wt. %. As the off-gas temperature increased, the loss raised with 15.3 wt. % at 313.15 K and 27.2 wt. % at 333.15 K. Due to the increased loss, the amount of collected extract decreased from 29.6 wt. % (at 293.15 K) to 11.1 wt. % (at 333.15 K). The residues ranged between 61.0 wt. % and 62.6 wt. %. Based on the residues, the VPL without losses was calculated between 7.8 and 8.1 g extract/kg scCO₂. The VPL is highly dependent on the feed used and on the extraction parameters. In a review the range of VPLs for the extraction of bio-oils varied widely, from 0.7 to 99.7 g extract/kg of scCO₂ (Montesantos and Maschietti 2020). According to Montesantos and Maschietti (2020), the VPL can be increased by raising both the density of scCO₂ and the temperature at constant density. This is also supported by the solubility curves of acetic acid and guaiacol (see Figure 1). Due to the strong dependence of the VPL on process parameters and feed material, it is advisable to make comparisons under similar extraction conditions. A similar VPL of 10.6 g extract/kg scCO₂ for the extraction of SPL at the same extraction temperature and pressure but with a lower solvent-to-feed ratio of 24 was reported (Möck et al. 2023). In the experiments conducted with setup 1, the solvent-to-feed ratio was 48. The slightly lower VPL observed in setup 1 suggests that increasing the solvent-to-feed ratio results in a reduction of the overall yield. This can be attributed to the decreasing amount of easily soluble compounds in the feed during extraction, so that toward the end of the process, the scCO₂ no longer provides such a high extraction efficiency. Using the loss values, the amount of extract remaining in the heated off-gas after depressurisation (VPL of the off-gases) was calculated as 2.9 g extract/kg CO₂ at 293.15 K, 3.2 g extract/kg CO₂ at 313.15 K and 5.7 g extract/kg CO₂ at 333.15 K. This indicates that, as the off-gas temperature increased, the solubility of the extract in the gaseous CO₂ increased. A previous study shows that the losses were reduced to less than 5 wt. % by cooling the off-gas temperature to 273 K, which corresponded to a VPL of the off-gases of less than 2 g extract/kg scCO₂ (Möck et al. 2023). The VPL in the experiments conducted with setup 1 was 2.9 g extract/kg scCO₂. It can therefore be assumed that a further decrease in the off-gas temperature would lead to an even lower VPL, thereby reducing the losses caused by extract components dissolved in the off-gas. Furthermore, cooled water baths were used in other studies to cool the collection vessels and thereby reduced losses (Montesantos et al. 2019, 2020).

The SPL, the residue and the extracts produced at different off-gas temperatures were characterized using GC-MS/FID to investigate the effect of off-gas temperature on the extract composition. Figure 7 shows the composition as the mean of duplicates, with error bars indicating the range. As not all compounds are detectable by GC, the columns do not sum up to 100 wt. %.

Figure 7:

GC-detectable composition of the residues, the SPL and the scCO₂ extracts obtained at different off-gas temperatures in setup 1 (extraction parameters: 12.5 MPa, 313.15 K, ρscCO₂ = 731 kg/m³, 4 h, flow = 5 g scCO₂/min, 25 g SPL).

The scCO₂ extraction enriched the extracts with syringols, guaiacols, ketones, acids, furans, phenols, and pyrans. By increasing the off-gas temperature the amount of substances detectable by GC decreased. Whereas at 293.15 K, the content was 57.7 wt. %, it was 44.7 wt. % at 333.15 K. This suggests that in gaseous CO₂ well-soluble substances were removed from the extract, resulting in an enrichment of non-detectable or high molecular weight substances in the scCO₂ extract. Poorly soluble sugars and high molecular weight substances largely remained in the residue. The high proportion of residues can be attributed to the fact that slow pyrolysis produces large amounts of high molecular weight substances (Baumbach et al. 2016), which are poorly soluble in scCO₂ (Mukhopadhyay 2000). Possible applications for the high molecular weight components of the residue include the production of activated carbon or energetic use. The sugars present in significant amount in the residue, especially levoglucosan, which is a platform chemical in the chemical industry could also be utilized for the production of biopolymers, biofuels, surfactants, and pharmaceuticals (Itabaiana Jr et al. 2020).

Increasing the off-gas temperature particularly affected the content of acids and ketones. At an off-gas temperature of 293.15 K, the extract consisted of 6.8 wt. % acids and 8.1 wt. % ketones. As the off-gas temperature increased, the acid content decreased to 2.9 wt. % at 313.15 K, dropping further to 1.8 wt. % at 333.15 K. The ketone content decreased from 8.1 wt. % at 293.15 K to 5.9 wt. % at 313.15 K and then to 3.2 wt. % at 333.15 K. In contrast, the syringol content first increased with the rise in off-gas temperature and then decreased slightly. It was 22.7 wt. % at 293.15 K, 26.7 wt. % at 313.15 K and 26.4 wt. % at 333.15 K. This is due to the reduction in acids and ketones, which increased the mass fraction of syringols. A similar behavior was observed for guaiacols. Their content decreased significantly from 10.1 wt. % at 313.15 K to 7.1 wt. % at 333.15 K off-gas temperature. The shift in extract composition was due to different solubilities in gaseous CO₂. For example, acetic acid has a solubility of 44.5 g/kg in gaseous CO₂ at 0.21 MPa and 353.15 K (based on data from Bamberger et al. 2000). Guaiacol, on the other hand, has a solubility of only 2.3 g/kg under similar conditions (353.15 k, 0.20 MPa) (based on data from Lee et al. 1999).

Using supercritical CO₂ extraction in combination with a hot off-gas temperature reduced the proportions of acids and ketones in the extract. Because these extracts contain high proportions of monomeric phenolic compounds, they are promising for the production of phenol resins and for wood preservation (Karthäuser et al. 2024; Lourençon et al. 2016; Sukhbaatar et al. 2009).

In order to capture as much of the extract as possible, it was essential to maintain the off-gas temperature as low as feasible. In industrial processes, scCO₂ is not usually expanded to atmospheric pressure, but rather depressurized into the subcritical state in a pressure separator. This allows CO₂ to be recycled with reduced compression work. Future work would benefit from using a pressure separator to decrease pressure only to the CO₂ density at which the solubility of the contained substances is minimized. For acetic acid, this CO₂ density is approximately 100 kg/m³ (based on data from Bamberger et al. 2000; see Figure 1). The extract obtained under these conditions would still contain a high proportion of acids and ketones. To further reduce the content of acids and ketones, the extract collected in the pressure separator could be transferred into a hot CO₂ gas stream, where the solubility of acids and ketones is maximized. This would enable their separation as demonstrated in experimental setup 2.

3.3 Fractionation of SPL extract with hot gaseous CO₂ (experimental setup 2)

Setup 1 showed that especially the content of acids and ketones in the extract can be reduced by using higher temperatures during the depressurization of the CO₂. Setup 2 investigated whether the extract could be fractionated by adding it to a hot, gaseous CO₂ stream at atmospheric pressure and using separators at different temperatures. For this, an SPL scCO₂ extract was pumped (0.025 g/min) into a hot CO₂ stream (2.5 g/min) and passed through five vials. The temperature in the vials gradually decreased from 353.15 K to 283.15 K, and five different fractions were collected. The colors of the fractions ranged from dark red (353.15 K) to orange (333.15 K) to shiny yellow (313.15 K, 293.15 K) to almost colorless (283.15 K). Figure 8 shows the extract and its different fractions. Figure 9 shows the amount of extract collected in the five vials as the average of the duplicates, with error bars indicating the range.

Figure 8:

Image of the scCO₂ extract and the five fractions obtained during subsequent gaseous CO₂ fractionation at different temperatures in setup 2 (fractionation parameters: 353.15 K, 333.15 K, 313.15 K, 293.15 K, 283.15 K, 4 h, flow = 2.5 g CO₂/min and 0.025 g extract/min).

Figure 9:

Yield of the five SPL scCO₂ extracts obtained during subsequent gaseous CO₂ fractionation at different temperatures in the vials in setup 2 (fractionation parameters: 353.15 K, 333.15 K, 313.15 K, 293.15 K, 283.15 K, 4 h, flow = 2.5 g CO₂/min and 0.025 g extract/min).

The majority of the extract was collected in the first vial (353.15 K), accounting for 60.2 wt. % of the feed. At 333.15 K, 9.3 wt. % was collected in the second vial and at 313.15 K, 7.0 wt. % was collected in the third vial. Upon further cooling, 9.1 wt. % was collected in the fourth vial (293.15 K), and an additionally 4.2 wt. % was collected in the fifth vial at 283.15 K. The total loss was 10.3 wt. %. Therefore, it can be assumed, that at a CO₂ temperature of 283.15 K, some volatile substances were carried out by the CO₂ gas stream.

Subtracting the cumulative yield from the amount of pumped extract gives the amount of extract in the CO₂ gas stream at each temperature. Furthermore, using this value and the amount of used CO₂ (600 g), the VPL of the gaseous CO₂ was calculated (Table 2).

The amount of extract in the gaseous CO₂ decreased as the temperature in the five vials decreased. The amount varied from 2.4 g at 353.15 K to 0.6 g at 283.15 K. Consequently, 2.4 g of extract dissolved in 0.6 kg hot gaseous CO₂ at 353.15 K. This corresponded to a VPL of 4.0 g extract/kg CO₂. When cooled to 333.15 K, 30.5 wt. % (1.8 g extract) of the SPL extract remained dissolved in the CO₂, yielding a VPL of 3.1 g extract/kg CO₂. As the temperature decreased further, the amount of extract dissolved in the gaseous CO₂ decreased to a VPL of 1 g extract/kg CO₂ at 283.15 K. Compared to setup 1, the VPL at 333.15 K in setup 2 was significantly lower (setup 1: 5.7 g extract/kg CO₂; setup 2: 3.1 g extract/kg CO₂). During decompression in setup 1, there was an atomization of the extract leading to the assumption that either the atomized extract was carried out or the extract dissolved better in the gaseous CO₂ due to the atomization. Moreover, in setup 2, extract separation occurred in the first vial, indicating that the gaseous CO₂ was no longer fully saturated in the second vial. The five different fractions in the vials were analyzed by GC-MS/FID. Figure 10 shows the composition of the scCO₂ extract, the five fractions and the SPL as the mean of duplicates, with error bars indicating the range. As not all compounds are detectable by GC, the columns do not sum up to 100 wt. %.

Figure 10:

GC-detectable composition of SPL, the scCO₂ SPL extract, and the five SPL scCO₂ extracts obtained during subsequent gaseous CO₂ fractionation at different temperatures in setup 2 (fractionation parameters: 353.15 K, 333.15 K, 313.15 K, 293.15 K, 283.15 K, 4 h, flow = 2.5 g CO₂/min and 0.025 g extract/min).

During scCO₂ extraction, the main part of the high molecular weight substances and sugars contained in the SPL were not extracted. This resulted in an enrichment of syringols (19.4 wt. %), guaiacols (9.3 wt. %), ketones (9.0 wt. %), acids (10.0 wt. %), and furans (4.2 wt. %) in the scCO₂ extract. The total proportion of GC-detectable substances was 37.2 wt. % in the SPL and 56.6 wt. % in the scCO₂ extract. In the first vial (353.15 K), the GC-detectable amount was only 42.8 wt. %, which is lower than in the scCO₂ extract. This decrease is attributed to the removal of substances that are highly soluble in gaseous CO₂, which increased the proportion of high molecular weight substances that are not GC-detectable but still soluble in scCO₂. In subsequent vials, the amount of GC-detectable substances increased, which indicates a lower proportion of non-detectable high molecular weight substances.

Through fractionation with heated gaseous CO₂, the amount of acids was reduced to 1.2 wt. % in the first vial (353.15 K). The amount of ketones decreased to 2.7 wt. %, while the amount of syringols increased to 25.5 wt. %. As the gaseous CO₂ in the vials was cooled further, the syringol content gradually decreased while the content of ketones, acids, and furans increased. No syringols were detected in the extract of the last vial (283.15 K). However, high amounts of ketones (25.2 wt. %), acids (15.8 wt. %), and furans (14.7 wt. %) were present. Initially, the guaiacol content increased before decreasing, reaching its highest concentration (19.2 wt. %) in the extract of the third vial at 313.15 K. The overall content of monomeric phenolic substances (guaiacols, syringols, phenols) was 15.1 wt. % in the SPL and 30.3 wt. % in the scCO₂ extract. The content of monomeric phenolic substances in the five vials varied between 33.7 wt. % (353.15 K) and 4.7 wt. % (283.15 K). Fractionation with hot gaseous CO₂ slightly increased the content of monomeric phenolic substances while drastically reducing the content of acids and ketones in the extract. Another study produced scCO₂ extracts with up to 41.1 wt. % monomeric phenolic substances using pressure separators (Möck et al. 2024). However, the content of acids was 4.3 wt. %, and the content of ketones was 8.0 wt. %. Extracts with low acid and ketone content could be promising for producing phenol resins and preserving wood (Karthäuser et al. 2024; Lourençon et al. 2016; Sukhbaatar et al. 2009). Further purification processes, such as distillation, can be used to obtain bio-based platform chemicals from extracts with high levels of acids and ketones. These chemicals can be used to produce biopolymers, etching agents, or solvents (Murali et al. 2017).

3.4 Limitations and outlook

Due to the limited heating capacity, the temperature of the off-gas in setup 1 was restricted to 333.15 K, and the temperature of the gaseous CO₂ in setup 2 was restricted to 353.15 K. Higher temperatures showed to improve the separation of acids and ketones. Therefore, future research should explore using higher temperatures to enhance separation efficiency. Setup 2 experienced losses amounting to 10.8 wt. %, they are suspected to be due to volatile substances carried away with the gaseous CO₂. Since the cooling capacity was limited to 283.15 K, applying colder cooling stages could reduce these losses and facilitate further investigation. In setup 1, the ratio of CO₂ to extract could not be controlled or varied. Setup 2 tested a ratio of 1:100 (extract to CO₂). Future studies should vary this ratio and examine its influence on fractionation to provide valuable insights. Additionally, the extract in setup 2 remained in the vials throughout the experiment, thus allowing for prolonged contact with CO₂. It would be beneficial to investigate whether effective fractionation can be achieved with short contact times between the extract and hot, gaseous CO₂. Furthermore, combining pressure separators with hot CO₂ fractionation appears promising for producing extracts with high concentrations of monomeric phenolic substances and low concentrations of ketones and acids. These extracts could be valuable to the adhesive industry as an alternative to petrochemical, phenol-based fractions. One possible applications of the residue could be the production of activated carbon, while sugars could be utilized as bio-based platform chemicals. Additional processing of fractions rich in acids and ketones may be employed to convert them into viable bio-based platform chemicals.