Characterization of Liquefaction Products from Lignocellulosic and Aquatic Biomass

Abstract

Biomass liquefaction is a promising thermochemical route to convert lignocellulosic residues into bio-oil. This study evaluates the liquefaction behavior of 13 biomasses with varying particle sizes (0.3–2.0 mm) and moisture contents (5–11%) under mild solvolysis conditions. High-performance liquid chromatography (HPLC-RID) and thermogravimetric analysis (TGA) were used to characterize bio-oil composition and biomass properties, respectively. Maximum conversion (72%) was achieved for Miscanthus, while Ulva lactuca reached only 23% due to its low carbohydrate content. Hemicellulose-rich feedstocks showed higher yields, whereas high lignin content generally reduced conversion. Furfural was the main compound identified in the aqueous phase (up to 51 g/L), reflecting extensive pentose degradation. Laboratory and industrial-scale liquefaction of cork and eucalyptus revealed scale-dependent differences. Industrial cork bio-oil showed increased xylose (0.70 g/L) and furfural (0.40 g/L), while industrial eucalyptus exhibited elevated levels of acetic (0.46 g/L) and formic acids (0.71 g/L), indicating enhanced deacetylation and demethoxylation reactions. These findings offer valuable insights for optimizing feedstock selection and process conditions in biomass liquefaction. The valorization of lignocellulosic residues into bio-oil contributes to the development of scalable, low-carbon technologies aligned with circular economy principles and bio-based industrial strategies.

1. Introduction

The shift toward renewable energy sources has opened new opportunities for a more sustainable and innovative future, with biomass standing out as one of the most versatile and promising resources. This renewable feedstock offers a dual advantage: it can be used directly as a fuel for energy production or as a raw material for synthesizing high-value chemicals [1]. Biomass is a promising and versatile resource in the search for sustainable alternatives to fossil fuels. As global concerns about energy security and environmental impact grow, biomass has emerged as a renewable source of energy and high-value chemical [2,3,4].

Through biochemical, thermochemical, or physico-chemical processes, biomass can be transformed into a wide range of sustainable products. Among these approaches, thermochemical conversion shines for its efficiency, leveraging high temperatures under limited or no oxygen to transform organic materials [5]. Processes like gasification, pyrolysis, and liquefaction not only yield energy-rich fuels but also generate key compounds for industrial and environmental applications [6,7]. Liquefaction has gained attention for its ability to produce bio-oils and other high-value chemicals under mild conditions compared to pyrolysis [8,9,10].

Liquefaction, the primary focus of this study, is a thermochemical process that involves the transformation of organic materials at high temperatures, typically in the presence of limited or no oxygen [1,4,5]. There are various methods of liquefaction, such as high-temperature hydrogen pressure processes, hydrothermal upgrading (HTU), and solvolysis [11,12,13]. The latter, applied in this study, is performed under moderate temperatures (100–250 °C) and atmospheric pressure with a catalyst to facilitate the reaction. Solvolysis liquefaction is influenced by several factors, including the type of biomass, solvent selection, catalyst properties, temperature, and reaction time. Recent studies have demonstrated the effectiveness of solvent-based liquefaction in improving bio-oil yields and reducing solid residues [1,4].

Lignocellulosic biomass, which includes materials like wood, agricultural residues, and various plant types, is mainly composed of cellulose, hemicellulose, and lignin. The composition of lignocellulosic biomass can vary depending on its source, such as hardwoods, softwoods, or herbaceous plants. Cellulose, the primary polymer in lignocellulosic biomass, is typically crystalline and resistant to chemical breakdown, limiting its reactivity [5]. In contrast, amorphous cellulose and hemicellulose, with their simpler linear and branched structures, are more reactive and easier to convert during liquefaction [11]. Lignin, a complex and highly branched polymer, further complicates the liquefaction process, as it tends to form solid residues through repolymerization at high temperatures, thereby §reducing bio-oil yields [12]. This is particularly evident in softwoods (coniferous trees), which contain higher lignin content than hardwoods (deciduous trees) and herbaceous plants [13]. Herbaceous biomass has chemical properties more similar to hardwoods than to softwoods, with a low lignin content that makes it easier to decompose [9]. In addition, studies have highlighted the challenges of depolymerizing lignocellulosic biomass, emphasizing the importance of pretreatment and catalyst selection in optimizing bio-oil production [14,15].

Among the biomass types studied, aquatic biomass such as algae offers distinct advantages due to its unique composition and reaction pathways during liquefaction. For example, the alga Ulva lactuca used in the experiments has a high content of fibers (54 wt%) and minerals (20 wt%), along with notable amounts of proteins (8.5 wt%) and lipids (7.9 wt%). These characteristics influence its liquefaction behavior and product distribution, making it a promising alternative feedstock for bio-oil production [16].

Bio-oils produced through liquefaction are complex mixtures of sugars, organic acids, alcohols, aldehydes, ketones, and other organic compounds, with their specific composition heavily influenced by the biomass source and processing conditions. Beyond their application as biofuels, bio-oils often contain valuable industrial chemicals such as furfural, levulinic acid, and formic acid [17]. For instance, furfural is formed when C5 sugars like xylose undergo dehydration, while levulinic acid results from the acid hydrolysis of hexoses at high temperatures [5,18,19,20]. These compounds can be extracted from bio-oils, producing an organic phase with a higher heating value and an aqueous phase enriched with high-value chemicals, thereby enhancing the economic and environmental viability of liquefaction processes [19].

Biomass composition, particularly the relative proportions of cellulose, hemicellulose and lignin, plays a crucial role in determining the efficiency of the liquefaction process and the resulting bio-oil yields. Higher lignin content is associated with lower bio-oil yields; as lignin’s complex macromolecular structure undergoes thermal decomposition at temperatures exceeding 250 °C, phenolic free radicals are formed. These radicals can repolymerize and condense, leading to the formation of solid residues and a decrease in bio-oil yield [21,22]. In contrast, cellulose and hemicellulose, with their simpler and more linear structures, are more readily broken down, contributing to higher bio-oil production [7]. In addition to the chemical composition of biomass, physical parameters such as particle size, porosity, and moisture content can influence liquefaction performance by affecting heat and mass transfer, solvent diffusion, and catalyst accessibility. Although these variables were not intentionally varied in this study, their potential role is acknowledged, and their values were recorded to ensure experimental consistency. Recent studies have emphasized the importance of optimizing operational parameters such as biomass-to-solvent ratio, reaction time, and catalyst type to enhance bio-oil yield and selectivity [23,24].This work contributes to the existing literature by systematically comparing multiple biomass types under standardized conditions and exploring the influence of feedstock composition and scale-up effects on bio-oil production and composition.

2. Materials and Methods

2.1. Materials

The biomasses studied in this work included softwoods (maritime pine saw-dust, and pine nut shells), hardwoods (eucalyptus sawdust, peach stone shells, olive stones, olive bagasse, spent coffee grounds, and cork residues), herbaceous (Miscanthus, thistle, and rice husk), and one aquatic biomass (Ulva lactuca). All biomass samples were sourced from local suppliers and industrial partners in Portugal. The solvent and catalyst used in the biomass liquefaction experiments were 2-ethyl-hexanol (provided by Resiquímica, Sintra, Portugal) and p-toluene sulfonic acid (98.5%, from Sigma-Aldrich), respectively. Acetone (≥99%) from Honeywell was used to wash the solid residues from liquefaction.

2.2. Methods

2.2.1. Liquefaction Reactions

The procedure used was optimized in previous studies on pinewood liquefaction [25,26]. All liquefaction experiments were carried out in duplicate at 160 °C and ambient pressure. The catalyst used was 3 wt% versus the organic matter content of the biomass; the biomass (50 g) to solvent ratio was 1:5 and the reaction time was 120 min. For maritime pine sawdust, a reaction time of 180 min was also studied. The reactions were carried out in a four-neck reactor with a mechanical agitator (Figure 1). One of the necks was attached to a Dean–Stark separator/condenser to collect the reaction condensates and a thermopar was placed in another neck.

To start each experiment, the solvent quantity was added to the reactor containing the biomass, the thermostat was set to 100 °C, and the stirrer was switched at a speed of around 180 rpm. When the temperature reached 140 °C, the desired quantity of the catalyst was added to the reactor, the stirrer velocity was increased to 330 rpm, and the reaction timer was set to begin. Every 30 min of reaction, the condensates were removed from the Dean–Stark separator for subsequent analysis. After the reaction, the liquid fraction was separated from the larger residues by filtration, with a nickel mesh with retention of 90 µm particles. Afterwards, the liquid obtained in this filtration was centrifuged at 4000 rpm for 15 min to separate the smaller solid particles from the bio-oil. The solid residues were washed with acetone, dried in an oven at 65 °C to constant weight, weighed, and stored for analysis.

The bio-oil yield was calculated using Equation (1):

$$C(\%)={\displaystyle \frac{m_i-m_{lr}-m_{sr}}{m_i}}\times 100$$

(1)

where m_i is the initial biomass dry mass, m_lr and m_sr are, respectively, the final dry mass of the larger and smaller particles residues [6].

2.2.2. Water Extraction of Bio-Oil

To extract the water-soluble components, the bio-oil was mixed with distilled water in a settling funnel, using a bio-oil to water volumetric ratio of 1:2. After stirring for 10 min, the mixture was allowed to settle and was centrifuged at 4000 rpm for 15 min to attain a better separation of the organic and aqueous phases.

2.2.3. Biomass Characterization

The particle size determination of the biomasses was carried out using an Endecotts sieves shaker, with diameters sieves pore from 150 µm to 3.55 mm. The average Sauter diameter (D_S) was calculated by the following Equation (2):

$$\overline{D_S}={\displaystyle \frac{1}{{\sum }_{i=1}^n\frac{x_i}{D_{pi}}}}$$

(2)

where x_i is the mass of sample retained in each sieve and D_pi is the average diameter of those particles.

The chemical characterization of the biomass and of the solid residues was carried out by thermogravimetric analysis (TGA) in Hitachi model STA 7200. The thermogravimetric analysis (TGA) was performed using a heating rate of 10 °C/min under a nitrogen atmosphere, following the ASTM E1131-20 standard for the compositional analysis of organic materials [25]. The chemical composition of the biomass and solid residues in terms of cellulose, hemicellulose, and lignin was determined via acid hydrolysis, following the NREL protocol (NREL/TP-510-42618) [13]. This methodology enables the differentiation of polysaccharide fractions based on the release of monosaccharides: glucose, primarily from cellulose; and xylose, arabinose, and acetyl groups, predominantly from hemicellulose. Although glucose may originate from both cellulose and hemicellulose, hemicellulose content is estimated by summing xylose, arabinose, and acetyl groups, as recommended in the protocol. This allows a reliable approximation of the biomass polysaccharide profile.

The thermogravimetric curves (TG) of lignocellulosic biomass allow the identification of several well-defined temperature levels that correspond to water evaporation (<125 °C), hemicellullose degradation (250–325 °C), cellulose degradation (325–375 °C), and lignin degradation (>375 °C) [15]. In quantitative acid hydrolysis (HAQ), acids can decompose the cellulose and hemicellulose biopolymers of biomass into monosaccharides that are afterwards analyzed using HPLC [26].

The ash content of the biomass and liquefaction residues was determined by calcination in a Nabertherm P330 oven where the samples were heated from ambient temperature to 400 °C in 2 h, then maintained at 400 °C for 3 h, then heated up to 1000 °C in 3 h and finally maintained at 1000 °C for 3 h. The calcinated samples were weighed after cooling down to the ambient temperature.

2.2.4. Bio-Oils Characterization

The complete chemical and physical characterization of bio-oils is extremely difficult due to the high complexity and diversity of compounds. HPLC analyses were performed using an The Agilent 1100 Series HPLC System was sourced from Agilent Technologies, based in Waldbronn, Germany. It was equipped with a Diode Array Detector (DAD) and a Fluorescence Detector (FLD), both from Agilent. The chromatographic column used was an Aminex HPX-87H, manufactured by Bio-Rad Laboratories, Hercules, California, USA.), with detection performed via refractive index (RID) and UV-visible absorbance. This equipment was used to analyze the aqueous phases obtained in the water extraction of bio-oil and also the aqueous phases collected in the Dean–Stark separator during the liquefaction experiments. The mobile phase was 5 mmol/L dilute sulfuric acid solution with a flow rate of 0.6 mL/min and the column temperature was kept at 50 °C. Although chromatograms are not included in the supplementary material, compound identification and quantification were conducted by comparison with authentic standards under consistent retention times, in accordance with established protocols [24].

3. Results and Discussion

3.1. Characterization of Biomass Feedstocks

3.1.1. Particle Size, Moisture, and Ash Contents

While particle size and moisture content were measured, they were not intentionally varied as this study focused on the effect of biomass composition. Their potential influence on conversion efficiency is acknowledged and represents an opportunity for future research.

Table 1. Particle size and moisture content of the biomass.

Biomass	Mean Sauter Diameter (mm)
Pine sawdust	0.5
Pine nut shells	2.0
Eucalyptus sawdust	0.7
Peach stone shells	1.0
Olive stones	1.5
Olive bagasse	1.2
Spent coffee grounds	0.3
Cork residues	1.3
Miscanthus	0.9
Thistle	0.6
Rice husk	0.9
Ulva lactuca	0.3

presents the Sauter mean diameter and moisture content of the different biomasses. Although these parameters were not deliberately varied, their influence on the liquefaction process is acknowledged. Smaller particle sizes generally promote better heat and mass transfer, facilitating solvent penetration and catalytic access, which can enhance biomass conversion. For instance, spent coffee grounds and Ulva lactuca, which exhibited the smallest particle sizes (0.3 mm), also showed relatively high and low bio-oil yields, respectively. This contrast highlights that, while particle size may aid liquefaction efficiency, chemical composition remains the dominant factor. Conversely, pine nut shells, with the largest particle size (2.0 mm), still achieved a moderate bio-oil yield (43.5%), suggesting that despite potential mass transfer limitations, their relatively high cellulose content supported significant conversion. These observations underscore that particle size can influence reaction dynamics but must be considered alongside compositional attributes to fully understand liquefaction performance.

Table 2. Moisture content of the biomass.

Biomass	Moisture (w %)	Ash (w %)
Pine sawdust	9.06	0.8
Pine nut shells	10.16	0.9
Eucalyptus sawdust	8.77	6.8
Peach stone shells	9.96	0.5
Olive stone	8.37	1.5
Olive bagasse	9.07	4.2
Spent coffee ground	2.54	1.6
Cork residues	7.26	3.8
Miscanthus	9.59	1.9
Thistle	9.11	2.6
Rice husk	10.84	16.9
Ulva lactuca	12.23	35.1

shows the moisture and ash contents of the biomasses. With the exception of spent coffee grounds, which had been previously dried, most samples exhibited moisture contents around 10%, typical for air-dried biomass. Regarding ash content, significant variability was observed among the samples. As expected, rice husk presented a high ash content (16.9%), which is consistent with its well-documented abundance of silica and other inorganic constituents derived from soil absorption during growth. Similarly, Ulva lactuca, an aquatic biomass, showed the highest ash content (35.1%), reflecting its marine origin and high mineral load. In contrast, hardwoods such as peach stone shells and olive stones exhibited low ash contents (0.5–1.5%), indicative of their denser, lignocellulosic structure with lower inorganic fractions. These variations are relevant because higher ash contents generally reduce the proportion of biopolymers available for liquefaction and may also interfere with catalyst efficiency. Therefore, both the moisture and ash profiles of the biomasses are important parameters to consider when evaluating liquefaction potential.

3.1.2. Thermogravimetric Analysis

The TGA and DTG curves for pine sawdust are shown in Figure 2. This analysis shows an initial mass loss of 6.5% up to 125 °C, which corresponds to the moisture removal; the second plateau corresponds to a mass loss of 23.6%, in the 125–325 °C temperature range, which corresponds to the degradation of hemicellulose; the peak at 364.5 °C in the dTG curve is associated with the mass loss due to the degradation of cellulose, corresponding to a total loss of 38.9% between 325 and 375 °C; the loss of mass recorded from 375 °C onwards, which represents 12.4% of the total mass, is associated with the degradation of lignin.

The TGA curve presented for pine sawdust is representative of the general thermal degradation behavior observed for all biomass samples. Only minor variations in degradation onset temperatures and peak intensities were detected, and the overall trend—comprising moisture loss, hemicellulose decomposition, cellulose degradation, and lignin breakdown was consistent across all materials.

Table 3. Biomass samples composition based on TGA.

Biomass Type	Raw Material	Cellulose (%)	Hemicellulose (%)	Lignin (%)	Residue (%)
Softwood	Pine sawdust	41.6	25.2	13.2	19.9
	Pine nut shells	39.3	12.7	14.1	33.9
Hardwood	Eucalyptus sawdust	35.2	18.1	14.7	31.9
	Peach stone shells	27.4	29.3	13.0	30.3
	Olive stones	40.4	15.4	11.1	33.1
	Olive bagasse	36.3	14.7	16.1	32.9
	Spent coffee grounds	14.4	41.4	20.6	23.7
	Cork residues	18.8	20.0	34.1	27.2
Herbaceous	Miscanthus	49.5	17.6	9.7	23.1
	Thistle	51.9	19.1	8.5	20.5
Crops	Rice husk	36.9	11.0	12.0	40.1
Aquatic	Ulva lactuca	10.6	23.7	10.2	55.6

shows the biomass composition obtained by TGA, including estimated contents of cellulose, hemicellulose, lignin, and residual mass. As expected, herbaceous biomasses such as Miscanthus and thistle presented the highest cellulose contents (49.5% and 51.9%, respectively), which may contribute to their high bio-oil conversion. Conversely, aquatic biomass (Ulva lactuca) and spent coffee grounds exhibited significantly lower cellulose levels (10.6% and 14.4%), correlating with their reduced liquefaction performance. Cork residues showed the highest lignin content (34.1%), which is consistent with the known structure of cork cell walls and helps explain the lower conversion efficiency of this biomass. The residual mass was particularly high for pine nut shells (33.9%) and rice husk (40.1%), indicating a substantial proportion of thermally resistant or inorganic material. These variations confirm that biomass type and, in particular, the relative content of cellulose, hemicellulose, and lignin plays a critical role in the liquefaction behavior.

The use of TGA to estimate the relative fractions of cellulose, hemicellulose, and lignin in lignocellulosic biomass is consistent with previous studies, including the study by Wang et al. [26], which compared thermal degradation patterns across multiple biomass types.

It is worth noting that the ash content in Table 2 is much lower than the solid residue resulting from the TGA (Table 3), which was carried out under N₂ atmosphere. In TGA, the degradation of the organic compounds occurs but their removal from the crucible may not be complete, whereas the ash content in Table 2 is the inorganic residue resulting from the combustion of the biomass with air that produces CO₂ and water, which are completely removed from the crucible.

Table 4. Biomass samples composition based on acid hydrolysis.

Biomass	Raw Material	Cellulose (%)	Hemicellulose (%)	Xylan (%)	Arabinan (%)	Acetyl Groups (%)	Lignin (%)	Ash (%)	Others (%)
Softwood	Pine sawdust	33.4	21.7	18.0	1.9	1.8	33.8	0.5	10.6
	Pine nut shells	25.9	21.0	16.3	2.7	2.0	42.6	3.0	7.4
Hardwood	Peach stone shells	22.9	29.6	24.0	0.0	5.6	41.6	0.2	5.7
	Olive stones	29.4	32.4	29.2	0.0	3.2	34.2	0.9	3.1
	Olive bagasse	19.2	22.1	14.3	3.0	4.9	38.1	8.8	11.8
	Spent coffee grounds	9.3	39.9	37.5	2.1	0.3	28.9	1.7	20.2
Herbaceous	Miscanthus	44.5	27.2	20.8	4.2	2.2	23.4	1.8	3.1
	Thistle	39.3	25.3	21.8	1.0	2.5	15.4	5.4	14.6
Crops	Rice husk	36.3	22.7	18.4	1.9	2.4	22.0	11.5	7.5
Aquatic	Ulva lactuca	10.3	16.1	11.4	0.0	4.7	7.7	38.2	27.8

presents the chemical analysis of the biomasses based on the results of acid hydrolysis. This method allows us to determine the contents of the polysaccharides of glucan (Gn), xylan (Xn), arabinan (Arn), and acetyl (Ac) groups by analyzing the aqueous solution, whereas the Klason lignin (KL) and the ash contents can be determined by analyzing the waste. Thus, the chemical composition of the samples on a dry basis in terms of cellulose, hemicellulose, and lignin can be estimated by the following Equations (3)–(6):

Cellulose (%) = Gn

(3)

Hemicellulose (%) = Xn + Arn + Ac

(4)

Lignin (%) = KL

(5)

Others (%) = 100 − Celullose − Hemicellulose − Lignin − Ash

(6)

It should be noted that the chemical composition in terms of lignin is an approximation, as only Klason lignin (i.e., the acid-insoluble fraction) was considered. This applies to all biomass samples listed in Table 4. The contribution of acid-soluble lignin, typically present in small amounts, was not quantified in this study.

On the other hand, the hexoses present in hemicellulose (rhamnose, galactose, and mannose) were quantified as xylose, since they were all coeluted in the operating conditions used in the HPLC analysis. This co-elution was taken into account during quantification. Compounds with overlapping peaks that could not be resolved under the applied chromatographic conditions were either grouped accordingly or excluded. This approach was consistent with recent literature and ensures comparability across biomass types. The identification and quantification were based on external standards for glucose, xylose, arabinose, furfural, HMF, formic, acetic, and levulinic acids.

Comparing the results presented in Table 3 and Table 4 it is possible to conclude that the cellulose contents determined by TGA are highly correlated with the ones obtained by acid hydrolysis (R² = 0.91). Hemicellulose contents presented a lower correlation coefficient (R² = 0.585), mainly due to the very bad correlation for softwoods and aquatic biomass as the correlation for hardwood biomasses and herbaceous crops was higher. Therefore, it is possible to conclude that cellulose and hemicellulose contents of biomass can be estimated from the TGA results. On the contrary, the correlation coefficient between the lignin and ash contents based on the two methods is low and so it is possible to conclude that TGA is not suitable for the determination of lignin and ash contents.

As shown in Table 4, Miscanthus and thistle exhibited high cellulose contents (44.5% and 39.3%, respectively) and moderate hemicellulose levels (27.2% and 25.3%), which correlate with their relatively high bio-oil conversions of 72.2% and 50.2% (

Table 5. Biomass conversion in bio-oil (0.1%).

Raw Material	Bio-Oil Conversion (%)
Ulva lactuca	23.2
Olive bagasse	31.3
Pine nut shells	43.5
Rice husk	48.3
Eucalyptus sawdust	49.7
Thistle	50.2
Pine sawdust	55.4
Spent coffee grounds	58.4
Cork residues	58.6
Olive stones	66.8
Peach stone shells	67.2
Miscanthus	72.2

). In contrast, Ulva lactuca, with the lowest cellulose (10.3%) and hemicellulose (16.1%) contents, showed a conversion of only 23.2%, confirming the limited liquefaction potential of aquatic biomass. Olive bagasse and olive stones, although both classified as hardwoods, demonstrated markedly different conversions (31.3% vs. 66.8%), which can be attributed to their differing biopolymer profile olive stones containing higher cellulose (29.4%) and hemicellulose (32.4%) than olive bagasse (19.2% and 22.1%, respectively). These examples support the conclusion that bio-oil yield is more dependent on chemical composition than on biomass classification.

3.2. Liquefaction Experiments

Table 5 presents the conversion of each biomass into bio-oil. These results allow us to verify that the biomass category does not seem to influence the conversion into bio-oil, which depends only on its chemical composition. Thus, for example, olive bagasse and peach stone shells, classified as hardwoods, have a difference in the conversion value greater than 35%. On the other hand, the low content of biopolymers of Ulva Lactuca explains its low conversion into bio-oil and indicates that this biomass is not suitable for liquefaction.

Lignin-rich biomasses typically exhibited lower liquefaction yields, which may be attributed to the high chemical resistance of lignin due to its aromatic, highly cross-linked structure. Lignin tends to form recalcitrant solid residues (char) during thermal treatment, limiting its conversion into liquid products. Recent studies have shown that depolymerization of lignin requires specific catalysts or pretreatment conditions to break ether and carbon–carbon bonds, making it less reactive under mild solvolysis conditions [27,28].

The correlation between the conversion and the initial content of cellulose, hemicellulose and lignin obtained from the acid hydrolysis test was calculated and it was found that the highest correlation coefficient (R²-0.872) was obtained for hemicellulose, whereas the correlation coefficient for cellulose was 0.673. Therefore, it is possible to conclude that biomasses with a high content of hemicellulose are easily liquefied.

As mentioned in Section 2.2, the reaction condensates were also collected in a Dean–Stark separator/condenser for analysis. These condensates are formed by an aqueous phase and an organic phase. The aqueous phase, characterized in the next section, includes the biomass moisture plus the water formed in the liquefaction reaction, while the organic phase consists mainly of the liquefaction solvent (2-Ethylhexanol) which forms an azeotrope with water (40% water and 60% solvent, w:w).

3.3. Characterization of the Liquefaction Products

3.3.1. Condensates Aqueous Phases

The aqueous phases of the reaction condensates were analyzed using high-performance liquid chromatography (HPLC) to quantify monosaccharides (glucose, xylose, and arabinose), aliphatic acids (formic, acetic, and levulinic acids), and furan derivatives (furfural and 5-hydroxymethylfurfural (HMF)). The results for pine sawdust liquefaction are presented in Figure 3.

As seen in Figure 3, furfural concentrations were significantly higher than the other identified compounds, peaking at 17.20 g/L after 60 min of reaction. Furfural is generated via the dehydration of hemicellulose-derived pentoses, primarily xylose and arabinose. The absence of these pentoses in the reaction condensates, alongside the elevated levels of furfural, suggests their complete conversion during the reaction. This agrees with prior research emphasizing the efficient dehydration of hemicellulose into furfural under similar conditions [23].

While glucose dehydration is expected to produce HMF, this compound was not detected in the condensates. This absence is likely due to HMF’s rehydration into formic and levulinic acids under the reaction conditions. In fact, formic acid, with a maximum concentration of 1.09 g/L that remained constant throughout the reaction, was the most abundant aliphatic acid identified. This behavior suggests that any HMF formed was quickly converted to formic and levulinic acids, as noted by Mukherjee et al. [29]. Acetic acid and levulinic acid were also detected, albeit in lower concentrations of approximately 0.1 g/L and 0.03 g/L, respectively. The continuous presence of acetic acid, a byproduct of hemicellulose breakdown, further supports the progressive degradation of hemicellulose.

The absence of glucose and HMF, combined with the low concentrations of formic and levulinic acids, implies that cellulose degradation occurred at later stages of the reaction, while hemicellulose degradation dominated the early stages. This aligns with the findings of Wang et al. [30] who observed that, when compared to the crystalline structure of cellulose, the amorphous structure of hemicellulose facilitates its rapid degradation.

Interestingly, levulinic acid can also be synthesized from pentoses via furfural, preserving the carbon atoms and without the carbon loss as formic acid. This pathway involves the hydrogenation of furfural to furfuryl alcohol, which is subsequently hydrated to levulinic acid [31]. However, the steady concentration of levulinic acid observed here suggests that furfural degradation followed alternative pathways.

The decline in furfural concentration during the second hour of reaction suggests its transformation into other compounds. The simultaneous reduction in acetic acid concentration supports the hypothesis that hemicellulose was nearly completely liquefied during this phase, corroborating insights from Hu et al. [27]. Their study highlights the rapid liquefaction of hemicellulose during initial reaction stages due to its amorphous structure, with crystalline cellulose liquefaction occurring later.

3.3.2. Biomass-Specific Trends

Condensates from various biomass sources demonstrated consistent identification of glucose, furfural, formic acid, acetic acid, and levulinic acid, with furfural exhibiting the highest concentrations. For example, the liquefaction of Miscanthus yielded a maximum furfural concentration of 51 g/L, although the overall furfural yield relative to initial biomass was only 1.33% (w/w) [26]. In contrast, condensates from Ulva lactuca displayed a lower furfural concentration of just 1.6 g/L, likely reflecting the unique chemical composition of this marine biomass and its reduced bio-oil yield (32.6%). This difference underscores the lower availability of cellulose and hemicellulose in Ulva lactuca for liquefaction, as reported by Kumar et al. [24].

The previous findings highlight the dominance of hemicellulose degradation in the early stages of liquefaction and its significant contribution to furfural production. The subsequent degradation of cellulose at later reaction stages underscores the sequential nature of biomass liquefaction processes. Further optimization of reaction parameters could enhance the yield of value-added products, such as furfural and levulinic acid, which are critical intermediates for biorefinery applications.

3.3.3. Bio-Oils Extraction Waters

The bio-oils produced in the liquefactions were subjected to a water extraction step to isolate the water-soluble compounds, which were subsequently analyzed using high-performance liquid chromatography (HPLC).

Table 6. Concentration of monosaccharides, aliphatic acids, and furans in the aqueous phases of bio-oils water extraction, based on the HPLC analysis with a RID detector.

Biomass	Glucose (g/L)	Xylose (g/L)	Furfural (g/L)	HMF (g/L)	Acetic Acid (g/L)	Formic Acid (g/L)	Levulinic Acid (g/L)
Pine sawdust	0.24	0.16			0.07	0.41	0.37
Pine nut shells	0.18	0.19	0.06		0.21	0.07
Eucalyptus sawdust	0.19	0.13			0.10	0.20	0.08
Peach stone shells	0.20	0.27			0.20	0.21	0.08
Olive stones	0.16	0.31	0.35		0.13	0.15	0.05
Olive bagasse	0.12	0.22	0.59		0.04	0.15
Spent coffee grounds	0.13	0.27				0.03
Cork residues	0.10	0.13	0.09		0.09
Miscanthus	0.30	0.28			0.14	0.23	0.14
Thistle	0.19	0.19	0.15		0.12	0.13	0.03
Rice husk	0.18	0.22	0.06		0.06	0.16	0.06
Ulva lactuca	0.09	0.06		0.06

presents the identified compounds in the extraction waters of the various bio-oils and their respective concentrations.

The variations in the composition of the several aqueous phases can be attributed to the differences in (i) the initial composition of the biomasses, (ii) the extent of biomass degradation during the liquefaction process, and (iii) the efficiency of the mass transfer of the water-soluble compounds from the bio-oil to aqueous phase. Previous studies have emphasized that the chemical composition of feedstock and the reaction parameters are critical in determining the yield and composition of liquefaction products [32].

As presented in Table 6, glucose and xylose were the only compounds identified in all aqueous phases. The presence of xylose in these extracts suggests that, contrary to expectations based on the analysis of the reaction condensates, hemicellulose-derived pentoses were not entirely degraded into furfural and acetic acid. The absence of arabinose in all extraction waters indicates that arabinose was either fully degraded during liquefaction or insufficiently transferred during extraction to surpass the detection limit of the refractive index detector (RID). Similar findings have been reported by Zhang et al. [23], who highlighted the rapid degradation of arabinose under hydrothermal conditions.

Although furfural was produced in all liquefaction reactions, it was identified only in the extraction waters of six biomass bio-oil samples. This suggests that furfural predominantly accumulates in the gaseous phase formed in the liquefaction reaction or remains within the bio-oil matrix. This observation aligns with previous research that indicated the high volatility of furfural during hydrothermal reactions [30]. If significant degradation of furfural into levulinic acid had occurred, the levulinic acid concentrations would have been substantially higher than those of formic acid. However, formic acid was consistently present in higher concentrations in all bio-oils, corroborating studies that identify formic acid as a primary degradation product of HMF and furfural [24].

Levulinic acid, derived from the degradation of HMF and furfural, was detected in lower concentrations than expected based on its degradation pathways and molar mass. This discrepancy may be attributed to competing parallel reactions involving levulinic acid, its subsequent degradation, or its lower solubility in the aqueous phase during extraction. Recent studies suggest that levulinic acid may participate in side-reactions, forming less water-soluble products such as oligomers or polymers [31].

On the other hand, the initial glucan and xylan contents of the biomass significantly influenced the composition of the resulting bio-oil. Thus, for instance, Ulva lactuca, which has the lowest glucan and xylan contents, produced a bio-oil with lower glucose and xylose concentrations. Conversely, biomasses with higher xylan content, such as peach stone shells, olive stones, and Miscanthus (24.0%, 29.2%, and 20.8%, respectively), yielded bio-oils rich in xylose. This confirms that the initial composition of the biomass plays a critical role in both the liquefaction yield and bio-oils composition, as previously demonstrated by Li et al. [28].

These findings show that the initial biomass chemical composition, particularly the glucan and xylan contents, strongly influences the efficiency of the liquefaction process and the composition of the products. The presence of key compounds, such as furfural and aliphatic acids, highlights the predominant degradation pathways of hemicellulose and cellulose, which are shaped by the reaction conditions and intrinsic properties of the biomass.

3.3.4. Solid Residues

The solid residues from the liquefaction experiments were analyzed through quantitative acid hydrolysis (

Table 7. Chemical composition of the larger particle residues obtained in the several biomasses liquefaction (w % on a dry basis), obtained by quantitative acid hydrolysis.

	Raw Material	Cellulose (%)	Hemicellulose (%)	Xylan (%)	Arabinan (%)	Acetyl Groups (%)	Lignin (%)	Ash (%)	Others (%)
Softwood	Pine sawdust	34.6	0.0	0.0	0.0	0.0	54.8	1.1	9.5
	Pine nut shells	38.7	0.5	0.0	0.5	0.0	54.6	3.9	2.2
Hardwood	Eucalyptus sawdust	21.2	0.1	0.0	0.0	0.1	61.6	7.0	10.1
	Peach stone shells	4.4	0.0	0.0	0.0	0.0	92.7	0.7	2.2
	Olive stones	47.3	0.4	0.0	0.0	0.4	41.8	3.0	7.6
	Olive bagasse	18.7	7.2	6.5	0.0	0.7	56.9	3.5	13.6
	Spent coffee grounds	20.8	29.4	27.8	1.6	0.0	32.6	1.9	15.4
	Cork residues	22.6	0.4	0.0	0.4	0.0	51.9	9.1	16.0
Herbaceous	Miscanthus	53.2	0.0	0.0	0.0	0.0	32.1	5.5	9.3
	Thistle	55.4	0.3	0.0	0.0	0.3	21.2	6.2	16.9
Crops	Rice husk	33.7	0.2	0.0	0.2	0.0	23.7	34.9	7.5
Aquatic	Ulva lactuca	5.7	0.0	0.0	0.0

). A direct comparison of the initial composition of each biomass with the chemical composition of the residues reveals that, as expected, hemicellulose was almost entirely degraded in nearly all cases. This is supported by the absence of xylan, arabinan, and acetyl groups in the residues. The notable exception is spent coffee grounds, where the conversion rate of 58.4% appears to be more closely linked to its high lipids content rather than to the degradation of hemicellulose or cellulose. This observation aligns with prior studies emphasizing the unique behavior of lipid-rich biomasses during liquefaction [12].

The reduction in the percentage values of “others”, which includes the extractives and other unidentified compounds, in some of the residues when compared with the initial biomass highlights their possible degradation and subsequent incorporation in the bio-oil. However, the Soxhlet washing step of the residues used in these experiments could have also contributed to the removal of some extractives from the residues. Previous research has documented that the washing steps of the solid residues can remove polar extractives, impacting residues’ composition [32].

As seen in the previous Table, in softwood biomasses residues, such as pine sawdust and pine nut shells, an increase in the cellulose content was observed, indicating that cellulose degradation occurred to a much lesser extent than hemicellulose degradation. This observation is further supported by the increased lignin and ash contents in the residues, consistent with the reduction in biopolymers such as hemicellulose and other volatile organic components. As mentioned above, the selective degradation of hemicellulose over cellulose is well-documented in the literature and attributed to its lower thermal stability and amorphous structure [33].

In hardwood biomasses, such as peach stone shells, cellulose degradation appears to be more pronounced. This correlates with the higher bio-oil conversion rates observed in these biomasses. The significant degradation of both cellulose and hemicellulose in peach stone shells supports its higher liquefaction efficiency. Studies have shown that hardwood biomasses, with their lower lignin content compared to softwoods, often exhibit greater reactivity during hydrothermal liquefaction, leading to higher bio-oil yields [31].

Herbaceous crops displayed distinctive behavior. Miscanthus, for instance, achieved the highest bio-oil conversion rate among the analyzed biomasses. The residues of Miscanthus liquefaction showed higher lignin and cellulose content compared to the initial biomass, suggesting that hemicellulose was entirely degraded while cellulose degradation was only partial. This behavior reflects the structural differences between hemicellulose and cellulose, as the former is more susceptible to hydrothermal breakdown. Rice husks, on the other hand, are characterized by their high initial ash content, which remains elevated in the residues due to the limited degradation of mineral components. This observation aligns with previous findings that highlight the resistance of inorganic materials in rice husks to thermal and chemical breakdown during liquefaction [26].

The only aquatic biomass studied in this work, Ulva lactuca, demonstrated a low conversion rate into bio-oil. However, this was not due to incomplete degradation of hemicellulose or cellulose, but rather to the low initial content of these biopolymers in the biomass. The limited availability of glucan and xylan in aquatic biomasses inherently restricts their liquefaction potential, as noted in recent studies [29]. Consequently, achieving high bio-oil yields with this type of biomass is certainly very challenging.

The analysis of the reaction condensates further confirms that during liquefaction, hemicellulose degradation occurs to a greater extent than cellulose degradation. It is likely that the degraded cellulose should be predominantly the amorphous cellulose, which constitutes only around 20% of the total cellulose in biomass. This explains the high cellulose contents remaining in the residues, which should certainly be the stable crystalline cellulose. As mentioned above, crystalline cellulose, due to its highly ordered structure, is known to be resistant to thermal and chemical degradation compared to its amorphous counterpart [34].

3.4. Influence of the Reaction Time on Liquefaction

The reaction time is a critical variable influencing biomass conversion, as it determines the extent of biomass degradation and the relative yields of bio-oil, gaseous products and residues. Previous studies have emphasized that longer reaction times generally promote higher biomass conversion, although they may also enhance side reactions, such as repolymerization, which reduce bio-oil yields [26].

The influence of the reaction time on liquefaction of sawdust biomass is presented in

Table 8. Influence of the reaction time on the products of outer layer of pinewood sawdust liquefaction.

Reaction Time (min)	120	180
Condensates volume (mL)	11.6	13.8
Large particles residues (g)	20.6	18.1
Small particles residues (g)	1.0	0.9
Ash content (% w/w)
Large particles residues	0.8	1.0
Small particles residues	2.5	3.3
Bio-oil	0.0	0.0
Conversion (% w/w)
Bio-oil Conversion	52.4	58.4
Bio-oil Conversion (dry basis without ash)	52.4	58.5

. As seen, there was an increase in the condensates volume with reaction. In fact, there was an augment of 2.2 mL of condensates in one hour of reaction since biomass degradation is favored, resulting in the release of more volatile products. It is worth mentioning that reaction time positively influences biomass conversion into bio-oil until a critical threshold value is reached, beyond which repolymerization reactions become significant, which tend to increase the production of solid residues. In the experiments presented in

Table 9. Influence of the reaction time on the composition of the extraction waters of pine sawdust bio-oils.

Time (min)	Glucose (g/L)	Xylose (g/L)	Acetic Acid (g/L)	Formic Acid (g/L)	Levulinic Acid (g/L)
120	0.24	0.16	0.07	0.41	0.37
180	0.23	0.15	0.05	0.23	0.12

, this critical reaction time threshold was not reached, since the mass of residues decreased when the reaction time increased from 120 to 180 min. This finding is aligned with prior research that suggests that the optimum reaction time depends on both the feedstock type and the operating conditions [12].

As mentioned above, the analysis of the condensates collected during pine sawdust liquefaction presented in Figure 2 shows that furfural concentration decreased significantly after the first hour of reaction. This observation suggests that hemicellulose degradation occurs in the initial stages of liquefaction. The products derived from cellulose liquefaction, including glucose, formic acid, and levulinic acid, were also analyzed. In the later stages of liquefaction, glucose was no longer detected and formic acid concentration decreased, confirming that cellulose degradation occurs in the later stages of liquefaction.

As seen in Table 8, an increase of 6% conversion in liquefaction was attained with the increase in the reaction time from 120 to 180 min. This conversion increase in the last hour of reaction can be attributed to the final degradation of cellulose. However, Table 9 shows that the concentration of the analyzed compounds in bio-oils extraction waters was lower after 180 min compared to 120 min. This reduction may be due to the occurrence of repolymerization and/or degradation reactions that alter the composition of bio-oils for longer reaction times [31,34].

3.5. Characterization of Two Industrial Bio-Oils

The aqueous phases obtained in the extraction of two industrial bio-oils were analyzed by HPLC and the results are presented in

Table 10. Concentration of monosaccharides, aliphatic acids, and furans in the aqueous phases of lab-scale and industrial bio-oils water extraction.

Biomass	Scale	Glucose (g/L)	Xylose (g/L)	Arabinose (g/L)	Furfural (g/L)	Acetic Acid (g/L)	Formic Acid (g/L)	Levulinic Acid (g/L)
Eucalyptus	Lab	0.19	0.13			0.10	0.20	0.08
Eucalyptus	Ind	0.07	0.07	0.03	0.46	0.46	0.71	0.89
Cork	Ind	0.22	0.7	0.30	0.40	0.78

. The composition of a bio-oil produced in the lab-scale liquefaction of Eucalyptus is also presented for comparison. In fact, considering that the industrial bio-oils were produced under operating conditions distinct from the ones used at laboratory scale, most notably a higher biomass-to-solvent mass ratio, it is interesting to compare their composition.

According to the results in Table 10, it is possible to conclude that similar trends were observed in the composition of the extraction waters of lab-scale and industrial bio-oils. In fact, furfural was not detected in Eucalyptus bio-oils extraction waters, possibly because it was emitted with gaseous products formed during the liquefaction process. These findings align with studies that indicate that gaseous phase reactions play a critical role in the volatilization of furan derivatives during liquefaction [23].

On the other hand, the higher concentrations of acetic, formic, and levulinic acids in the extraction waters of the industrial eucalyptus bio-oil should be the result of the higher biomass-to-solvent mass ratio used in the industrial tests. Table 10 also shows that cork bio-oil extraction waters did not contain formic or levulinic acid and, in the absence of HMF, this may suggest a lower degree of glucose degradation in cork biomass compared to other feedstocks.

4. Conclusions

This study demonstrates that biomass liquefaction under mild solvolysis conditions is strongly influenced by feedstock composition, particularly the contents of hemicellulose and cellulose. Among the 13 biomasses evaluated, Miscanthus achieved the highest bio-oil yield (72%) due to its high hemicellulose content, while Ulva lactuca, with low glucan and xylan levels, showed the lowest conversion (23%). These results confirm the key role of hemicellulose in depolymerisation efficiency and in the formation of value-added compounds such as furfural.

Cellulose also contributed to bio-oil production, although its effect was less pronounced, likely due to its crystalline structure. Lignin-rich biomasses generally showed lower conversion, but exceptions like peach stone shells (67% yield with 42% lignin and 30% hemicellulose) suggest that a holistic view of biomass composition is essential.

Reaction time significantly influenced product yields and distribution. Longer residence times increased overall conversion but led to reduced concentrations of soluble products such as glucose and acetic acid in the aqueous phase, suggesting secondary degradation or repolymerisation reactions.

While process parameters such as solvent, stirring, and reactor configuration were kept constant, their influence on scale-up and efficiency should be addressed in future work. The valorisation of lignocellulosic residues into bio-oils supports the development of sustainable, carbon-neutral processes aligned with the circular bioeconomy.

The insights from this study provide practical guidance for feedstock selection and optimization of liquefaction conditions. Further research should focus on elucidating product molecular profiles and improving recovery strategies to advance the industrial the applicability of biomass liquefaction.