Investigation of Component Interactions During the Hydrothermal Process Using a Mixed-Model Cellulose/Hemicellulose/Lignin/Protein and Real Cotton Stalk

Abstract

Converting agricultural and forestry waste into high-value-added bio-oil via hydrothermal liquefaction (HTL) reduces incineration pollution and alleviates fuel oil shortages. Current research focuses on adjusting HTL parameters like temperature, time, catalyst, and pretreatment. Few studies explore raw material composition and its interactions with bio-oil properties, limiting guidance for future multi-material hydrothermal co-liquefaction. In view of the above problems, the lignocellulosic model in this paper used cellulose, hemicellulose, lignin, and protein as raw materials. At a low hydrothermal temperature (220 °C), the yield and properties of hydrothermal bio-oil were used as indicators to explore the influence of the proportional content of different model components on the interaction in the hydrothermal process through its simple binary blending and multivariate blending. Then, compared with the hydrothermal liquefaction process of cotton stalk, the interaction between components in the hydrothermal process of real lignocellulose was explored. The results demonstrated significant interactions among cellulose, lignin, and hemicellulose in cotton stalks. The relative strength of component interactions was ranked by yield (wt.%) and property modulation as follows: cellulose–lignin (C-L, 6.82%, synergistic enhancement) > cellulose–hemicellulose (C-X, 1.83%, inhibitory effect) > hemicellulose–lignin (X-L, 1.32%, non-significant interaction). Glycine supplementation enhanced bio-oil yields, with the most pronounced effect observed in cellulose–glycine (C-G) systems, where hydrothermal bio-oil yield increased from 2.29% to 4.59%. Aqueous-phase bio-oil exhibited superior high heating values (HHVs), particularly in hemicellulose–glycine (X-G) blends, which achieved the maximum HHV of 29.364 MJ/kg among all groups. Meanwhile, the characterization results of hydrothermal bio-oil under different mixing conditions showed that the proportion of model components largely determined the composition and properties of hydrothermal bio-oil, which can be used as a regulation method for the synthesis of directional chemicals. Cellulose–lignin (C-L) interactions demonstrated the strongest synergistic enhancement, reaching maximum efficacy at a 3:1 mass ratio. This study will deepen the understanding of the composition of lignocellulose raw materials in the hydrothermal process, promote the establishment of a hydrothermal product model of lignocellulose, and improve the yield of bio-oil.

1. Introduction

Biomass from agricultural and forestry wastes constitutes a renewable resource widely regarded as a future primary energy source. At present, biomass energy accounts for about 10% of the global energy supply, and there is a huge room for appreciation [1,2]. On the other hand, China is the largest cotton producer in the world, accounting for about 30% of the annual global cotton output. The traditional cotton stalk treatment is still dominated by composting and incineration now, and a large number of solid particles and smoke are released from the cotton stalk during the incineration process, causing environmental pollution. Therefore, the high-value transformation of cotton stalk has a positive significance for the optimization of China’s energy structure. Through the mild hydrothermal method, agricultural and forestry waste is converted into bio-oil and used for biofuels, hydrogen production raw materials, main sources of value-added chemicals, biological fertilizers, carbonaceous materials (coke, activated carbon, and graphite), and adhesives of boilers or automobile engines [2], which has broad application space.

The traditional research on hydrothermal bio-oil of lignocellulose mainly focuses on the reaction temperature, time, and catalyst. Biomass is primarily composed of three polymers—cellulose, hemicellulose, and lignin—along with minor constituents such as pectin and protein [3]. The main components of lignocellulose, cellulose, hemicellulose, and lignin components form a relatively solid complex three-dimensional network structure through chemical bonding (ester bond, ether bond, glycosidic bond, and other covalent bonds), which contributes to its physical and chemical persistence and results in certain interactions in the process of hydrothermal transformation. These factors will inevitably affect the composition and yield of bio-oil and also have a certain impact on the correlation model of related properties of hydrothermal bio-oil. Moreover, the component contents of different lignocelluloses are different, which will have a certain influence on the properties of co-feeding hydrothermal bio-oil. Therefore, it is necessary to study the interaction of lignocellulose components in the hydrothermal process.

Recently, most of the studies on the interaction in the thermal conversion process of biomass composition focus on the interaction between cellulose, hemicellulose, and lignin under pyrolysis conditions [4,5,6]. However, relatively systematic studies on the hydrothermal liquefaction (HTL) of lignocellulose biomass and the mechanism analysis under real lignocellulose biomass composition are relatively few. Meng et al. [7] systematically investigated the properties and formation mechanisms of MHTC-derived products from crop straws under varying media and temperatures. Their work demonstrated that magnesium acetate facilitates the formation of ordered carbon structures with nitrogen- and oxygen-containing surface functional groups while reducing hydrocarbon crystallinity. Huang et al. [8] employed catalytic hydrothermal liquefaction to convert traditional Chinese medicine residues into liquid bio-oil. They investigated the effects of temperature, water/biomass mass ratio (W/B), and duration on reaction performance, evaluating diverse catalyst combinations (iron, nickel, and ZSM-5 molecular sieves) for bio-oil quality enhancement. Wu’s team [9] analyzed thermodynamic parameters during reed straw waste (RSW) hydrothermal conversion, revealing Gibbs free energy (153.45–158.39 kJ mol⁻¹) and enthalpy changes (248.10–292.04 kJ mol⁻¹), which underscored RSW’s potential as a feedstock for bioenergy and value-added chemicals. Also, the proportion of bio-oil in the total three-phase products, as well as the categories and proportions of the main components of bio-oil. Mahadevan et al. [10] conducted hydrothermal liquefaction (HTL) of biomass mixtures under subcritical, near-critical, and supercritical conditions and predicted oil yields. Under the supercritical conditions, the oil yield of HTL from lignin, cellulose, and starch was lower than that from HTL at lower temperatures. On the other hand, HTL of protein can also provide high oil yield even at supercritical temperatures.

It can be seen from the above studies that some multivariate blending system models in the hydrothermal liquefaction (HTL) process mostly take lipids, proteins, and carbohydrates as the research objects, and the interaction between proteins and polysaccharides is particularly important, so it has good prediction and explanation for the hydrothermal liquefaction process of algae, soybeans, or other biomass materials rich in proteins and carbohydrates. However, the lignocellulose-based agricultural and forestry wastes have poor universality due to a lack of research on the interaction among cellulose, hemicellulose, and lignin. Therefore, in view of the above problems, the cotton stalk was used as the real lignocellulosic biomass material to prepare bio-oil under low-temperature hydrothermal liquefaction conditions (220 °C, 4 h); meanwhile, the single component, binary, ternary, and multivariate blending of cellulose, hemicellulose, lignin, and protein were carried out under the same hydrothermal conditions. Then, compared with the actual bio-oil of cotton stalk, it was characterized and analyzed by infrared spectroscopy, EA elemental analysis, thermogravimetric analysis, and GC-MS. This study will further deepen the understanding of the composition of lignocellulose raw materials in the hydrothermal process, which has a certain guiding role in the industrialization process of hydrothermal bio-oil conversion of agricultural and forestry waste under multi-material co-feeding conditions.

2. Experimental Materials and Methods

2.1. Experimental Materials

Cotton stalks came from Karamay, Xinjiang, China. The collected cotton stalks were smashed into powder and dried at 105 °C for 12 h in an air atmosphere in an oven for further use. Reagents used in the experiment were as follows: cellulose ((C₆H₁₀O₅)_n) from Beijing Innochem Science & Technology Co., Ltd. (Beijing, China); sodium lignosulfonate (C₂₀H₂₄Na₂O₁₀S₂) from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan); xylan ((C₅H₈O₄)_n) from Adamas Reagent Co., Ltd. (Shanghai, China); glycine (C₂H₅NO₂, ≥98.5%) from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Ultrapure water (DI water, 18.25 MΩ·cm) used in this study was prepared by WPUP-UV-20 ultrapure water machine from Sichuan Water Technology Development Co., Ltd. (Chengdu, China), and the polytetrafluoroethylene liner hydrothermal reactor (CLF-50) was prepared by Shanghai Yushen Instrument Co., Ltd. (Shanghai, China).

2.2. Experimental Methods

In this experiment, hydrothermal liquefaction experiments were performed on cellulose, hemicellulose, lignin, and protein, respectively; and the interaction mechanism between different components in cotton stalk was explored by binary, ternary, and multivariate blending under the same hydrothermal conditions.

2.2.1. Intercomponent Blending Process

Monocomponent: Hydrothermal liquefaction experiments were carried out on the cellulose model component (microcrystalline cellulose/C), hemicellulose model component (xylan/X), lignin model component (lignosulfonate/L), and protein model component (glycine/G), respectively; the total mass of raw materials in each group was 4 g.

Binary blends: Hydrothermal liquefaction of the cellulose + hemicellulose model (C-X), cellulose + lignin model (C-L), and hemicellulose + lignin model (X-L) was carried out, respectively, and each group was mixed evenly according to the proportion of 3:1, 2:1, 1:1, 1:2, and 1:3. The total mass of each group was 4 g.

According to the actual proportion of individual cotton stalk components, 13:1, 7:1, and 7:1, the mixture was evenly mixed, and the total mass of each group was 4 g. Hydrothermal liquefaction was carried out on the cellulose + protein model (C-G), hemicellulose + protein model (X-G), and lignin + protein model (L-G), respectively.

Multicomponent blends: According to the actual proportion of individual cotton stalk components, cellulose:hemicellulose:lignin:protein is 13:7:7:1; the hydrothermal liquefaction experiments of the cellulose + hemicellulose + lignin ternary model (C-X-L) and cellulose + hemicellulose + lignin + protein quaternary model (C-X-L-G) were carried out, and the total mass of each group was 4 g.

2.2.2. Hydrothermal Liquefaction

Based on our team’s previous studies [11,12,13], optimal selection of temperature, reaction duration, and solid-to-liquid ratio enables enhanced yield attainment. Specifically, elevated solid-to-liquid ratios exhibit multiple adverse effects: impaired heat/mass transfer efficiency triggers side reactions and compromises product selectivity, whereas reduced ratios lead to a dilution of reactant concentration, suppressing reaction rates and ultimately decreasing target product yield while increasing solid char byproducts. The standardized experimental procedure under optimized conditions proceeded as follows.

The experimental process is shown in Figure 1. Hydrothermal conversion experiments were conducted using 4 g of raw material (cotton stalk or model components) mixed with 40 mL ultrapure water at a solid–liquid ratio of 1:10. The mixture was transferred into a sealed hydrothermal reactor and heated to 220 °C under continuous stirring at 400 rpm, maintaining this temperature for 4 h. After natural cooling to ambient temperature, gaseous products were vented under atmospheric pressure without collection. The reactor contents were subsequently subjected to three successive extractions with 100 mL dichloromethane (DCM), each extraction lasting 4 h with thorough wall rinsing. The combined extracts underwent vacuum distillation for 4 h until complete DCM evaporation. The resulting slurry was filtered through a Büchner funnel under reduced pressure. The filter cake (solid residue) was dried to constant weight at 105 °C for gravimetric analysis and storage. The filtrate was allowed to phase-separate in a separatory funnel, from which the DCM phase containing bio-oil was collected and subjected to vacuum distillation (40–45 °C, 4 h) for solvent recovery. Bio-oil yield was determined gravimetrically by mass difference calculation.

Yield of different bio-oil:

$$Y_{bio–oil}\left(wt.\%\right)={\displaystyle \frac{M_{bi}}{M_f}}\times 100\%$$

(1)

Yield of different biochars:

$$Y_{solid}\left(wt.\%\right)={\displaystyle \frac{M_{si}}{M_f}}\times 100\%$$

(2)

Gas and water yield of different hydrothermal liquefaction products:

$$Y_{others}\left(wt.\%\right)=100\mathrm{\%}-Y_{bio–oil}-Y_{solid}$$

(3)

2.2.3. Characterization of Hydrothermal Bio-Oil

The surface functional groups of hydrothermal bio-oil samples were characterized by Fourier transform infrared spectroscopy (TENSOR-27, BRUKER, Billerica, MA, USA). The bio-oil elements were measured and analyzed by a Vario EL cube element analyzer from Elementar Company (Langenselbold, Germany), and the calorific value was calculated according to the mass fraction of the measured bio-oil elements; the calculation formula is shown in (4). The composition of bio-oil was determined by gas chromatography-mass spectrometry (GC-MS, Agilent 7890B/5977, Santa Clara, CA, USA). Thermogravimetric analysis of hydrothermal bio-oil was carried out using an STA409 comprehensive thermal analyzer from Netzch, Selb, Germany. The test atmosphere was nitrogen, and the heating rate was 10 °C/min.

The heat value calculation formula [14] is shown in Equation (4), here C, H, S, and O are the element mass percentages of bio-oil samples in EA analysis.

$$\mathrm{H}\mathrm{H}\mathrm{V}\left(\mathrm{M}\mathrm{J}/\mathrm{K}\mathrm{g}\right)=0.338\mathrm{\ast }\mathrm{C}+1.428\mathrm{\ast }\left(\mathrm{H}-\mathrm{O}/8\right)+0.095\mathrm{\ast }\mathrm{S}$$

(4)

3. Results and Discussion

3.1. Distribution of Bio-Oil Products

3.1.1. Interaction Between Single Components in Cotton Stalk

The yield and product distribution of hydrothermal bio-oil from single-component and actual cotton stalks are shown in Figure 2. It can be seen that the interaction between different components has different effects on the yield of hydrothermal bio-oil. The yield of monocomponent cellulose hydrothermal bio-oil (wt.%) is 2.29% for cellulose (C), 1.77% for hemicellulose (X), and 0.59% for lignin (L). It can be seen that the yield of cellulose hydrothermal bio-oil from single-component cotton stalks is low, but the yield of cellulose hydrothermal bio-oil from a three-component blend (C-X-L) can be increased to 5.88%, indicating that the coexistence of multiple components in cotton stalk has a great influence on the oil phase [15]. The three-component blend (C-X-L) has little difference compared with the real cotton stalk (CS) hydrothermal bio-oil yield, indicating that cellulose (C), hemicellulose (X), and lignin (L) are the main sources of cotton stalk hydrothermal bio-oil.

Figure 3 shows the yields of binary and ternary blend hydrothermal bio-oils. The mixing ratios for cellulose + hemicellulose (C-X), cellulose + lignin (C-L), and hemicellulose + lignin (X-L) were 2:1, 1:1, and 1:1, respectively, matching the actual proportion of each single component in cotton stalk. Compared with monocomponent cellulose, the hydrothermal bio-oil yield of cellulose + lignin binary blend is increased from 2.29% to 6.82%, and its hydrothermal bio-oil yield is significantly increased, or even higher than that of ternary blend (C-X-L) by 5.88%, indicating that there is a significant mutual promotion between the two. In the hydrothermal process, lignin inhibited the trend of levoglucan polymerization into coke or converted into volatile substances, and promoted the formation of light compounds in cellulose, thereby improving the yield of hydrothermal bio-oil [5]. Compared with lignin, the binary blending effect (C-X) of cellulose and hemicellulose is weak and even shows an inhibitory effect on cellulose. This is because cellulose is covered by the hemicellulose layer before pyrolysis, so the main pyrolysis product of cellulose, levoglucan, is inhibited by the relatively volatile hemicellulose derivatives [16]. The yields of hemicellulose–lignin (X-L) and cellulose (X-L) binary blends hydrothermal bio-oil are 1.83% and 1.32%, respectively, which are not significantly affected by cellulose and lignin. The yield results of lignin binary blend hydrothermal bio-oil showed that it interacted with cellulose (L-C) and hemicellulose (L-X), and both promoted the pyrolysis of lignin. This is because the presence of cellulose and hemicellulose inhibits the polymerization of lignin derivatives and promotes the formation of light compounds (such as guaiacol) in lignin [4].

The results of hydrothermal bio-oil yield by blending the three components with protein (glycine) (Figure 4) further showed that there is a mutual promotion between cellulose (C-G), hemicellulose (X-G) and protein; compared with the single component, the hydrothermal bio-oil yield of the two components is improved, and the interaction with cellulose is the most obvious, and the hydrothermal bio-oil yield is increased from 2.29% to 4.59%. The interaction between cellulose, hemicellulose, and protein is mainly due to the Maillard reaction, which is caused by the reaction between amino groups of amino acids and carbonyl groups in carbohydrates, and this reaction will lead to high bio-oil production [17]. Cellulose is a linear polysaccharide composed of glucose, while hemicellulose is a branched hybrid of various monosaccharides including xylose and pectinose. The different structural composition between the two leads to the difference in reactivity with proteins. Compared with hemicellulose, the single structure sugar monomer of cellulose can react with protein by Maillard reaction more harmoniously, and its macroscopic performance is the higher bio-oil yield.

3.1.2. Effect of Single Component Content on Binary Interaction

The above results have a certain understanding of the component interaction between cotton stalks, but it is only the interaction between components under a single proportion, and there is a lack of a relatively complete interaction study within the single component content range. In the final analysis, the contents of cellulose, hemicellulose, and lignin in different lignocellulosic crops are different. Therefore, it is of great significance to further improve the yield of low-temperature hydrothermal bio-oil and to the industrialization of hydrothermal conversion of biomass raw materials by exploring the effect of different proportions of single component content on binary interaction.

Hydrothermal liquefaction of cellulose + hemicellulose (C-X), cellulose + lignin (C-L), and hemicellulose + lignin (X-L) was carried out according to the ratio of 3:1, 2:1, 1:1, 1:2, and 1:3, respectively. The results of hydrothermal bio-oil are shown in Figure 5. Cellulose and lignin (C-L) promote each other most obviously. With the increase of cellulose content, the interaction effect is enhanced, but the growth rate is gradually slowed down. At a 3:1 ratio, the degree of interaction reaches the maximum, so there may be a maximum interaction point near this ratio. It can be seen that a small amount of lignin can fully promote the formation of light compounds in cellulose. The strength of the inhibition between cellulose and hemicellulose (C-X) increases with the increase of cellulose and lignin content, and the mutual inhibitory effect of them is the weakest only at the ratio of 1:1. There was no obvious interaction between hemicellulose and lignin (X-L) in most content ranges, but there was a weak mutual promotion at 1:1. Based on the above results, it can be seen that the content of single component not only significantly affects the interaction between components but also determines the strength of the interaction.

This study is compared with recent works from the past two years on hydrothermal liquefaction (HTL) for bio-oil production. As previously mentioned, Guan et al. [7] achieved a high bio-oil yield of 24.7 wt.% through catalytic conversion at 320 °C using metal and ZSM-5 catalysts, specifically targeting herbal residues with high protein and low lignin content. In contrast, the present investigation focuses on natural cotton straw, a typical lignocellulosic feedstock, to elucidate the interaction mechanisms among components under non-catalytic mild conditions at 220 °C, with the primary objective of revealing the synergistic effects between the three major components (cellulose, hemicellulose, lignin) and protein at lower temperatures rather than pursuing maximum yield. While Guan et al.’s study provides catalytic process solutions for high-value utilization of specialty biomass (herbal residues), this research explores the interaction parameters of multiple components in abundant low-cost cotton stalk feedstock under mild (220 °C) non-catalytic conditions. Although the combination of high temperature (>300 °C) with iron, nickel, and ZSM-5 catalysts can significantly promote lignin depolymerization and deoxygenation reactions (achieving bio-oil HHV of 29.39 MJ/kg in Guan’s study), it faces challenges of high energy consumption and catalyst costs. Notably, under low-temperature non-catalytic conditions, through component recombination strategies like cellulose + lignin systems, this study achieved bio-oil HHV ranging 26.581–29.364 MJ/kg, approaching the level of catalytic high-temperature processes, demonstrating that low-temperature synergistic strategies can partially substitute harsh conditions to enhance oil quality.

3.2. Characterization of Biomass Oil

3.2.1. FTIR

The results of infrared spectroscopy (FTIR) analysis of different components of hydrothermal bio-oil are shown in Figure 6. In the range of 3200~3600 cm⁻¹, most of the spectra show obvious broad and scattered absorption peaks, which is attributed to the stretching vibration of O-H, indicating the presence of amines, amides, alcohols, and phenols. Most of the spectra show sharp and large stretching vibration peaks of carbonyl (C=O) with absorption frequencies in the range of 1850~1660 cm⁻¹, indicating that ketones, carboxylic acids, lipids, anhydrides, and other substances exist in bio-oil [18], while no obvious stretching vibration peak appears in lignin as a single component. The characteristic C=O stretching vibration at 1708 cm⁻¹ exhibits a broadened absorption band in lignin, contrasting with the sharp peaks observed in cellulose and hemicellulose. This spectral distinction primarily arises from lignin’s inherently low intrinsic carbonyl content and its predominant thermal decomposition pathway involving aromatic ring condensation, which generates minimal C=O-containing products. In contrast, cellulose and hemicellulose hydrolysis yield aldehyde derivatives and oligosaccharides (including disaccharides) through cleavage of their abundant carbohydrate structures. This mechanistic difference elucidates the inhibitory effect observed in binary component systems (C-X) during bio-oil production. Excessive carbonyl group formation creates steric hindrance effects that impede the polymerization of small molecular intermediates into higher molecular weight bio-oil constituents. In the absorption frequency range of 1660~1540 cm⁻¹, the amino acid (glycine) component shows an obvious N-H stretching vibration peak, indicating the existence of amides and nitro compounds. In the absorption frequency range of 1500~1450 cm⁻¹, the stretching vibration of the benzene ring carbon skeleton (C=C) appeared in lignin components, indicating that the pyrolysis and polymerization of lignin components produced aromatic substances. The deformation vibration peaks of methyl (-CH₂) and antisymmetric deformation vibration peaks of methyl (-CH₃) appeared in the absorption frequency range of 1450~1370 cm⁻¹, further indicating the presence of alkanes [19]. The absorption frequency is in the range of 1150~900 cm⁻¹, and the stretching vibration peak of the ether group appears in most components. Besides the cellulose monocomponent, the peaks of other components at 735 cm⁻¹ indicated the existence of heterocyclic derivatives in bio-oil [20]. The ternary blend (C-X-L) exhibited a weak absorption band at 735 cm⁻¹ in FTIR analysis, whereas both protein-supplemented samples and native cotton stalk (CS) demonstrated prominent sharp peaks at this wavenumber. This phenomenon correlated strongly with nitrogen incorporation: glycine supplementation introduced exogenous nitrogen, while native CS contained residual nitrogen primarily from proteinaceous components in plant cell walls. Nitrogen enrichment facilitated the formation of heterocyclic derivatives, whose characteristic vibrational modes were directly associated with the peak intensity at 735 cm⁻¹ [20].

The results of infrared spectrum analysis showed that cellulose components liquefied to produce alkanes, esters, organic acids, alcohols, and other substances; the liquefaction products of hemicellulose components were basically consistent with cellulose, and the lignin produced multi-heterocycles, aromatic hydrocarbons, and other benzene-containing structural substances. However, with the binary mixture of different components, their compounds gradually varied, such as the amides and nitro compounds produced in bio-oil under the action of group amino acids and single components. When the single component was further mixed, the hydrothermal bio-oil contained a large number of different types of compounds, which showed that the cellulose, hemicellulose, lignin, protein components, and the interaction between the components made the bio-oil components complex and diverse.

3.2.2. Elemental Analysis

In this section, the composition and thermal properties of bio-oil obtained by hydrothermal liquefaction of different components were determined to understand its composition and basic properties, and the quality of bio-oil was indirectly reflected by the composition content and calorific value of elements. The results have important reference value for subsequent bio-oil extraction, processing, and transportation.

It can be seen from

Table 1. Hydrothermal bio-oil element analysis results.

Samples	C (wt.%)	N (wt.%)	H (wt.%)	S (wt.%)	O (wt.%)	H/C	O/C	HHV (MJ/kg)
cellulose	62.51	0.06	6.374	0.073	30.983	1.215	0.372	24.707
hemicellulose	64.19	0.03	5.511	0.208	30.061	1.023	0.352	24.220
lignin	64.60	0.37	6.325	5.227	23.478	1.167	0.273	27.173
cellulose + hemicellulose	66.04	0.33	6.341	0.276	27.013	1.144	0.307	26.581
cellulose + lignin	66.48	0.34	6.331	3.784	23.065	1.135	0.260	27.753
hemicellulose + lignin	64.04	0.14	6.123	2.682	27.015	1.139	0.317	25.822
cellulose + glycine	63.99	10.32	6.575	0.394	18.721	1.224	0.220	27.713
hemicellulose + glycine	66.49	10.62	6.760	0.425	15.705	1.212	0.177	29.364
lignin + glycine	64.45	7.93	6.634	1.705	19.281	1.227	0.225	27.978
mixed component	65.32	0.17	6.263	3.152	25.095	1.143	0.288	26.842
mixed component + glycine	63.72	0.53	6.021	3.076	26.653	1.126	0.314	25.381
cotton stalk	61.52	0.62	5.734	0.262	31.864	1.120	0.390	23.320

O(wt.%) = 100% − (C% + H% + S%).

that the contents of bio-oil and biochar obtained by each component are higher than the carbon content of cotton stalk (61.52%), and all the carbon content values are higher than that of cotton stalk. Oxygen content ranged from 15% to 30%, while hydrogen content was approximately 6.0–6.3%. The nitrogen content and sulfur content in liquefied bio-oil of cotton stalk are very low, 0.62% and 0.262%, respectively, while other components are less than the nitrogen content in bio-oil of cotton stalk. The sulfur content of the lignin component is the largest, which may be the main source of sulfur content in cotton stalks. The hydrothermal bio-oil oxygen content of cellulose and hemicellulose in single components is higher than that of lignin; because cellulose and hemicellulose are polysaccharides, their structure contains a large number of light groups, resulting in higher hydrogen and oxygen content but lower nitrogen and sulfur content than lignin. From Table 1, it can be seen that the bio-oil calorific values of each component are higher than those of cotton stalk (23.32 MJ/kg), and the bio-oil calorific values of lignin and glycine components are higher than 27 MJ/kg; the calorific values of each bicomponent are greatly improved compared with their single component, and the L-G of hemicellulose and glycine group is the largest, which is 29.36 MJ/kg. The above results showed that the hydrothermal bio-oil of cotton stalks has great potential to become high-quality biofuel. The lignin and glycine monocomponent in cotton stalks are the main sources of the hydrothermal bio-oil calorific value of cotton stalks. Meanwhile, the interaction between cotton stalk monocomponent will further improve the hydrothermal bio-oil calorific value.

While carbon content fundamentally governs the high heating value (HHV), the distinct HHV disparity between the cellulose–glycine composite (C-G, 63.99 wt.% C, 27.713 MJ/kg) and lignin (L, 64.60 wt.% C, 27.173 MJ/kg) arises from synergistic effects of oxygen distribution, heteroatom interplay, and molecular architecture. Regarding oxygen content and O/C ratio, C-G demonstrated superior combustion efficiency with lower oxygen content (18.72 wt.% vs. 23.48 wt.% in L) and reduced O/C ratio (22.0% vs. 27.3% in L), effectively minimizing energy loss through water formation during combustion. The heteroatom analysis revealed sulfur’s stronger negative correlation with HHV compared to nitrogen. Despite C-G’s higher nitrogen content (10.32 wt.% vs. 0.37 wt.% in L, attributed to glycine incorporation), its 13-fold lower sulfur content (0.394 wt.% vs. 5.227 wt.% in L) proved decisive. Molecular characterization showed that C-G’s elevated H/C ratio (1.224 vs. 1.167 in L) indicated greater aliphatic hydrocarbon content (ketones, esters), which exhibits higher energy density per unit mass than lignin-derived aromatic hydrocarbons. Furthermore, Maillard reaction products in C-G demonstrated enhanced combustion efficiency compared to phenolic oligomers from lignin. The marginal absolute HHV difference (~0.6 MJ/kg, <2.5% relative variation) reflects the balanced competition between these opposing physicochemical factors.

Compared with the binary blend of cellulose and hemicellulose, the oxygen content decreased by 10%, while the carbon and nitrogen content increased slightly. As shown in Figure 7, the cellulose + lignin binary blend (C-X) moves towards dehydration and decarboxylation compared with single component cellulose (C) and hemicellulose (X), indicating that the interaction between cellulose and the groups generated by hemicellulose pyrolysis will remove some oxygen-containing groups [16]. Compared with the single component, the oxygen content of cellulose + lignin binary blend decreased greatly to 23.065%, with a decrease of 23%. It is shown in the figure that H/C and O/C shift downward, indicating that the depolymerization reaction between components will be further strengthened during the pyrolysis of cellulose and lignin, so as to obtain oligomer components such as furfural, 5-hydroxymethylfurfural (5-HMF) and acetic acid, and make them have high carbon content and low oxygen content [21]. The hydrogen and oxygen content of hemicellulose + lignin binary blend (X-L) do not change significantly, and the interaction between hemicellulose and lignin is not very obvious. The interaction between cellulose, hemicellulose, and protein (glycine) in binary blends is obvious, and the oxygen content decreased by more than 35%, which shows a significant shift to decarboxylation in the diagram. Amino acids can react with carbonyl compounds by Maillard reaction, while cellulose and hemicellulose are rich in reducing monosaccharides. In the hydrothermal liquefaction process, it breaks carbonyl compounds and then reacts with amino compounds through the process of carbonyl ammonia condensation and molecular rearrangement, releasing CO₂, which makes the oxygen content in bio-oil decrease greatly [20].

In the multi-components, the nitrogen and hydrogen contents of the three-component C-X-L and the four-component C-X-L-G added with protein are basically similar to those of cotton stalk CS. It can be seen from the figure that they gradually shift to cotton stalk CS, but the oxygen content is still lower than that of the real cotton stalk. It may be a complex spatial structure of cotton stalk, which hinders the hydrothermal liquefaction reaction and makes it difficult to separate the oxygen-containing groups into the gas phase.

In hydrothermal conversion processes, decarboxylation and dehydration emerge as two dominant reaction pathways whose dynamic equilibrium governs bio-oil yield, compositional distribution, and fuel properties. The competition between these pathways predominantly depends on the substrate’s chemical structure and reaction conditions. Cellulose and hemicellulose, rich in hydroxyl groups and ether linkages, preferentially undergo dehydration. For instance, cellulose undergoes β-elimination reactions during hydrothermal treatment to form 5-hydroxymethylfurfural (HMF) with substantial H₂O release. Dehydration intermediates (e.g., HMF) may subsequently undergo decarboxylation to generate furanic compounds, where decarboxylation acts as a secondary step following dehydration. However, lignin’s rigid aromatic structures constrain dehydration pathways, resulting in predominantly decarboxylation-derived aromatic compounds as evidenced by GC-MS results. Notably, inter-component interactions can modulate reaction priorities. During the co-processing of cellulose and lignin, reactive intermediates (e.g., H⁺) from cellulose dehydration may catalyze side-chain decarboxylation in lignin, establishing synergistic effects that enhance yields. Substrate composition regulates reaction equilibrium cellulose content: When exceeding 50%, cellulose dehydration dominates, producing abundant furans/ketones. Protein incorporation: Nitrogenous components like glycine engage in Maillard reactions with dehydration products (e.g., reducing sugars), forming nitrogen-containing heterocycles as evidenced by N-H vibration peaks at 1660–1540 cm⁻¹. This process consumes dehydration intermediates, forcing system equilibration through enhanced decarboxylation to maintain carbon balance. Consequently, the C-G system exhibits elevated nitrogen content (10.32 wt.%) and reduced oxygen content (18.72 wt.%) in bio-oil, accounting for its marginal HHV advantage (27.713 vs. 27.173 MJ/kg).

3.2.3. GC-MS Analysis of Bio-Oil

In this paper, the hydrothermal bio-oil was characterized by gas chromatography-mass spectrometry (Agilent 7890A GC-5975C MS), and the organic components of the bio-oil were obtained. Quantification of bio-oil components was performed based on the relative peak area percentage of each compound within the total chromatographic profile. Given the structural complexity and diversity, compounds were systematically classified into 12 categories: aliphatic hydrocarbons, aromatic hydrocarbons, polyheterocycles, phenols, amines, carboxylic acids, esters, ketones, aldehydes, ethers, alcohols, and others (Figure 8). In cotton stalk (CS)-derived hydrothermal bio-oil, aromatic compounds, and polyheterocycles constituted approximately 25% of components, while non-aryl species (aliphatics, acids, aldehydes, ketones, esters, alcohols, amines) accounted for ~35%, aligning with Zheng et al.’s reported compositional profile [22]. Acidic, aldehydic, ketonic, ethereal, and aliphatic compounds primarily originated from cellulose (C) and hemicellulose (X) through depolymerization, dehydration, and cyclization of polysaccharides [23,24], whereas phenolic, polyheterocyclic, and aromatic components were derived from lignin (L) via condensation and cyclization pathways [25].

As shown in Figure 9, Co-hydrothermal treatment of cellulose and lignin (C + L) elevated non-aryl compound content (ketones, aldehydes, esters, aliphatics) compared to individual components. Aryl compound abundance decreased relative to lignin alone, though phenolic content increased, demonstrating synergistic interactions between C and L. This non-aryl enrichment stems from lignin-derived intermediates participating in chain-end decomposition of levoglucosan (a cellulose pyrolysis product), redirecting reaction pathways toward low-molecular-weight species (esters, aldehydes, cyclic ketones) rather than levoglucosan formation [26]. Concurrently, cellulose promoted lignin-derived guaiacol derivatives [5], while hydroxyl-rich monosaccharide decomposition products facilitated lignin demethoxylation, suppressing highly polymerized aromatics [27]. Cellulose–hemicellulose (C + X) co-treatment maintained compound diversity but reduced ketone, aldehyde, ether, and alcohol contents versus individual components. This reduction arises from hemicellulose-mediated inhibition of glycolaldehyde and levoglucosan formation [28]. Early-stage hemicellulose derivatives (generated at lower pyrolysis temperatures) physically coated cellulose surfaces, suppressing volatile ether/alcohol release. The compositional profile of lignin–hemicellulose (L + X) systems exhibited minimal divergence from individual components, corroborating Yu et al.’s findings [21] and confirming negligible chemical interaction.

Glycine-modified systems displayed distinct nitrogen incorporation patterns. Cellulose/glycine (C + G) and hemicellulose/glycine (X + G) bio-oils exhibited predominant amine and nitrogen-containing compounds, with concentrated ketones, aldehydes, and acids. Enhanced effects in C + G systems confirmed carbohydrate-glycine interactions via carbonyl-amine condensation and molecular rearrangements between glucose units and glycine, yielding structurally diversified products (ketones, aldehydes, nitrogenous compounds, diketones) [29]. Lignin–glycine (L + G) compositions remained lignin-dominated, indicating non-reactivity. Multicomponent systems (C + X + L and C + X + L + G) displayed hybrid compositional profiles combining aryl compounds (aromatics, phenols) and non-aryl species (aldehydes, ketones, esters, acids, aliphatics). The quaternary system (C + X + L + G) demonstrated elevated amines, aldehydes, ketones, and acids, closely approximating authentic CS bio-oil composition. This confirms that ternary component (C + X + L) interactions are the primary driver of compositional diversity in CS bio-oil, with glycine supplementation amplifying non-aromatic compound abundance to better replicate real biomass characteristics.

4. Conclusions and Prospects

In this paper, hydrothermal bio-oil was prepared from cotton stalk under mild conditions (220 °C, 4 h, using 4 g of cotton stalk and 40 mL of water). Various parameters of the bio-oil, including yield, functional group composition, elemental composition, calorific value, and compound composition, were investigated under different single-component, binary blending, and multivariate blending conditions. The results revealed interactions among cellulose, lignin, and hemicellulose in cotton stalks, with the most pronounced interaction observed between cellulose and lignin, promoting oligomer formation and enhancing bio-oil yield. Conversely, cellulose and hemicellulose inhibited each other, reducing bio-oil yield, while the interaction between lignin and hemicellulose was less evident. The content of individual components significantly influenced the intensity of inter-component interactions: Higher cellulose content (>50%) enhanced the interaction between cellulose and lignin, resulting in a yield increase exceeding 150%, with the maximum interaction intensity observed at a cellulose-to-lignin ratio of 3:1, while the interaction between cellulose and hemicellulose remained dominated by mutual inhibition across all content ranges. Hemicellulose and lignin exhibited a weak promotion effect only when their content was in a 1:1 ratio, and at this ratio, the bio-oil yield increased by 0.8 wt.%. The lesser protein components in cotton stalk also interacted with these three elements, with cellulose and hemicellulose reacting with protein via the Maillard reaction to improve bio-oil yield. The interaction between cellulose and protein demonstrated greater significance, with yields increasing from 2.29% to 4.59%, whereas the lignin–protein interaction showed limited effectiveness. Notably, although the lignin–protein system exhibited minimal interaction, the lignin–glycine combination achieved an HHV of 29.36 MJ/kg. Characterization results from infrared spectroscopy and elemental analysis of single-component, binary blend, and multivariate blend bio-oil indicated that the interaction among the three elements in cotton stalk was the primary source of bio-oil yield and contributed to the complexity and diversity of its compounds. Specifically, the interaction between cellulose and lignin promoted the formation of oligomers such as ketones, aldehydes, esters, and aliphatics while inhibiting the production of aromatic groups and multiple hybrid compounds. The interaction between cellulose and hemicellulose reduced the content of ketones, aldehydes, ethers, and alcohols, but there was no significant difference between hemicellulose, lignin, and their individual components. The interaction between these three elements and protein components further diversified the types of hydrothermal bio-oil compounds, with cellulose and hemicellulose increasing the contents of aldehydes, ketones, and acids, and significantly enhancing the formation of nitro-containing compounds.

Future research should focus on the following directions to enhance the economic viability and sustainability of biomass conversion technologies. First, priority should be given to improving oil quality and impurity removal. The practical application of bio-oil is constrained by its chemical stability and contaminant content. Therefore, systematic oil aging tests are required to clarify the effects of storage temperature, oxidation conditions, and impurity migration on oil quality (e.g., viscosity increases, phase separation). To address the negative impacts of N and S heteroatoms in bio-oil, efficient catalytic denitrogenation/desulfurization processes should be developed. Second, an in-depth investigation of scale-up challenges and solutions is necessary. Current research primarily operates at the laboratory scale, while scale-up processes may face mass and heat transfer limitations, such as uneven temperature distribution leading to localized coking and intensified side reactions like coke blockage, which could significantly reduce target product yields and system energy efficiency. Furthermore, research on comprehensive high-value utilization of all hydrothermal conversion components needs strengthening. Existing technologies predominantly focus on bio-oil products while neglecting the resource potential of aqueous-phase organics (e.g., sugars, organic acids) and solid residues (e.g., coke, unreacted lignin). Simultaneously, life cycle assessment (LCA) should be employed to quantify the carbon footprint and economic benefits of full-component utilization. In conclusion, by integrating oil quality improvement, process scale-up, and full-component valorization strategies, it is possible to overcome technical bottlenecks in biomass energy conversion, providing economically viable and environmentally friendly solutions for carbon neutrality goals.