Structural features of xylan dictate reactivity and functionalization potential for bio-based materials

Abstract

Background

See More

Plant-based materials have the potential to replace some petroleum-based products, offering compostability and biodegradability as critical advantages. Xylan-rich biomass sources are gaining recognition due to their abundance and underutilization in current industrial applications. Research of potential xylan applications has been complicated by the complex and heterogeneous structure that varies for different xylan feedstocks. Acylation is a broadly used reaction in functionalization of polysaccharides at an industrial scale. However, the efficiency of this reaction varies with the xylan source. To optimize xylan valorization, a systematic understanding of structure–reactivity relationships is essential.

Results

This study explores, characterizes, and compares various xylan feedstocks in the acylation process. Xylan feedstocks were analyzed for their chemical composition, degree of polymerization, branching, solubility, and presence of impurities. These features were correlated with xylan glycotypes’ reactivity toward functionalization with succinic anhydride in an optimized DMSO/KOH condition, achieving carboxyl contents of up to 1.46. We used principal component analysis and hierarchical clustering to identify key structural features of xylan that promote its reactivity. Our findings reveal that xylans with higher xylose content and lower degrees of branching exhibit enhanced reactivity, achieving higher carboxyl content and yields. Structural analyses confirmed successful modification, and light scattering analyses showed dramatic changes in the solution properties. Succinylation improves the solubility and film-forming properties of native xylans.

Conclusions

This study shows key structure–reactivity relationships in xylan succinylation, establishing that low branching, high xylose content, and reduced lignin impurity enhance chemical functionalization. The results offer a framework for selecting optimal biomass feedstocks and support future efforts in genetic and synthetic biology to design plants with tunable xylan architectures. These findings advance the hemicellulose valorization for applications in coatings and packaging.

Background

In recent years, essential efforts have been put into implementing plant-derived feedstocks to alleviate environmental concerns arising from the high consumption of fossil fuel-based plastics. Lignocellulosic biomass has gained traction due to its abundance, renewability, and the presence of reactive hydroxyl groups that can be targeted for various chemical processing methods. Currently, a wide range of products generated from biomass, such as paper, textiles, biofuels, and viscose fibers, are ubiquitous in today’s markets [1]. However, most valorization efforts have been dedicated to using enriched cellulose, while other naturally occurring biopolymers, such as hemicelluloses and lignin, remain relatively less explored and are frequently disregarded during processing [2]. In recent years, the valorization of hemicelluloses and lignin has become an increasingly attractive area of research in biorefining. Numerous studies have focused on their extraction, structural modification, and conversion into biofuels, platform chemicals, and functional materials [3, 4].

Despite constituting nearly 30% of plant biomass, hemicellulose is often considered a waste material in the typical Kraft pulping process [5]. The production of rayon via the viscose process is expanding rapidly, with a compound annual growth rate (CAGR) projected at 6.8% from 2026 to 2032 due to increasing demand for bio-based and biodegradable textiles [6]. Similarly, corn ethanol production is growing at a CAGR of 5.8% from 2024 to 2032 due to a rise in global fuel demand, abundant corn availability, and price competitiveness [7]. These processes consistently generate significant xylan-rich side streams, such as sulphite liquor in rayon production and distillers grain in corn ethanol production. As we move toward a circular economy model by producing an ever-increasing number of products from agricultural residues, combined with the forecasted expansion in these industries, the volume of associated waste streams also increases. Currently, these valuable polysaccharides and lignin are predominantly degraded into black liquor and used as fuel to recover some energy, thereby squandering a potentially rich carbon source. However, this underutilization underscores the need for further exploration and optimization of hemicellulose as a feedstock for the production of new materials, a gap that this study aims to address.

Various strategies have been explored in recent years to functionalize xylan-rich hemicelluloses for value-added applications. A recent review article by our group provided various pathways for enhanced xylan utilization by chemical and enzymatic modifications [8]. Most recently, chemically modified xylans have been developed for downstream uses as moisture barrier coatings [9], adhesives [10], hydrogels [11], and bioplastics [12]. Indeed, hemicelluloses are valuable feedstocks for the production of biomaterials, but one drawback is that they are structurally heterogeneous in the degree of polymerization, composition, and solubility, which has limited the development of efficient valorization technologies. Factors such as composition and structure related to different species, cultivation strategies, harvesting times, and plant tissue (e.g., corn cobs vs. bagasse) impact polysaccharide structural features, including the degree of acetylation, the amount and distribution of glycosyl substituents, and the accessible surface [8]. Taken together, we and others have hypothesized that structure and composition influence the parameters required to achieve chemical modification of these polysaccharides at desired yields [13]. Furthermore, the assumption that one universal technique is applicable to all hemicellulose feedstocks to reach identical outputs has proven both impractical and limiting. Thus, it is clear that better functionalization strategies are urgently needed, specifically approaches that are adaptable to structurally diverse saccharide feedstocks.

The succinylation of polysaccharides has been deemed an effective chemical functionalization method for producing bio-based materials. This method involves the introduction of carboxylic acid groups to constituent glycosyl residues. They can then be further modified while often maintaining the inherent biocompatibility and biodegradability of the starting material. It has been shown that the solubility and processability of some hexose-based plant polysaccharides, including starch and cellulose, are improved by modification with succinic anhydride [14]. In contrast, most prior research describing the succinylation of xylan reports low degrees of functionalization, which is primarily due to hydrolysis of succinic anhydride in the reaction mixture [15,16,17,18]. Sun et al. reported succinylation of water-soluble xylan from wheat straw by functionalization with succinic anhydride in DMF/LiCl with pyridine as a catalyst. However, they found that using a high amount of pyridine and succinic anhydride was a limiting factor for scalability [19]. The same conditions were later used to prepare succinylated sugarcane bagasse using a much lower content of N-bromosuccinimide, again resulting in a low degree of functionalization [20]. In another study, acid catalysts were evaluated for succinylation of xylans from beechwood. Here, inclusion of Lewis acids, such as FeCl₃, ZnCl₂, AlCl₃, SbCl₃, and SnCl₂, led to high degrees of succinylation in DMSO [21]. More recently, the use of ionic liquids as an effective solvent for xylan modification has been reported. In this approach, wheat straw hemicellulose was functionalized with succinic anhydride in imidazolium-based ionic liquids; however, the main drawback of this method is slow degradation of the xylan backbone in the ionic liquid [22, 23].

In this work, we evaluated several xylan glycoforms from both natural and industrial sources for their amenability to chemical functionalization with succinic anhydride. Initially, we optimized the process for one xylan glycotype. Then, we applied our workflow to xylans that were systematically analyzed to determine their structural characteristics and examined the effect of these factors on reaction yield and carboxyl content (CC). Taken together, we identified structure–functionalization relationships for xylan glycotypes, providing insights to guide future selection and bio-design strategies of plant-based feedstocks for advanced polysaccharide-based materials. This work contributes to the broader objective of valorizing abundant agricultural waste streams, taking a major stride toward optimizing the production of bio-based materials.

Methods

Materials

Xylan was obtained from 10 different sources (Table 1). Lenzing Inc. xylan was a kind gift of Lenzing AG (Austria), Beechwood, birchwood, corn cob and oat spelt xylans were purchased from Millipore Sigma (USA). Low, medium, and high degree of polymerization (DP) xylan noted as LDP, MDP, and HDP, respectively, was a kind gift of RYAM Inc. (USA). Chia (Salvia hispanica; Simply Nature Chia Seeds) seed mucilage was isolated according to methods used for Arabidopsis [24, 25]. Xylose and xylobiose, purchased from Sigma (USA), were used as a reference. Succinic anhydride (SA), potassium hydroxide (KOH), sodium borohydride (NaBH₄), ethanol, chloroform, acetone, isopropyl alcohol (IPA), and dimethyl sulfoxide (DMSO) were purchased from Thermo Fisher Chemical and used as received. Liquozyme (#AUP61163) and Spirizyme (#NAPFM084) were a kind gift from Novozyme.

Table 1 Full name of Xylan Feedstocks, the abbreviations used in this study, their product identifier, and their degree of polymerization (DP) measured by a viscometry method

Extraction of xylan from poplar wood

Woody biomass from poplar was suspended in ethanol with a homogenizer and filtered through 50 µm nylon mesh, followed by thorough washing with a 1:1 toluene/acetone mixture to remove wax and organic soluble impurities. The mixture was filtered again through nylon mesh, washed generously with acetone, and dried at room temperature to obtain alcohol-insoluble residue (AIR). Prepared AIR (~ 150 g) was resuspended in 1% (w/v) ammonium oxalate (3 L) supplemented with 11 mL of Liquozyme and 2.3 mL of Spirizyme for pectin and starch removal, respectively. The mixture was stirred for 2 days at 50 °C. The buffer was filtered through a Whatman grade GF/A glass microfiber filter overlayed with 50 µm nylon mesh. Oxalate/amylase-treated AIR was resuspended in a solution of 1 M KOH with 1% (wt) NaBH₄, and gently stirred overnight at room temperature. The obtained slurry was filtered, and the filtrate was neutralized with glacial acetic acid. The final mixture was concentrated with a rotary evaporator and dialyzed against DI water in 3.5 kDa dialysis tubing (SpectraPor) for 10 days at 4 °C with frequent changes of DI water, then freeze-dried.

Analysis of the feedstocks

Neutral sugar composition was analyzed using the alditol acetate method [26]. Briefly, AIR was hydrolyzed in 2 M trifluoroacetic acid (TFA) at 120 °C for 2 h, then dried under air. Next, to prepare alditols from monosaccharides, 1 M ammonium hydroxide containing NaBH₄ was added, then the residue was acetylated with acetic anhydride and pyridine as a catalyst. After acetylation, the mixture was washed with toluene, and the chloroform-soluble fraction of the product was removed from the mixture. Finally, the residue was dissolved in pure acetone for analysis with an Agilent 7890A gas chromatography system using an Rtx-2330 column (30 m, 0.25 mm I.D., 0.2 µm; Restek). Analyses were done in triplicate.

While our intent was to use ‘pure’ xylan samples, we did not rule out the possibility of lignin contaminants. The method described by Littunen et al. was used to estimate the lignin content of feedstocks [27]. Briefly, 10 mg of each feedstock was mixed with 0.1 M NaOH, and the Ultraviolet (UV) absorbance of the soluble fraction was measured at 280 nm. A calibration curve was made using a serial dilution of alkali lignin (purchased from Sigma-Aldrich) prepared in 0.1 M NaOH.

Glucuronic acid (GlcA) content was determined using high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC–PAD) (Dionex chromatography) according to published methods with minor modifications [28]. Briefly, hydrolysates (2 M TFA, 2 h, 120 °C) were injected into a Dionex CarboPac PA1 ion chromatography column (Thermo Fisher Scientific). A 100 mM NaOH solution was used as the mobile phase. A calibration curve was prepared by making a serial dilution of GlcA in water.

To enrich soluble xylans from each feedstock, a DMSO dispersion of each native hemicellulose sample with a known concentration was prepared and mixed for 3 h at 80 °C. The dispersions were cooled to room temperature and centrifuged at 10,976×g for 10 min. This procedure was repeated by adding DMSO to the sediments until the supernatant was a clear solution. To calculate the insoluble fraction, the mass of the obtained sediment was divided by the mass of the initial sample.

To measure the molecular weight of xylan feedstocks, reduced viscosities of xylan solutions in a 1 M cupriethylenediamine were plotted against concentration [29]. Five different concentrations in the range of 1–0.1 g/dL were analyzed for each xylan sample, and the degree of polymerization (DP) was estimated based on the Mark–Houwink equation [30] as follows:

$$\left[\eta \right]=2.2\times {10}^{-2}{DP}^{0.72}$$

Synthesis of succinylated hemicelluloses

All reactions were conducted on feedstock samples without any additional fractionation. In a typical protocol, ground xylan powder was dried overnight at 100 °C and then transferred to a round-bottom flask equipped with a magnetic stirrer. Anhydrous DMSO was added to the flask and mixed for 1 h at 80 °C. KOH was added to the mixture, stirred for 10 min, then succinic anhydride was added. Reactions were conducted for 3 h under vacuum distillation. After that, the mixture was cooled to room temperature and poured dropwise into a flask containing cold IPA while stirring to promote precipitation. The resulting mixture was centrifuged at 10,976×g for 12 min. The obtained sediments were mixed with dilute HCl, transferred to dialysis tubing (12.4 kDa or 2 kDa), and dialyzed against DI water for 7 days with frequent refreshments of DI water. Subsequently, the solutions were concentrated using a rotary evaporator and dried in a vacuum oven to obtain flakes of modified xylan.

Characterization of modified samples

Carboxyl content and yield values

To estimate the carboxyl content (CC) of the modified samples, a known amount of material was dissolved in DI water and titrated with 0.01 M NaOH. The CC was calculated using the following formula, where CC is the carboxyl content of the modified sample, n is the number of moles of NaOH used to neutralize the solution, and m is the mass of the analyte in grams:

$$CC=\frac{132 n}{m-100n}$$

The CC value for modified xylan samples is between 0 and 2, where 0 represents an unmodified sample and 2 represents a sample with both hydroxyl groups on a xylose backbone unit (O-2 and O-3) converted to succinic acid groups. The theoretical mass of fully functionalized xylan is estimated to be CC = 2.0. To estimate the yield of the reaction, the mass of the final product is divided by the calculated value for CC = 2. The reported yield values are represented as percentages.

Chemical structure characterization

Attenuated total reflectance–Fourier transform infrared (ATR–FTIR) analysis was conducted to study the chemical structure of the succinylated samples using a PerkinElmer spectrometer. Each sample was placed on a diamond ATR crystal, and spectra were recorded with a resolution of 4 cm⁻¹ in the 4000–650 cm⁻¹ range. The data were analyzed using PerkinElmer Spectrum IR software.

Solution-state nuclear magnetic resonance (NMR) spectroscopy was performed at 298 K to study the chemical composition of the unmodified and modified samples. Approximately 10 mg of each sample was dissolved in 0.5 mL of DMSO-d₆. ¹H NMR and ¹³C NMR spectra were recorded on a 600 MHz Bruker Avance Neo using standard operating conditions, and spectra were analyzed using Mnova (Spain).

Matrix-Assisted laser desorption/ionization time-of-flight (MALDI–TOF) mass spectrometry (MS) was carried out using a Bruker smartfleX spectrometer in positive ion mode. 2,5-Dihydroxybenzoic acid (DHB) matrix prepared in water (10 mg/mL) was used to crystallize xylo-oligosaccharides or succinylated reaction products in a 1:1 ratio directly on the plate. A minimum of 200 laser shots were summed to generate each spectrum, and the data were analyzed using FlexAnalysis (Bruker).

Characterization of thermal properties

The thermal stability of all samples was evaluated by thermogravimetric analysis (TGA) using a TGA 8000 (PerkinElmer). For each experiment, 5–10 mg of powder was heated from room temperature to 800 °C with a heating rate of 10 °C/min. The data were analyzed using Pyris series software.

Dynamic scanning calorimetry (DSC) was conducted further to characterize the thermal properties of modified and unmodified samples using an 8000 PerkinElmer DSC. A dried sample (5 mg) was put in aluminum disks and heated up to 50 °C below the onset of the degradation of the sample through a heating–cooling–heating cycle. The reported thermographs are from the second cycle of heating. The data were analyzed with Pyris series software.

Light scattering characterization of xylan particles

Changes in the hydrodynamic radius before and after modification were analyzed with a Malvern Zetasizer Nano-ZS. A 1 mg/mL solution was put in a quartz cell, and the R_h and PDI were measured at 90°. The data were analyzed with Zetasizer software.

The hydrodynamic radius (R_h) and polydispersity index (PDI) for each xylan glycotype and the succinylated xylan samples were measured with a Brookhaven BI-200SM Goniometer at a 90° angle. Static light scattering analysis was done to obtain the gyration radius (Rg) over the angular range of 35°–135°. Each sample was dissolved in DMSO in a 1–5 mg/mL concentration and tuned to ensure a minimum 50kcps scattering while preventing multiple scattering. Each sample was filtered through a 0.45 µm nylon syringe filter to remove any dust particles in the sample. The data were analyzed using Particle Explorer software.

UV–Vis spectroscopy

Ultraviolet–visible (UV–Vis) spectroscopic analysis was performed using a Shimadzu UV-2401PC spectrophotometer in transmittance mode to evaluate the optical properties of the native and modified xylan solutions. Measurements were carried out over wavelengths of 400–700 nm using quartz cuvettes with a 1 cm path length. Solutions were prepared at four different concentrations, ranging from 0.5 to 0.05 wt%, to assess concentration-dependent transmittance behavior. Deionized water was used as the reference blank for all measurements. The data were analyzed with UVProbe software.

Characterization of xylan films

Aqueous dispersions were used to prepare xylan-based films using a spin coater. Briefly, a 2.5 wt% DI water dispersion of each native and modified xylan sample was mixed at 70 °C for 6 h. Then, the 2.5 wt% solution in water was spin-coated onto an Si-wafer (1 × 1 cm² square plasma treated and cleaned with ethanol and water to remove any residues from the surface) at 4000 rpm with an acceleration of 300 rpm/s for 60 s to prepare thin films. The samples were dried at room temperature for 24 h before imaging.

The films were characterized by scanning probe microscopy (SPM); a Bruker Dimension Icon microscope was used to measure thickness and evaluate film morphology in the dry state. Samples (~ 10 × 10 mm²) were placed on the sample support table and held in place by vacuum. Measurements were conducted using Bruker PeakForce Tapping mode in air using Bruker RTESPA-300 probes (rectangular probe, nominal resonance frequency 300 kHz, spring constant 40 N/m, tip radius 8 nm, aluminum reflective coating). The films were scratched with a razor blade to obtain the basal surface level (silicon substrate) to serve as a reference for height measurement. The scan resolution was fixed at 512 × 512 pixels, with the scan areas set at 30 μm². The collected scans were leveled and analyzed using Gwyddion software.

Statistical analyses

One-way analysis of variance (ANOVA) was carried out to optimize reaction parameters based on calculated CC values. Experiments were conducted in triplicate for each reaction parameter, including the reagent and catalyst amount, time, and temperature.

To investigate the key structural features that affect xylan reactivity, we used principal component analysis (PCA) to uncover underlying patterns. PCA is a multivariate statistical tool that reduces the dimensionality of the data set by transforming a set of variables to another set of independent and orthogonal variables called principal components (PC), which result from linear combinations of the original variables. The data set comprising chemical composition, molecular weight, particle size, and the CC values from xylan feedstocks was standardized before analysis. The purpose is to determine the components that explain the most variations in the data, with the smallest possible dimensions. The first two PCs were used to generate a score plot to visualize sample clustering and a loadings plot to identify variable contributions.

Results and discussion

Compositional characterization of xylan samples

The chemical composition of hemicellulose feedstocks is summarized in Table 2. The neutral sugar composition of initial materials was analyzed by gas chromatography of alditol acetate derivatives after hydrolysis of polysaccharides. HPAEC–PAD was used to quantify GlcA content. Lignin content was estimated using a spectrophotometric method. All hemicellulose feedstocks consisted mainly of xylans, with xylose being the dominant component. BIRC-X, LMW-X, RYH-X, BEW-X, and RYM-X are the purest hemicellulose samples in terms of xylose content; with a composition of more than 90% xylose. POP-X had a significant amount of rhamnose (Rha), which is likely due to the presence of the reducing end tetrasaccharide sequence in hardwood glucoronoxylans that has the following structure: Xylp-1,4-β-D-Xylp-1,3-α-L-Rhap-1,2-α-D-GalpA-1,4-D-Xyl [31]. The results for POP-X composition indicate a low selectivity of alkaline extraction toward xylans, as the sample contains a relatively high amount of other heteropolysaccharides, such as mannans (mannose, Man) and pectins (galactose, Gal). In addition, glucose (Glc) was present in CORN-X and OAT-X, which we contributed to starch contamination as well as hemicellulosic mixed-linkage (1–3; 1–4)-β-glucan, which can be abundant in cereals. As expected, OAT-X and CORN-X samples contain a high amount of arabinose (Ara), as hemicellulose in cereal grains and grasses are reported to be mainly composed of arabinoxylans and glucuronoarabinoxylans, respectively [32, 33]. Moreover, almost all samples contained a small amount of fucose (Fuc), which is not a constituent of xylan but can be found in other cell wall polysaccharides, including pectins, arabinogalactans, and xyloglucans [34]. Noteworthy, industrial hemicellulose streams obtained from RYAM Inc., including RYL-X, RYM-X, and RYH-X, consisted of highly pure xylans. In addition, HPAEC-PAD analysis indicated that some feedstocks contain GlcA substituents. Our data showed that CHIA-X has the highest content of GlcA side chains, followed by OAT-X, BIRC-X, and POP-X. We estimated the degree of branching by dividing the total amount of side-chain constituents, namely, GlcA and Ara, by that of the backbone xylose residues. This value is highest for CHIA-X, OAT-X, CORN-X, and POP-X, indicating the presence of highly substituted xylan chains. On the other hand, some feedstocks, including all of the industrial hemicellulose waste streams (i.e., LMW-X, RYL-X, RYM-X, and RYH-X), consist of homoxylans that have been stripped of all side chains. Thus, the results of our chemical composition analysis confirm that the selected feedstocks represent a broad spectrum of xylan glycotypes (Table 2).

Table 2 Chemical composition of xylan feedstocks

Furthermore, our data showed that most samples contained lignin contaminants. Indeed, lignin and hemicellulose are capable of forming covalent lignin–carbohydrate complexes (LCCs), which present challenges in complete separation during hemicellulose extraction [35]. Lignin content was highest in POP-X, CORN-X, and CHIA-X samples (Table 2).

The results obtained with capillary viscometry indicate significant variations in the degree of polymerization (DP) in the feedstocks, ranging from 3 for CORN-X to 3320 for CHIA-X (Table 1). LMW-X consists mostly of short xylan chains with a relatively low DP and low amount of residual lignin. The values reported herein for other hardwoods (BEW-X and BIRC-X) agree with molecular weight values reported in the literature [36, 37]. Although CORN-X was found to have a low DP (DP3), consistent with the value reported by the supplier (M_w = 300–900 Da), we could not confirm this value with mass spectrometry. This observation can be explained by the high amount of glucose in this commercial feedstock, likely complicating the crystallization with the matrix necessary for MS analysis.

Xylans in general can be broadly categorized into two major groups based on their degree of branching: homoxylans, which are linear β-(1–4)-d-xylopyranose backbones lacking decoration, and heteroxylans, which contain various glycosyl substituents, such as Ara and GlcA residues. In this study, we did not consider O-acetylation, as the majority of xylans were isolated using alkali, resulting in de-esterification. Heteroxylans include glucuronoxylans (GX), arabinoxylans (AX), and glucuronoarabinoxylans (GAX), each distinguished by their specific side-chain composition and patterning. Our chemical compositional analysis across the xylan feedstocks confirms that we incorporated a representative range of glycotypes to study their functionalization potential. Specifically, feedstocks such as LMW-X, RYL-X, RYM-X, and RYH-X are classified as homoxylans, due to their minimal branching. In contrast, feedstocks such as OAT-X (GAX), CORN-X (AX), and POP-X and CHIA-X (GX) represent heteroxylans, with varying degrees of branching. Furthermore, the wide range of measured degrees of polymerization, from as low as 3 for CORN-X to over 3000 for CHIA-X, further substantiates the structural diversity among these feedstocks.

Light scattering analysis

All solutions of native xylan prepared in DMSO show intense light scattering signals. This observation indicates the presence of large particles. DMSO has been reported as an effective solvent for xylan [38]. However, our results reveal significant variations in the particle sizes of xylan feedstocks when DMSO is used as a solvent (Table 3). While we found that DMSO can effectively solubilize xylan to some extent, it does not achieve complete molecular dissolution of the feedstocks. To mitigate this, all solutions were centrifuged to remove insoluble material, which we determined to be between 10 and 50%. For further analysis, a colloidal solution (soluble fraction) of LMW-X in DMSO (48 mg/mL) was selected as an example to be fractionated using different syringe filters (5, 0.45, 0.2, and 0.1 µm in pore size). The distribution of particles by size (cumulative and differential) is shown in Fig. 1a.

Table 3 Dynamic and static light scattering analyses of xylan feedstocks

Fig. 1

Solution properties of hemicellulose feedstocks in DMSO. a Particle size distribution in the soluble fraction of LMW-X. b FITR spectra of soluble, insoluble fractions, and original LMW-X. c Magnified FTIR spectra in the 2000–1500 cm⁻¹ region of fractionated LMW-X. d Correlation between the value of the shape factor and the soluble fraction for each hemicellulose feedstock. e Appearance of soluble and insoluble fractions of LMW-X feedstock. f Weisner staining shows the characteristic pink-colored stains due to the reaction of cinnamaldehyde end groups of lignin with phloroglucinol

The soluble and insoluble fractions were analyzed separately using FTIR and compared to that of the starting feedstock. FTIR analysis revealed no essential compositional differences between fractions (Fig. 1b). This observation led us to conclude that, within the sensitivity of FTIR, the insoluble and soluble fractions have similar chemical compositions. The only minor difference between the three spectra is a small band at around 1700 cm⁻¹ that is present only in the insoluble fraction. This spectral signal can be attributed to stretching vibrations of C=O bonds at the β location and in COOH groups of lignin. In a typical Kraft process, xylan isolation from black liquor may induce chemical cross-linking of xylan chains, forming large aggregates in DMSO. On the other hand, the large number of hydroxyl groups on the xylan backbone can form extensive hydrogen bonding networks. In addition, lignin has been proven to cross-link with polysaccharides [39]. As evidenced by the Weisner staining method (Fig. 1f) [40], lignin is present in the feedstocks used herein, which adds another layer of complexity to biomass both analysis and chemical transformation.

The ratio R_g/R_h shape factor (ρ) for the soluble fraction of xylan provides qualitative information on the conformation of soluble xylan molecules and the structure of their aggregates (particles). For a compact spherical particle, the theoretical value for ρ is 0.775, while the ρ value for a random polymer coil is 1.54 [41, 42]. Large variations are apparent between the ρ values of xylan feedstocks (Table 3). Taken together, these data indicate there is a correlation between ρ value and the amount of xylan in the soluble fraction; the lower the ρ value, the lower the soluble fraction (Fig. 1d). For samples with the highest solubility, the ρ value is close to 1.5. This relationship is attributed to the compactness of the particles.

Also, comparing the shape factor of samples with different degrees of polymerization (DP) shows that glycotypes with a higher DP are less soluble and form more compact particles (the experimental ρ value for RYL-X is 1.06 and decreases to 0.58 for RYH-X). As DP increases, we hypothesize that polymer segments are more entangled and engage in increased hydrogen bonding with each other. Furthermore, increased inter- and intramolecular interactions of polymer chains would cause them to coil more tightly, thus forming more compact particles. While DMSO can form hydrogen bonds with xylan chains, these interactions are not strong enough to overcome some fraction of the hydrogen bonds between the polymer segments.

Succinylation

Succinylation of hemicellulose samples was conducted in DMSO with KOH as a catalyst. The reaction scheme is shown in Fig. 2. KOH deprotonates hydroxyl groups on the xylan backbone, converting them into nucleophilic alkoxide ions that can open succinic anhydride rings, ultimately forming ester linkages with succinic acid moieties. Succinylation of xylobiose was used as a reference reaction for comparison with more complex glycotypes. Analysis of succinylated xylobiose by MALDI–TOF–MS is illustrated in Fig. 2. The products of the reaction with succinic anhydride had distinct masses of 520, 620, 720, 820, and 920 Da, representing xylobiose with two or more succinyl substitutions (Fig. 2b). These data show that succinic acid moieties can modify all of the hydroxyl groups on a xylobiose molecule. On the other hand, no visible signals of unreacted xylobiose (m/z 304 and 320) in the product spectrum confirm the effectiveness of purification steps.

Fig. 2

Reaction of xylan with succinic anhydride. a Reaction scheme, b MALDI–TOF mass spectra of DHB matrix, xylobiose, and xylobiose succinate. c–f Effect of c reaction time, d temperature, e succinic anhydride (SA) to xylan ratio, and f catalyst amount (**p < 0.001) on the carboxyl content (CC). Data are represented as means of 3 technical replicates with individual data points shown

We opted to use LMW-X to optimize reaction parameters, with a goal of achieving high CC, due to its abundance and accessibility. LMW-X is an industrial feedstock that comprises short xylan polymers with an average DP of 29, making it a suitable feedstock for valorization. The effect of temperature, time, KOH, and succinic anhydride concentration were optimized to yield the highest CC. Our results demonstrated that reaction parameters significantly impact CC. The reaction proceeds rapidly, approaching the highest CC within a few hours (Fig. 2c), which we attributed to the high consumption rate of succinic anhydride molecules. After 3 h, slow hydrolysis begins to dominate the reaction, as indicated by the observed decrease in CC. This slow hydrolysis was more evident when we studied the effect of reaction temperature on CC. At low temperatures, the reaction proceeds slowly (Fig. 2d). The highest CC was 1.3, which occurred at 70 °C (Table 4). A higher reaction temperature favors a higher rate of hydrolysis, thus lowering the CC. Compared to reported results [15, 20], succinylation of xylans in DMSO using KOH as a catalyst does not require a high amount of succinic anhydride to reach high degrees of substitution. Our data shows that the equimolar ratio of succinic anhydride to xylan results in the highest CC (Fig. 2e). LMW-X consists of short, unsubstituted xylan polymers that form large, compact aggregates, which likely limit the access of succinic anhydride molecules to the hydroxyl groups on the xylan backbone of this feedstock. Indeed, the surface of these large particles is functionalized rapidly, while reaction of the hydroxyl groups that are not exposed to the outer shell of particles remains kinetically controlled. However, we observed that increasing reaction time does not improve the CC dramatically, likely due to the competing hydrolysis reaction.

Table 4 Carboxyl content and reaction yield of succinylation of different feedstocks measured by titration

Next, we evaluated the concentration of KOH on the formation of succinylated xylan. Our data confirms that the effect of KOH is nonlinear, and we observed CC initially dramatically increases and then rapidly decreases (Fig. 2f). KOH can cause hydrolysis of succinic anhydride in the presence of trace level of humidity. On the other hand, KOH can effectively hydrolyze the ester bonds formed during the initial stages of the reaction. This hydrolysis reaction competes with the succinylation reaction and reduces the overall CC.

Structural analysis of succinylated feedstocks

To verify the success of the modification and study the resulting structural changes in xylan feedstocks, we performed spectroscopic analyses on representative samples. FTIR spectra of native and modified xylan feedstocks with high CC (CC = 1.3 for LMW-X) are shown in Fig. 3a. The unmodified xylans display a broad band at around 3300 cm⁻¹ due to O–H stretching vibrations from hydroxyl groups involved in extensive hydrogen bonding. The sharp band at 1033 cm⁻¹ is a characteristic feature of xylan feedstocks corresponding to C–O–C stretching vibrations of glycosidic linkages between xylosyl residues. The band at 897 cm⁻¹ also corresponds to out-of-plane bending vibrations of C–H bonds in β-glycosidic linkages. The distinctive sharp band at around 1735 cm⁻¹ in the modified samples can be related to carbonyl stretching from the carboxylic acid and ester functional groups. Furthermore, C–O stretching at approximately 1300 cm⁻¹ is pronounced in the spectra of modified xylans. The bands below 3000 cm⁻¹ also intensify after modification due to the newly added methyl groups in the succinic acid moiety. Furthermore, the broad band attributed to absorbed moisture has shifted to higher wavenumbers, which is a characteristic of carboxylic acid groups present in the molecule. On the other hand, two distinctive signals of succinic anhydride at 1860 and 1780 cm⁻¹, corresponding to symmetric and asymmetric C=O stretching vibrations, respectively, are not present in the modified samples. This is an indication that the purification step has successfully removed any residual unreacted succinic anhydride.

Fig. 3

FTIR and NMR spectroscopy of modified xylan: a FTIR spectra of native and modified LMW-X, b ¹H NMR spectra of native and modified LMW-X feedstock, and d ¹³C NMR spectra of native and modified LMW-X

The ¹H NMR spectrum of the modified xylan revealed a distinct new chemical shift at around 2.45 ppm, which corresponds to the methylene groups introduced by the succinic acid moieties. The carboxylic acid pendant groups in the succinic acid moieties created a unique broad chemical shift at approximately 12.5 ppm, reflecting their specific chemical environment (Fig. 3b). Similarly, the ¹³C NMR spectra displayed characteristic signals associated with the succinic acid moieties (Fig. 3c). New peaks were observed at 29 ppm, corresponding to the carbons in the methylene groups; at 171 ppm, representing the carbon in the carbonyl group of the ester bonds; and at 176 ppm, assigned to the carbons in the carboxylic acid groups. Taken together, these data provide direct evidence of the successful functionalization of the xylan backbone with succinic acid.

Thermal properties

To evaluate the effect of modification on the thermal properties of xylan feedstocks, we investigated the thermal stability and transitions of native and modified samples using TGA and DSC analysis. All of the xylan-rich hemicellulose feedstocks show similar thermal degradation behaviors, with a degradation onset at around 200–250 °C (Fig. 4). Notably, CORN-X shows a lower temperature degradation onset compared to other samples, which we attribute to the lower molecular weight. In addition, a slight weight loss is observed in the 60–100 °C temperature range due to water evaporation from the sample, and is most apparent in water-soluble POP-X. This can be due to a higher GlcA content in POP-X, which has a higher hydrophilicity. Modified xylans showed slightly lower thermal stability relative to native xylan, with a degradation onset at around 180 °C. Pendant carboxylic acid groups can undergo autocatalytic decomposition that can lower the thermal stability of succinylated xylan samples. In addition, the newly added carboxylic acid functional groups can impose the risk of intramolecular backbiting, resulting in the formation of volatile compounds and cyclic oligomers, reducing the onset of thermal degradation. We did not observe a correlation between different degrees of succinylation and thermal stability.

Fig. 4

Characterization of the thermal properties of native and modified xylans. a TGA thermograms of native xylan feedstocks, b TGA thermograms of modified xylans, c DSC curves of native xylan feedstocks from 30 to 50 °C below degradation onset, and d DSC curves of modified xylans from 30 to 50 °C below degradation onset

To further assess the thermal properties of native xylan feedstocks and their functionalized counterparts, we performed DSC analysis (Fig. 4b).We found that there are no clear thermal transitions in any of the native and modified xylan feedstocks. This observation could be an indicator of a very broad and weakly detectable thermal transition. These results reinforce the notion that xylan and its modified derivatives are largely amorphous without clear crystalline transitions or sharp thermal events.

Impact of xylan composition and conformation on xylan reactivity with SA

To understand how structural variations in xylan feedstocks influence their chemical reactivity, we examined the relationship between chemical composition, degree of polymerization, solution properties, and their reactivity with succinic anhydride. CC and yield values differ based on the type of xylan feedstock (Table 4). According to CC values estimated in this work for BIRC-X (CC = 1.46), DMSO appears to be an effective solvent for the modification of birchwood xylan. Compared to results reported by Abdus Salam et al. [43], succinylation of birchwood xylan in an ethanol/water mixture leads to a CC of 0.7. A similar observation was made by Hettrich et al. [17], where the degree of succinylation of oat spelt xylan in a water dispersion was only 0.19, whereas in this work, we obtained a value for OAT-X of 0.95. Water can effectively hydrolyze succinic anhydride rings before the reaction, thereby decreasing the CC. In this study, we mitigated this issue by running the reaction under vacuum distillation to remove water from the reaction mixture, using pre-dried feedstock, as well as anhydrous solvents for reactions.

To comprehensively analyze the main structural features that dictate the reactivity of xylan-rich hemicellulose feedstocks, we performed hierarchical clustering analysis (Fig. 5) [44]. First, all data were normalized to 1 to ensure consistency and comparability across the data set. Our analysis revealed a notable correlation between increasing reactivity and xylose content. Evident from the mass spectrum of xylobiose succinate (Fig. 2b), all of the hydroxyl groups on a xylobiose molecule are reactive with succinic anhydride. As the complexity of the molecule increases, we observe a lot more variability in reactivity with succinic anhydride, as evidenced by the range in CC values from 0.2 to 1.46 (Table 4). In addition, we noted a relevant correlation between the degree of branching of hemicellulose chains and the carboxyl content. The degree of branching was estimated by the sum of arabinose and GlcA content divided by xylose content. Interestingly, while a higher degree of branching is often associated with improved water solubility, we did not see a positive correlation between this value and CC. Branching creates a more complex network that can influence the accessibility of different regions of the polymer to succinic anhydride. In highly branched saccharides, the diffusion of succinic anhydride into interior regions of the polymer matrix can be reduced, resulting in a lower overall reactivity despite the higher theoretical number of reactive sites. CHIA-X, CORN-X, and POP-X are highly branched heteroxylans, and show lower reactivity compared to samples comprising homoxylans, such as LMW-X, RYH-X, and RYM-X. On the other hand, the degree of polymerization follows a variable trend, suggesting that the reactivity of xylan feedstocks cannot be predicted by a single parameter.

Fig. 5

Structure–functionalization relationship of different xylan feedstocks. a Heat map sorted based on the CC from left to right, b hierarchical clustering analysis of the same data set

We also identified that many of the xylan feedstocks evaluated herein contain traces of lignin, which can impact the final degree of functionalization as measured by CC. Samples containing larger amounts of lignin show relatively lower reactivity. Polyphenolic polymers can negatively impact the access of succinic anhydride from interacting with the xylan’s hydroxyl groups, thereby reducing the overall reaction efficiency [27]. In addition, traces of lignin in hemicellulose samples may contain some hydroxyl groups that may compete with xylan for succinic anhydride, potentially leading to a reduction in overall CC values. Interestingly, clustering of the soluble fraction, lignin content, and ρ highlights their inhibitory effect on succinylation, likely due to steric hindrance. These results make it clear that successful functionalization results from an interplay of main xylan chain content (i.e., xylose content and degree of polymerization), branching complexity (GlcA/Xyl and Ara/Xyl), and impurity levels (lignin content and other heteropolysaccharides). Impurity levels can be reduced through improved and optimized extraction strategies, and we hypothesize that more effective lignin extraction strategies will continue to emerge, making hemicellulose streams more amenable to chemical processing.

To analyze the complex influence of compositional and structural features of xylan reactivity, we conducted Principal Component Analysis (PCA) to reveal hidden patterns and correlations within the data set (Additional file 1: Figs. S1 and S2). PCA was performed on the standardized data set using the StandardScaler function to ensure equal weighting of all variables using the scikit-learn package in Python [45]. The first two principal components (PCs) explained 49% and 23% of the variation, respectively (Fig. 6). A clear clustering pattern was observed, where hardwood feedstocks with high xylose content show higher CC values, indicating greater chemical modification potential. It is evident that industrial waste streams enriched in homoxylans, with low levels of impurities (i.e., lignin, other heteropolysaccharides, and side chains) are great candidates for valorization, as they show high reactivity toward acylation. The observation that they cluster together suggests that industrial processing produces more accessible xylan structures for chemical modification. On the other hand, feedstocks with more complex branching, such as CHIA-X and CORN-X, as well as those with higher lignin impurities, were less reactive. This observed trend underscores the impact of feedstock composition and structural features on their chemical reactivity. In other words, xylan feedstocks extracted with harsh alkaline conditions exhibit higher reactivity, leading to higher CC values, whereas samples that were fractionated with enzymes, mild alkali, or mechanical treatments display lower CC, indicating reduced suitability for chemical modification.

Fig. 6

Principal component analysis (PCA) of structural features of xylan-rich hemicellulose feedstocks that had been modified with succinic anhydride. The first two principal components are plotted

The proposed PCA-based mapping offers a robust framework for guiding the selection of feedstocks in future studies. Furthermore, these data-driven approaches shed light on how feedstock composition and processing conditions affect chemical reactivity, providing practical insights for improving the production of chemically modified, xylan-rich hemicelluloses. Ultimately, such statistical analyses enable rational design of feedstock for improved utilization, streamlining their applications in industrial chemical processes.

Furthermore, this knowledge can also be applied in the field of synthetic biology. As we deepen our understanding of structure–functionalization relationships, it becomes increasingly possible to engineer plants to produce specific hemicellulose glycotypes. Furthermore, it may be possible to modulate xylan structure at the genetic level to reduce side-chain branching, minimize lignin–carbohydrate complexes, and enhance the accessibility of hydroxyl groups for chemical modification. Using synthetic biology tools, one can develop systems with optimized cell wall architectures that are optimized for downstream extraction and processing. By bridging the genetic blueprint of biomass with its chemical reactivity potential, this approach can align plant design and material functionality, advancing future biomass valorization and bio-design strategies.

To study the effect of succinylation reaction on particle size, DMSO solutions of modified xylans were prepared at low concentrations and analyzed by dynamic light scattering (DLS). Our data showed that modification improves the solubility of the samples in DMSO, but they still contain some particulate aggregates (Additional file 1: Fig. S3). However, the majority of xylan succinate products showed a dramatic increase in particle size. The only exceptions are BIRC-X succinate and CHIA-X succinate, which both fall into the extremes of the CC range (CC for BIRC-X succinate and CHIA-X succinate is 1.46 and 0.2, respectively). The average particle size in the CHIA-X succinate did not change dramatically, and the distribution was largely influenced. On the other hand, BIRC-X shows the highest CC. Succinylation disrupts the xylan particles, leading to higher solubility and lower average particle size (Table 5).

Table 5 SLS and DLS analysis on native and modified xylan-rich hemicellulose feedstocks

Comparing the results obtained through dynamic and static light scattering analysis of the feedstocks, we observed a clear shift toward higher values of the shape factor for most of the samples. We can also see that some particles (LMW-X, BEW-X, BIRC-X, and OAT-X) decreased in size after functionalization with succinic anhydride. We speculate that the introduction of succinic acid moieties induces swelling and improves the solubility of the modified xylan. Native xylan particles are compact due to the extensive network of hydrogen bonds within xylan segments, pulling the segments tightly together, resulting in a denser structure [46]. On the other hand, succinylated xylan particles adopt a more expanded structure when hydrogen bonds are disrupted. Further disruption of hydrogen bonds may lead to the dissolution of some modified xylan molecules, leading to a decrease in particle size. This observation invites future studies to employ molecular dynamics simulations to analyze the impact of succinic acid moieties on the interaction of the functionalized xylan with water.

Properties of xylan succinate films

To evaluate the film-forming and optical properties of xylan succinate, we selected BIRC-X as a representative sample. Among the various investigated feedstocks, BIRC-X exhibited the highest carboxyl content and reactivity (CC = 1.46), which makes this feedstock a promising candidate for further material testing. Native BIRC-X feedstock (Fig. 7a) exhibits poor water solubility, forming a cloudy dispersion (Fig. 7b), resulting in low transmittance values across visible spectrum in UV–Vis analysis (Fig. 7c).This limited solubility limits the film formation properties of native xylan, evidenced by opaque and brittle films prepared by solvent casting (Fig. 7d). In contrast, sucinylation of BIRC-X significantly improves its water solubility, forming flakes (Fig. 7e) that show much higher water solubility compared to the native feedstock (Fig. 7f). The effect of succinylation on the optical properties of BIRC-X was examined, and showed succinylation dramatically improves light transmission in the 400–700 nm wavelength range (Fig. 7g). Adding succinic acid moieties to xylan improves solubility in water, resulting in homogeneous transparent solutions for modified BIRC-X and several other feedstocks (Additional file 1: Fig. S4). The improved solubility and reduced aggregation enabled the formation of uniform, transparent, self-standing films from BIRC-X succinate (Fig. 7h). This is likely due to the introduction of succinic acid moieties anchored to the xylan backbone, as succinic acid moieties are known to act as internal plasticizers.

Fig. 7

Optical properties of BIRC-X and BIRC-X succinate. a Appearance of native BIRC-X feedstock, b 5 mg/mL solution of native BIRC-X, c UV–Vis spectrum of BIRC-X solutions at different concentrations (wt%), d film formation of BIRC-X by solvent casting, e appearance of BIRC-X succinate flakes, f 5 mg/mL solution of BIRC-X succinate, g UV–Vis spectrum of BIRC-X succinate solution at different concentrations (wt%), h film formation of BIRC-X succinate by solvent casting

SPM imaging was used to evaluate surface morphology, roughness, and film thickness of spin-coated thin films prepared from 2.5 wt% solutions of both native and modified xylan (Fig. 8). To assess film thickness, a scratch was manually introduced to the surface using a sharp blade, and line profiles were extracted from AFM scans across the scratch or a representative deposited aggregate. Most native xylan feedstocks deposit in the form of colloidal aggregates and discontinuous patches on the substrate. Unmodified xylan does not form a uniform film; however, large aggregates are found on the surface. These particles vary in size from 0.1 to 2 µm (Fig. 8). These results align well with the light scattering results. In native BIRC-X, POP-X, and CORN-X, we observed a deposited film on the Si-wafer substrate; however, these films are heterogeneous and contain inclusions due to the presence of particles that are not fully dissolved (Table 6). The RYH-X sample contains particles that are deposited intact in their native form on the surface with an average size of 500 nm. Considering previously reported enzymatically synthesized xylan microcrystalline particles [47], we can speculate that RYH-X samples contain microcrystalline particles. On the other hand, modified xylan forms uniform, well-adhered films with reduced surface roughness (Table 6). In all of the cases, a formed film from modified feedstocks resulted in a lower roughness and aggregate size compared to the native xylan samples. For OAT-X, RYH-X, and BIRC-X, the aggregate size is noticeably decreased (Fig. 8). For example, OAT-X deposits in the form of large aggregates (> 20 µm) on the surface, while after succinylation, the average aggregate size drops to 3.6 µm. This observation once again substantiates the effect of introduced succinic acid moieties to the xylan backbone, acting as internal plasticizers that disrupt intermolecular hydrogen bonding between xylan chains and increase the chain’s mobility. This plasticizing effect leads to enhanced free volume within the film matrix, which facilitates molecular rearrangement during film formation and reduces brittleness in the final material. As a result, modified xylan feedstocks form smoother films with optical clarity, which are key properties for coatings, packaging, and bioplastic films.

Fig. 8

SPM scans of deposited native and modified xylan samples

Table 6 Surface morphology of native and modified xylan films made by spin-coating on Si-wafer substrate

Conclusion

This study systematically explores the structural and compositional features of xylans derived from plant biomass and their impact on reactivity during succinylation with succinic anhydride. We demonstrated that feedstocks with higher xylose content and lower degrees of branching exhibit the highest chemical reactivity, achieving higher carboxyl content (CC) and yields. Conversely, lignin impurities and complex branching negatively impacted reactivity, which we attribute to reducing accessibility to reactive hydroxyl groups. These findings support our hypothesis that structural accessibility, combined with reduced steric hindrance at the polymer backbone, are essential in achieving high-efficiency functionalization. Our findings indicate that industrial hemicellulose waste streams, such as those from textile production, predominantly composed of homoxylans, are promising candidates for scalable functionalization and materials production. The use of DMSO and KOH as reaction media and catalysts, respectively, enabled efficient functionalization, producing succinylated xylans with a carboxyl content of up to 1.46. Furthermore, we showed that multivariate analysis with hierarchical clustering and principal component analysis (PCA) is a robust approach to readily evaluate feedstock reactivity based on compositional and physicochemical parameters. The findings in this work provide a framework for selecting and engineering hemicellulose feedstocks with optimized structures and support the development of a design schema for future genetic engineering efforts in Populus aimed at creating multi-use poplar optimized for biomaterials production [48].

The introduction of succinic acid moieties induced conformational changes of xylan molecules in the particles, causing a transition from compact to expanded structures due to disrupted hydrogen bonding. These structural modifications not only improved solubility but also enabled the production of flexible, optically transparent films with potential applications in biodegradable packaging. Furthermore, the thermal properties of modified xylans allow for moderate processing temperatures, further supporting their application potential. Taken together, our approach establishes a framework for selecting and optimizing hemicellulose for chemical modification, advancing efforts for valorization of underutilized biomass and agricultural waste streams for sustainable material development. We envision future studies that will focus on molecular simulations and the structure of insoluble fractions of xylan (colloidal particles) to further elucidate the dynamic interplay between xylan structure and its reactivity and material properties.