β-Irradiation in the presence of 1,3-dialkylimidazolium ionic liquids causes covalent cellulose derivatization with simultaneous nitrogen incorporation

Pretreatment of biomass with ionic liquids (ILs) has become an important element of biorefinery scenarios (for recent reviews see: Zhang et al. 2014; Amini et al. 2021; Quesada-Salas et al. 2022; Meng et al. 2023). In most cases, 1,3-dialkylimidazolium ionic liquids are used due to their good availability and comparatively low prices. Although most recently deep eutectic solvents (DES) have increasingly entered the field of biomass pretreatment, they are still mainly at the research stage, while IL pretreatments are more mature (Bhaskar Reddy et al. 2019; Roy et al. 2020; Roy and Chundawat 2023). The beneficial effects of IL pretreatment have been described as increasing the accessibility of wood or biomass components to reagents or enzymes, swelling or partial dissolution. In this way, subsequent processes, such as hydrolytic or enzymatic saccharification, delignification and pulping, or extraction steps, are generally facilitated or accelerated. At the same time, it had to be recognized that there are inherent problems with IL purification and recycling in these biomass pretreatment approaches (Tu and Hallett 2019; Ejaz and Sohail 2020; Mora-Pale et al. 2011). It has been recognized that complete separation of the IL from the biomass is not possible because of its strong adhesion to the complex biomass matrix and, in some cases, chemical reactions with biomass components. In many cases, however, complete removal of the IL is neither necessary nor desired because the residual IL does not interfere with the immediately following processing steps. The biomass is simply washed with process water to remove most of the loosely adhering IL. Residual IL, either more strongly adsorbed or as a component in adhering process fluids, is typically present in concentrations between 2 % and 5 % when the IL-treated biomass enters the next step, such as saccharification, pulping, pyrolysis, or extraction (Tadesse and Luque 2011; Zhang et al. 2014; Elgjarbawy et al. 2016; Lin et al. 2022).

Exposure of biomass to electron beams (β-irradiation, “e-beaming”) has become a frequent biomass pretreatment option as well. Industrial electron accelerators operate in the low (80–300 keV), medium (300 keV–5 MeV), and high energy (above 5 MeV) ranges. The irradiation dose is usually measured in Gray (Gy), defined as one Joule of adsorbed irradiation energy per kilogram of matter (SI unit: m² s^-2). It is influenced by irradiation intensity and irradiation time. Several studies have demonstrated that β-irradiation enhances the rates of acidic and enzymatic hydrolysis of renewable feedstocks, such as bagasse, hemp, and woody plants (Kumakura and Kaetsu 1983; Duarte et al. 2012), in particular with regard to the hemicellulose fractions (Shin et al. 2008). As early as 1952, β-irradiation was reported to cleave cellulose chains and decrease cellulose crystallinity (Saeman et al. 1952). It became clear further on that β-irradiation causes a depolymerization of polysaccharides in general and an introduction of oxidized groups along the polymer chains (Ershov 1998; Dubey et al. 2004; Bouchard et al. 2006). For the specific case of cellulose, this can be employed to enhance reactivity in derivatization reactions (Iller et al. 2022, Ekman et al. 1984), increase the adsorption capacity of microcrystalline cellulose in food applications (Nemtanu et al. 2007) or facilitate the functionalization by grafting of polymers onto cellulosic carriers (Sekine et al. 2010; Desmet et al. 2011; Wojnarovits et al. 2010).

Thus, it was only logical to combine both pretreatments. While synergistic effects have not been observed, an additive effect on the rate of acidic hydrolysis of polysaccharides and a tar reduction in biomass pyrolysis have been noted (Jusri et al. 2019; Zhang et al. 2014; Amini et al. 2021), as well as an improved dissolution and faster derivatization (acetylation) of cellulose (Lee et al. 2013). Electron beaming in combination with polymerizable ionic liquids (such as N-allyl-imidazolium derivatives) has been used for micropatterning of synthetic and natural polymers in photonics (Rola et al. 2019 and 2021). Hao et al. (2012) reported progressive cellulose depolymerization by β-irradiation in 1-butyl-3-methylimidazolium chloride (BMIm-Cl), Kimura et al. (2016) a crosslinking effect for intensively β-irradiated polysaccharides, and Croiteru (Croiteru et al. 2014; Croiteru and Patachia 2016) described hydrophobization of cellulose and wood surfaces by β-irradiation in long-chain 1,3-dialkylimidazolium ILs. Apart from these few phenomenological observations, possible chemical processes induced by the combination of electron beaming and imidazolium ILs have never been conjectured, nor have in-depth studies of mechanistic aspects or the type of chemical modifications been performed.

Electron beam (EB) irradiation was carried out at NHV Corporation (Kyoto, Japan) on an Curetron^® EBC300-60 electron beam generator at 300 keV, and at Kremsmünster (Austria) on a Rhodotron IBA TT-100 accelerator at 10 MeV. Powdered cellulose samples were evenly distributed inside PET plastic bags for irradiation. Also cellulose sheets (pulp, bacterial cellulose) were inserted into plastic bags before irradiation for optimum comparability. The samples went irradiated at doses of 30 kGy, 60 kGy and 90 kGy (30.7 mA). For detailed procedures see: Ehrhardt et al. (2005), Henniges et al. (2012, 2013). Cellulose sources were available from previous work: bacterial cellulose (Sulaeva et al. 2020), cellulose II gels from Lyocell processing (Beaumont et al. 2017)

In a first set of experiments, different types of lignocellulosic biomass that had been studied in previous work, such as wood chips, wood flour, corn stalks, oil palm empty fruit bunches or Rambutan fruit peel, all containing 3 wt% of an imidazolium ionic liquid (IL), had been subjected to β-irradiation at a dose of 60 kGy. The used IL concentration was in the range of IL concentrations in biomass that was pretreated with ionic liquid and then pressed or washed once with water, as used in different biorefinery scenarios for better saccharification of the polysaccharides (Taherzadeh and Karimi 2008; Azizan et al. 2020; Haldar and Purkait 2021) or preparation of cellulose nanocrystals (Hao et al. 2012; Zhang et al. 2021). Either 1-ethyl-3-methylimidazolium acetate (EMIm-OAc) or 1-butyl-3-methylimidazolium acetate (BMIm-OAc) was used. Several blank runs were carried out: first, by employing either 1-methylimidazolium acetate (MIm-OAc), morpholine or dimethylsulfoxide (DMSO) instead of the IL, second, by carrying out the irradiation under inert gas (Ar) instead of ambient air so that atmospheric oxygen was excluded, and third, by replacing the irradiation step by a heating step (50°C) of equal time length. All irradiation treatments or blank trials were followed by a thorough soaking / washing sequence with water, performed similarly in all cases, and a drying step (80°C, to weight constancy).

Most starting materials had a minor natural nitrogen content, about 0.10 – 0.20 %. All nitrogen contents were determined by microanalysis, the limit of quantification being 0.04 %. After processing, a general increase in nitrogen content – by up to 0.2 % – was seen as soon as N-containing compounds, either IL or methylimidazolium acetate, were involved, which reflects the adsorption of exogenous nitrogen species that are washed out incompletely. However, it was evident that β-irradiation in the presence of IL boosted the nitrogen levels of the products far more, which ranged between 2.5 and 4.2 % above that of the corresponding non-irradiated material (see Fig. 1). This increase was observed in the case of all substrates and independent of the IL type (ethyl or butyl derivative) being used. e-Beaming did not have any significant effect in the case of MIM-OAc or DMSO, and also the type of atmosphere, air or argon gas, did not influence the N-uptake. Only the combination of irradiation and IL caused a significantly increased N-content, i.e., an uptake of IL or IL-derived species into the biomass structure in a form that was not removable by washing, or at least much harder than without irradiation. It was impossible to say at this point whether the nitrogen incorporation was due to the covalent binding of N-containing species in a chemical reaction or to a physical alteration by the irradiation that favored an entrapment. Previous research on irradiation of cellulosic pulps had shown that both effects can occur (Henniges et al. 2012, 2013).

When the experiments were conducted with cellulose instead of the whole biomass with an otherwise identical setup, the results were very similar. Only the combination of β-irradiation and the presence of IL caused a significantly increased nitrogen content of the cellulose (see Fig. 2), but other N-containing compounds, such as morpholine or 1-methylimidazole, did not.

The N-content increased linearly with the irradiation dosage at a constant IL concentration (3 %), and also linearly with the IL concentration at a constant irradiation dosage (60 kGy), see Fig. 3. There was no obvious influence of the crystallinity (determined by solid-state ¹³C NMR spectroscopy) and of the type of cellulose used. However, there seemed to be increasing nitrogen incorporation in pulps which were oxidatively damaged and had an increased content of carbonyl groups (Fig. 4 a). The degree of derivatization of cellulose (such as in cellulose acetates, carboxymethyl celluloses or methyl celluloses with different degree of substitution), however, seemed to have no systematic influence, and neither did the choice of the cellulose allomorph I or I, i.e., native cellulose vs. regenerated cellulose (data not shown).

Two different celluloses containing 3 wt% EMIm-OAc were β-irradiated and thoroughly washed according to the above protocol. Afterwards, the celluloses were dissolved in ZnCl₂*3H₂O (Chen et al. 2020; Lara-Serrano et al. 2020; Burger et al. 2020), reprecipitated in water and thoroughly washed until chloride-free, and this dissolution/reprecipitation/washing sequence was repeated twice. This was done to address the binding mode of the nitrogen to the cellulose matrix. The celluloses´ N-content should not change significantly if nitrogen was covalently linked, while physically bound (adsorbed) nitrogen should – at least to some part – be released into the solvent medium, which would be reflected by a decrease in N-content after each dissolution/reprecipitation/washing cycle. It is obvious that for this approach a nitrogen-free cellulose solvent had to be used – and there are astonishingly few cellulose solvents that do not contain N (Liebert 2010), which is the reason for the use of the less common ZnCl₂*3H₂O solvent. As seen from Fig. 4 b, the nitrogen content of the cellulose decreased only insignificantly upon dissolution/reprecipitation, which was a strong indication that the nitrogen was indeed covalently linked to the cellulose, and not just tightly physically adsorbed. The ionic liquid must have been the nitrogen source since no other N-source had been present, and it must have been converted into a reactive species by the β-irradiation since no N-fixation was observed without this radiation exposure.

It is known that imidazolium ionic liquids often contain impurities, such as degradation products, different counterions or byproduct traces from synthesis (Liebner et al. 2010), which might give rise to side reactions. In the present case, at least the common impurities 1-methylimidazole and imidazole seemed to be innocuous because artificially increased concentrations of these byproducts in the IL – 10 vol% of a 1:1 (v/v) mixture of 1-methylimidazole and imidazole – were obviously without influence. Similarly, changes in the IL, using EMIm-Cl, BMIm-OAc or BMIm-Cl instead of the BMIm-OAc, did not significantly alter the outcome of nitrogen fixation (data not shown).

The independence of the N-incorporation process on the addition of stabilizers that act as antioxidants, such as α-tocopherol or propyl gallate, seems to argue in favor of a heterolytic, i.e., non-radical reaction mechanism. This agrees with the fact that all attempts failed to trap possible radical intermediates by EMPO derivatives, which usually work well with oxygen-, nitrogen- and carbon-centered radicals (Stolze et al. 2003).

Current studies now try to address the reaction mechanism in detail, especially the reactive species and the exact nature of the binding to the cellulose matrix. The presence of oxidized functionalities, carbonyl, and carboxyl groups along the cellulose polymer chain (Potthast et al. 2005; Ahn et al. 2019), which can also cause crosslinking effects and different solubilities (Henniges et al. 2011), seems to be a crucial structural element in this regard. We currently follow several mechanistic options, such as subsequent reactions of β-alkoxy-elimination processes (Hosoya et al. 2018), reactions of the cellulosic key chromophore 2,5-dihydroxy-benzoquinone (Korntner et al. 2015) which is readily formed from carbonyl functions in celluloses and responsible for the reaction with byproduct amines (Goto et al. 2020), or formation of reactive species from the IL cation, such as keteniminium ions (Potthast et al. 2002) or N-heterocyclic carbenes (Jahnke and Hahn 2017).

The combination of β-irradiation and imidazolium-based ionic liquids should be avoided in the treatment of biomass. It leads to the fixation of nitrogen from the ionic liquid into the cellulosic matrix. Experiments strongly suggested that this binding was covalent and caused by reactive species formed from the ionic liquid by β-irradiation. In any case, the nitrogen contents of up to 3 % are so high that it is no longer possible to regard them as trace amounts or their formation processes as unimportant side reactions. Unintended cellulose derivatization is of particular importance if the pulp is to be separated and used for applications on human beings (from textile fibers to medical products). Nitrogen fixation depended on the concentration of the ionic liquid, the intensity of irradiation and the oxidative pre-damage of the pulp, but not on the source of the cellulose, the type of allomorph and the crystallinity.

The detailed reaction mechanism with the cellulose polymer is currently being investigated based on several options that seemed plausible based on the data available so far. Also the extent to which analogous reactions occur with hemicelluloses or lignin and the question of whether the presence of these polymers can minimize or prevent N-bonding to cellulose, needs to be scrutinized. Provided that the N-modification of cellulose follows a well-defined pathway and is not a rather random or multi-path process, it should eventually be studied whether it can be used for a controlled derivatization, which would offer the advantage of a solvent-less and chemical-less – apart from the small content of IL – reaction system.