Enhancing the Antioxidant Activity of Technical Lignins by Combining Solvent Fractionation and Ionic‐Liquid Treatment

Abstract A grass soda technical lignin (PB1000) underwent a process combining solvent fractionation and treatment with an ionic liquid (IL), and a comprehensive investigation of the structural modifications was performed by using high‐performance size‐exclusion chromatography, 31P NMR spectroscopy, thioacidolysis, and GC–MS. Three fractions with distinct reactivity were recovered from successive ethyl acetate (EA), butanone, and methanol extractions. In parallel, a fraction deprived of EA extractives was obtained. The samples were treated with methyl imidazolium bromide ([HMIM]Br) by using either conventional heating or microwave irradiation. The treatment allowed us to solubilize 28 % of the EA‐insoluble fraction and yielded additional free phenols in all the fractions, as a consequence of depolymerization and demethylation. The gain of the combined process in terms of antioxidant properties was demonstrated through 2,2‐diphenyl‐1‐picrylhydrazyl (DPPH.) radical‐scavenging tests. Integrating further IL safety‐related data and environmental considerations, this study paves the way for the sustainable production of phenolic oligomers competing with commercial antioxidants.


Introduction
Industryi si ncreasingly demanding of biobasedp henolic compounds, to be used as building blocks for polymer synthesis or valued for their antiradicala ctivity,e specially in the field of polymers, materials, and cosmetics. Phenolics derived from plant biomass are in particular potentiala lternatives to synthetic commercial antioxidants such as Bisphenol A [1] and tert-butylated hydroxytoluene( BHT). [2] Amongt hem, ferulic acid is already knownf or its potential for various applications, [3] but its low availability from plant sourcesm ight hinder its industrial development. As polymers of phenyl propanoids representing up to 30 %o fthe plant biomass, ligninsr epresent one of the major potential sources of biobased phenolics. [4] Moreover, the valorization of technical lignins generated as industrialb yproducts would increase the sustainability of lignocellulosebased biorefineries. [5] Various strategies driven by green chemistry principles have been designed to convert technical lignins into functional moleculeso ra ssemblies while mitigating their variability processing from botanical origin and transformationsd uring industrial treatments. [6] These strategies include fractionation and depolymerization. Fractionation can be performed by ultrafiltration, potentially directly integrated into the pulping process, [7] or by solvente xtraction, whichi sa dvantageously performed at ambient temperature andp ressure and does not requires pecific equipment. [8] In contrast, among the depolymerization strategies (thermochemical, biological, or chemocatalytic), [9] the recently reported implementationo fm ethyl imidazolium bromide ([HMIM]Br) ionic liquid (IL) providesaw ay to produce lignin oligomers with increasedf ree phenol content (PhOH) and decreased polymerization degree, as ac onsequence of the selectivecleavage of aryl-alkyl ether bonds (Figure 1). [10] In particular,t his was shown to generate new phenol groups from the methoxy groups of lignin syringyl and guaiacyl units.S uch Ag rass soda technical lignin (PB1000) underwent ap rocess combining solvent fractionation and treatment with an ionic liquid (IL), and ac omprehensive investigation of the structural modifications was performed by using high-performance sizeexclusion chromatography, 31 PNMR spectroscopy,t hioacidolysis, and GC-MS. Three fractionsw ith distinct reactivity were recoveredf rom successive ethyl acetate (EA), butanone, and methanol extractions.I np arallel, af raction deprived of EA extractives was obtained. The samples were treated with methyl imidazolium bromide ([HMIM]Br) by using either conventional heatingo rm icrowavei rradiation. The treatment allowed us to solubilize2 8% of the EA-insoluble fraction and yieldeda dditional free phenols in all the fractions,a saconsequence of depolymerization andd emethylation. The gain of the combined processi nt erms of antioxidant properties was demonstrated through2 ,2-diphenyl-1-picrylhydrazyl (DPPHC)r adical-scavenging tests. Integratingf urtherI Ls afety-related data and environmentalc onsiderations,t his study paves the way for the sustainable production of phenolic oligomers competing with commercial antioxidants. phenolic oligomers have great potentiala sa ntioxidant additives for the formulation of materials because they are likely to combine miscibility in the polymer matrix and limited migration towards the environment owing to their oligomeric structure. [11,12] Despite this potential, the process has, until now, only been applied to lignin models, and the antioxidantp roperties of the oligomers have not been assessed yet. The advantages of using [HMIM]Br for the transformationo fl ignins are foreseen to potentially ensure the homogeneity of the reaction medium, to avoid the use of additional chemical reagents, and to perform the reactioni nm ild conditions (T < 110 8C, t < 40 min) compared with most lignin catalytic conversion processes. Moreover,i nt his context of use, the IL can be recycled. All these advantages also render the process attractive from an environmental point of view.H owever,t he sustainability assessment of the process requires further health and safety information on [HMIM]Br,w hich has not been provided so far.T he present paper aims at assessing the possibility of transferring the newly developedI Lp rocess [10] to technical lignin fractions. The objective was to recover antioxidant extracts soluble in ethyl acetate (EA), one of the conventional solvents recommended for their relativelyl imited health hazard and environmental footprint as assessed from the Health and Safety Executive (HSE) criteria compiled from variouss ources. [13] EA was selected here to separate extractives from the IL/waterl ayer.B ecausea lkali grass lignin Protobind 1000 (PB1000)f rom GreenValue LLC is available at an industrial scale and is already knownf or its antioxidant properties, [14] it was selected as the starting material for this study.T hisl ignin was subjected to at hree-step semi-continuouss olvent fractionation process to recover three structurally distinct soluble fractions (F1-F3). In parallel, the same lignin sample was submitted to an extensive EA washing to recover an EA-insoluble residue (F4) and validate the possibility of recovering soluble compounds through chemical treatment of this fraction. PB1000 and the four fractions were submitted to the previously optimized IL treatment by using microwavei rradiation (MW) or conventional heating (CH) as heating processes. The advantage of MW is the short reactiont ime (10 s), whereas CH was used to generate more severe modifications relevant to investigation of the structure-properties relationships. The phenolic monomers and oligomers were extracted from the reaction media with EA, and the ethyl acetate extracts( EAE) were analyzed by chromatographic and spectroscopic methods to assess the efficiency of the conversion, elucidate mechanisms, and identify molecules of interestf or furtherd evelopments. The antioxidant properties of PB1000 and the EAE as well as those of the insoluble residues recoveredf rom the most drastic process (CH, 40 min) were assessed through 2,2-diphenyl-1picrylhydrazyl (DPPHC)r adical-scavenging tests. The half-maxeffective concentrationE C 50 was used as the criterion to compare the performance of the different samples.T he study demonstrates the technical advantage of implementing the [HMIM]Br-based treatment in ac ascadinga pproachc ombining fractionation and depolymerization. In addition, sustainability considerations of the process are further discussed according to safety data obtainedf or the IL.

Results and Discussion
Recovery of different PB1000 fractions and their contrasted reactivitytowards [HMIM]Br treatments

PB1000f ractionation
PB1000w as fractionated at as emi-pilot scale (1 kg dry powder) through as emi-continuous process intended for future industrial development. [8] This process consisted of a three-step sequential extraction with solvents of increasing polarity ( Figure 2) and yielded 31 wt %f or the EA-soluble fraction F1, 19 wt %f or the butanone-(MEK)-solublef raction F2, and 23 %f or the MeOH-soluble fraction F3. Because F2 and F3 contained residual EA-soluble compounds (37 and 7%,r espectively), PB1000 was also submitted, at lab scale [g],t oadrastice xtensivew ashing with EA yieldinga nE A-insoluble fraction F4 (50 %o ft he weighto ft he startingP B1000) subsequently used only to validate the depolymerization process.
One main characteristic of this set of four fractionsw as the increasing proportion of lignin inter-unit aryl-alkyl b-O-4 linkages from F1 to F4, as reflected by the increasing thioacidolysis yields (Table 1). [15] Because our previouss tudy performed on dioxan-isolated model lignins showedt he b-O-4 linkages to be the mostp rivileged target of the IL treatment allowing partial depolymerization, [10] these fractions were thus expected to show increasing reactivity towards the IL treatment.
To determine the influence of structural variations on the efficiency of the depolymerization process, PB1000 and its four derived fractions were subjected to heating treatments in [HMIM]Br,e ither under MW (10 s, 110 8C) or under CH [110 8C, 20 (CH20) or 40 min (CH40)].I nl ine with the objective of the treatment to recover EA-soluble functional extracts, the efficiency of the treatments was assessed on the basis of the recovery yields of the EAE after treatment and the characteristics of the extracts, selecting average molar masses and PhOH as the major criteria for furthera pplications (Table 2). Moreover, to estimate the solubility gain, the proportion of EA-soluble compounds with respect to the total products recovered was calculated and compared with the EA solubility of the initial sample ( Figure 3).

Effect of the [HMIM]Br treatment on the recovery of EAE
As ac onsequence of the fractionation process, the four fractions exhibited initial contrasted solubility in EA:F 1w as 100 % EA-soluble, whereas F4 was poorly soluble, and F2 and F3 only partially (37 and 7%,r espectively). The[HMIM]Br treatments induced ad ecrease in the proportion of EA-soluble compounds of all samples, except for F3 and F4,w hich conversely exhibited an increase. The highest effect of the [HMIM]Br treatment was observedf or F4 under the MW conditions. Indeed, 15 %o f this insoluble fraction could be solubilized, and 28 %o ft he products formed was soluble. The EAE represented 8% of PB1000.T hree-fold andt wo-fold lower solubilities wereo bserved with CH20 and CH40 conditions, respectively.
In contrast, the treatment of the totally soluble fraction F1 led to the decrease of the solubilityw ith production of insoluble material (16-40 %d epending on the treatment). The maximum solubility decrease was observed with the CH conditions, indicating that these conditions potentially favored recondensation reactions of the EA-soluble compounds owing to the longer reaction time. In the case of the only partially EA-soluble fractionsF 2a nd F3, two contrasting behaviors were observed:t he F2 behavior was similart oF 1w ith as olubility decrease enhanced in CH conditions, and the F3 behavior was similar to F4 with solubility increase (up to 19 %s olubility increase)u nder MW conditions. The EAE recovered from F3 accounted for 21 %o ft he fractiona nd 5% of PB1000.T he treatment of PB1000w ithout any previousf ractionation led to an intermediate behavior with as olubility decrease observed only in the most severe conditions, CH40. In conclusion, the  [HMIM]Brt reatment could produce EA-soluble compounds from fractions showingp oor initial EA solubility,a nd the presence of EA-soluble compounds in the initial sample led to an apparent decrease in EA solubility,w hich was most probably owing to recondensation reactions favored by CH conditions. Besides initial EA solubility,t he fraction-dependent effect of [HMIM]Brt reatment on the amount of EAE recovered could be related to the structurald ifferences of the fractionsi nt erms of b-O-4 bond content. Indeed, in the case of F1 and F2, which exhibit the lowest amount of b-O-4 bonds according to thioacidolysis yields, the proportion of EA-soluble compoundsi n the reaction mixturesa fter treatment decreased, whereas in the case of F3 and mainly F4, this proportion significantly increased. The resultsa re in agreement with our previously reported statements [10] and suggestthat b-O-4 linkages are effectively the most privileged targets of the IL treatment. This also demonstrates that the efficiency of the depolymerization process is able to counterbalance the effect of recondensation reactions occurring during the treatment, and that may be particularly important in the case of long reaction times (CH compared with MW).
In all cases, and whatever the treatment, the yield of the reaction was good but never reached 100 %o wing to losses of compounds, mainly volatiles (not identified nor quantified), during the reactiona nd soluble andi nsoluble compounds during the separation process (precipitation of the products and removal of the water/ILl ayer).

Molar mass distributiono fthe EAE
Whatever the sample, the weight-average molar mass (M w )o f the EAE, before or after treatment, did not exceed 2000 gmol À1 ,i ndicating that the EAE were essentially composed of oligomers (less than ten phenylpropaneu nits). Nevertheless, some variations in molar distribution appeared between the samples upon treatments. Increasing EAE average molar masses was observed from F1 to F3, in agreement with the use of solventw ith increasing polarities for the PB1000 fractionation process. [16,17] The effect of the [HMIM]Br treatment on the molar masses depended on the fraction and conditions used. In the case of F3, all treatments induced ad ecrease in average molarm asses with an increasing effect from MW to CH20 and CH40. For F1 and F2, ad ecrease in the molar mass was only observed in CH conditions,a nd the use of MW by contrastl ed to as light increase. In thec ase of F4, the EAE produced by the treatments exhibitedm olar masses 1.4-to 1.6fold lower than that of the initial F4 insoluble fraction,a nd of the same range as the non-modified EAE of the other fractions. As light increasew as observed in the CH conditions. This result supported the hypothesist hat recondensation reactions were favored by the CH conditions.

Evidence of demethylation and depolymerization upon [HMIM]Br treatment
GC-MS analysis was performed to investigate the phenolic monomers present in the EAE (Supporting Information, S1). It revealed the presence of am ixture of compounds (phenolic acids, ketones, and aldehydes accounting in total for 1.2 %o f PB1000)i nF 1, the absence of monomeric compounds in F2 and F3 before treatment, and the formation of new compoundsf or all fractionsa fter treatment. The chromatograms of the EAE recovered after treatment revealed the presenceo f phenolicm onomers diagnostic of demethylation (all fractions) and/or depolymerization (F2 and F3). In F1, the treatment led to the total conversion of acetosyringone, the major EAE phenolic monomer before treatment, into its once-and twice-demethylated counterparts [1-(3,4-dihydroxy-5-methoxyphenyl)ethan-1-one C and 1-(3,4,5-trihydroxyphenyl)-ethan-1-one D] and to the disappearance of all other phenolic extractives, most probablyi nvolved in recondensation reactions. In F2 and F3, the treatmentl ed to the production of two ketones A and B previously identifieda sa cidolysis ketones from experiments on lignin models. [10] In F2, only the demethylated form B was detected whereas in F3 both were formed. The analysiso ft he thioacidolysis products also indicated that demethylation took place within lignin units linked through b-O-4 bonds (Supporting Information, S2).I ndeed,c atechol, 5-hydroxyguaiacyl, and 3,4,5-trihydroxyphenyl thioethylated derivatives were detected after thioacidolysis of all treated samples. The proportion of these demethylated units was higheri nt he CH-treated samples, with an increased effect after 40 min of CH. An important feature arising from theseresults is that demethylation and depolymerization throught he cleavage of b-O-4 bondss eem to be concomitantb ut independentp rocesses. Both the demethylation and depolymerization reactions diagnosed herein allow the appearance of new phenolicg roups initially involved in ether bondsa nd could thus account for the increase of PhOH.

PhOH functionality gained by IL treatment combinedwith EA extraction
To assess the functionality gained throughI Lt reatment combined with subsequent extraction, the PhOH content of the EAE after treatment were compared with the initial PhOH content of the samples. As shown by the higherP hOH content of F1 (the PB1000 EA-soluble fraction) compared with PB1000,E A extraction alone provided aw ay to recover compounds with higher functionality.H owever,t his functionality was even higher when the IL treatment was applied beforer ecovery of the EAE (3.95-11.94 versus 3.88 mmolg À1 ), except for F3-MW (2.78 mmolg À1 ). Moreover,w hatever the sample and the conditions, the [HMIM]Br treatment led to an increase of PhOH compared with the startings ample. This effect increased from F1 to F3, for which am aximum ten-fold increase was obtained ( Table 2). In the case of F3, the increase in PhOH was concomitant to the decrease in average molar masses, suggesting that the treatment induced the cleavageo fe ther bonds with subsequent release of phenol groups, as previously demonstrated on lignin models. [10] The PhOH enrichment (DPhOH, mmol g À1 )t hrough IL treatment and subsequentE Ae xtractioni ncreased with the thioacidolysis yield of the initial fraction, whatever the treatment  Figure 4). This was consistent with the major contribution of depolymerization reactions through b-O-4 bond cleavage to form new phenol groups. However,P B1000 and its residual F4 fraction did not followed the same tendency as the other three fractions. In particular, the DPhOH upon CH treatments was lower than expected from the behavior of these fractions. This suggested that ah igherp roportion of phenolsf ormed by depolymerization was present in the EA-insoluble residues compared with the other fractions.M oreover,t he weak differences in terms of mass distribution, and in fact the unclear correlationb etween thioacidolysis yield of the starting fraction and the M w of the obtained EAE,o nce more indicated that in the case of fractions for which the depolymerization process could be highly efficient (F3 or F4), recondensation reactions also occurred. Interestingly,t hese recondensations reactions seemed not to impact the global PhOH content. Thisc ould be also proofo fa synergistic effect between both demethylation and depolymerization processes. In all cases, the efficiency of the treatment conditions in terms of phenols production was increased from MW to CH20 and CH40. As am ajor feature, the [HMIM]Br treatment was shown to induce an increaseo fP hOH content, which may be the consequence of both depolymerization and transformation of methoxy groups into phenol groups.
To assess the effect of the PhOH increase on the antioxidant properties, the EAE showingt he highest PhOH content (CH40s amples) were selected fort he furtherr adical-scavengingt ests. The corresponding EA-insoluble residues werea lso tested in view of the potentialuse of all fractions of PB1000.
Interesti nt he products of fractionation and [HMIM]Brtreatment as potential antioxidants

Antioxidant properties (AOP) of the phenolic monomers compared withreference antioxidants
The main phenolic monomers detectedi nt he EAE after treatment, namely acidolysis ketones (compounds A and B, Figure 5) and acetosyringone demethylated once or twice (compounds C and D,F igure 5), were tested for their DPPHC radical-scavengingc apacity according to at est previously used to compare the AOP of lignin modelsa nd technical lignins. [14] EC 50 (concentrationo ft ested sample necessary to reduce 50 % of the radicals) was used for this comparison.
A, B, C,a nd D showed higher AOP (EC 50 < 0.2 gL À1 )t han the other lignin model compounds tested (ferulic acid, coniferyl alcohol, and p-coumaryl alcohol;F igure 6). The lowest EC 50 was reachedw ith the twice-demethylated acetosyringone (compound D)-four times lower than that of ferulica cid (0.21 gL À1 ), which is ar eference for natural antioxidants. The radical-scavenging capacity of lignins,and phenolics in general, relies on the ability of phenols to trap free radicals in the mediuma fter the loss of ap roton [18] followed by the stabilization of this radicalb ym esomerism.T he best performance of compounds C and D is consistent with the presence of highly electron-withdrawing groups conjugated with the aromatic ring and with the presence of several phenols carriedb ya djacent carbons. [19] Interestingly,a ll tested compounds were competitive with a commercial antioxidant such as BHT (EC 50 = 0.19 gL À1 ). Thus, these compounds might be advantageously purified or synthesized and used as antioxidants for commercial applications. However,t oa void costly purification steps and to profit from the possible synergies between phenolic compounds, the AOP of the mixtures of products recovered from the treatments [EAE and EA extractionr esidues (EAR)] was considered.

AOP of EAE andEAR compared with PB1000 and its fractions
All fractions (F1-F3) and the products recovered from the CH40 IL treatment of these fractionse xhibited DPPHC radicalscavenging capacities similar to or higher than that of the PB1000 reference sample, according to the EC 50 data (Table 3).
Concerning the untreated fractions, the resultsa re in accordance with previous studies on the fractionation of other technical lignins (organosolv lignin BIOLIGNIN and LignoBoost Kraft lignins), showingthe generally higherradical-scavenging activity of the EA-soluble fraction compared with the other ones. [20] The IL treatment of the fractions led to the production of EAE with enhanced radical-scavenging activity compared with PB1000 and the fraction considered. This enhancement could be explained by the increased PhOH contenta fter treatment and the presence of the highly antioxidant phenolicmonomers in EAE. In addition, EAR also exhibited interestingly higher activity than their corresponding startingf ractions, except for F1. In the case of F3, the AOP of the insoluble residue was even twice that of the EAE, which confirmed that some new-formed phenol groups were carriedb ys ome compounds of higher molar mass insoluble in EA after treatment. According to the EC 50 ,F 2a ppeared as the most interesting fraction with respect to the production of antioxidantst hroughI Lt reatment because the radical-scavenginga ctivity of both the EA-soluble and -insolublef ractions after treatment were the best.

Benefit of combining fractionation and IL treatment
Owing to PB1000 fractionation and the subsequent IL treatment of the fractions, as et of functional antioxidant products with distinct characteristics wasg enerated.T he mapping of the different samples according to their antioxidant property (EC 50 ), molar mass distribution (M n , M w )a nd phenolic content (Figure 7) highlights that the EAE recovered after IL treatment of the fractions combined severala dvantages:l ower molar masses (M n % 700 gmol À1 ), higher PhOH content (6-12 mmolg À1 ), andh igherr adical-scavenginga ctivity( EC 50 = 0.10-0.17 gL À1 ).
Moreover,i ts hows that the molar mass differencesb etween the fractionsw ere levelled by the IL treatment. These characteristics make them good candidates as antioxidant ingredients for the formulation of plastics or cosmetics or for the synthesis of new biobased polymers and multifunctional molecules. Based on these results, an integrated process could be proposed,and its sustainability was discussed.

Towards as ustainable cascadeintegrated process
Ta ken together,t he results showed that IL treatment could be advantageously applied to lignin fractionsf or different purposes:p roduction of competitive antioxidant extracts, increaseo f lignin free phenol content, and partial dissolution of insoluble residues.I na ll cases, the demethylation and depolymerization induced by the treatmenti sb eneficial. To design ac ascade approach and preliminarily assess its sustainability,i ti sn ecessary to consider the technical performances of the productsa long with the recovery yields andsafety aspects.

Recoveryyields and technical gain
The process proposed in Figure 8a llows the productiono fa total of 67.7 %o fp roductsw ith enhanced antioxidant activity compared with PB1000,i ncluding 35.5 %E A-soluble oligomers enriched in PhOH.T oo ptimize the gain in AOP and PhOH,t he IL treatmentc onditions selected are the most drastic ones (CH40), which on the contrary do not favor the formation of soluble material. However,b ecause the insoluble products exhibit high antioxidant properties, they could be advantageously incorporated directly as fillers in plasticso rf unctionalized by graftingo ft he PhOH. In the first step of the process, af unctional fraction is directlyo btained from PB1000 by EA extraction. In the second step, the residue is submitted to aM EK extraction combined with IL treatment of the extract to recover both EA-soluble and -insoluble highly antioxidant products.I n the last stage, the MEK extraction residue undergoes aM eOH extraction combined with IL treatment to recover EA-soluble oligomers highly enriched in phenol groups togetherwith antioxidanti nsoluble compounds. Owing to its lowest contenti n lower-molar-mass phenolic extractives, the final residue of the process might find applications,for instance, as afiller in materials. [21] Figure 6. DPPHC radical-scavenging capacity of phenolic compoundsp roduced by IL treatment of PB1000 fractions (compounds A-D)c ompared with monolignols and acommercial synthetic antioxidant (BHT). Error < 1%.

Safety-and environment-related aspectsa nd further sustainability considerations
Sustainabilitya ssessment of the developed value chain was furthere xamined for physicochemical hazards and environmental impactsp otentially arisingf rom the key chemicals involved.W ith regard to the fire hazard, the two main critical points of the process in terms of safety are the use of organic solvents (EA, MEK, and MeOH) for the extraction steps and the use of an ew ionic liquid ([HMIM]Br). Of course, the well-estab-lished flammability of the mentioned solvents has to be handled from supply to process end-use, which leads in this particular context to limited issues according to mild operations conditions. Despite their flammability properties, such materials remains till recommendable as extractions olvents, with regard to HSE criteria taken into consideration with some prioritizing rules. [22] In particular, these substances are not mentioned in any list of the EU REACH regulation designating substances of particular concern owing to their adverseh ealth or environmental impacts. As ummary of resultsp ertaining to physical hazards potentially associated to the use of [HMIM]Br is provided in Ta ble 4. Moreover, EA and MeOH could potentially be substitutedw ith biobased recommendeda lternatives, bio-ethyl lactate and bioethanol, respectively (the latter is nowadays totally biosourced). MEK, if not so easily substituted, and despite some toxicityh azard, might at least be produced from biomass in the future,i nacost-effective way as an intermediate to biobutanol. [23] Although often considereda sg reen solvents, ILs remain controversial for safety issues [24] and chiefly requireacase-by-casea nalysisi nt he context of use because safety assessments are not based only on intrinsic properties but essentially on apparatusa nd testing procedure-dependentp roperties( flash point, thermals tability, metal corrosivity). [25] Ap reliminary assessment of the [HMIM]Br safety profile hasb een performed and is providedh erein (with technical details in the Supporting Information, S3), integrating both physicochemical hazards (fire, corrosivity to metals)a nd eco-toxicological properties of the IL, in line with REACH regulation safety data neededf or future registration. This studya lso provides new insights regarding [HMIM]Br thermals tability and fire behavior,s howing in particular ar emarkable resistance to ignition and af lame retardancy property. A first-order evaluation of the corrosivity potentialo f the neat IL has indicated that further investigation on the matter will be required for the appropriate selection of the reactor materialo wing to practical conditions for use of the IL.
Regardingt he environmentalp roperties of the IL, [26] the selected test battery includes the tests required by annex VII of the REACH regulation (substancem anufactured or imported into the European Union in quantities between 1a nd 10 tons per year) and immunotoxicity tests on the three-spined stickleback (Gasterosteus aculeatus). The results (Table5)s howed that [HMIM]Br has low the toxicity for Daphniam agna and Pseudokirchneriella subcapitata (EC 50 > 100 mg L À1 ). No biodegradation was observed in the manometric respirometryt est, leading us to conclude that [HMIM]Br is not rapidlybiodegradable. These results are consistent with those already obtainedo nc ompounds of the imidazolium family,w hich demonstrate that the toxicity increases It was concluded that 75 %o fc leanI Lc an be recovered at the end of the treatment andc ould be used again for similarr eactions for several cycles. [10] All these resultsc ontribute to as afeby-design biorefineryi nvolving the studied value chain.

Conclusions
The possibility to transfer the methyli midazolium bromide ([HMIM]Br)t reatment to technicall ignins was demonstrated, by using alkali grass lignin Protobind 1000 (PB1000) as ac ommercial reference sample. Safety assessment of [HMIM]Br highlighted the effective flame-retardant property of this ionic liquid (IL) and provided data on its eco-toxicological footprint,w hich does not departf rom that of other ILs of the imidazolium family and is useful for future REACH registration. The treatment induced both depolymerization andd emethylation of lignin, leading to the formation of additional free phenols. Based on theser esultsa nd after checking that similar effects were obtained with other commercial technical lignins including Kraft lignin (data not shown), an integrated cascade process combining IL treatment and solvent extractionsw as designed to optimize the recovery yield of ethyl acetate (EA)-soluble extracts with enhanced performance comparedw ith the technical lignin. The first step directly provides af unctional EA extract, whereas the second and third steps combine extractions with IL treatment to further improve functionalities for applicationsa sa ntioxidantso rb uildingb locks. Indeed,t he extracts consisted of free-phenol-rich oligomers with antioxidant properties favorably competing with ferulic acid and tert-butylated hydroxytoluene (BHT). Besides their high antiradical activity,t hese extracts have the advantages of standardized average molar masses( 872-937 gmol À1 ), low polydispersity (1.2-1.3), and   high free-phenol content(up to 11.9 mmolg À1 ), which provides opportunities for green biobased innovation in plastics and cosmeticsformulations.

Experimental Section
General materials and methods PB1000 was purchased from GreenValue LLC (USA). [27] [HMIM]Br was synthesized by following ap reviously reported procedure. [28] EA was purchased from Carlo Erba Reagents (France) and used as received. All other reagents as well as compounds A (ref 410659) and B (ref 796883), were purchased from Sigma-Aldrich Chemical Co. (USA) and were used as received. Thin-layer chromatography (TLC) experiments were performed with aluminum strips coated with Silica Gel 60 F 254 from Macherey-Nagel, revealed under UV light (254 nm), then in the presence of a5%w/w ethanolic solution of phosphomolybdic acid. Evaporations were conducted under reduced pressure at temperatures below 35 8Cu nless otherwise stated. Column chromatography (CC) was performed with an automated flash chromatography PuriFlash system and pre-packed INTERCHIM PF-30SI-HP (30 mms ilica gel) columns. 1 Ha nd 13 CNMR spectra were recorded in CD 3 OD at 400 or 100 MHz, respectively,w ith aB ruker Ascend 400 MHz instrument. Chemical shifts are reported in ppm relative to internal references (solvent signal).
Lignin fractionation and EA solubility tests EA extraction (1):P B1000 (1 kg) was fractionated by at hree-step sequential solvent extraction process, according to ap reviously published approach. [8] The following solvents were used sequentially in as emi-continuous process:E A, MEK, MEOH. Lignin was loaded in the first solvent in the glass column. After settlement, EA was pumped by HPLC pump at af low rate of 2-4 mL min À1 into the column. The solubilized fraction was collected at different heights in the column. The solvent was removed by vacuum evaporation, and the final solvent was removed by vacuum drying. The recovered solvent was reused in the process. When the concentration of solubilized lignin was very low,t he second solvent was added by the pump into the column. For each solvent, the procedure was repeated until, after three solvent extractions, the residual lignin fraction was collected from the column. This fraction was dried at amaximum temperature of 40 8C. EA extraction (2):P B1000 (6.5 g) was dissolved in EA (250 mL), and the mixture was stirred at room temperature for 30 min. The resulting solid residue was filtered, and the filtrate was recovered. The procedure was repeated nine times to obtain an exhaustive extraction. The combined EA extracts were concentrated under reduced pressure, and the final solvent was removed by vacuum drying. The residual lignin fraction was collected and dried at am aximum temperature of 40 8C. EA solubility test:L ignin solubility was determined gravimetrically by dispersing lignin (100 mg) in EA (10 mL, room temperature, 30 min), then centrifuging the suspension (20 8C, 20 min, 4000 g), and drying the solid residue at 40 8Cf or 48 h. Solubility was determined in duplicate, based on the amount of solid residue.

[HMIM]Brtreatments
For all treatments, the IL was vacuum-dried at room temperature before use. [HMIM]Br (2 g) were placed in an Anton Paar 30 mL reaction tube equipped with am agnetic stirrer.T he mixture was irradiated with P = 300 W, T max = 110 8C, ramp 30 s, hold 10 s, with full air cooling and stirring. At the end of the reaction, the solid residue was filtered and washed with water (20 mL) and EA (20 mL). The filtrate was recovered, the layers were separated, and the aqueous layer was extracted with EA (2 20 mL). The combined EA extracts were dried over MgSO 4 and concentrated under reduced pressure below 35 8C. The crude soluble mixture was analyzed by 31 PNMR spectroscopy,H PSEC, and thioacidolysis. CH treatment:T he sample (200 mg) and [HMIM]Br (2 g) were placed in an Ace pressure tube equipped with am agnetic stirrer under an inert atmosphere, then flushed with Ar.T he tube was closed, and the mixture was stirred in an oil bath at 110 8C. After 20 or 40 min, the solid residue was filtered, and the same downstream procedure as for MW irradiation was applied to the solid residue and the filtrate.
Synthetic procedure for compounds Ca nd D Acetosyringone (500 mg, 2.5 mmol) and [HMIM]Br (2.5 g, 6equiv.) were placed in an Anton Paar 30 mL reaction tube equipped with am agnetic stirrer.T he mixture was irradiated with P = 300 W, T max = 110 8C, ramp 30 s, hold 10 s, with full air cooling and stirring. At the end of the reaction, water (20 mL) and EA (20 mL) were added. The layers were separated, and the aqueous layer was extracted with EA (2 20 mL). The combined EA extracts were dried over MgSO 4  Chemical analysis of lignin and lignin-derived products GC-MS analysis:E As olutions (20 mL, 1mgmL À1 )p reviously dried with Na 2 SO 4 were silylated with bistrimethylsilyl-trifluoroacetamide (BSTFA, 100 mL) and GC-grade pyridine (10 mL). The silylation was completed within af ew minutes at room temperature. GC-MS analyses were performed in splitless mode with an Agilent 7890A GC coupled to an Agilent 5977B MS, with ap oly(dimethylsiloxane) column (30 m 0.25 mm;Rxi-5Sil, RESTEK), working in the temperature program mode from 70 to 330 8Ca t+ 30 8Cmin À1 ,o ver 20 min, with helium as the carrier gas. The chromatographic system was combined with aq uadrupole MS operating with electron-impact ionization (70 eV) and positive-mode detection, with a source at 230 8Ca nd an interface at 300 8C, and with a5 0-800 m/z scanning range. [29] Quantitative 31 PNMR spectroscopy and sample preparation:D erivatization of the samples with 2-chloro-4,4',5,5'-tetramethyl-1,3,2- dioxaphospholane (TMDP,S igma-Aldrich, France) was performed according to ar eported procedure. [30] Lignin samples (20 mg) were dissolved in am ixture of anhydrous pyridine and deuterated chloroform (400 mL, 1.6:1 v/v). Then, as olution (150 mL) containing cyclohexanol (6 mg mL À1 )a nd chromium(III) acetylacetonate (3.6 mg mL À1 ), which served as internal standard and relaxation reagent, respectively,a nd TMDP (75 mL) were added. NMR spectra were acquired without proton decoupling in CDCl 3 at 162 MHz, with aB ruker Ascend 400 MHz spectrometer.Atotal of 128 scans were acquired with ad elay time of 6s between two successive pulses. The spectra were processed by using To pspin 3.1. All chemical shifts were reported in ppm relative to the product of phosphorylated cyclohexanol (internal standard), which has been observed to give ad oublet at 145.1 ppm. The content in hydroxyl groups (in mmol g À1 )w as calculated on the basis of the integration of the phosphorylated cyclohexanol signal and by integration of the following spectral regions:a liphatic hydroxyls (150. 8 Thioacidolysis:T hioacidolysis of lignins (5 mg) was performed according to al iterature protocol, [31] by using heneicosane (C 21 H 44 , Fluka) as internal standard. Lignin-derived p-hydroxyphenyl (H), guaiacyl (G) and syringyl (S) thioacidolysis monomers were analyzed as their trimethylsilyl derivatives by GC-MS (Saturn 2100, Varian) equipped with ap oly(dimethylsiloxane) column (30 m 0.25 mm;S PB-1, Supelco) and by using the following heating program:4 0-180 8Ca t3 08Cmin À1 ,t hen 180-260 8Ca t28Cmin À1 .T he MS was an ion trap with an ionization energy of 70 eV and positive-mode detection. The determination of the thioethylated H, G, and Sm onomers was performed from ion chromatograms reconstructed at m/z = 239, 269 and 299, respectively,c ompared with the internal standard signal measured from the ion chromatogram reconstructed at m/z = (57 + 71 + 85). The molar yield of the detected thioethylated monomers was calculated on the basis of the Klason lignin content of the sample, determined according to a published procedure. [32] Molar mass distribution: M n and M w of the samples were estimated by HPSEC using as tyrene-divinylbenzene PL-gel column (Polymer Laboratories, 5 mm, 100 ,6 00 mm 7.5 mm inner diameter) with ap hotodiode array detector (Dionex Ultimate 3000 UV/Vis detector) set at 280 nm, and by using BHT-stabilized THF (1 mL min À1 ) as eluent. The samples were solubilized in THF and filtered through ap olytetrafluoroethylene membrane (0.45 mm) before injection. The molar mass averages were assessed from the apparent molar masses determined by ac alibration curve based on polyethylene oxide standards (Igepal, Aldrich) and lignin model dimers. [33] Assessmento fantioxidant properties Preparation of the solutions:Alignin sample was weighed into a 2mLm icrofuge tube, and the solvent (90:10 v/v dioxane/water mixture) was added to obtain concentrations between 0.1 and 0.5 mg mL À1 .T he dispersion was homogenized by using av ortex (Heidolph TOP-MIX 94323, Fisher Scientific Bioblock, Vaulx Milieu, France) for 30 sa t2 0000 Hz. The resulting solutions were tested for their radical-scavenging activity. Measurement of the free-radical-scavenging activity by DPPHC test:T he free-radical-scavenging activity of the samples was evaluated by measuring their reactivity toward the stable free radical DPPHC according to ap ublished method. [14] In aq uartz cuvette, the sample dioxane/water solution (77 mL) was added to 3mLo fa6 10 À5 mol L À1 DPPHC solution, prepared daily in absolute ethanol. The absorbance at 515 nm of each sample was monitored by using an UV/Visible double-beam spectrophotometer (Shigematsu Scientific Instrument, USA), until reaching ap lateau. Ab lank was prepared under the same conditions, by using 77 mLo ft he solvent instead of the sample solution. All kinetics were obtained from at least six solutions, prepared from three different lignin preparations. The kinetics of the disappearance of DPPHC were obtained by calculating at each time the difference between the absorbance of the blank solution and the absorbance of the sample. When the absorbance reached ap lateau, the percentage of residual DPPHC was calculated and plotted versus the concentration of soluble lignin in the sample tested. The concentration of antioxidant extract needed to reduce 50 %o ft he initial DPPHC (EC 50 ,w ith EC standing for efficient concentration) was determined from this linear curve.

Safetya ssessment of [HMIM]Br
The overall multicriteria safety analysis has been inspired by the global strategy developed by some of the authors of this manuscript for greener use of ILs in general, as exemplified by Eshetu et al. [34] in the case of energy storage in electrochemical devices. Fire hazard:E xamined by fire calorimetry testing based on the use of the most polyvalent fire calorimeter designated as the Fire Propagation Apparatus, following ISO 12136. See also the Supporting Information for further details on procedures and complementary details on achieved experimental data. Corrosivity screening tests:C arbon and stainless-steel specimens, partially immersed in plastic cells containing neat IL and IL added with 10 %w ater,f ollowing ah ome-made procedure (exposure in an oven regulated at 100 8Cf or 8days), with mass loss determined before and after exposure with ac alibrated balance, inspired by the procedure developed for IL corrosivity assessment by German ILs producer IO-LI-TEC. Ecotoxicity tests:E cotoxicity tests required by annex VII of the REACH regulation (substance manufactured or imported into the European Union in quantities between 1a nd 10 tons per year) have been performed. They are presented briefly in Ta ble 6. In addition to these regulatory tests, fish immunomarker tests were conducted to study possible long-term effects on aquatic ecosystems. The protocol is detailed in aprevious paper. [26]