Oxidation level and glycidyl ether structure determine thermal processability and thermomechanical properties of arabinoxylan-derived thermoplastics

: Herein we present arabinoxylan (AX)-based thermoplastics obtained by ring opening oxidation and subsequent reduction (dA-AX) combined with hydrophobization with three different glycidyl ethers [n-butyl (BuGE), isopropyl (iPrGE) and 2-ethylhexyl (EtHGE) glycidyl ether]. We also present the relationship of structural composition, thermal processing and thermomechanical properties. The BuGE and iPrGE etherified dA-AXs showed glass transition temperatures (T g ) far below their degradation temperatures and gave thermoplastic materials when compression-molded at 140˚C. The BuGE (3 mole) etherified dA-AX films at 19 and 31 % oxidation levels exclusively exhibit 244 % (±42) and 267 % (±72) elongation. In contrast, iPrGE-dA-AX samples with shorter and branched terminals in the side chains had maximum 60 % (±19) elongation. The dramatic difference in elongation is assumed to be due to the presence of longer alkoxide chains, higher molar substitution and dual T g for the BuGE samples. Such superior elongation of AX thermoplastic films and its relationship with molar substitution and T g has not been reported before.


INTRODUCTION
Sustainable resource supply and its efficient use for materials and chemicals is central to the success of the bioeconomy. Polysaccharides are promising candidates for this purpose owing to their renewability, abundance and functionality. Among polysaccharides, arabinoxylans (AX) are heteroxylans present in cereal grains and are characterized by backbones comprised of beta-1,4linked xylopyranosyl residues that are substituted mainly with arabinofuranosyl residues and less frequently with arabino-linked feruloyl esters and glucuronosyl residues, depending on the biological origin 1,2 . Both substitution degree and pattern, and extraction method affect structure imparting differential reactivity and physical properties to the polymer.
Material scientists are highly interested in exploiting structural functionality and physical properties of polysaccharides for antioxidants 3 in multifunctional food application or for film, coating, emulsion 4 , additives 5 , hydrogels in materials applications and for stretchable substrates in electronic devices [6][7][8] . Use of AX in films for packaging and stretchable electronics is restricted by brittleness and poor stretchability. Organic solvents can dissolve and plasticize biopolymers by reducing inter-and/or intra-molecular H-bonding and thus enable processing. However, upscaling of solvent processing for materials is unattractive due to long processing time and bulk solvent requirement. Therefore, commercially viable processing techniques are required to harness the potentials of polysaccharides to superior materials. Conversion of polysaccharides to materials is challenging because of poor thermal processability and inaccessible melt phase owing to higher melting (Tm) and/or glass transition temperature (Tg) compare to their degradation temperature (Td) [9][10][11] .
Strategies to render polysaccharides thermoplastic involves 1) use of external plasticizers, e.g. involatile glycerol and sorbitol or water etc.; 2) chemical modification. Use of external plasticizers, fillers and stabilizers enhance the processability, mechanical properties and environmental stability of polymers 12,13 . Thermal processability has previously been achieved for starch 14,15 and cellulose acetate 16 using external plasticizers and this method has been extensively explored. However, the addition of plasticizers can also cause unwanted side-effects, e.g. weak mechanical properties, plasticizer migration, recrystallization over time etc. 12,17 . Numerous studies have underscored the significance of chemical modification of starch [18][19][20][21][22] , proteins 23,24 and cellulose acetate [25][26][27] to improve processability and product performance. Despite this progress, the chemical modification strategy is rather less investigated for thermoplastic arabinoxylans in contrast with cellulose and starch. Thus, it is highly desirable to develop more means within this strategy.
Stronger intra-and intermolecular interactions through H-bonding and rigid backbone of the polymer are responsible for the poor thermal processability and make AX difficult to process via mature industrially available polymer thermal processing techniques such as compression molding, injection molding, extrusion etc. The glass transition temperature (Tg) depends on an interplay of intermolecular forces 28,29 , chain stiffness 30 , pendant groups 31,32 , molecular weight 33 .
Tuning these interactions/factors by chemical modification leads to decrease in Tg below the degradation temperature. In this context, it has been shown that the Tg of xylan decreased to 70 °C , 124 °C and 134 °C upon derivatization with propylene oxide 32 and butyl glycidyl ether 34 and upon oxidation 35 , respectively. Tg of glucomannan was also shown to decrease upon acylation and the glucomannan acylates could be thermo-processed 11 .
A recent study showed that grafting of n-butyl glycidyl ether (BuGE) on OH groups of sequentially periodate oxidized and reduced arabinoxylan (AX) and/or ring-opening polymerization of BuGE monomers on sequentially periodate oxidized and reduced AX as the initiation site of polymerization results into a thermally processable stretchable thermoplastic material 33 . Despite these studies, there remain moot questions a) how thermomechanical properties may relate to the structural composition (molar substitution) and thermal processing of the derivatized polymer, b) whether or not other reactants can be used to enable thermal processability and how their structure may relate to thermal processing and c) whether or not lowering Tg necessarily enables thermal processing. Further research advances are needed in modification strategy to make polysaccharides thermally processable thermoplastics and get tunable physiochemical properties. Chemical modification of polysaccharides using ester and glycidyl ether were shown to affect polymerization and/or processability 33,34,36,37 . A variety of factors, such as structure and composition of native AX and oxidation level can contribute to properties and processability [31][32][33] and are included in this study together with different etherification agents.
The objective of this study is to design and synthesize thermally processable AX with differential physical properties and to investigate the possibility of using different glycidyl ethers (etherification agents) after activating AX through successive periodate oxidation (at two levels) and reduction to have AX melt processable. This is done to probe the correlation between property and structure. Two oxidation levels were used as a mean to adjust the number of open carbohydrate units. Three etherification agents namely n-butyl (BuGE), isopropyl (iPrGE) and 2-ethylhexyl (EtHGE) glycidyl ether were judiciously selected that vary significantly from one another in structure, chemical and physical properties (see Table S1) and were varied in amount to shield and/or to use the reactive OH groups as coupling points on AX. This will help in adjusting intermolecular forces and molecular weight. This broad experimental design offers the possibility of tuning the properties of the obtained materials. The neat and functionalized AXs were analyzed for structure (IR and NMR), chemical composition (HPLC), thermal and mechanical properties (DSC for Tg, TGA, Tensile tests) to provide information about the correlation between properties and polymer structure.

Syntheses of activated arabinoxylan ethers
Syntheses of di-alcohol/activated arabinoxylans (dA-AX). The activated AX (dA-AX) was synthesized from AX through successive oxidative and reduction in Step 1 (Figure 2a). (Table S1 in SI), obtained oxidation levels and sample codes (sample coding detail in Fig. S2 in SI) are given in Table 1.

Information on etherification agents, reaction parameters
Periodate oxidation of arabinoxylan. Di-aldehyde arabinoxylan (dAl-AX) was prepared according to the reported literature 40,41 . Briefly, AX (4g dry basis, 30.30 mmol) was suspended in 10 mL 2-propanol and 165 mL water and the mixture was stirred at 50 °C for 1h. Different predetermined amounts of sodium meta periodate (vid Table 1) were dissolved in 25 mL water and the solutions were added to the flask giving 2 % concentration. The reaction was carried out Reduction of oxidized arabinoxylan. A premixed suspension of 2 g sodium borohydride NaBH4 (reducing agent) and 0.3 g of sodium mono phosphate NaH2PO4 (used as pH buffer to keep the pH constant) in 50 mL water was added slowly to the reaction mixture from the oxidation step giving 1.6 % concentration. The reaction was performed at room temperature (20 °C) for 4 h. The product   h. Then the mixture was cooled down and neutralized to pH 7.0 with 2M CH3COOH. The BuGE-dA-AX product was separated by centrifugation (while the iPrGE-dA-AX and EtHGE-dA-AX product were first precipitated in methanol and diethyl ether mix (1:0.7) and subsequently separated by centrifugation) and then dried under vacuum at room temperature for 14 h. The product yields are given in Table S3.
Film fabrication. Compression molding or solvent casting was used to fabricate films from neat AX and activated AX ethers. The processing method detail is given in Table S3.
Compression molding. The GE-dA-AX samples were cut into small pieces to get homogenous melt and compression molded into films using molds of 80 x 80 x 0.5 or 50 x 50 x 0.5 mm dimension (L x W x T) between two metal plates on a manual press set to 150 °C for melting. After melting the sample without any pressure between the plates for 3 min, the sample melt was compression molded under 50 kPa pressure at 140 °C for 3 min to get films.
Solution casting. This method is used for making films of the samples which were not compression moldable. The samples were dissolved either in H2O or ethanol water mix (3:1, ethanol 95 %) at 20 g/L and stirred at 50 °C for 1 h. The solution after cooling was poured into petri dish and dried at room temperature.
Characterization Oxidation level measurement. Oxidation level was indirectly determined from the periodate consumption (eq. 1 in SI) and the sugar content reduction (eq. 2 in SI). The periodate consumption in the oxidation reaction was determined by UV-Vis method 33 (eq. 1 in SI).
Carbohydrate composition and NMR spectroscopy. The neat and functionalized AXs were hydrolyzed by the 2-stage sulfuric acid hydrolysis according to the NREL procedure 38

RESULTS AND DISCUSSION
The synthesis strategy is depicted in Figure 2. To investigate structural composition-propertyprocessing correlation, different activated AX ethers were synthesized which varied in structure and composition. Firstly, thermal processability is determined followed by structural and molar mass characterizations and thermomechanical properties determination. How molar substitution of different glycidyl ethers in activated arabinoxylan ethers relates to thermal processing and thermomechanical properties is discussed in the end.
Thermal processing. Compression molding, injection molding and extrusion are the most common thermal processing techniques used for synthetic polymers. In the present work, compression molding (140 °C and 50 kPa) was used to determine whether or not the activated AX ethers were thermally processable. All activated AX ethers synthesized using BuGE and iPrGE were compression moldable materials and the representative images of films are shown in Figure   3. These results suggest the possibility of achieving thermal processability in arabinoxylan both using other reactant (i.e. iPrGE) and using different combinations of activation and amount of BuGE and iPrGE. The activated AX ethers synthesized using EtHGE could not be compression molded. This might be attributed to the low water solubility of EtHGE as compare to other etherification agents (Table   S1). Ethanol and water (30% and 70%) mix was used as solvent to increase the solubility of EtHGE, but the product was still not compression moldable. The BuGE5.0dA31%AX material yielded gluey and heterogenous film. The films were transparent except the one fabricated from iPrGE5.0dA31%AX material. The yellowness of the films is an indication of degradation of low molecular carbohydrates and/or proteins or non-structural compounds during compression molding 33,42,43 .
Chemical and structure characterization. The neat AX, dA-AX and activated AX ethers are structurally and chemically analyzed.
Film not available FTIR. To verify coupling of GEs on the activated AX, ATR FTIR was employed to compare the changes in functional groups. The spectrum of neat AX and the spectra of representative samples from dA-AX and GE-dA-AXs are in Figure 4 and the spectra of all the samples are in Fig. S3.
The OH stretching band centered at ∼3300 cm −1 shifted leftward for BuGE-dA-AX samples,  Table S3) which change with the degree of modification.  Intact carbohydrate substitution (ICS). ICS solely determines the substitution of the etherification agents on OH groups of intact carbohydrate in dA-AX polymer (Eq. 3 and Fig. S1 in SI) and the results are presented in Table 2. Up to 58 % (ICS-0.58) sugar units were BuGE etherified on dA19%-AX sample (Entry 2 in Table 2 ) contrary to 37 % (Entry 3 in  This has to do with the branched alkoxide chain in EHGE which may inherently create steric hindrance to reach on the available OH group reacting sites on di-alcohol AX samples compare to BuGE having a linear carbon chain (Figure 2). This argument together with low water solubility of EtHGE may not only be responsible for rendering almost nil MS for EtHGE-dA-AX samples and also in turn making these samples thermally incompressible. In order to find whether or not OH groups of the epoxide ring-opened of the glycidyl ether moieties may serve as initiation sites for the polymerization of glycidyl ethers to make polyether on the di-alcohol AX, 3 and 5 mole equivalence of BuGE, iPrGE and EtHGE were reacted under same etherification reaction conditions but without dA-AX. 1 H NMR (Fig. S4-S6) spectra did not provide any support to this possibility because the glycidyl ethers were converted to their corresponding di-ols. Polyethers were shown to formed from different glycidyl ethers using t-Bu-P4 (superbase) and n-butanol 47 .
However, a separate in-depth investigation is needed to prove this.

Molar mass distribution. Average molar masses and polydispersity indices of neat AX and GE-
dA-AX samples are presented in Table 2. Mn and Mw of neat AX are close to that of wheat AX (low viscosity) from a previous study using DMSO as solvent 48 , but lower than that of another study 35 . DMSO based SEC (used in this study) generally provide lower average molar masses compare to water-based SEC 48 . The molar masses could not obtain for di-alcohol samples because they were not soluble in DMSO. The molecular weights were decreased after oxidation due to degradation of the AX polymers 35 (Table S4). Likewise, the Tg of AX (Araf/Xylp 1.1) was shown to remain unchanged upon oxidation in a previous study 35 . Considering our results and previous study results, we hypothesize that Tg does not alter when the Araf/Xylp is >0.7 because arabinosyl units create steric hindrance limiting the oxidation of the xylan chains keeping the xylan backbone stiff and restrict the rotational freedom of the xylan backbone for being bulky pyranose rings. In contrast, Tg of xylan was found to decrease upon oxidation for xylan 35 due the pyranose ring-opening in the xylan backbone.
It becomes clear that Tg of the activated AX ethers (Tg ranged between -67 to 143 °C) reduced compared to their counterparts, i.e. neat AX and dA-AX samples (Tg ~189 °C) ( Table 3) Fig. S9 in SI). Remember that a new band at ∼740 cm −1 (CH2 rocking vibration) plus its counterpart at ∼1465 cm −1 (stronger CH2 vibration) is specific to BuGE-dA-AX samples was overserved and is indicative of long-chain linear aliphatic structure 45 (Figure 4 and Fig. S3b).
Previously coupling of n-butyl glycidyl ether and propylene oxide directly (without oxidation and reduction) on hardwood hemicellulose lowered Tg to 124 and 175 °C 34 indicating that subzero endotherm transition arise from the longer alkyl or alkoxide side chains. Contrary to two endotherms in the BuGE-dA-AX samples, iPrGE-dA-AX samples possess only one endotherm above zero, i.e. Tg 100, 69, 70, 54 °C ( Table 3 and Fig. S10-S11 a' and b'). This is also in agreement with a study on evaluating relations between thermal transition, structure and Table 3. DSC, tensile (Tensile test) and thermogravimetric (TGA) data of neat AX, dA AXs and GE-dA-AXs Sample DSC data [a] Tensile data [b] Thermogravimetric data [ [a] Tg was obtained from DSC second heating scan and data in parentheses are the standard deviations of two replicates. [b] data in parentheses are standard deviations of 4-5 replicates.
[c] data in parentheses are standard deviations of two replicates; E Young's modulus, σt max tensile stress, εb tensile strain at break; Tonset initial decomposition temperature, Tinflection temperature at the maximum decomposition rate, Tendset final decomposition temperature, Final residue (FR)-char content at 500 °C. * data not available (one replicate). -data not available as solvent casted films were too fragile/brittle. n.d. not determined because the film was too soft to be handled at room temperature.  solvent casted films were used for tensile tests. n.a. second endotherm was not observed in the DSC scans morphology of the conjugated polymers 49 , reinforces that longer alkyl/alkoxide side chains create a dominating subzero endotherm apart from the backbone endotherm. This necessitates the use of multi-techniques such as broadband dielectric spectroscopy, variable temperature ellipsometry and DMA etc. to probe the different transitions arising from the backbone and long side chains etc. as is the case here.
Thermal stability. Thermal stability of neat AX, dA-AX and GE-dA-AX samples was investigated using TGA and the TGA results are presented in Table 3. The TG thermograms (Fig.   S14a) reveal that all samples essentially possess one step decomposition profiles. The major decomposition (Tinflection) of BuGE5.0 dA31% AX sample take place at ∼237 °C with a shoulder at ∼200 °C ( Table 3 and Fig. S14c) contrary to the other samples which decompose between 280 and 300 °C ( Table 3, Fig. S14c and d). This has to do with gluey nature of this film upon compression molding. The di-alcohol AXs (dA-AXs) (dA19%AX and dA31%AX) exhibit a higher thermal stability compared to the neat AX, as the onset decomposition temperature of the former samples (269 °C) is higher than the latter (261 °C).
Thermal stability of the BuGE etherified di-alcohol AX (BuGE-dA-AX) samples decreases with the combination effect of increase in oxidation level and in mole equivalence of GE/ASU compared to their di-alcohol AX counterparts. This is reflected from the significant reduction in onset decomposition temperature (Tonset) ( Table 3). This trend remains also valid with iPrGE etherified di-alcohol AX (iPrGE-dA-AX) samples, but the reduction is less pronounced. The lower thermal stability of BuGE-dA-AX samples most probably occur from the alkoxide side chains' ability to depolymerize back to epoxide. As content of alkoxide side chains increases (Fig. S1c in   SI), the possibility to depolymerization increases and may speed up thermal decomposition. The chemical modifications of celluloses fibrils 50 and arabinoxylan 33,35 tend to lower the thermal decomposition temperature. The endset decomposition temperatures (Tendset) of GE-dA-AX samples (BuGE-dA-AX 320 °C and iPrGE-dA-AX 316 °C) are higher than dA-AX samples (301 and 311 °C) indicating the higher thermal stability ( Table 3), similar to a previous report 33 .
Mechanical properties. The mechanical properties of the neat AX and GE-dA-AX samples were examined by the tensile test. Young's modulus, maximum tensile stress and elongation at break data are presented in Table 3. As expected, the materials with sub-zero Tg, i.e. BuGE3.0 dA19% AX and BuGE3.0 dA31% AX are the most stretchable films with 244 (42) and 267 % (72) elongation at break (εb). The εb of iPrGE-dA-AX samples can hardly reach up to 50 % (Figure 5).
In a previous study, a maximum of 185 % (32) εb was achieved 33   BuGE3.0dA19%AX and BuGE3.0dA31%AX samples were very low at room temperature, i.e. above their sub-zero Tg ( Table 3). It is noteworthy that augmentation in tensile strain comes with loss in tensile stress and Young's modulus. Strength and toughness properties are mutually exclusive resulting into trade-off between these and this poses a challenge in designing materials 51

Structure-property-processing correlation
The qualitative structure-property relationship of the different samples i.e. the relationship between molar substitution, tensile strain and glass transition temperature, was evaluated and is shown in  Thermal processing was enabled when BuGE and iPrGE were introduced in the activated AXs. It also supports the fact that it is not possible to enable thermal processing without covalently bonded internal thermal plasticizer as it was shown for EtHGE. It is noteworthy though surprising that Tg of EtHGE3.0 dA19% AX is well lowered to 143 °C from 180 °C compare to its starting material, i.e. dA31% AX; however, does not necessarily enables thermal processability (Figure 7a, c and e). This result plausibly has to do with the degradation/chemical modification of the polymer during the etherification reactions. In addition, Tg is measured as the inflection point of the endotherm part ( Fig. S13a'). In fact, Tg is normally spread over a range of temperature. Conversely, it was argued that reducing Tg may render arabinoxylan and/or xylan capable of thermal processing 31,35 . Our results (Tg, molar substitution and IR data) show lowering Tg through chemical modification with elongating side chains in AX may be sufficient to enable thermal processing. Figure 6. Molar substitution-property (Tensile strain and Tg) correlation. a) Evolution of tensile strain and Tg against molar substitution of BuGE etherified dA-AX samples. Red square represents logical expected tensile strain of BuGE5.0dA31%AX material (experimental data unavailable because the film was soft to be free standing film.) Linear fit for tensile strain (R 2 =0.99999) and 2 nd order polynomial for Tg (R 2 =0.99988); b) Evolution of tensile strain and Tg against molar substitution of iPrGE etherified dA-AX samples. 2 nd order polynomial fit for tensile strain (R 2 =0.99416) and 2 nd order polynomial fit for Tg (R 2 =0.97871). The solid (green, for εb) and dashed (blue, for Tg) lines are used for visual guide for fit. Note that sub-zero endotherm calculated Tg values of BuGE etherified samples are used for plotting and curve fitting.   Figure 7. Thermal processability of the activated AX ethers; b) structures of the etherification agents; c) molar substitution (NMR determined) of the activated AX ethers; d) weight average molecular masses and polydispersity indices of the neat AX and the activated AX ethers; e) Glass transition temperatures (Tg) of the activated AX ethers; f) Tensile strain at break of the neat AX and the activated AX ethers. These plots were grouped to better compare and find out the relationship among them. Note that BuGE5.0 dA31% AX in Fig. f does not have tensile strain value because the film was too soft to be free standing.

CONCLUSIONS
Exploring the news ways to achieve thermoplasticity for polysaccharide-based polymer by chemically modification and understanding the relationship among their structure, processing and properties are crucial both for their current application areas such as packaging, films etc. and for their new application areas namely substrates for stretchable electronics. Our results show that the successful introduction of n-butyl (BuGE), isopropyl (iPrGE) glycidyl ethers in the activated arabinoxylans render thermal processability via compression molding. The thermal processability is found to be enabled for all evaluated combinations of activation (periodate oxidation and reduction) and etherification reaction conditions for n-butyl (BuGE), isopropyl (iPrGE) glycidyl ethers. It was also established that both the structure and composition of the native AX and modified AX, and the structure, chemical and physical properties of the etherification agents have strong implications on the thermal processability and final properties of the obtained materials.
The findings of this study contribute to the continued development of the chemical modification strategies to enable thermoplasticity and processability via commercial polymer techniques and provide new insights into structure-processing-property relationship of the AX-based thermoplastic materials. The findings are pertinent for the material scientists working on polysaccharide-based materials. We propose to use dynamic mechanical analysis and broadband dielectric spectroscopy to better understand the molecular mobility of the materials, i.e. main and secondary relaxations.
ASSOCIATED CONTENT

Supporting Information.
The supporting information is available in a separate pdf.
Experimental methods, characterization tools, spectroscopic data, thermomechanical data are included.

Notes
The authors declare no competing financial interest.