Study on spinnability of arabinoxylan extracted from barley husks

Valorisation of bio-based materials derived from agricultural and industrial side-streams or waste-streams is a basis of circular economy. However, the success of it depends on the full understanding of materials and finding their optimal way of processing. Barley husk is a side-stream waste material derived from the starch and ethanol production. This study is focused on the processability of the arabinoxylan extracted from barley husk using the electrospinning technique to produce thin xylan-poly(vinyl alcohol) fibres. As a comparison, lignin-free xylan of beech wood was used. The properties of spinning solutions and resulting nanofibrous mats were assessed by using rheological measurements, FTIR spectroscopy, scanning electron microscopy and contact angle measurements. It was found that solubility plays a crucial role in the spinnability of xylan extracts. Decrease in viscosity of arabinoxylan achieved by decreasing its concentration was found to improve the jet stability but at the same time, to reduce the diameter of spun fibre. Hydrophilicity of nanofibrous mats were strongly affected by the type of xylan and solvent used. The xylan-based nanofibres have specific properties that could be used for developing value-added applications, such as material for scaffolds for tissue engineering, drug delivery, medical implants, biosensors, wound dressing, water filtration and packaging or further process into carbon nanofibres.


Introduction
Circular economy has increased the usage of the biopolymers derived from various side streams and industrial waste materials both in agriculture and wood processing. The importance of circular Abstract Valorisation of bio-based materials derived from agricultural and industrial side-streams or waste-streams is a basis of circular economy. However, the success of it depends on the full understanding of materials and finding their optimal way of processing. Barley husk is a side-stream waste material derived from the starch and ethanol production. This study is focused on the processability of the arabinoxylan extracted from barley husk using the electrospinning technique to produce thin xylanpoly(vinyl alcohol) fibres. As a comparison, ligninfree xylan of beech wood was used. The properties of spinning solutions and resulting nanofibrous mats were assessed by using rheological measurements, FTIR spectroscopy, scanning electron microscopy and contact angle measurements. It was found that solubility plays a crucial role in the spinnability of xylan extracts. Decrease in viscosity of arabinoxylan achieved by decreasing its concentration was found to improve the jet stability but at the same time, to reduce the diameter of spun fibre. Hydrophilicity of nanofibrous mats were strongly affected by the type of xylan and solvent used. The xylan-based nanofibres have specific properties that could be used for developing value-added applications, such as material economy in the fight against microplastic and carbon dioxide emissions has been well acknowledged in recent years (Forni 2020). Biopolymers originated from industrial and agricultural side streams can be divided into three groups such as polysaccharides (cellulose, hemicellulose, starch, chitin), proteins (silk, collagen) and phenol-based polymer lignin (Moohan et al. 2019). Hemicelluloses are of specific interest being a second most abundant group of biopolymers after cellulose and representing 30-40% of agricultural residues (Sun et al. 2011), but compared to cellulose, they have been barely utilized as raw material (Peng et al. 2014). Among all types of hemicelluloses, xylans are the most abundant type, which can be further divided into subclasses: homoxylans, glucuronoxylans, (arabino)glucuronoxylans, arabinoxylans, (glucurono)arabinoxylans and heteroxylans (Ebringerová 2006). The fractionation and purification of some specific xylans from plant-based materials is complicated, costly, and time-consuming process (Ferrari et al. 2015). Use of crude (nonrefined) xylan-containing extracts wherever it is possible can significantly reduce the cost of product.
Increased interest in barley husks can be explained by significance of barley (Hordeum vulgare) as a crop plant. Finland produces barley in amount of 1.4 million tons per year (Luke 2020), whereas worldwide annual production is 160 million tons (FAO 2019). Husk is the outermost layer of the grain which forms 10-20% of the grain weight (Holopainen-Mantila 2015; Bhardwaj et al. 2020), and a side-stream waste material from the manufacture of starch and ethanol (Köhnke et al. 2009). Traditionally, barley husk has been peeled off and burned. Extraction method used can affect chemical composition of barley husk material (Höije et al. 2005(Höije et al. , 2006Börjesson 2018), however, mostly it consists of hemicelluloses, i.e. arabinoxylan, and cellulose with smaller amounts of lignin, protein, starch and fat. Recently, possibilities to exploit the fractions of barley husk in high-value product for different applications have been studied. Lähde et al. (2020) has described synthesis of nanostructured silicon carbide and graphene-like carbon from the silica extracted from barley husk. The transformation of cellulose fraction of barley husk into nanocellulose by sulphuric acid hydrolysis was studied by Börjesson et al. (2018). Höije et al. (2005) have published research on development of oxygen barrier coatings from arabinoxylans extracted of barley husks. Berglund et al. (2018) have studied the formation of composite aerogels from lignin-containing arabinoxylan and cellulose nanofibers which could be applied in the soft tissue engineering.
Electrospinning is a distinguished method for the preparation of nanofibres consisting of dissolved or melt polymers and biopolymers or their blends (Frenot and Chronakis 2003;Subbiah et al. 2005;Agarwal et al. 2009Agarwal et al. , 2016. In electrospinning process, a polymer solution or melt is placed into a syringe with a millimeter-size nozzle and is subjected to high electric field. Under the applied electrostatic force, the polymer is ejected from the nozzle and transported and deposited on a collector, which also serves as the ground for the electrical charges. Electrospun fibres have been widely used in several applications including scaffolds for tissue engineering, drug delivery, medical implants, biosensors, wound dressing (Rogina 2014), water filtration and packaging materials (Torres-Giner et al. 2018). Usually, the electrospun fibres have unique properties, such as high surface area-to-volume ratio, high pore interconnectivity and easy surface functionalization.
Several studies have been published on spinnability of various xylans. Krishnan et al. (2012) and Venugopal et al. (2013) have studied the beech-wood derived glucuronoxylan and Duan et al. (2019) have electrospun the arabinoglucuronoxylan extracted from straw fibres. In these studies, different solvents were used: 1 N sodium hydroxide and hot water, respectively. Most often, poly(vinyl alcohol) (PVA) and poly(ethylene oxide) (PEO) have been used to ensure the adequate entanglements and to facilitate the electrospinning of the xylan solutions. Due to the excellent properties of PVA, such as water-solubility, non-toxicity, biodegradability, it has been widely used in different nanofibre composite materials, for example, together with nanoclay (Ristolainen et al. 2006), chitosan (Park et al. 2009), aloe vera (Abdullah et al. 2014) and graphene oxide (Wang et al. 2012). Lignin has also been successfully electrospun with PVA (Ago et al. 2012). Additionally, the electrospun wheat xylan nanofibres (Duan et al. 2019) and also plantbased lignin (Kumar et al. 2019) have been processed further to fabricate carbon nanofibers.
Limited dispersion and solubilization of biopolymers are factors which make their electrospinning process challenging (Peresin and Rojas 2014). It has been shown in numerous studies (Ebringerová and Heinze 2000;Ebringerová 2006;Glasser et al. 2000;Roos et al. 2009) that solubility of xylans and their rheological behaviour depend on their plant origin, molecular structure, and method of extraction. Therefore, solubility is one of the subjects which will be addressed in the present study.
Objective of this work was to produce arabinoxylan-based nanofibres focusing first on the different dissolving methods (water/alkaline) of arabinoxylan extract and then on the variables affecting the spinnability. The concentration of arabinoxylan in the arabinoxylan/PVA solutions was varied and surfactant addition was studied also. The properties of spun fibres were explained in terms of the physico-chemical characteristics (e.g., viscosity, conductivity, surface tension) of the solutions. Special attention was paid to the role of lignin in spinning of arabinoxylan extract. The xylan from beechwood was used as a reference. The quality of fibres was assessed by scanning electron microscopy and their chemistry was studied by FTIR spectroscopy and contact angle measurements. Based on the knowledge of the authors, this is the first time that arabinoxylan of barley husk with PVA has been electrospun into nanofibres.

Preparation of spinning solutions
Two types of approach were used to dissolve xylan, referred in this study as Method A and B. In the Method A described by Duan et al. (2019), the xylan powder is dissolved in water under vigorous magnetic stirring at 80 °C for 24 h in tightly sealed flask as illustrated in Fig. 1. Thereafter, the 7 g of PVA powder was added into prepared solution and heated at 80 °C under stirring at 500 rpm for 3 h to complete the dissolution of PVA followed by the overnight stirring without heating. The 10 mmol of cationic surfactant, benzalkonium chloride (BAC) was added to the AXyl 1 _PVA 7 solution and thoroughly stirred overnight.
In Method B shown in Fig. 1, the arabinoxylan powder was dissolved in 1 N NaOH solution and mixed for 24 h at room temperature. The 7 g of PVA was added using the same procedure as in Method A. The composition of the spinning solutions, including type of xylan used, solvent, content of components and final spinning dope concentration have been summarized in Table 1. The final spinning dope concentration was calculated using formula: where m xylan , m water and m PVA are the masses of xylan, water and PVA added to solution.

Analyses applied to characterize spinning solutions
Rheological properties of the solutions were measured using Anton Paar Modular Compact Rheometer MCR 302 equipped with concentric cylinder CC27 system at temperature of 20 °C. The pHs and conductivities of the solutions were determined with the multi parameter meter (VWR MU 6100L). The surface tension was measured by the Du Nouy Ring method using a platinum ring and surface tension tester (Krüss GmbH, Hamburg, Germany).

Electrospinning
Electrospinning was performed using a modular bench top Spinbox instrument (Bioinicia, Valencia, Spain). The distance between nozzle and collector plate was 13 cm, and pumping speed was 0.3 mL/h. The current applied varied from 25 to 12 kV depending on the composition (conductivity) of the electrospun solution. The relative humidity inside the chamber varied from 27 to 41% depending on the day of the measurement.
SEM Hitachi SU3500 scanning electron microscope was applied to analysed gold-coated samples. Images were taken at the accelerated voltage of 10 kV using various magnifications: 1000 × and 10,000x. Diameters of spun fibres were analysed using ImageJ software. At least 100 fibres were measured to determine the mean diameter and standard deviations.

FT-IR analysis
A PerkinElmer FT-IR Spectrometer Frontier equipped with ATR sampling accessory was used to acquire spectra of electrospun mats. Spectra were recorded in the range of 4000 to 400 cm −1 at resolution 4 cm −1 . Each spectrum consisted of 10 scanning runs.

Contact angle measurements
The contact angles of the electrospun mats were measured using a sessile drop test method. The electrospun mat was fixed on the sample glass with double-sided tape in order to obtain as smooth surface as possible. 10 µL of deionised water was placed on the surface of the sample by syringe and three images were immediately taken by camera and handled using CAM 2008 software (KSV instruments, Finland). Out of the three images, the software calculates the contact angles of the water drop on the surface. At least 5 different locations were measured per each mat.

Results and discussion
Electrospinning of Xyl 10 _PVA 7 _w and AXyl 10 _ PVA 7 _w solutions prepared using method A In Method A, hot water (80 °C) was used to prepare 10 wt% xylan (Xyl and AXyl) solutions according to the procedure described by Duan et al. (2019). At 10 wt% concentration, the beechwood xylan powder was soluble in hot water at 80 °C, whereas the barley husk xylan powder formed a dispersion. Indeed, xylans isolated from biomass are known to have poor solubility in aqueous and aprotic solvents (Jain et al. 2001). PVA concentration of 7 wt% was chosen based on the preliminary electrospinning trials and easy spinnability at room temperature. Increase in PVA concentration to 10 wt% resulted in formation of thick gel in the syringe. In general, no disturbances in flow coming from nozzle (so-called, Taylor cone) were observed during spinning of the solution containing beechwood xylan and PVA mixture. The SEM images of the electrospun mat containing beechwood xylan (10 wt%) and PVA (7 wt%) have been shown in Fig. 2. The Xyl 10 _ PVA 7 _w electrospun mat had a homogeneous physical appearance. However, SEM images (Fig. 2, right) show that nanofibers were containing inclusions such as local thickenings (knots) and beads.
In the case of the AXyl 10 _PVA 7 _w solution prepared by Method A, separation of flow coming from the nozzle into white and dark brownish parts became evident immediately after the formation of Taylor cone. More continuous colourless part of the flow resulted in the white or slightly beige mat on the collector plate, which was interrupted by the dark brownish part as shown in Fig. 3. The brownish part flied to collector plate in the form of droplets. Some of these brownish droplets were falling under gravity force in the interspace between nozzle and collector. Several attempts were made to increase the electrostatic force, for example, by increasing voltage and by reducing distance between nozzle and collector plate to minimum. However, it was impossible to overcome the effect of the gravitational force.
The morphology of the AXyl 10 _PVA 7 _w mat was studied using SEM analysis and the images have been shown in Fig. 4. A typical morphology of brownish droplets is indicated with boundary line in Fig. 4 Fig. 2 SEM images of Xyl 10 _PVA 7 _w fibres: magnification 1000x (left) and 10,000x (right) Cellulose (left) showing as fibreless spot. Instead, the fibrous part of the AXyl 10 _PVA 7 _w electrospun mat was containing inclusions, such as local fiber thickenings and large fibre joints which are visible in Fig. 4 (right).
Electrospinning of arabinoxylan/PVA solutions prepared using method B During electrospinning of the AXyl 10 _PVA 7 _w solution, it was clear that phase separation in PVA-rich phase and hemicellulose-rich phase was taken place. Most probably, the insolubility of arabinoxylan extract in water was the main reason for flow instability. After a careful literature review on electrospinning of hemicelluloses, it was decided to use 1 N NaOH solution to dissolve arabinoxylan and decrease the concentration to 2.5 wt%. While the flow of AXyl 2.5 _PVA 7 _alk solution was improved compared to AXyl 10 _PVA 7 _w solution, the formation of droplets caused by flow instabilities was still noticeable. The electrospun fibres have been shown in Fig. 5. The thickest fibres of the study were spun from AXyl 2.5 _ PVA 7 _alk solution (Fig. 5, right).
In order to improve flow, the further decrease of arabinoxylan content and addition of the cationic surfactant were implemented. SEM images of the electrospun fibres have been shown in Figs. 6 and 7. As expected, both decrease in the arabinoxylan concentration and the addition of the cationic surfactant, benzalkonium chloride, resulted in improvement of the flow of the solution used in electrospinning and, simultaneously, the quality of the electrospun mat. Moreover, comparison of mats in Figs. 5, 6 and 7 and results shown in  Table 2 revealed that the electrospun mat containing cationic surfactant was composed by the thinner fibres and had the denser structure compared to the others. Nevertheless, the large fiber joints were not found in mats consisting of AXyl 1 _ PVA 7 _alk (Fig. 6, right) and AXyl 1 _PVA 7 _alk_ BAC (Fig. 7, right). None of the applied strategies, however, led to the complete disappearance of defects such as beads and local fibre thickenings. According to the results obtained by Lee et al. (2003) in their study on electrospinning of polystyrene fibres, beads in electrospun fibres are inevitable with below 15 wt% solutions of polystyrene. The morphology of beads has been found to depend on polymer concentration, applied voltage and nozzle-to-collector distance.

Electrospinning of AXyl 1 _PVA 7 solutions prepared by method A with and without surfactant
There was also an interest to see if by applying similar strategy, i.e., decrease in hemicellulose concentration (i.e., from originally used 10 wt% to 1 wt%) and addition of cationic surfactant will lead to improvements in the case of water-based solution of arabinoxylan. Analysis of SEM images of AXyl 1 _PVA 7 _w and AXyl 1 _PVA 7 _w_BAC mats shown in Fig. 8 and Fig. 9 revealed trends similar to ones found for alkalidissolved xylan with PVA (AXyl 1 _PVA 7 _alk) and with cationic surfactant (AXyl 1 _PVA 7 _alk_BAC) containing mats (shown in Figs. 6 and 7). The electrospun mats were more uniform compared to the initial AXyl 10 _PVA 7 _w mat (Fig. 4) and fibres were finer. Somehow, the exchange of alkali to water led to the denser structure of the resulting mat.

Size of electrospun fibres
Average diameters of the fibres in the structure of the electrospun mats were calculated using Image J software and the results have been presented in Table 2. Results shown in the Table 2 are average of calculations made for 100 fibres per each type of mat. The 7% aqueous solution of PVA spun at pumping rate of 0.3 mL/h, distance of 13 cm, voltage of 20 kV was used as a reference. From Table 2 it is evident that solution containing xylan extracted from beechwood, Xyl 10 _PVA 7 _w, resulted in finer fibres than solution containing arabinoxylan Operating conditions during spinning were identical for both types of xylan. Similarity between fibre size of PVA 7 _w and AXyl 10 _PVA 7 _w mats and flow (phase) separation observed in the case of the AXyl 10 _PVA 7 _w solution electrospinning can indicate that the fibre mat consisted mostly of PVA while arabinoxylan was disposed as droplets.
Chemical composition of the two phases and the results of the FTIR analysis of the mats will be discussed in the following section. Generally, the mats spun from the aqueous solutions had thinner fibres than their analogues spun from the alkaline solutions. Moreover, both the decrease in concentration of arabinoxylan from 2.5 to 1.0 wt% and use of cationic surfactant resulted in thinner fibres. More specifically, the 42% decline in the fibre diameter was calculated after addition of cationic surfactant to AXyl 1 _PVA 7 _w solution.
Effects of solution properties on spinnability and appearance of the fibres The appearance of the fibres collected depended on many factors (Jarusuwannapoom et al. 2005). One of them is the viscoelastic force, which depends on the concentration of the solution, average molecular weight of the polymer, viscosity, and surface tension of the solution. Another one is the gravitational force which depends on the density and the conductivity of the solution and electrostatic forces. Recognising this, conductivity, viscosity, and surface tension of solutions were measured, and results have been shown in Table 3. The solution conductivity is known to be one of the main parameters in the electrospinning process since the viscous polymer solution is stretched due to repulsion of the charges present on its surface. As expected, presence of sodium hydroxide drastically increased the conductivity of the solutions. Increased  conductivity resulted in decrease of voltage at which Taylor cone is formed (e.g., voltage decreased from 25 to 20 kV for the AXyl 1 _PVA 7 _w and AXyl 1 _ PVA 7 _alk solution, respectively). However, high conductivity is not always guarantee spinnability of the solution. For example, solution of AXyl 10 _PVA 7 _w had higher conductivity than AXyl 1 _PVA 7 _w solution but its spinnability was problematic due to extremely high viscosity. Viscosity curves of the solutions as a function of shear stress including 10 and 1 wt% AXyl_w, 2.5 and 1 wt% AXyl_alk and 7 wt% PVA, are shown in Fig. 10. Both xylans used, beechwood (Xyl) and barley husk (AXyl), contains of 4-O-methyl-glucuronic acids (Köhnke et al. 2009;Teleman et al. 2002), and, thus, they are negatively charged in aqueous solutions. Dissolution of the negatively charged groups is highly dependent on pH (Sjöström 1989). In the solutions studied (Table 3), these uronic acid groups are known to be fully dissociated at pH 6-8.5. At higher pH values when pH exceeds 12, also carbonyls and phenolic hydroxyl groups have been fully dissociated. According to manufacturer, arabinoxylan of barley husk contains lignin (Klason lignin 19.6%) bound to xylan, thus, it contains these phenolic hydroxyl groups. At pH 13, only small amount of hydroxyl groups in alcohols have reported to be dissociated (Sjöström 1989). Comparison of conductivity of AXyl 10 _PVA 7 _w and Xyl 10 _PVA 7 _w solutions shown in Table 3 indicates that Xyl contains higher charge (higher amount of 4-O-methyl glucuronic acid) compared to AXyl. In Table 3, increase of the pH up to 13 shows clearly that conductivity increase drastically showing increase of Na + an OH − ions and possible increase of the dissociating groups.
When cationic surfactant, i.e. benzalkonium chloride, is added, complexation of charges between xylan and surfactant is evident. The driving force for the complexation between oppositely charged polyelectrolytes and surfactants are electrostatic and hydrophobic forces (Holmberg et al. 2002). The head groups of the surfactant molecules are attracted by the charged polymer segments, and also the hydrophobic surfactant tails can interact with the hydrophobic backbone of the polyelectrolytes, such as lignin attached to xylan in AXyl. According to Table 3, surface tension of the solutions decreases from 41.9 to 37.5 (pH 13) and 49.0 to 37.0 (pH 6.5), respectively. Concentration of 10 mM for cationic surfactant (BAC) was used in the solutions, which is higher than the critical micelle concentration (CMC) of BAC reported, 3 mM (Turku and Sainio 2009). Schematic plot of the concentration dependence of the surface tension for mixed polymer solution described by Holmberg et al. (2002) shows that in the concentration of benzalkonium chloride used in the studies, CMC and critical association concentration CAC have been reach and micelles of surfactant have been formed, seen as lowered surface tension values of the solutions. Maximum effectiveness of benzalkonium chloride have been reported to exist between pH 4-10 (Handbook of Pharmaceutical Excipients 2006), indicating that the at pH 13 the quaternary ammonium compounds are not fully dissociated. Because PVA is also present in the solutions, the behavior of surfactant-polymer aggregations becomes more complex, which needs to be studied further. Viscosity as a function of shear rate expressed as log (η) v's log (γ) for different spinning and reference solutions were shown in Fig. 10. As indicated in Fig. 10, viscosities of AXyl(Xyl)_PVA solutions decreased with increasing shear rate, indicating a shear-thinning or non-Newtonian flow behaviour. This non-Newtonian behaviour was characteristic especially for solutions having high content of polysaccharides (xylans). Viscosities of the reference solutions including 1 wt% AXyl (both aqueous and alkaline), 7 wt% PVA, and AXyl 1 _PVA 7 solutions were less dependent on shear rate.
The power-law model was applied to evaluate relation between viscosity and shear rate. The viscosity of a power law fluid changes proportionally to K·γ n−1 , where K is the flow consistency index and n is the flow behaviour index (Mewis and Wagner 2011).
As shown in Table 4, the consistency index, K gradually decreases as the content of arabinoxylan decreases. This is seen especially in the case of reference arabinoxylan solutions (e.g., 8.068 v's 0.002 for 10 wt% and 1 wt% of AXyl_w, respectively), and in the case of AXyl_PVA solutions (e.g., 44.11 v's 0.157 for AXyl 10 _PVA 7 _w and AXyl 1 _PVA 7 _w, respectively). At the same time, the flow behaviour index, n, increases and becomes closer to 1. In more detail, the n ≈ 1 indicates near-Newtonian flow behaviour, while n < 1 indicates flow behaviour of pseudoplastic fluid. Among solutions studied, 7 wt% PVA and 1 wt% AXyl solutions showed behaviour close to Newtonian in wide range of shear stresses. As was discussed earlier, 7 wt% PVA solution was easily spinnable at ambient temperature. Thus, electrospinning solutions having viscosities close to viscosity of pure PVA 0.08 Pa·s (e.g., AXyl 1 _PVA 7 _w and AXyl 1 _PVA 7 _alk with and without surfactant) had good spinnability. According to results of rheological measurements not shown here, viscosity of the AXyl 10 _PVA 7 _w and AXyl 2.5 _PVA 7 _alk solutions made at higher temperatures (i.e., 50 and 80 °C) was drastically reduced. It was therefore concluded that decrease in the viscosity of the solution is considered as an efficient way to improve the spinnability.
The consistency index, K, is an indicator of the viscous nature of the system. The consistency index of the AXyl 10 _PVA 7 _w solution was significantly higher compared to the Xyl 10 _PVA 7 _w solution. Since equal amount of PVA was used in both solutions, the difference seen was connected to the nature of xylan. Both chemical composition and molecular mass of xylan can be a reason for that. Molar mass of the xylan extracted from beechwood (M w = 13.24 kg/ mol) is lower than molar mass of arabinoxylan extracted from barley husk (34.00 kg/mol). Chemical compositions of xylan and arabinoxylan extracts will be discussed later. Addition of PVA in AXyl solutions increased their viscosities (consistency indexes). This effect was independent on the concentration or alkalinity of the AXyl solution. Increase in the viscosity of twocomponent mixture may signify the polymer's interactions (or formation of larger particulates). Addition of PVA caused formation of hydrogen bonds between hydroxyl groups of PVA and of arabinoxylan and lignin molecules in arabinoxylan-lignin complex, and as a consequence, increased the viscosity of the solution. Lignin is supposed to be closely bound (in non-dissociated complex) to arabinoxylan in aqueous dispersion of barley husk extract. Indeed, a single step extraction, which commonly has been used to isolate xylan from barley husks (Glasser et al. 2000) resulted in xylan entrapped with lignin. According to the information available, 16-h extraction with 4 wt% sodium hydroxide solution at room temperature was applied to the dry cake of barley husks (Nygren 2009). Three types of linkages are known to exist between lignin and polysaccharides in lignin-carbohydrate complexes (LCC), i.e., benzyl ester-, benzyl ether-and glycosidic linkage, however only ester type is readily hydrolysed in alkali (Lai 1991).
The consistency index of the AXyl 1 _PVA 7 _w solution was lower compared to the AXyl 1 _PVA 7 _alk solution (e.g., 0.157 v's 0.316). The difference in viscosity could be explained by difference in chemical structure of arabinoxylan extract. Combination of alkali with heat treatment (i.e., at 80 °C for 3 h) used to prepare AXyl_PVA_alk solutions might lead to hydrolysis of the glycosidic linkages between lignin and arabinoxylan, which resulted in release of new phenolic hydroxyl groups of lignin (Lai 1991). Shifts associated with phenolic hydroxyl groups were identified in the FT-IR spectra of the electrospun AXyl_ PVA_alk mats. These new phenolic hydroxyl groups and the arabinoxylan released can participate into formation of additional hydrogen bonds with PVA. It was suggested to be a reason for increased viscosity of AXyl 1 _PVA 7 _alk solution.
Increase in viscosity (consistency index) was found also in the case of addition of surfactant to alkaline AXyl 1 _PVA 7 solution. A strong polymer-surfactant interaction could alter the rheological properties of the polymer solution, e.g., if the polymer can associate with the surfactants (Lin et al. 2004). In alkaline AXyl 1 _PVA 7 solution, an addition of cationic surfactant is assumed helping to form large associates between positively charged micelles of surfactant and negatively charged groups of arabinoxylan and lignin. It was very interesting to note a difference in effects caused by surfactant on viscosities of AXyl 1 _ PVA 7 _w and AXyl 1 _PVA 7 _alk solutions. Viscosity of AXyl 1 _PVA 7 _w_BAC solution was lower compared to the AXyl 1 _PVA 7 _w solution, in contrast to AXyl 1 _PVA 7 _alk_BAC solution. The difference can be explained by the difference in the dissolution state and by the charge of the arabinoxylan in pH-neutral and alkaline solutions. Most probably, at neutral pH there was not enough anionic groups in arabinoxylan/ lignin-PVA complex available to participate into electrostatic interactions with benzalkonium cations associated in micelles. Formation of free micelles nonbound electrostatically to the polymer backbone, can be responsible for decrease in viscosity in the case of AXyl 1 _PVA 7 _w_BAC solution. However, this question requires more detailed study.
Surface tension of the solution is the factor, which can affect the spinnability. According to the Taylor equation, there is a relationship between voltage applied and the surface tension of the solution: where V c is the critical electrical voltage, H is the separation between the tip of nozzle (capillary) and collector plate, L is the length of the capillary, R is the radius of the capillary and γ is the surface tension of the solution (Yao et al. 2003). Voltage required to obtain a stable Taylor cone in a solution with or without cationic surfactant at the same working distance and pumping feed rate was noticeable different. Addition of cationic surfactant clearly reduced the surface tension of AXyl 1 _PVA 7 _alk and AXyl 1 _PVA 7 _w on 10.5 and 24.5% (Table 3), respectively, and, as consequence, resulted in lower applied voltage.
There is a strong connection between polymer concentration and thickness of fibres. As was discussed earlier, the average thickness of AXyl 2.5 _PVA 7 _alk fibres was higher than thickness of AXyl 1 _PVA 7 _alk fibres (212.3 ± 55.7 v's 156.3 ± 43.0 nm) ( Table 2). The decrease in thickness of fibres with the decrease of polymer concentration was previously described by researchers using polystyrene (Uyar and Besenbacher 2008; Lee et al. 2003). Most often, high polymer concentration means higher viscosity. According to the studies on polystyrene, the high polymer concentrations is prerequisite for formation of stable jet during electrospinning. Stable jet should reduce the number of beads (defects) significantly. As stated earlier, the spinnability of AXyl_PVA 7 system with increase of arabinoxylan concentration to 10 wt% was impossible due to too high viscosity of the solution. Based on the results of this work, it is preferred to increase the concentration of polymers and use of higher temperatures to improve the spinnability of solution and to obtain defect-free fibres.

Chemical composition of nanofibrous mats
The FT-IR spectroscopy was used to assess the chemical composition of the electrospun mats, and the spectra have been presented in Fig. 11. In order to interpret the FT-IR spectra of the electrospun mats, the spectra of pure powders of PVA, arabinoxylan extracted from barley husk and xylan extracted from beechwood were measured and peak assignments have been shown in Table 5. Following conclusions were drawn based on data shown in Table 5. First, the main difference between spectra of xylan and arabinoxylan is absence of peaks at 1510 and 1595 cm −1 which can be assigned to the aromatic skeletal vibrations of lignin residue (Sun et al. 2005). Second, the common features for xylan and arabinoxylan are peaks associated with xylopyranose backbone: peak at 1035-1041 cm −1 , which is attributed to the C-O stretching, peaks at 2919 cm −1 and 3341-3343 cm −1 assigned to C-H and O-H stretching, respectively (Duan et al. 2019). The peak at 1035-1041 cm −1 in xylans has two shoulders at 1160 and 987 cm −1 of low-intensity. The peak  at 1603 cm −1 in xylan spectra can be explained by absorbed water. In general, the spectra of xylans and PVA are close to each other, for example the peaks 1377-1378, 1414-1418 and 1252-1238 cm −1 corresponding to the C-H and C-O stretching and bending vibrations in xylans and PVA molecules (Sun et al. 2005; Bary et al. 2019). The characteristic peak of PVA at 1086 cm −1 has shoulder at 1142 cm −1 , which determines polymer crystallinity (Mallapragada et al. 1996, Tretinnikov et al. 2012. There are differences between the xylan and PVA peaks obtained in region above 2000 cm −1 . The peaks at 3272 and 2940 cm −1 correspond to O-H and C-H stretching of PVA, respectively. Analysis of spectra of AXyl 10 _PVA 7 _w droplets and fibres showed that fibres had more pronounced peak at 1083 cm −1 that is closer to the characteristic peak of PVA, while droplets had more pronounced peak at 1039 cm −1 and no peak at 1083 cm −1 . Use of alkaline for arabinoxylan dissolution caused the formation of the peaks at 1434 and 1565-1574 cm −1 . They are considered as shifts from peaks with original positions 1421 and 1595 cm −1 which correspond to aromatic skeletal vibrations. It is known that in an alkaline solution, ionization of the phenolic hydroxyl groups is responsible for the shifts of the phenolic absorption bands of compounds (Morohoshi 1991). The arabinoxylan extract from barley husks contained 19.6% of Klason lignin. Lignin was thought to be a reason why AXyl formed the dispersion in water. Addition of surfactant to the AXyl 1 _PVA 7 _w and AXyl 1 _PVA 7 _alk solutions was appeared as a sharp peak at 2855 cm −1 , which is typical for benzalkonium chloride.

Hydrophilicity of electrospun mats
Contact angles of the electrospun mats as a function of composition has been shown in Table 6, where the mean values and standard deviations have been calculated out of 15 (3 × 5) measurements. Two-sample t-test and Tukey's simultaneous test were applied to determine if changes in contact angle from the PVA 7 _w was statistically significant at P-value less than 0.05.
The contact angle measurements are very sensitive to roughness and porosity of surface. The difference in microstructure of spun mats can be assessed from SEM images presented in earlier section. As can be seen, equal times (means equal fused volumes) of spinning was not warranty that fibre mats produced had similar microstructure (i.e., porosity). As can be seen from Table 6, the contact angle of Xyl 10 _PVA 7 _w mat was significantly lower than the contact angle of PVA 7 _w and AXyl 10 _PVA7_w mats. Hydrophilic nature of former mat can be explained by hydrophilicity of beechwood xylan. Presence of lignin in extract of arabinoxylan can explain its higher hydrophobicity. Decrease in content of arabinoxylan extract (from 10 to 1 wt%) and consequently in lignin concentration led to predictable decrease in hydrophobicity of mat. Usage of alkaline for arabinoxylan dissolution led to formation of highly hydrophilic mats. This hydrophilicity is in a good agreement with results of FTIR analysis which revealed ionization of the phenolic hydroxyl groups of lignin at alkaline conditions.

Conclusion
This study was focused on the spinnability of the arabinoxylan extracted from barley husk. Two different solvents were applied to produce the spinning solutions. It was found that alkaline conditions were required to solubilize the arabinoxylan extract. Among other components of the arabinoxylan extract, lignin was believed to be a major one to determine solubility. In contrast to the arabinoxylan extract, the xylan extracted from beechwood was easily soluble in hot water. The comparison of flow consistency indexes, K, of Xyl 10 _PVA 7 _w and AXyl 10 _PVA 7 _w solutions (5.322 v's 44.11) clearly showed that the former solution was less viscous. Moreover, the Xyl 10 _PVA 7 _w solution had Table 6 Contact angles of electrospun mats S corresponds to significant change (at probability level 0.05), NS corresponds to non-significant change (at probability level 0.05), and arrows show the trends (↑ corresponds to improvement and ↓ corresponds to decline)
In addition, lignin was found to be responsible for increased hydrophobicity nanofibrous mats; the values of contact angle were determined as 30.2° and 54.3° for Xy1 10 _PVA 7 _w and AXy1 10 _PVA 7 _w mats, respectively. In future studies, the lignin-rich solution can be used to improve the water resistance of the PVA-based nanofibres. However, when comparing the AXyl 1 _PVA 7 _w and AXyl 1 _PVA 7 _alk mats having contact angles 48.1° and 18°, respectively, it became obvious that solvent nature can also drastically change the hydrophilicity of the mats.
The rheological measurements demonstrated clearly that the type of solvent was the key factor affecting the interaction between polymers (i.e., PVA and arabinoxylan/lignin) and surfactants. However, more work is needed to understand the interactions between molecules and molecular structure of the nanofibrous mats. The 75-200 nm fibres electrospun in this study have potential applications in the fields where high aspect ratio (or high surface area) of fibres is required (e.g., air filtration).