Structural disorganization of cereal, tuber and bean starches in aqueous ionic liquid at room temperature: role of starch granule surface structure

: The structural disorganization of different starches in a 1-ethyl-3-methylimidazolium acetate ([Emim][OAc])/water mixture (1:6 mol./mol.) at room temperature (25 °C) was studied. For normal cereal starches, which have pinholes randomly dispersed on the granule surface or only in the outermost annular region (wheat starch), the aqueous ionic liquid (IL) completely destroyed the granule structure within 1–1.5 h. Pea starch (PeS) granules with cracks were destroyed by the aqueous IL within 6 h. High-amylose maize starch (HAMS), as well as potato and purple yam starches (PoS and PYS), which have a dense and thick outer granule layer, were even more resistant to the action of the solvent. Structural disorganization was accompanied by increased viscosity and controlled the binding of water molecules with starch chains. From this study, we concluded that the surface characteristics of starch granule are an important factor affecting starch structural disorganization in an aqueous IL.


Introduction
There has been huge interest in developing biodegradable polymeric materials based on natural biopolymers due to the desire for achieving sustainability and taking advantage of the unique properties of these natural biopolymers (Seoud, Koschella, Fidale, Dorn, & Heinze, 2007). Starch is an abundant biopolymer widely used in both food and non-food applications due to its desirable properties such as low cost, biocompatibility, renewability, and biodegradability (Pérez & Bertoft, 2010). The shape (sphere, lobular, polyhedron, disk, or irregular tube) and size (the length of the long axis ranging from 0.5 to 100 μm) of starch granules vary largely with botanical source (Jane, et al., 1994;Wang & Copeland, 2015). Based on the X-ray diffraction (XRD) patterns, starches can be divided into A-, B-and C-types, which have different physicochemical properties (Lopez-Rubio, Flanagan, Gilbert, & Gidley, 2008;Wang, Yu, & Yu, 2008). C-type starches (for example Chinese yam starch) are shown to have both A-and B-type crystalline polymorphs, with the B-type polymorph in the center surrounded by the A-type polymorph Wang, et al., 2008;Wang, Yu, Yu, Chen, & Pang, 2007). The A-and B-type crystallites are considered to have the same double-helical configuration, but the way of packing and the water content in the crystallites are different (Pérez, Baldwin, & Gallant, 2009). The periphery of starch granules is the first barrier against hydrolysis and chemical actions (Nierle, Baya, Kersting, & D. Meyer, 1990). For maize and millet starches, there are pores distributed randomly on the granule surface, often in clusters and to different degrees, leading to high susceptibility to attack by enzymes or chemical reagents (Fannon, Hauber, & BeMiller, 1992). Large granules of wheat starch have pores along the equatorial groove, while there are wrinkles on the pea starch surface. For starches without pores on the granule surface, J o u r n a l P r e -p r o o f such as potato and yam starches, their surface is corroded much more evenly with the action of amylases or chemical reagents (Gallant, Derrien, Aurnaitre, Guilbot, & Massy, 1973).
The dissolution behavior of starch (rate and degree of granule disintegration) in ILs-based solvents J o u r n a l P r e -p r o o f are affected by several factors, but not limited to the solvent effects only (Zhao, et al., 2015). Firstly, the type and structure of cations and anions of ILs can highly affect starch dissolution. For example, ILs with alkyl imidazolium and [OAc − ] or [Cl − ] have proved to be most effective at dissolving polysaccharides including starch (Lappalainen, Kärkkäinen, & Lajunen, 2013;Tan & Macfarlane, 2010). The difference in size between anion and cation (Papanyan, Roth, Wittler, Reimann, & Ludwig, 2013), the alkyl chain length of cation (Khan, Taha, Ribeiro-Claro, Pinho, & Coutinho, 2014;, and the hydrogen-bonding capacity of anion  all determine the dissolving capability of ILs. Secondly, ILs, when mixed water, could be more powerful for starch dissolution than pure ILs. Previous studies showed that 1-ethyl-3-methylimidazolium acetate ([Emim][OAc])/water mixtures have a critical ratio for the most efficient dissolution of starch (Zhang, et al., 2017). The [Emim][OAc]/water ratio influences viscosity, the interaction between water and IL ions, and the mobility of water molecules, which all impact starch dissolution (Liu & Budtova, 2013;. However, the dissolution mechanism of starch in aqueous ILs at room temperature is still under debate. Recently, Wang et al. found that the damage of potato starch structure was less severe than maize starch in [Emim][OAc])/water mixtures (irrespective of mass ratio, 2:8, 4:6, 5:5, 6:4, 8:2, or 10:0), which could be due to its distinct surface morphology . Among six starches to be dissolved in 78 % NMMO, cereal starches (A-type) were found to dissolve more rapidly and completely compared with B-and C-type starches (Kogantia, Mitchell, MacNaughtan, Hill, & Foster, 2015).
To better understand the mechanism of starch dissolution in aqueous ILs, this work focuses on cereal J o u r n a l P r e -p r o o f (millet, maize, wheat, waxy maize, and high-amylose maize), tuber (potato and yam) and bean (pea) starches, which are widely used in food and non-food applications. Meanwhile, a [Emim][OAc]/water (1:6 mol./mol.) mixture was used as it is proved to be the most effective at dissolving maize starch Zhang, et al., 2017). The changes in morphology, crystalline structure, and rheological properties of these eight starches during room-temperature treatment by aqueous IL were monitored, based on which the underlying mechanisms are discussed.
By examining the relationship between starch architecture and dissolution behavior, our findings can improve our understanding of starch structures and provide insights into the processing and industrial application of starch-based materials.

Analysis of chemical composition and particle size distribution of starches
The moisture, protein and lipid contents of the isolated starches were determined according to AOAC Official Methods 925.10 (AOAC, 1925), 954.01 (AOAC, 1955), and 920.39 (AOAC, 1920), respectively. Amylose content was determined by the iodine-binding colorimetric method (Qi, et al., 2019). The phosphorous content was determined by inductively coupled plasma mass spectrometry (ICAPQ, Thermo Fisher Scientific, German). The particle size distribution of starch granules was measured using a Bettersize 2000B Intelligent Laser Particle Size Analyzer (Better size Instruments Ltd, China). For particle size measurement, starch was evenly dispersed in a sample cell filled with distilled water to attain an obscuration of 15-20%.

Sample preparation
The [Emim][OAc]/water (1:6 mol./mol.) mixture was pre-made based on the following equation: where nIL and nwater are the moles of the IL and water, respectively, and nIL:nwater = 1:6; mIL and mwater are the weights of the IL (95% purity) and water added, respectively; MIL and Mwater are the molar masses of the IL and water (170.21 g/mol and 18 g/mol), respectively.
Mixing [Emim][OAc] and water generates heat. After the aqueous IL prepared was cooled to room temperature, different starch suspensions (5 wt%), i.e. 0.5 g of starch (wet basis) mixed in 9.5 g of J o u r n a l P r e -p r o o f the aqueous IL, were prepared and stirred magnetically (300 rpm) at room temperature (25 °C) for 0.5 h, 1 h, 1.5 h, 6 h, and 24 h, after which five volumes of absolute ethanol was added. The samples were centrifuged at 4500 g for 15 min. The precipitates were washed using ethanol and centrifuged for three more times to remove residual [Emim] [OAc]. The obtained starches were dried in an oven at 30 °C overnight and ground to pass through a 100-mesh sieve.

Differential scanning calorimetry (DSC)
The thermal properties of starch samples were examined by a differential scanning calorimeter (200 F3, Netzsch, Germany) equipped with a thermal analysis data station. Starch samples (approximately 3 mg) were weighed accurately into 40 μL aluminum pans, where deionized water was added to obtain a starch/water ratio of 1:3 (w/w). The starch/water mixtures were left to stand overnight at room temperature before DSC measurement. The samples were heated from 20 °C to 135 °C at a heating rate of 10 °C/min and an empty aluminum pan was used as the reference. The onset temperatures (To), peak temperatures (Tp), conclusion temperatures (Tc), and melting enthalpy changes (ΔH) were calculated using Proteus analysis software.

X-ray diffraction (XRD)
The crystallinity of starch samples was analyzed using an X-ray diffractometer (D8 Advance, Bruker, Germany) operating at 40 kV and 40 mA. Starch samples were equilibrated over a saturated NaCl solution for 7 days before measurement. Starch powders were packed tightly in a round glass cell J o u r n a l P r e -p r o o f 9 and scanned from 5° to 35° (2θ) at a rate of 1 °/min and a step size of 0.02°. Relative crystallinity was calculated as the ratio of the crystalline area to the total area between 5° and 35° (2θ) using Eva software provided with the instrument.

Attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR)
The ATR-FTIR spectra for native and treated starch samples were determined using a Thermo Scientific Nicolet IS50 FTIR spectrometer (Thermo Fisher Scientific, USA). Starch samples (150 mg) were pressed into round tablets before being scanned in the range of 4000-400 cm −1 at a resolution of 4 cm −1 , and with an accumulation of 32 scans. The spectrum of an empty cell against air was recorded as the background. The ratio of absorbances at 1047/1022 cm −1 was calculated using OMNIC 9.2 software to estimate the short-range molecular order of starch samples.

Laser confocal micro-Raman spectroscopy
A Renishaw Invia Raman microscope system (Renishaw, Gloucestershire, UK) equipped with a Leica microscope (Leica Biosystems, Wetzlar, Germany) was used to obtain the Laser confocal micro-Raman (LCM-Raman) spectra for starch samples in the range of 3500-100 cm −1 at a resolution of 7 cm −1 . A 785 nm green diode laser source was used. The full width at half maximum (FWHM) of the band at 480 cm −1 , which is usually used to characterize the short-range molecular order of starch, was obtained using WIRE 2.0 software . The morphology of starch samples was examined using a scanning electron microscope (SU1510, Hitachi, Japan). Starch powders were scattered on the conductive tape stuck on aluminum stubs and then sputter-coated with gold before imaging at an accelerating voltage of 2 kV.

Light microscopy
A light microscope (DM-400 M-LED, Leica, Germany) was used to observe the changes in granule birefringence under normal and polarized lights. During the treatment of starches (5 wt%) by the aqueous IL in glass vials, a drop of the starch/IL/water suspension was quickly taken out to a glass slide and covered with a coverslip for imaging.

Rheological measurement
The viscosity of starch/IL/water suspensions was measured at 25 °C on an MCR 302 rheometer (Anton Par GmbH, Austria). The measuring system had a cone-plate geometry with a 4° angle and a 50 mm diameter. Before testing, starches were mixed with the aqueous IL in glass vials, and then the suspensions were applied onto the platform rapidly at different treatment time points. The shear rate was fixed at 300 s −1 for 2 h for measurement. To prevent any changes in water content in the suspensions, silicone oil was placed around the edge of the measuring cell and the insulation cover of the rheometer was applied.
J o u r n a l P r e -p r o o f

Low-field nuclear magnetic resonance (LF-NMR)
A 20 MHz NMI20-025 V-I NMR analyzer (Niumag Co., Ltd, Suzhou, China) was used to analyze hydrogen proton migration during treatment of starch with the aqueous IL. At different time points (0 h, 0.5 h, 1 h, 1.5 h, 6 h, and 24 h), 5 g of a starch/IL/water suspension (5 wt% starch in the aqueous IL) was placed in a 15 mm NMR glass tube and inserted into the NMR-probe. The T2 transverse relaxation time of samples was measured by a Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence at 32 °C with a τ-value of 300 μs between 90° and 180° pulses (Li, et al., 2020). Data from 8000 echoes were acquired based on 4 scan repetitions, and the repetition time between two successive scans was 7 s.

Statistical analysis
All analyses except for XRD were performed at least in triplicate, and the results are reported as mean values and standard deviations. Analysis of variance (ANOVA) followed by post-hoc Duncan's multiple range tests (p < 0.05) was conducted to determine the significant differences between mean values using SPSS 19.0 software (SPSS Inc. Chicago, IL, USA). Table 1 lists the compositions (moisture, protein, lipid, amylose and phosphorous contents) of the different starches. HAMS had the highest amylose content (69.5%) and WMS the lowest (0.7%). The amylose contents of PoS and PYS were 34.4% and 34.9% respectively while those of the rest J o u r n a l P r e -p r o o f starches were about 25%. The moisture, lipid and protein contents of eight starches were in the range of 7.5-12.6%, 0.1-0.6%, and 0.1-0.5%, respectively. Organophosphates were detected in all starches, with the greatest amount observed for potato starch mainly due to the presence of phosphate groups (Blennowa, Nielsen, Baunsgaard, Mikkelsen, & BEngelsen, 2002). The average granule size of the eight starches, as reflected by mean particle diameters (D4,3), was ranked in the order PoS (56.3 μm) > PYS > PeS > WS > WMS > NMS > HAMS > MiS (10.3 μm). Values are represented as mean ± standard deviation; Data followed by the different lower-case superscript letters in the same column were significantly different (p < 0.05).

Short-range molecular order of starch samples treated with aqueous [Emim][OAc]
The FTIR spectra for the different starch samples after treatment and the ratios of absorbances at 1047/1022 cm −1 (R1047/1022) are shown in Fig. S1 and Table 2  Values are represented mean ± standard deviation; Data followed by the different lower-case superscript letters in the same row were significantly different (p < 0.05). Values are represented as mean ± standard deviation; Data followed by the different lower-case superscript letters in the same row are significantly different (p < 0.05).     Values are represented as mean ± standard deviation; Data followed by the different lower-case superscript letters in the same column were significantly different (p < 0.05); ND, not determined.

Granule morphology of starch samples treated with aqueous [Emim][OAc]
The effect of the aqueous IL on starch granule morphology at room temperature was observed under light microscopy and SEM (Fig. 3). MiS, WMS, and NMS all had a small granule size, angular or irregular shapes, and a rough surface containing many randomly distributed indentations with pores in the center of indentation. Upon treatment of these three starches by the aqueous IL, the pores on the surface became larger and cracks were formed at 0.5 h and 1 h. The granules sagged inward until finally, the granules were broken completely at 1.5 h, exposing the internal cavity. Among these three starches, WMS was the most susceptible to treatment, for example, the granules lost its morphology completely and agglomerated into more dense lumps after 1 h of treatment. Under polarized light, the birefringent patterns (Maltese crosses) of WMS and MiS almost disappeared at 0.5 h, while at 1 h for NMS.
SEM images show that WS is composed of big discoid-like granules (A-type) and small spherical granules (B-type). The granule surface was mostly smooth and the pinholes could only be found at the equatorial circular depression of the discoid granules. At 0.5 h, the outermost annular region near the equatorial trough was most sensitive to solvent corrosion and the peeling of the granules occurred.
With prolonged treatment, this tangential degradation led to the cutting of the outermost annular region and further inwards tangential destruction until the stable central nucleus was exposed. By 1.5 h, the central nucleus was completely destroyed. Light microscopy images also show that the

Rheological properties of starch samples in aqueous [Emim][OAc]
The viscosity of the IL/water (1:6 mol./mol.) mixture at a constant shear rate (300 s −1 ) and J o u r n a l P r e -p r o o f temperature (25°C) was around 12 mPa·s, which remained essentially unchanged during the shearing process . The viscosity of starch (5 wt%) in the same aqueous IL as a function of treatment time was monitored (Fig. 4). During the first 2 h (Fig. 4A), the viscosity of the WMS/IL/water suspension rose rapidly from 14.5 mPa·s to 685.9 mPa·s at around 0.5 h of treatment, followed by a slow increase up to 1184.8 mPa·s at 2 h. Similar viscosity changes were also observed for MiS, NMS, and WS, although the increase occurred at a different rate and to a different degree.
The changes indicate that the granule structure was disrupted gradually during treatment, leading to the dissolution of starch and increased viscosity of the solutions. In contrast, there were no apparent changes for the suspensions of HAMS, PoS, PYS, and PeS over 2 h of treatment, meaning that the starch structure was not disrupted in the aqueous IL/water within the first 2 h (Zhang, et al., 2017). J o u r n a l P r e -p r o o f

T2 distribution of aqueous [Emim][OAc] and starch samples treated with aqueous [Emim][OAc]
The state and distribution of water in different systems can be characterized by low field-NMR, with a longer T2 relaxation time reflecting a higher degree of molecular mobility (Li, et al., 2020). From the results of the LF-NMR measurement (Fig. 5), there were three different proton groups from the relaxation signals, which can be ascribed to strongly bound water (0.01-1 ms), weakly bound water (1-100 ms), and free water (100-10000 ms) (Li, et al., 2015;Tang H, 2001). At room temperature, there were small changes in the T2 relaxation time of proton groups over time ( Fig. 5a and f  J o u r n a l P r e -p r o o f

Further discussion
In this study, the dissolution behavior of the eight starches in aqueous [Emim][OAc] was investigated by characterizing the structural changes of starch and the viscosity of starch suspensions during treatment. The results showed that the surface structure of starch granules is key to affecting the dissolution behavior of starch in an aqueous IL. Two different patterns for starch dissolution are proposed based on the results of granule morphology and thermal transition (Fig. 6). For MiS, WMS, NMS, and WS, the pinholes on the surface could allow the IL/water to penetrate into the granule easily, resulting in endo-corrosion of starch granules. As shown in Fig. 6A(a), MiS, WMS, and NMS, which have holes randomly distributed on the surface, only needed 1 h for the full destruction of the granules, while it took a longer time (1.5 h) for WS, which has holes concentrated in the outermost annular region while having other surface areas being smooth, to be fully destroyed as illustrated in Fig. 6A(b).
For MiS, WMS, NMS, and WS, less stable crystallites of the starch granule were first destroyed, followed by more stable structures, resulting in the endothermic transition shifting to a higher temperature. Further treatment would destroy all the helical structures and crystallites in starch as shown by DSC and XRD results. With the effective destruction of the starch granule structure, amylose was liberated from the granule, which significantly increased the viscosity of the solution and hindered the mobility of water molecules (Zhang, et al., 2017).
Starch granules (PoS, PYS, and PeS) with a relatively smooth surface and a thick outer layer are J o u r n a l P r e -p r o o f 33 more resistant to the attack by the aqueous IL. In this case, the corrosion of starch granules by the solvent started from the surface, leading to the formation of cracks (see Fig. 6B and SEM images).
As the action of the solvent occurred firstly on the stable surface structures, the endothermic transition occurred at a lower temperature and the destruction process was slow. These starches still showed an endothermic peak and crystallinity after 24 h of treatment. In particular, for large starch granules such as potato starch granules, this exo-corrosion from the surface could not reach the entire starch granule within 24 h, leaving some starch crystallites intact. Therefore, the mobility of water molecules was hindered and the change in viscosity due to the leaching of amylose chains was small.
For maize starches, amylose content has an impact on the destruction process with the aqueous IL as shown by microscopy, DSC and XRD results. HAMS were more rigid and resistant to the solvent than NMS and WMS. The different rate and extents of granule destruction for WMS, NMS, and HAMS with the aqueous IL confirm that starch with a higher amylose content is more stable in the aqueous IL with less fragmentation of the granule.
J o u r n a l P r e -p r o o f the mobility of water molecules in the system are influenced by the dissolution process. Our results clearly indicate that the destruction process of starch by an aqueous IL is predominantly determined by granule surface structure, meanwhile being strongly impacted by amylose content. Our findings reveal the structural evolution of starch granules during aqueous IL treatment, which could provide an insight into the design of dissolution and modification processes for different starches.

Declaration of Competing Interest
There is no conflict of interests to declare.