Valorization of Rice Straw into Cellulose Microﬁbers for the Reinforcement of Thermoplastic Corn Starch Films

: In the present study, agro-food waste derived rice straw (RS) was valorized into cellulose microﬁbers (CMFs) using a green process of combined ultrasound and heating treatments and were thereafter used to improve the physical properties of thermoplastic starch ﬁlms (TPS). Mechanical deﬁbrillation of the ﬁbers gave rise to CMFs with cumulative frequencies of length and diameters below 200 and 5–15 µ m, respectively. The resultant CMFs were successfully incorporated at, 1, 3, and 5 wt% into TPS by melt mixing and also starch was subjected to dry heating (DH) modiﬁcation to yield TPS modiﬁed by dry heating (TPSDH). The resultant materials were ﬁnally shaped into ﬁlms by thermo-compression and characterized. It was observed that both DH modiﬁcation and ﬁber incorporation at 3 and 5 wt% loadings interfered with the starch gelatinization, leading to non-gelatinized starch granules in the biopolymer matrix. Thermo-compressed ﬁlms prepared with both types of starches and reinforced with 3 wt% CMFs were more rigid (percentage increases of ~215% for TPS and ~207% for the TPSDH), more resistant to break (~100% for TPS and ~60% for TPSDH), but also less extensible (~53% for TPS and ~78% for TPSDH). The incorporation of CMFs into the TPS matrix at the highest contents also promoted a decrease in water vapor (~15%) and oxygen permeabilities (~30%). Finally, all the TPS composite ﬁlms showed low changes in terms of optical properties and equilibrium moisture, being less soluble in water than the TPSDH ﬁlms. performance of biodegradable food packaging materials. Likewise, this study further conﬁrms that the DH process of starch represents a green physical method to enhance the water barrier capacity and tensile strength of starch ﬁlms. Future studies will deal with the hydrolysis resistance of these microcomposite ﬁlms and their practical application as packaging materials in food preservation.


Introduction
Today, the proper management of agro-industrial wastes is an essential socioeconomic and environmental issue. Most of these residues or by-products, generated in large quantities globally, are currently discarded or burned in the harvest fields [1]. In this sense, the use and valorization of agro-industrial waste into added-value materials are an emerging challenge that seeks to minimize environmental problems and promote the correct residue disposal. Agro-food and industrial waste derived biomass is essentially composed of lignocellulosic fractions, which are constituent biopolymers, such as cellulose, hemicellulose, and lignin [2,3]. Due to its renewability and biodegradability, the use of lignocellulosic-rich wastes can yield environmentally friendly materials through cost-effective processes with low environmental impact [4,5]. Several authors have reported the use of plant residues to obtain lignocellulosic fractions with potential application for biodiesel production [6], catalysts for biodiesel synthesis [7], antioxidant and antimicrobial materials [8,9] as well as reinforcing agents [10,11]. Particularly, cellulose and its derivatives extracted from biomass, such as cellulose microfibers (CMFs) and cellulose nanofibers, are highlighted renewable materials at the micro-and nanoscale, respectively, with low density, good compatibility, fibers were dried under vacuum (50 ± 2 • C, 0.5 mmbar) for 16 h and milled (3 cycles of 90 s each, IKA, model M20, Staufen, Germany), and sieved to <0.5 mm particle size.

Extraction of Cellulose Microfiber
A combination of ultrasound and heating treatments are known to lead to a certain degree of leaching out of hemicellulose and lignin present into the lignocellulosic RS matrix [8]. Based on this, CMFs were extracted according to our previous methodology [35]. Briefly, an aqueous dispersion of RS particles with 5% (w/v) solids was sonicated for 30 min at 25 • C (maintaining the temperature in an ice bath) using an ultrasonic homogenizer (750 W power, 20 kHz frequency, continuous mode) equipped with a high-intensity probe (Vibra CellTM VCX750, Sonics & Material, Newton, MA, USA). After sonication, the RS dispersion was heated in a reflux heating apparatus at 100 • C for 1 h. Then, the plant dispersion was filtered using a qualitative filter (Filter Lab, Vidra Foc, Barcelona, Spain) and the solid fraction was dried at 40 • C for 48 h. Afterwards, the bleaching step of the cellulosic material was carried out by mixing the dry powder residue and the bleaching solution at 5% (w/v). The bleaching solution consisted of equal parts of sodium hypochlorite solution (1.7%, w/v), acetate buffer solution (2 N), and distilled water. The dispersion was treated under reflux heating for 4 cycles of 4 h each. The supernatant was filtered and the residue was washed with distilled water several times to remove the residual bleach solution at each stage. Then, the bleached solid fraction was dried at 35 • C for 48 h and milled using a milling machine (pulses of 2 s for 20 min, model M20, IKA Werke GmbH & Co. KG, Staufen, Germany) to obtain the so-called CMFs.

Dry Heating Treatment
Prior to melt mixing, corn starch was submitted to a DH pre-treatment, according to Maniglia et al. [33]. To this end, 50 g of starch were spread (1 mm thick layer) and compacted in an aluminum foil envelope. Then, the packaged sample was heated in a hot air oven (J.P. Selecta, S.A., Barcelona, Spain) at 130 • C for 4 h. Thereafter, the resultant starch was placed in a dark bottle and stored in a desiccator containing P 2 O 5 at 6 ± 2 • C until further use.

Film Preparation
TPS and TPSDH films were prepared by melt blending and subsequent compression molding using glycerol as plasticizer 30 wt% with respect to the total starch mass. Green composites were obtained by addition of CMFs in concentrations of 1, 3, and 5 wt% in relation to the starch content based on the optimization performed in our previous study [36]. Table 1 summarizes the different sample compositions used to produce the films. The film components were previously dry mixed in a beaker and then melt-blended using an internal mini-mixer (HAAKETM PolyLabTM QC, Thermo Fisher Scientific, Karlsruhe, Germany) at 130 • C and 50 rpm for 10 min. The obtained dough was cold milled in the IKA's model and pre-conditioned at 25 • C and 53% RH using an Mg(NO 3 ) 2 over-saturated solution for one week.
The obtained doughs were compression molded to yield films in a hydraulic press (Model LP20, Labtech Engineering, Samut Prakan, Thailand). To this end, approximately 4 g of each dough was placed between two Teflon sheets and compressed in the hot plates. Control films (TPS) and those incorporating 1 wt% (TPS1), 3 wt% (TPS3), and 5 wt% (TPS5) of CMFs were preheated for 3 min at 160 • C in the plate press, then compressed at 160 • C for 2 min at 50 bars, followed by a second compression at 160 • C for 4 min at 130 bars, An optical microscope (Optika Microscope B-350, OPTIKA S.r.l., Ponteranica, Italy) equipped with a camera (Optikam B2) was used to evaluate the morphology of the untreated RS and the resultant CMF samples before and after the milling process. To this end, the samples were extended with a water drop on the holder glass plate and the images were taken at 10× and 40× magnification. Morphogeometric properties of CMFs, expressed as cumulative distributions, were obtained by measuring the particle length (l) in different regions of the sample, as shown in the imaging procedure of Figure 1. The Optika Vision Lite program was used to obtain a minimum of 60 measurements.
final cooling to 80 °C. All films were conditioned at 25 °C and 53% RH (M week before characterizations.  The morphologies of RS, CMFs, and the cross-sections of the cryo-fr tained by immersion in liquid nitrogen were evaluated using a Field Em Electron Microscope (FESEM) equipped with focused ion gun (AURIGA Oxford Instruments). The conditioned samples (P2O5 at 25 ± 2 °C for 1 we with a platinum layer (EM MED020 sputter coater, Leica Biosystems, B and the images were taken at 2.0 kV acceleration voltage. The SmartTiff p The morphologies of RS, CMFs, and the cross-sections of the cryo-fractured films obtained by immersion in liquid nitrogen were evaluated using a Field Emission Scanning Electron Microscope (FESEM) equipped with focused ion gun (AURIGA Compact, Zeiss, Oxford Instruments). The conditioned samples (P 2 O 5 at 25 ± 2 • C for 1 week) were covered with a platinum layer (EM MED020 sputter coater, Leica Biosystems, Barcelona, Spain), and the images were taken at 2.0 kV acceleration voltage. The SmartTiff program (version 2, Zeiss, Oxford Instruments, Abingdon, UK) was used to measure the particle width (w) a minimum of 60 times.
In which A 500 and l are the absorbance at 500 nm and the film thickness, respectively.

Equilibrium Moisture Content
The equilibrium moisture content of the films was determined by a gravimetric method. For this, film samples sizing 3 cm × 3 cm that were previously conditioned (25 • C and 53% RH for two weeks) were dried at 60 • C for 24 h using a drying oven. Thereafter, the dried samples were placed in a dissector containing P 2 O 5 at 25 • C for two weeks to eliminate bonded water. Moisture content was determined from the total mass loss of the conditioned film during the drying process. The measurements were carried out in triplicate.

Water Solubility
Water solubility was determined based on the Talón et al. [38] methodology. A known mass of dry film sample (2 cm × 2 cm), conditioned in P 2 O 5 , were placed on a mesh, and immersed in a crucible with distilled water at 25 • C for 24 h. Afterwards, the samplecontaining system was dried in an oven (J.P. Selecta, S.A.) at 60 • C for 48 h and, afterwards, conditioned in a desiccator with P 2 O 5 at 25 • C for 1 week. The mass of residual film was weighed and compared with the initial mass of dry film. The measurements were carried out in triplicate and the results were expressed as g of solubilized film per 100 g film.

Barrier Measurements
Water vapor permeability (WVP) of the films was determined according to ASTM E96/E96M (ASTM, 2005) gravimetric methodology following the modifications proposed by McHUGH [39]. Film samples (Ø = 3.5 cm) were placed and sealed in Payne permeability cups filled with 5 mL of distilled water (100% RH). Then, the cups were placed into desiccators containing Mg(NO 3 ) 2 over-saturated solution (53% RH) and weighed periodically (ME36S, Sartorius, ± 0.00001 g, Fisher Scientific, Hampton, NH, USA) every 1.5 h for 25 h. For each treatment, the WVP was calculated considering the water vapor transmission rate (WVTR), which was determined from the slope of the weight loss vs. time. The measurements were performed in triplicate.
Oxygen permeability (OP) of the films was determined using an Ox-Tran equipment (Model 1/50, Mocon, Minneapolis, MN, USA) at 25 • C and 53% of RH, according to ASTM D3985-05 (ASTM, 2010). The used area of films was 50 cm 2 and the oxygen transmission rate (OTR) was obtained every 15 min until equilibrium was reached. The measurements were carried out in triplicate.

Mechanical Characterization
The tensile properties of the films were obtained according to ASTM D882 (ASTM, 2012). A universal testing machine (Stable Micro Systems, TA.XT plus, Stable Micro Systems, Godalming, UK) was used to determine elongation at break (EB), tensile strength at break (TS), and elastic modulus (EM). Film samples with dimensions of 25 mm × 10 mm were grabbed by two grips initially separated by 50 mm and stretched at a crosshead speed of 50 mm.min −1 . Eight samples were evaluated for each formulation. Before the analysis, the films were conditioned at 53% RH (Mg(NO 3 ) 2 ) and 25 • C for two weeks.

Thermal Analysis
The thermal stability of the films was determined by thermogravimetric analysis (TGA) using a thermogravimetric analyzer (TGA 1 Star e System analyzer, Mettler-Toledo GmbH, Greifensee, Switzerland). Film samples, with a weight of 3-5 mg, were heated from 25 to 600 • C at a heating rate of 20 • C.min −1 under nitrogen atmosphere (10 mL/min). For each thermal event, the initial (T onset ) and final (T final ) degradation temperatures, temperature of maximum degradation rate (T peak ), and the residual mass were determined by analyzing the TGA curves and their first derivative thermogravimetry (DTG) curves. The measurements were carried out in triplicate.
The phase transitions of the samples were investigated by differential scanning calorimeter (DSC) according to the method described by Collazo-Bigliardi et al. [40] with a DSC 1 Star e System analyzer (Mettler-Toledo GmbH) operating under a nitrogen atmosphere (10 mL.min −1 ). Samples of about 5-6 mg were weighted in aluminum pans and heated from 25 to 160 • C, cooled to 25 • C, and then heated (second heating step) to 160 • C at 10 • C.min −1 using, in all cases, rates of 10 • C/min.

Statistical Analysis
The experimental data were submitted to multifactorial analysis of variance (ANOVA) at a confidence level of 95% using Minitab Statistical Program (version 17). Tukey's studentized range (HSD) test, considering the least significant difference of 5%, was applied to determine the influence of CMFs on the properties of the starch-based films. Figure 2 shows the FESEM micrographs of the untreated ground RS, the lignocellulosic residue after the combined ultrasound-heating treatment, the bleached cellulose fibers, and the resultant CMFs obtained after the milling step. It can be observed that the untreated RS (Figure 2a) presented heterogeneous particle sizes, with a predominance of rod-like shape, as seen in the magnified FESEM image. The RS particles showed mean sizes of width and length in the 20-60 µm and 16-270 µm ranges, respectively. Prior to the RS fiber extraction, the grinding step was applied to improve the effectiveness of the combined ultrasound-heating method and the bleaching treatment. According to Chen et al. [2], RS has a strong crystalline structure, making the extraction and purification of cellulose fibers difficult. Likewise, RS presents a high proportion of non-fibrous cells, such as parenchyma, epidermis, and vessel cells [41]. The traditional alkaline pre-treatment, which aims to eliminate hemicellulose and lignin [42], was replaced by a more environmentally friendly aqueous treatment based on the application of ultrasound, followed by reflux heating. Figure 2b reveals that the application of ultrasound and heating promoted a certain degree of fibrillar plane de-bonding and distorted the RS particles, as previously reported [8]. The acoustic cavitation produced by the ultrasound waves, combined with the thermal erosion generated by reflux heating, promoted the disruption of the primary structure of RS and improved the leaching out of non-cellulosic components from the cellulosic matrix [43][44][45].

Morphological Characterization of CMFs
combined ultrasound-heating method and the bleaching treatment. According to Chen e al. [2], RS has a strong crystalline structure, making the extraction and purification of cel lulose fibers difficult. Likewise, RS presents a high proportion of non-fibrous cells, suc as parenchyma, epidermis, and vessel cells [41]. The traditional alkaline pre-treatmen which aims to eliminate hemicellulose and lignin [42], was replaced by a more environ mentally friendly aqueous treatment based on the application of ultrasound, followed b reflux heating. Figure 2b reveals that the application of ultrasound and heating promote a certain degree of fibrillar plane de-bonding and distorted the RS particles, as previousl reported [8]. The acoustic cavitation produced by the ultrasound waves, combined wit the thermal erosion generated by reflux heating, promoted the disruption of the primar structure of RS and improved the leaching out of non-cellulosic components from the ce lulosic matrix [43][44][45]. As shown in Figure 2c, the bleached cellulose fibers exhibited a swollen, rough, and unregular appearance. The removal of non-fibrous components from the lignocellulosi matrix, such as hemicellulose, lignin, and waxes, gave rise to an entangled material, a observed by other authors [46]. The lignocellulosic matrix comprises cellulose fibrils or ganized in fibers arranged in a longitudinal orientation and inserted in a cementing matri consisting of hemicellulose and lignin [2]. Thus, after removing the amorphous compo nents, the hydroxyl (−OH) groups of cellulose could establish extensive hydrogen (−H bonds among their chains, leading to a more compact and disordered arrangement [47 This phenomenon can be observed in Figure 2c in the FESEM image taken at higher mag nification. Thereafter, the cellulose fibers were ground to break up the agglomerates an obtain thinner CMFs (Figure 2d). Image analysis using both optical microscopy and FESEM techniques served to determine the cumulative distributions of CMF lengths an As shown in Figure 2c, the bleached cellulose fibers exhibited a swollen, rough, and unregular appearance. The removal of non-fibrous components from the lignocellulosic matrix, such as hemicellulose, lignin, and waxes, gave rise to an entangled material, as observed by other authors [46]. The lignocellulosic matrix comprises cellulose fibrils organized in fibers arranged in a longitudinal orientation and inserted in a cementing matrix consisting of hemicellulose and lignin [2]. Thus, after removing the amorphous components, the hydroxyl (-OH) groups of cellulose could establish extensive hydrogen (-H) bonds among their chains, leading to a more compact and disordered arrangement [47]. This phenomenon can be observed in Figure 2c in the FESEM image taken at higher magnification. Thereafter, the cellulose fibers were ground to break up the agglomerates and obtain thinner CMFs (Figure 2d). Image analysis using both optical microscopy and FESEM techniques served to determine the cumulative distributions of CMF lengths and widths, respectively shown in Figure 3. Therefore, the mechanical defibrillation of the cellulose fiber bundles was evident, which was considered to occur by detaching CMFs into long fibrils with a major cumulative frequency of lengths below 200 µm (Figure 3a). Likewise, there was a marked reduction in the CMF thickness, showing predominant values between 5-15 µm. The magnified image of the ground material, shown in Figure 2d, revealed that CMFs were isolated with a smooth surface pattern, confirming that the milling step was an efficient method to disrupt the tangled microfiber arrangement. In this regard, Jiang et al. [48] observed that mechanical defibrillation was effective at detaching cellulose fibrils from RS treated with the 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO)-mediated oxidation method. The authors reported less uniform defibrillated particles, as also detected herein in Figure 2d. In this sense, the micrographs suggested that, due to the detachment of the fiber bundles into CMFs, the cellulosic material obtained could promote an efficient force transferring from the microfibers to polymer matrices, such as TPS [49].  Figure 4 shows the FESEM micrographs of the cross-sections of the TPS-based films incorporating different CMF concentrations (1, 3, and 5 wt%) obtained by cryo-fracture and their corresponding samples using DH-modified starch, that is, TPSDH. The TPS control film samples revealed a smooth and homogeneous cross-section surface pattern, indicating a good starch thermoplasticization, as also reported by other authors for starch films obtained in similar conditions [50,51]. In contrast, the TPSDH films exhibited remaining starch granules, which suggests changes in the starch thermoplasticization behavior after the DH treatment. In this context, Chandanasree et al. [31] reported that the treatment of native starch by DH provokes partial damages in the semi-crystalline and crystalline regions of the granules, which prevent the granule swelling and delays its gelatinization. Furthermore, the DH treatment can oxidize and convert the -OH groups present in the starch chains to carbonyl (C=O) and carboxyl (-COOH) groups, which alters the gelatinization profile [52]. A progressive increase in the number of non-gelatinized starch granules was also observed in the cross-sections of the films of non-modified starch when the CMF concentration increased. However, this number increased in the samples prepared with DH-modified starch.

Microstructure of the Films
The starch thermoplasticization occurs during heating, in the extruder or melt-mixing equipment, in the presence of a plasticizer, such as glycerol, at high temperatures (130 • C) and pressures [21]. The microstructural features of the microcomposite films revealed that the CMF incorporation interfered the starch gelatinization with glycerol. This could be explained by the competitive interaction of the plasticizer with both CMFs and starch through -H intermolecular bonds with the cellulose chains, decreasing their availability and subsequently provoking gelatinization of the granular structure of starch. This effect was mainly observed in the TPSDH samples, where the thermal treatment involved changes in its granular structure that already limits its gelatinization. The starch thermoplasticization occurs during heating, in the extruder or melt-mixing equipment, in the presence of a plasticizer, such as glycerol, at high temperatures (130 °C) and pressures [21]. The microstructural features of the microcomposite films revealed that the CMF incorporation interfered the starch gelatinization with glycerol. This could be explained by the competitive interaction of the plasticizer with both CMFs and starch through −H intermolecular bonds with the cellulose chains, decreasing their availability and subsequently provoking gelatinization of the granular structure of starch. This effect  Figure 4, which shows the fracture surfaces of film samples, revealed that the CMFs incorporated into the films exhibited good compatibility with both types of TPS matrices, that is, non-modified and DH-modified starch. This observation can be deduced from the good interfacial adhesion observed between the fibers and starch matrix due to the absence of gap between the fiber surface and the TPS matrix. Nevertheless, the film cryo-fracture does not seem to promote the breakage of the fibers, but their separation from the matrix, resulting in hollows that arise from the outgoing fiber during the fracture of the film samples. Similar findings were observed by Collazo-Bigliardi et al. [40] and Kargarzadeh et al. [28] in TPS films reinforced with CMFs from rice husk. The apparently good interfacial adhesion properties of fibers in the starch matrices could derive from the obtaining process of CMFs, without alkaline pre-treatment, which promoted the presence of exposed -OH groups on the surface of the fibers [53], thus enhancing the -H bond interactions with the -OH groups of starch.

Optical Properties: Color and Transparency
Film color and transparency are not only important factors in food packaging designs, but also very useful for evaluating the filler-polymer matrix compatibility. Table 2 summarizes the color coordinates in terms of L*, C ab *, and h ab * as well as the total color difference (∆E*) with respect to the neat TPS films without DH treatment and CMFs and the opacity coefficient of the different films. Likewise, the T i spectra of the different films are shown in Figure 5. Both the CMF incorporation and starch DH modification significantly (p < 0.05) affected the evaluated optical properties. Comparison of TPS and TPSDH films without fibers permits to elucidate that DH gave rise to slightly darker films with more saturated, that is, higher C ab * values, and a yellower color, which suggest that Maillard or sugar caramelization reactions could occur during thermal treatment of native starch. Likewise, as the CMF concentration increased, both TPS and TPSDH films exhibited a slight decrease in their lightness (L*) while color saturation (C ab *) increased and hue tended to be more yellowish. This was reflected on the ∆E* values, in which the color coordinates L*, a*, and b* of the TPS control film were used as a reference. In the TPS films without DH treatment, the ∆E* values attained after the incorporation of low contents of CMFs were below 1 indicating unnoticeable color changes, whereas only in the microcomposite with 5 wt% CMFs, that is, TPS5, an unexperienced observer can already observe color differences (2 ≤ ∆E* ≤ 3.5). In the TPSDH films, as opposite, the combined effect of dispersed phases, that is, CMFs and non-gelatinized starch granules, and colored compounds formed during thermal modification of starch [31,54], gave rise to higher color variations with respect to the neat TPS films. Thus, the TPSDH and its microcomposite with the lowest CMF content showed clear noticeable differences (3.5 ≤ ∆E* ≤ 5), whereas the samples with higher fiber contents yielded films with a noticeable different color (∆E* ≥ 5). Table 2. Optical properties in terms of lightness (L*), chroma (C ab *), hue angle (h ab *), color difference (∆E*), and opacity (O) of thermoplastic starch (TPS) and TPS modified by dry heating (TPSDH) films with different contents of cellulose microfibers (CMFs). persed phase also caused changes in the refractive index of the polymer matrix, pro light scattering and opacity in the films. Therefore, opacity increased due to the pres CMFs and the non-gelatinized starch granules. In particular, the opacity paramet included in Table 2, increased from 0.141 for the neat TPS film up to 0.171 for the T film.  Table 3 shows the equilibrium moisture content and water solubility of the d films prepared with TPS and TPSDH. One can observe that the DH treatment o slightly promoted water sorption capacity and water solubility of the films. In thi Lim et al. [57] reported that DH treatment of starch can damage its granular str which culminates in the exposure of its −OH groups, promoting the ability to intera water molecules and water sorption capacity. Oh et al. [32] also reported an incr water solubility of rice starch modified with the DH treatment. Furthermore, th incorporation slightly decreased the water sorption capacity of both TPS and T films, whereas it provided different effects on water solubility for each kind of star reducing solubility for the TPS films but increasing it in the case of the TPSDH film small changes in the water sorption capacity provoked by CMFs can be explained i of the lower water affinity of the fibers in comparison with TPS [58], reflected on th equilibrium water content (3%) at 53% RH [35]. In fact, the resulting water conten The T i spectra of the films within the wavelength range of 400-700 nm shown in Figure 5 revealed that the TPSDH films without CMFs exhibited lower transparency that the TPS films, with lower values of T i , mainly at the lowest wavelength. This can be explained by the presence of colored compounds, formed during the starch DH modification process and the presence of dispersed starch granules in the films, which were non-completely gelatinized. These dispersed particles, having refractive indexes different from that of the starch continuous phase, promoted light scattering and film opacity. In both kinds of films, a slight and progressive decrease in light transmission was observed as the CMF concentration increased. This result agrees with the observations reported by other authors for CMF-containing TPS samples [36,55,56], where the filler incorporation as dispersed phase also caused changes in the refractive index of the polymer matrix, promoting light scattering and opacity in the films. Therefore, opacity increased due to the presence of CMFs and the non-gelatinized starch granules. In particular, the opacity parameter, also included in Table 2, increased from 0.141 for the neat TPS film up to 0.171 for the TPSDH3 film. Table 3 shows the equilibrium moisture content and water solubility of the different films prepared with TPS and TPSDH. One can observe that the DH treatment of starch slightly promoted water sorption capacity and water solubility of the films. In this sense, Lim et al. [57] reported that DH treatment of starch can damage its granular structure, which culminates in the exposure of its -OH groups, promoting the ability to interact with water molecules and water sorption capacity. Oh et al. [32] also reported an increase in water solubility of rice starch modified with the DH treatment. Furthermore, the CMF incorporation slightly decreased the water sorption capacity of both TPS and TPSDH films, whereas it provided different effects on water solubility for each kind of starch film, reducing solubility for the TPS films but increasing it in the case of the TPSDH films. The small changes in the water sorption capacity provoked by CMFs can be explained in terms of the lower water affinity of the fibers in comparison with TPS [58], reflected on their low equilibrium water content (3%) at 53% RH [35]. In fact, the resulting water content of the microcomposites could be deduced from the mass balance considering the equilibrium water content of both starch matrix and fibers. The different effect of the fibers on water solubility of TPS and TPSDH films can be explained by the different interactions of CMFs with both starch matrices. The -OH groups present in CMFs could interact with the hydrophilic groups in starch, making it unavailable to interact with the water molecules [59]. This effect was evident for the water solubility of the TPS films, which decreased significantly (p < 0.05) after the CMF incorporation, as previously reported by other authors [56,60,61]. Furthermore, the lower water solubility of these microcomposites due to the CMF presence could make starch less susceptible to hydrolysis, though this property mainly depends on its structural characteristics and modifications [62]. However, in the TPSDH films, the CMF incorporation promoted film water solubility, except for the samples containing 5 wt% CMFs. This different effect could be ascribed to a poorer integration and dispersion of the fibers in the starch matrix, having lower adhesion forces at the interface and, thus, favoring water penetration through the junction zones. The higher presence of non-gelatinized starch granules could also contribute to the higher solubility in the TPSDH films since this dispersed phase can reduce the cohesion forces of the matrix, which are related to the interchain bonds that depend on their interfacial adhesion in the continuous matrix.

Barrier Properties to Water Vapor and Oxygen Gas
The values of WVP and OP of the TPS films are also shown in Table 3. The nonmodified starch films exhibited WVP values ranging from 5.3 to 6.3 g.mm/kPa·h·m 2 , close to the range reported by other authors for corn starch films [34,40]. Films of TPS with 3 and 5 wt% CMFs exhibited slightly lower but still significant (p < 0.05) WVP values than the neat TPS film, reaching percentages increases in the barrier performance of 6% and 15%, respectively. Since WVP of packaging materials is mainly affected by the solubility coefficient in the polymer matrix and also, but to a lesser extent, by the vapor diffusion rate [63], the permeability decrease attained in the film samples containing CMFs can be ascribed to the chemical interaction by -H bonds of cellulose with hydrophilic groups of the starch-glycerol matrix described above [4,49,64]. Other authors also reported a decrease in WVP of TPS films from different sources incorporated with cellulosic fractions [61,65]. Furthermore, the CMF addition did not influence significantly (p > 0.05) the WVP values of the TPSDH films, which already presented values remarkably lower (p < 0.05) than the TPS samples. The latter lower permeability to water vapor of DH-modified starch can be ascribed to the large presence of intact starch granules, previously observed in the FESEM images, which would increase the tortuosity factor of the modified matrix, further limiting the diffusion of water molecules through the films. Other authors also reported lower values of VWP for DH-modified rice starch films [66].
As concerns the OP values, the incorporation of CMFs at contents of 3 and 5 wt% promoted a significant (p < 0.05) OP decrease of approximately 30% with respect to the neat TPS film. This barrier enhancement can be attributed to the increase of the tortuosity factor associated with the fiber distribution along the TPS matrix, with a good compatibility and dispersion, which would limit the diffusion rate of oxygen molecules through the films. Other studies have also reported a decrease in the OP of starch-based films with cellulosic fractions [36,65], which are in agreement with the results reported herein. However, nonsignificant differences (p > 0.05) in the OP values were observed for the TPSDH films after fiber incorporation. These films exhibited significantly (p < 0.05) higher OP values than TPS microcomposite films with 3 and 5 wt% CMFs and similar to those observed for TPS films without and with 1 wt% CMFs. The lack of influence of the incorporated fibers on the barrier properties (both WVP and OP) of the TPSDH films suggests that the expected fiber promotion of the tortuosity factor is probably compensated by lower adhesion forces between the fibers and biopolymer matrix that promoted mass transfer through the union points. Moreover, the larger content of C=O and -COOH groups produced by depolymerization of heat-treated starch chains could promoted the solubility of oxygen molecules in the films, enhancing the transmission rate [67].
Therefore, incorporating CMFs from RS in TPS films represents an adequate strategy to enhance the oxygen barrier capacity of the biopolymer films, being suitable to reduce or modulate oxidation reactions in packaged foods. In contrast, a very low effect of CMFs on the barrier properties of the DH-modified starch films was observed, though these biopolymer films already exhibited a better water vapor barrier capacity (90% reduction of WVP with respect to untreated TPS films).

Mechanical Properties
The mechanical properties of the films, namely EB, TS, and EM were determined by tensile tests and are summarized in Table 4. Both TPS and TPSDH films exhibited similar stiffness or elastic modulus, though the TPSDH films were slightly more rigid and remarkably more resistant to break but also less extensible. The TPS control film had an EB value of approximately 30%, in line with that observed by Hernández-García et al. [50] for TPS films. Furthermore, the TPSDH film without fibers showed an EB value of 23%, being lower than the neat TPS film. This mechanical embrittlement can be attributed to the changes occurred in the starch chains, such as short-chain amyloses produced by heatinduced hydrolysis of starch [54] or the increase in the content of C=O and -COOH groups during the DH of starch [67]. Other authors [60,68] also suggested that the conversion of native -OH groups to C=O and -COOH ones in oxidized starch may promote -H bonds between amylose and amylopectin molecules, increasing the integrity of the biopolymer matrix and improving the tensile resistance of the films. Table 4. Mechanical properties in terms of elongation at break (EB%), tensile strength at break (TS), and elastic modulus (EM) of thermoplastic starch (TPS) and TPS modified by dry heating (TPSDH) films with different contents of cellulose microfibers (CMFs). The incorporation of CMFs at 3 wt% increased notably film stiffness in both kinds of films, being this effect less noticeable at other ratios of fibers. Several studies have also reported higher EM values of starch films containing cellulosic fractions from different sources [36,56,69]. In general, all the here-developed films became more resistant and less stretchable as higher loading of CMFs were incorporated. Higher fiber contents, of 3 and 5 wt% CMFs, decreased the TPS film stretchability by about 50%, whereas a similar decrease was observed for the TPSDH film with only 1 wt% CMFs. The lack of stretchability limits the overall extensibility of the films without markedly reducing the resistance to break. As shown in FESEM micrographs, CMFs were strongly bonded to the polymer matrix, which restricted the movement of starch chains, decreasing their stretchability but also reinforcing the TPS matrix [70]. The highest reinforcing effect was observed for the 3 wt% CMFs in both TPS and TPSDH films. These films particularly exhibited the highest EM (about 550 and 590 MPa for TPS and TPSDH, respectively) and resistance to break (approximately 7 and 10 MPa for TPS and TPSDH, respectively), but also with reduced elongation capacity (14 and 5% for TPS and TPSDH, respectively). In this regard, Fazeli et al. [65] also observed a threshold concentration in cellulose nanofibers from henequen to reinforce starch films.

Formulation EB (%) TS (MPa) EM (MPa)
Therefore, the obtained results in tensile behavior of TPS and TPSDH films demonstrated the reinforcing capacity of CMFs for starch films, which indicates good compatibility and dispersion along the TPS matrix. However, contents higher than 3 wt% led to lower mechanical performances, which can be attributed to a loss of cohesion forces or a lower degree of dispersion and certain agglomeration of the fibers in the TPS matrix. Figure 6 shows the TGA and first derivative (DTG) curves obtained for the TPS and TPSDH films and their microcomposites films prepared with CMFs. Table 5 gathers the thermal stability parameters, namely T onset , T final , and T deg as well as the mass loss at T deg . One can observe that fiber incorporation did not notably influence the thermal degradation properties of the TPS and TPSDH films. In particular, the TPS films showed the main degradation event ranging at approximately 150-455 • C, corresponding to a mass loss of nearly 80% with a T deg value of 341 • C. This event is associated with the thermal decomposition of glycerol [71], starch [28,72], and CMFs. Figure 6a shows the thermal decomposition profile of the CMF fractions, showing the previously reported mass loss steps [35]: evaporation of bonded water (30-140 • C), cellulose and hemicellulose decomposition (177-350 • C) and lignin degradation (350-500 • C). The second thermal degradation step is referred as the "active pyrolysis zone" since the mass loss rate is high, whereas the third one is called "passive pyrolysis zone" since the percentage of mass loss is smaller, and the mass loss rate is also much lower compared to that in the second zone [73]. In relation to the degradation profile of the TPS films, the sample containing 5 wt% CMFs exhibited slightly higher mass loss from 180 to 280 • C, which can be attributed to the thermal degradation of the cellulose fractions, more noticeable due to its high content. However, this TPS5 sample was more stable than the other TPS films in the thermal range of 280-310 • C (see Figure 6a,c), suggesting that the CMF incorporation delayed the thermal degradation of the shortest chains of starch produced during its thermo-processing [74].

Thermal Properties
Regarding the TPSDH films, these samples exhibited a slight initial mass loss (3-5%) in the thermal range from approximately 50 to 150 • C, which is ascribed to evaporation of water adsorbed in the inner regions of the non-gelatinized starch granules and it may also derive from thermal decomposition of low-molecular weight (M W ) components produced during the DH treatment [54]. This mass loss can be observed in Figure 6b, where the mass loss curves of the TPS and TPSDH films were compared. As shown in Table 4 and also Figure 6d, TPSDH films exhibited slightly higher T deg values than the TPS films. This thermal stability enhancement may be attributed to the presence of C=O and -COOH groups produced by the heat treatment of starch that allows for more -H bonds between the polymer chains [68]. Furthermore, the non-gelatinized starch granules, with more crystalline regions, could also contribute to the resultant thermal behavior. In the TPSDH films, the concentration increases in CMFs also implied a slight increase in the thermal stability, which could be due to the increasing presence of starch granules in the matrix. produced by the heat treatment of starch that allows for more −H bonds between the polymer chains [68]. Furthermore, the non-gelatinized starch granules, with more crystalline regions, could also contribute to the resultant thermal behavior. In the TPSDH films, the concentration increases in CMFs also implied a slight increase in the thermal stability, which could be due to the increasing presence of starch granules in the matrix.  Table 5. Onset, final, and peak temperatures (T onset , T final , and T deg ), and mass loss at T deg (∆m%) obtained from thermogravimetric analysis (TGA) and glass transition temperature (T g ) obtained from differential scanning calorimetry (DSC) for thermoplastic starch (TPS) and TPS modified by dry heating (TPSDH) films with different contents of cellulose microfibers (CMFs). Finally, DSC analysis was also performed to evaluate the effect of the starch modification by DH and CMF incorporation on the thermal phase transitions of starch. The first heating scans, gathered in Figure 7, showed the typical endotherm of starch gelatinization associating peak temperatures in the 55-110 • C range for the film samples of TPSDH and TPS with the highest CMF loadings (TPS3 and TPS5). This agrees with the above-reported FESEM observations that revealed the presence of non-gelatinized starch granules in the starch matrix. It can also be observed that the gelatinization enthalpy increased with the CMF ratio due to the aforementioned higher presence of starch granules in the films. In the second heating scan, the TPS films also exhibited the glass transition (T g ) of the already gelatinized starch, being this transition less appreciable, that is, having a lower change in specific heat (∆C p ) for TPSDH samples. The T g values (midpoint) of the TPSDH sample was slightly higher than that of TPS, which is coherent with the stronger interactions promoted in the matrix when starch was DH modified. This result also agrees with the more resistance but less ductile behavior of the films attained during the mechanical characterization. As expected, the addition of CMFs did not induce notable changes in the starch's T g , when compared with the corresponding control sample, since the fibers are not miscible with starch. The small variations of T g in the TPS composites with CMFs can be attributed to potential interactions between fibers or their remaining water contents and plasticizer that could modify the plasticization level of starch [31]. In this context, Collazo-Bigliardi et al. [40] found similar results for corn starch films filled with fibers derived from rice husk. associating peak temperatures in the 55-110 °C range for the film samples of TPSD TPS with the highest CMF loadings (TPS3 and TPS5). This agrees with the above-re FESEM observations that revealed the presence of non-gelatinized starch granules starch matrix. It can also be observed that the gelatinization enthalpy increased w CMF ratio due to the aforementioned higher presence of starch granules in the fi the second heating scan, the TPS films also exhibited the glass transition (Tg) of the a gelatinized starch, being this transition less appreciable, that is, having a lower cha specific heat (∆Cp) for TPSDH samples. The Tg values (midpoint) of the TPSDH s was slightly higher than that of TPS, which is coherent with the stronger interactio moted in the matrix when starch was DH modified. This result also agrees with th resistance but less ductile behavior of the films attained during the mechanical cha ization. As expected, the addition of CMFs did not induce notable changes in the s Tg, when compared with the corresponding control sample, since the fibers are not ble with starch. The small variations of Tg in the TPS composites with CMFs can tributed to potential interactions between fibers or their remaining water conten plasticizer that could modify the plasticization level of starch [31]. In this context, C Bigliardi et al. [40] found similar results for corn starch films filled with fibers d from rice husk.

Conclusions
CMFs with good reinforcing capacity for starch films were obtained from RS, a material derived from the agricultural and food industry, using a green process an chanical defibrillation. Fibrils with a major cumulative frequency of lengths below 2 and 5-15 µm of thickness were obtained. Incorporation of CMFs into starch matric wt%, the most optimal content, led to stiffer and more resistance to break films fo non-modified and DH-modified corn starch films. Likewise, fibers at 3 and 5 w

Conclusions
CMFs with good reinforcing capacity for starch films were obtained from RS, a waste material derived from the agricultural and food industry, using a green process and mechanical defibrillation. Fibrils with a major cumulative frequency of lengths below 200 µm and 5-15 µm of thickness were obtained. Incorporation of CMFs into starch matrices at 3 wt%, the most optimal content, led to stiffer and more resistance to break films for both non-modified and DH-modified corn starch films. Likewise, fibers at 3 and 5 wt% improved the oxygen and water vapor barrier capacity of TPS films but did not modify barrier properties of TPSDH films. Fibers interfered starch gelatinization during the starch-glycerol thermo-processing, more markedly in DH-modified starch with modified granular structure, giving rise to non-gelatinized starch granules into the film matrix. The dispersed phase (starch granules or fibers) also modified the optical properties of the TPS films, making them slightly less transparent. Therefore, it can be concluded that RS is a good source of CMFs that can be used to improve properties of starch films with different purposes, such as their use as food packaging material. This contributes to the valorization of the agricultural and food waste into added-value materials, useful for improving the performance of biodegradable food packaging materials. Likewise, this study further confirms that the DH process of starch represents a green physical method to enhance the water barrier capacity and tensile strength of starch films. Future studies will deal with the hydrolysis resistance of these microcomposite films and their practical application as packaging materials in food preservation.