The effects of crosslinking agents on faba bean flour–chitosan‐curcumin films and their characterization

Although glutaraldehyde (GLU) has been frequently preferred to improve some characteristics of bio‐degradable food packages, it is regarded as a hazardous chemical. Therefore, in this study, the possibility of replacement GLU with citric acid (CA) a non‐toxic compound was questioned in a film formation. Different ratios of GLU and CA on faba bean flour, chitosan, and curcumin film were examined. To determine the most favorable film composition, water solubility (WS), swelling degree (SD), water contact angle (WCA), water vapor permeability (WVP), and mechanical properties of the films were examined. In addition, the effects of crosslinking agent and their ratios on crystallinity (XRD), thermal properties (DSC and TGA), and antioxidant activity (DPPH and ABTS) were also studied. Time Domain Nuclear Magnetic Resonance (TD‐ NMR) relaxometry was also employed to explain distribution and mobility of the protons in faba bean flour, chitosan, and curcumin films. At the end of the analysis, it was concluded that CA and GLU reduced SD by 75% and 13% compared to the films without cross‐linker, respectively. Incorporation of crosslinking agent increased the WCA from 77.22° to 99.41° and caused formation of more hydrophobic surfaces. Increasing both crosslinking ratios restricted the permeance drastically which was more than 110%. Furthermore, the lowest WVP was measured for the films containing CA at 1.5% w/v ratio (FC‐1.5C). While GLU improved the tensile strength (TS) of the film, lower elongation of the films caused formation of brittle character which limited the utilization of the films. At the end of the analysis, it was concluded that FC‐1.5 had more feasible character with respect to the other films and CA was suggested as a crosslinking agent instead of GLU for faba bean‐chitosan films.


| INTRODUCTION
Nowadays, biodegradable and eco-friendly components have drawn attention because of increasing environmental concerns originated from petroleum-based packaging materials. In this sense, to replace traditional packaging, natural biopolymers such as polysaccharides, proteins, and lipids have been regarded as possible alternatives (Lin et al., 2019).
Because flours contain carbohydrates, lipid, and protein together, they have been an excellent source for producing food packages. For example, banana flour mixed with chitosan (Pitak & Rakshit, 2011), rice flour combined with fish gelatin (Ahmad et al., 2015), and quinoa flour (Pagno et al., 2016) have been studied to serve as possible food package alternatives. Among different flour types, faba bean, also known as broad bean, filed bean, and horse bean (Multari et al., 2015), is a good source of protein and carbohydrate. The protein content of faba bean is higher than other high protein flours such as pea flour (Yildiz, Bayram, et al., 2021) and lentil flour (Aydogdu et al., 2019). In addition to that, faba bean contains high amount of secondary metabolites, that is, phenols and flavonoids that enrich their antioxidant activity (Millar et al., 2019). However, film formation performance of faba bean flour has not been studied yet.
Chitosan, a natural polysaccharide obtained from chitin, is one of the most commonly used biopolymer for food packaging applications owing to its non-toxicity, biodegradability, biocompatibility, and antioxidant activities (Yujia Liu et al., 2016). However, the hygroscopic nature of chitosan affects both swelling and permeability characteristics of the chitosan films adversely (Pavoni et al., 2021).
Although films made from polysaccharides show good oxygen and carbon dioxide barrier properties due to their tightly packed network structure, they are generally permeable to water vapor due to their hydrophilic nature (Mohamed et al., 2020). On the other hand, proteins provide better mechanical features to the produced films compared to polysaccharides. However, the hydrophilic nature of proteins also leads to poor water vapor barrier characteristics with the films (Mohamed et al., 2020). To eliminate potential drawbacks of these biopolymer films, crosslinking these compounds to modify film structure is one of the promising solutions.
For example, sodium trimetaphosphate (STMP) (Aydogdu et al., 2020) and sodium sulphate crosslinking agents (Kalaycıo glu et al., 2017) were used in guar gum/orange oil emulsion and chitosan antimicrobial film, respectively. Further, the effect of graphene oxide (Grande et al., 2017) on crosslinking chitosan films and UV induced crosslinking (Zhou et al., 2008) on corn starch films were also analyzed.
Rather than these crosslinking agents, citric acid (CA) is a biobased tricarboxylic acid and is commonly used due to its low-cost, non-toxic nature. The carboxyl group of citric acid might react with the hydroxyl and amino group of different biopolymers (Nataraj et al., 2018;Yang et al., 2010). Thus, CA was utilized to improve swelling, mechanical, and gas barrier properties of many films such as potato starch/chitosan (Wu et al., 2019), cassava starch (Seligra et al., 2016) poly (vinyl alcohol)/starch/graphene (Jose & Al-Harthi, 2017), and wheat straw hemicellulose (Azeredo et al., 2015).
On the other hand, glutaraldehyde is one of the most commonly preferred crosslinking agents for proteins and polyhydroxy compounds. Because it has a rigid chemical structure without any hydrophilic group, the addition of glutaraldehyde decreases the water sensitivity of the material. The carbonyl group of glutaraldehyde reacts with the amino group (chitosan) and hydroxyl group of polysaccharide molecules (Gonenc & Us, 2019).
Curcumin is one of the most valuable ingredients in turmeric and has been long used in different fields due to its anticancer, antioxidant, antimicrobial, and anti-arthritic character (Gutierrez-Gonzalez et al., 2020). Antioxidant activities of curcumin were attributed to electron donor ability to neutralize free radicals by producing stable products. Furthermore, studies shown that scavenging hydrogen peroxide ability of curcumin was higher than α-tocopherol, butylated hydroxyanisole (BHA), and butylated hydroxytoluene (BHT), most commonly used antioxidants in food science. However, BHA and BHT were to blame for carcinogenesis and liver damage (Ak & Gülçin, 2008). Therefore, taking advantage of using natural and safer antioxidants has become more promising in food packages.
Although it is a well-known fact that glutaraldehyde is a toxic chemical, it is still frequently used as a crosslinking agent in food packaging applications. In this perspective, it was aimed to compare the performance of different crosslinking agents in the production of curcumin added active film. The solubility, water vapor permeability, mechanical property, thermal property, antioxidant activity, and contact angle of the films were studied. Furthermore, the films were also compared in terms of crystallinity, distribution, and mobility of the protons to interpret the different crosslinking agents and ratios.

| Film preparation
Chitosan (2% w/v) was dissolved in acetic acid solution (2% v/v) overnight at 45 C by using a magnetic stirrer at 750 rpm (MaxTir 500, Daihan Scientific, Seoul, Korea). A slurry of faba bean flour (4%, w/v) was prepared by stirring at 750 rpm and at 80 C for 45 min to complete starch gelatinization and protein denaturation. After cooling down the slurry to 45 C, chitosan and flour slurry were mixed in equal volumes. Then, 0.4% (w/v) glycerol was added to the solution as a plasticizer. Curcumin (5% w/v) was dissolved in ethanol and 1 ml of this solution was added into the mixture and film-forming solution (FFS) was obtained. Films without any crosslinking agent were regarded as a control (FC-C). To prepare citric acid (CA) cross-linked films (FC-0.5C, FC-1C, FC-1.5C), CA was added to the FFS in different ratios (0.5%w/v, 1% w/v, 1.5% w/v). To obtain glutaraldehyde (GLU) (FC-30GLU, FC-60GLU, FC-120GLU) cross-linked films, GLU was included in to the FFS in different ratios (0.03% v/v, 0.06 v/v, and % 0.12 v/v). Films were symbolized according to the presence of crosslinking agents and ratio. The nomenclature was given in Table 1. FFS was stirred at 1,200 rpm for 45 min and then degassed using an ultrasonic bath (Lab Companion, Jeiotech, Seoul, Korea) at 37 kHz for 25 min. Solutions of 15 g were poured into LDPE Petri plates and dried in a 55 C oven (Binder ED 115, Tuttlingen, Germany) for 6.5 h.
Films were kept in a climate chamber at a constant relative humidity of 52% before analysis.
2.2 | Moisture content, swelling degree, and water solubility of the films MC, SD, and WS of the films were measured with the method described by Wu et al. (2019). First, the films were cut into pieces (2 cm Â 2 cm), and their initial weight was recorded (W1). Then, the films were dried in 105 C oven until reaching a constant weight (W2).
After that, they were immersed in 25 ml of distilled water at room temperature for 24 h. After removal, the excess water on the surface was removed by the filter paper, and the films were weighed again (W3). Finally, they were dried in a 105 C oven for 24 h, and the last weight was recorded (W4). Moisture content, swelling degree, and water solubility of the films were calculated by using Equations 1-3 below:

| Water contact angle
The surface hydrophobicity of the films was analyzed by a goniometer (Attension Theta, Sweden). A drop of distilled water was placed on the surface of the films (4 cm 2 ), and static contact angle was measured.

| XRD
Crystalline properties of the films were analyzed by X-ray diffractometer (Rigaku Ultima-IV, USA) using copper (Cu) irradiation with 30 mA current and 40 kV energy. Samples were scanned at an angular range of 5 and 70 scanning range with 2 /min scanning rate.
The crystallinity degree of the samples was determined by using Equation 4, based on the method described by Demirkesen, Campanella, Sumnu, Sahin, and Hamaker (2014).
where I c is integrated intensity of crystalline phase and I a is the integrated intensity of amorphous phase.

| Water vapor permeability
To measure water vapor permeability (WVP) of the films, the method described by Yildiz, Bayram, et al. (2021) was used. First, the thickness of the films was measured from different random points with a digital micrometer (LYK 5202, Loyka, Ankara, Turkey). Then, the cylindrical test cups with 40 mm internal diameter were filled with 35 ml water.
After placing the films between cup and cap, it was screwed. The cups' initial weight was recorded and placed in a desiccator which was equilibrated to nearly 0% RH by using silica gels. The weight change of the cups was recorded through a 12-h period at 1 h intervals.
The relative humidity and temperature in the desiccator were controlled by using a humidity/temperature logger (EBI20-TH1, EBRO, Ingolstadt, Germany). Finally, WVP of the films were calculated using the Equation 5 below: Water vapor transmission rate (WVTR) can be calculated from weight changes versus time graph (g m À2 s À1 ). R1 and R2, respectively, represented relative humidity inside and outer sides of the cups. Δx is the thickness of the films (m). Finally, the Psat (Pa) is the saturated water vapor pressure at room temperature. 2.6 | Differential scanning calorimetry Thermal analysis of chitosan, curcumin, faba bean flour, and films was performed by using a differential scanning calorimeter (Pyris 6 DSC, PerkinElmer, Massachusetts, USA) with a method described by Yildiz, Sumnu, and Kahyaoglu (2021). About 5 mg was placed in an aluminum pan and heated with a heating rate of 10 C/min from 25 C to 350 C.

| Thermogravimetric analysis
Thermogravimetric analysis of the chitosan, curcumin, faba bean flour, and films were carried out using a thermo-gravimetric analyzer (Perkin Elmer Pyris 1) with a method described by Yildiz, Sumnu, and Kahyaoglu (2021). Sample of 5 mg was heated from 30 C to 600 C at a rate of 10 C/min and nitrogen flow rate of 30 ml/min.

| Mechanical properties
Tensile strength, Young's modulus, and percentage elongation at break were measured by a texture analyzer (Brookfield, Ametek CT3, Mid-dleboro, MA, USA, TA-DGA tension probe). The test was carried out with using 0.4 N load cell at a test speed of 0.40 mm/s.

| DPPH radical-scavenging activity
To determine the DPPH radical-scavenging activity, a method given by Yildiz, Bayram, et al. (2021) with some modifications was used with some modifications. Films ($0.0245 g) were dissolved in 15 ml ethanol: water (80:20) solution and centrifuged at 1,500g for 20 min.
Then, a diluted sample of 2 ml was mixed with 2 ml DPPH (2,2-diphenyl-1-picrylhydrazyl) solution. Absorbance value of samples was measured at 517 nm after 30 min. DPPH scavenging activity of the samples was calculated according to Equations 6 and 7 below: A control and A sample are the absorbance of the DPPH solution with and without sample.

| ABTS radical scavenging activity
To prepare the ABTS analysis, potassium persulfate (2.6 mM) and ABTS (7 mM) were mixed and stored at room temperature in the dark.
The diluted sample of 0.5 ml was mixed with 1.5 ml ABTS solution and incubated at room temperature for 30 min. The absorbance of the solution was measured by a spectrophotometer at 734 nm (Roy & Rhim, 2020 3 | RESULTS AND DISCUSSION

| Physical properties of FC films
The main aim of the crosslinking in biopolymer films was to decrease solubility and swelling for the improvement of water resistance and applicability. The solubility of the FC films was significantly lower than the films with similar compositions (Table 2) were between 25% and 42%. The main reason for this result might be the bond formation between the hydroxyl group of starch, amino groups of chitosan ( Figure S2). This might decrease the water affinity, which resulted in even a further decrease in water solubility of the FC films without any cross linker (Bajer et al., 2020). While the amide group of chitosan might function as a proton acceptor, the hydroxyl group of starch might behave as a weak proton donor. Therefore, the formation of hydrogen bonding became more favorable (Sudhakar, Sowmya, Selvakumar, & Bhat, 2012). Furthermore, the presence of curcumin might have a decreasing effect on water solubility due to its hydrophobic nature (Roy & Rhim, 2020).
The addition of both crosslinking agents caused a decreasing trend in water solubility ( Table 2). The hydrophobic ester bond formation between CA and polysaccharides ( Figure S3a) lowered the number of available hydrophilic hydroxyl groups and, in turn, decreased the solubility of the films (Wu et al., 2019). Similarly, the less available amino group of chitosan resulted in less soluble films ( Figure S4a).
However, only CF-1.5C films had significantly lower solubility than the control films. The main reason for that might be the higher crosslinking density of the CF-1.5C films.
Although chitosan is a polysaccharide, unlike many other biopolymers, it is insoluble in water. It dissolves in weakly acidic solutions and forms hydrogels. Since the hydrogels can absorb a high amount of water, this feature makes application chitosan-based films practically useless (Murray & Dutcher, 2006). Therefore, to prevent such an outcome, crosslinking of chitosan films becomes more critical. The main idea behind the crosslinking is to reduce the free volume by increasing the number of junctions and thereby restrict the possible hydrogen bond formation between chitosan and water molecules (Wu et al., 2019).
The swelling degree of all the films decreased drastically with increasing the ratio of the crosslinking agents (Table 2). Glutaraldehyde has three hydroxyl groups, while citric acid has three carboxyls and one hydroxyl group in the structure (Figure S1c,d). This means that citric acid has a higher chance to make covalent bonds as compared to glutaraldehyde. Therefore, the lowest amount of citric acid reduced swelling degree more than the highest amount of glutaraldehyde (Yoon, Chough, & Park, 2006).
Moisture content was expected to decrease in the crosslinked films; however, it had an increasing trend with the addition of the crosslinkers. Although GLU and CA reduced the starch and chitosan hydrophilic sides, both crosslinking agents might have unreacted hydroxyl or carbonyl groups. Because all the crosslinked molecules were bulky, steric hindrance might be the restrictive factor and the reason for unreacted groups. Therefore, water molecules might have attached to these available groups through hydrogen bonding and increased the moisture content of the crosslinked resulting films.
The surface hydrophilicity and hydrophobicity were determined by water contact angle where θ > 65 indicated hydrophobic surfaces (Narasagoudr et al., 2020). In the literature, the water contact angle of neat chitosan film was reported nearly as 103 (Silva et al., 2007).
Although faba bean flour increased surface wettability, FC films still had highly hydrophobic surfaces whose water contact angles were between 73 and 99 (Table 2). These findings were attributed to the strong covalent bonding between starch and chitosan (Zhao et al., 2018). Furthermore, as mentioned, curcumin is a hydrophobic

| Water vapor permeability
Water vapor permeability (WVP) of the films was shown in Table 2.
WVP of FC films was dependent on many factors such as crystalline- amorphous ratio, hydrophilic-hydrophobic ratio, the integrity of the film, polymer chain mobility (Balasubramanian et al., 2018), solubility in water, and moisture content. Therefore, the permeability was shaped under multiple effects, not based on a single parameter. For example, moisture affected the intermolecular forces between polymer chains by increasing free volume and enhancing chain mobility (Yeamsuksawat & Liang, 2019). Thus, films with higher moisture content were expected to have higher permeability. Because molecule alignments were more compact and regular in the crystalline region compared to the amorphous part, transferring rate of water molecules from one side to the other became relatively slow (Uygun et al., 2020). Furthermore, higher WCA could be interpreted as a reduction in potential interaction between water vapor and film surface. Therefore, films with higher WCA were expected to have lower WVP.
As seen in Table 2

| Mechanical properties
Enhancement in TS was generally expected with introducing crosslinking agents due to increasing and tightening molecular interactions.
As shown in Table 3, although GLU addition resulted in increasing TS of the film, incorporation of citric acid decreased TS. Increasing TS might be interpreted as more resistance to mechanical stress. On the other hand, increasing EAB was the indicator of the higher stretchability and more deformable films. High TS and low EAB were the indicators of brittle materials as in the example of GLU incorporated films. This showed that the motion of FC film matrices was more restricted after incorporation of GLU. Similar findings were also reported in the study of gelatin films cross-linked with GLU (Martucci et al., 2012).
On the other hand, FC-1C and FC-1.5C films showed a yieldlike deformation similar to the thermoplastic films with lower TS and high EAB. As can be seen, after an elastic deformation, a large deformation region was observed, which indicated enhancement of ductility and extensibility (Martucci et al., 2012). Although in this study, CA was preferred as a crosslinking agent, some studies also reported plasticizer character of unreacted CA at high ratios by reducing the interaction between components (Wu et al., 2019).
The present research showed that both effects might coexist, although these impacts seemed to be inconsistent with each other. TS of FC-1C and FC-1.5C films tended to decrease more than 30% compared to FC-C. EAB of FC-1C (4.23 AE 0.66%) and FC-1.5C (12.93 AE 1.06%) films were also significantly different than the FC-C (1.67 AE 0.30%). Therefore, it might be concluded that citric acid played a role as a plasticizer for the mechanical properties of the films. On the other hand, especially for the FC-1.5C films, WS, SD, and WVP significantly decreased, reflecting the behavior of the crosslinking agent. Similar outcomes were also reported in the study of Azeredo et al. (2015) and Sebti et al. (2003).
These controversial outcomes were also explained by the higher heterogeneity of the spacing between cross-links (Sebti et al., 2003).
This heterogeneity might distribute the stress on the chains, which were not stabilized by H bonding. After non-H bonding chains had been broken, they transferred the stress to the other chains by forcing them to either break or slip to release the stress.
Although high TS was desirable for the packaging films, stretchability was another important character for the film application.
Therefore, GLU incorporated films with high TS, but low elongation and rigid structure restrict its application as a packaging material. On the other hand, lower TS and higher EAB of CA cross-linked films were more feasible in practice.

| Thermal properties
The weight loss versus temperature curve of films, chitosan, faba bean flour, and curcumin was shown in Figure 1. In general, the thermal decomposition of FC films followed multistep decomposition. The first step centered at 100 C was due to the evaporation of water and other volatile compounds. The second stage degradation appeared around 200 C was because of evaporation of glycerol (Bagheri et al., 2019). Furthermore, for CA cross-linked films, evaporation of citric acid derivatives also overlapped at these values (190-200 C;Qiao et al., 2021). This was followed by a decomposition process which was the highest weight loss observed for all samples. This step corresponded to the depolymerization and decomposition of acetylated and deacetylated units of the polymer (Wu et al., 2019), in other words, cleavage of covalent bonds formed during crosslinking between starch, protein and chitosan (Sharma et al., 2018). Although the peak temperature of this stage generally was positioned around 300 C, this stage shifted gradually to higher temperatures (nearly 315 C) with the increasing CA ratio in films which was the indicator of the strong intermolecular interactions between components. However, the degradation temperature of GLU incorporated films had a similar decomposition peak temperature to FC-C, which was located approximately at 290 C.
Differential scanning calorimetry results of all films were shown in Table 3. All films exhibited an endothermic peak between 170 C and 180 C. XRD analysis showed that crosslinking agents had a decreasing impact on the crystallinity of the films (Table 3). Therefore, it was expected the reduction in Tm and melting enthalpy (ΔHm; Jose & Al-Harthi, 2017). On the other hand, it was reported that introducing cross-linker slightly increased melting temperature by stabilizing bond between all components (Sharma et al., 2018). In addition to that, crosslinking agents increased molecular weight of polymeric matrix through intermolecular interactions (Sharma et al., 2018).
Therefore, under two counteracting effects, reduction of crystallinity and stabilizing bond strength, Tm did not change significantly.
Although the melting enthalpy of GLU cross-linked films nearly the same as FC-C, CA-introduced films exhibited a noticeable increase in the melting enthalpy with an increasing CA ratio. In addition to bond stabilizing effect, melting of CA in the films might also have an effect on increasing enthalpy because the melting temperature of CA ($190 C) overlapped the Tm of the films (Shafie et al., 2019).

| Antioxidant activity
FC-C films showed the highest DPPH and ABTS activity because of curcumin and the active antioxidant biopolymer, chitosan. The hydroxyl and amino groups of chitosan interacted with free radicals and exhibited the radical scavenging activity (Roy et al., 2021). In addition to this, antioxidant activity of curcumin was originated from phenolic hydroxyl group and methylene group of β-diketone moiety (Priyadarsini et al., 2003).
As can be seen in Table 4, the addition of cross-linking agent decreased the antioxidant activity of the films by nearly 50%. First, the cross-linking reaction might be responsible for this result. Carboxyl group of CA and GLU might have reacted with the hydroxyl and amino group of chitosan and active sides of curcumin during crosslinking reaction causing a decrease in the number of active sides required for the antioxidant activity. In addition to that, releasing of the active agent, curcumin was inversely correlated with the crosslinking degree. Because the releasing rate was expected to be lower in the cross-linked polymer matrix, cross-linked films achieved relatively lower antioxidant activity (Valizadeh et al., 2019).
In parallel with this finding, releasing of active components was also linked to the swelling degree of polymer matrix (L opez de Dicastillo et al., 2016). As shown in Table 2, the highest swelling degree was observed for FC-C, as well as the highest DPPH and ABTS activity.
Although CA incorporated films achieved a relatively lower swelling degree, they showed higher antioxidant activity than GLU. This F I G U R E 1 TGA curves of FC films result was probably due to the antioxidant activity contribution of CA (Sauraj et al., 2018).

| XRD
The relative degree of crystallinity (RDC) was measured by the ratio between area of crystallinity and total area under the XRD diffraction pattern . RDC and XRD pattern of the all films were shown in Table 4 and Figure 2. There might be two main reasons behind the crystallinity of the FC films. The first one was the retrogradation of starch (Seligra et al., 2016), and the second one was the presence of NH 2 (amine) group in chitosan which promotes the formation of hydrogen bond between polymer chain (Pavoni et al., 2021). Introducing crosslinking agent showed a decreasing tendency on the crystallinity of the films because the carboxyl group of CA and GLU reacted with the hydroxyl and amine groups. Therefore, this interaction restricted the formation of crystal structure in starch and chitosan molecules. As seen, CA added samples had a significantly lower degree of crystallinity compared to FC-C. As mentioned before, CA has more sides to make crosslinking than GLU. Therefore, CA might limit the chain segment movements and thereby interfere with crystallization (Ren et al., 2017).
The characteristic peaks of chitosan were observed around 10 and 20 (2θ). The first peak was the indicator of hydrated crystalline structure because of water molecules attached to the crystal lattice.
The peak around 20 (2θ) was attributed to the amorphous part of chitosan (Giannakas et al., 2014). The diffraction peaks around 15.42 , 17 , and 20 (2θ) corresponded to the B-type starch diffraction pattern in faba bean flour (Liu et al., 2013). As seen diffraction pattern of FC films, the intensity of peaks decreased and even some disappeared (15.42 ), which was supported by the degree of crystallinity. This F I G U R E 2 X-ray diffraction patterns of the FC films result showed that interaction between components (starch-chitosan and crosslinking agents) restrained the intramolecular interactions.
However, the peak at 17.54 was observed in FC films for all formulations. This peak indicates the formation of a double helical B-type crystalline structure and most sensitive to hydration (Zhong et al., 2011). These results were consistent with the previous research findings related to the amylose and amylopectin films (Myllärinen et al., 2002). It should be noted that although cross-linking decreased the solubility or water sensitivity of the films, the main components of these films were very hydrophilic (chitosan and faba bean). Therefore, a little amount of water interaction or water present in these films (MC) might have an ongoing impact on their structures.

| Time domain NMR
NMR relaxometry is a noninvasive and user-friendly method preferred to investigate the change in mobile protons of food systems. Because NMR signal is obtained with the contribution of all protons in a sample, it is possible to get insight into the mobility and distribution of different proton compartments, which provides detailed information about the internal structure and complex associations (Ozel, Uguz, et al., 2017;Pocan et al., 2019). It is known that the relaxation times are affected by the presence of free hydroxyl groups; thus, the crosslinking of the films can be detected with the help of differences in the water mobility (Seligra et al., 2016). Therefore, it could be deduced that relaxation times could explain crosslinking, water mobility, or crystallinity of the food products. There are several types of research conducted on the use of relaxation times to characterize proton mobility and distribution on films (Picchio et al., 2018;Seligra et al., 2016;Xiao, 2018). T1, called longitudinal relaxation time, is defined as the time takes for spins to give back energy from the radiofrequency pulse. It is related to the mobility of water protons (Kirtil & Oztop, 2016). Therefore, it could be concluded that the T1 spinlattice time may provide insight into the moisture content of food products (Le Botlan et al., 1998). The limited movement of the water molecules results in shorter T1 times. Therefore, it can be said that samples with short T1 times had lower water mobility. T1 spin-lattice times of films are given in Figure 3. Results showed that control film (FC-C) had the longest T1 spin-lattice relaxation times, and it decreased as the CA ratio increased, indicating lower water mobility.
On the other hand, significant changes were not observed when GLUcontaining films were considered. This finding could be explained with the chemical structure of crosslinking agents. Glutaraldehyde has three hydroxyl groups, whereas citric acid has three carboxyls and one hydroxyl group in the structure. It is stated in a study that citric acid added films was better in terms of crosslinking than glutaraldehyde added films due to enhanced inter/intramolecular hydrogen bonding in the presence of carboxyl group and hydroxyl group together (Yoon et al., 2006). Since FC-C films did not contain any crosslinking agents, it was expected to observe the highest water mobility and consequently shorter T 1 spin-lattice times.
It is also worth to mention that in addition to water mobility, information about the crystallinity could be obtained from the spinlattice relaxation times. According to the study of Le Botlan et al. (1998), longer T 1 relaxation times were associated with the more crystalline regions. When moisture content of the samples were the same, then longer T 1 values might be associated with the effect of crystalline regions . According to moisture content measurement, FC-1C, FC-1.5C, FC-120GLU, and FC-60GLU had similar moisture content. When these results were re-evaluated considering the T1 values, it was seen that the FC-1.5 CA sample had the shortest T1 value, indicating lower crystalline regions compare to other films. This outcome was also consistent with the X-ray diffraction experiments because it demonstrates that the total crystallinity of FC-1.5 CA was lower than other films (Table 3).
In addition to spin-lattice time T 1 , spin-spin relaxation time T 2 were also measured. T 2 , transverse relaxation time, is the time required for transverse magnetization to reach equilibrium value (Ozel, Dag, et al., 2017). It was possible to get insight on the mobility F I G U R E 3 T1 spin-lattice times of FC films of hydrogen molecules depending on length of T2, where short and T2 indicated immobile structure whereas long T2 represented mobile components. Food products generally show multi-compartmental proton distribution, and inverse Laplace transformation can be used to obtain one-dimensional distribution, relaxation spectra, from decaying magnetization curves. To inspect the proton populations in detail, relaxation spectrum analyses were also performed. Representative 1D NMR T2 relaxation spectra of the FC films ( Figure 4) showed two distinct proton populations in all FC films. Relative areas (RA) of the corresponding peaks were also given in Table 5.
Depending on the magnitude of T 2 values, slowly relaxing component can be associated with non-exchangeable hydrogen of more mobile subgranular polymer components, whereas fast relaxing component can be ascribed to water molecules trapped in the crystalline phase (Ritota et al., 2008). It was stated that fast transverse relaxation time was dominated by a fast chemical exchange between OH groups F I G U R E 4 1 H NMR relaxation times for all the films of compartments of polymer chain, in other words solid-solid interaction (Ilhan et al., 2020;Ritota et al., 2008). Solid-solid interaction covers both inter and intramolecular interaction of polymers.
Therefore, RA1 was shaped under the effect of both starch-starch, chitosan-chitosan (intramolecular-crystallinity), and starch-chitosan (intermolecular-crosslinking). When the results were examined, the relative area (RA) of the first peak of the FC-C and glutaraldehyde containing films (except FC-30 GLU) were lower than the films containing citric acid as the crosslinking agent. As mentioned before, the crystallinity value of FC-30 GLU was the highest according to X-ray diffraction experiments. Based on this, it was hypothesized that the high RA1 might not have been resulted from crosslinking of chitosan starch but from a higher tendency of polymers to interact among themselves.
On the other hand, ascending RA1 with the addition of citric acid might have indicated the enhanced crosslink between chitosan and starch. It is worth noting that, because the second peak is related to water tightly bound to polymers, the decrease in RA2 supports the hypothesis about higher crosslinking between polymers and less interaction with water. Here, it can be concluded that NMR relaxometry and relaxation times might be considered as promising indicator to define quality parameters of films including crystallinity, crosslinking, or water mobility.

| CONCLUSION
This study aimed to compare GLU and CA crosslinking performance on faba bean flour-chitosan-curcumin films. Increasing the crosslinking concentration generally improved the film characteristics (WVP and SD) especially for the films containing CA. FC-30 GLU usually exhibited similar features with the FC-C, particularly WVP and RDC. This might be due to lower crosslinking concentration. On the other hand, CA-1.5C showed advanced attributes including SD, WCA, and WVP. Addition of CA lowered the mechanical strength and brought generally high elongation. However, GLU added films received high TS and lower elongation, which resulted in more brittle character. Antioxidant activity (DPPH and ABTS) of all films was lower than FC-C, which might be related to difficulty in swelling out of curcumin through the film matrix and interaction between crosslinking agents curcumin chitosan. At the end, this might be concluded that FC-1.5C film generally exhibited the most acceptable properties compared to the films with and without a crosslinker. CA might be recommended instead of GLU to improve the film characteristics of water sensitive films. Overall, FC-1.5C films might be suggested as a promising active food packaging material.

CONFLICT OF INTEREST
The authors declare no conflict of interest with any one or any other organizations.

FUNDING INFORMATION
There are no funders to report for this submission.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author up on reasonable request.