Multi-scale structural changes of starch and proteins during pea flour extrusion

Multi-scale structural changes of starch and proteins during pea Abstract Dehulled yellow pea flour (48.2% starch, 23.4% proteins, d.b.), was processed by a twin-screw extruder at various moisture contents MC (18-35% w.b.), product temperature T (115-165°C), and specific mechanical energy SME (50-1200 kJ/kg). Structural changes of extruded pea flour were determined at different scales by measurements of density (expansion), crystallinity (X-ray diffraction), gelatinisation enthalpy (DSC), starch solubility in water and protein solubility in SDS and DTE (SE-HPLC). Foam density dropped from 820 to 85 kg/m 3 with increase in SME and T (R 2 >0.78). DSC and XRD results showed that starch was amorphous whatever extrusion conditions. Its solubility in water augmented up to 50%. Increasing temperature from 115 to 165°C decreased proteins soluble in SDS from 95 to 35% (R 2 =0.83) of total proteins, whereas the proteins soluble in DTE increased from 5 to 45% (R 2 =0.75) of total proteins. These trends could be described by sigmoid models, which allowed determining onset temperatures for changes of protein solubility in the interval [125, 146°C], whatever moisture content. The SME impact on protein solubility followed similar trends. These results suggest the creation of protein network by S-S bonds, implicating larger SDS-insoluble protein aggregates, as a result of increasing T and SME , accompanied by creation of covalent bonds other than S-S ones. CSLM images suggested that extruded pea flour had a composite morphology that changed from dispersed small protein aggregates to a bi-continuous matrix of large protein aggregates and amorphous starch. This morphology would govern the expansion of pea flour by extrusion. T with SME . Data dispersion is due to differences in die configuration, imposed die temperature, and total mass flow rate.


A C C E P T E D M A N U S C R I P T
τ measured torque (in % of maximum torque) τ empty measured torque when extruder is empty (in % of maximum torque)

Introduction
Pulse legumes (pea, lentil, faba bean) are excellent source of proteins (20-30% dry basis), dietary fibers (10-30%), starch (40-55%), vitamins and minerals, and they are low in sugar, sodium and fat (El-Adawy, Rahma, El-Bedawey, & El-Beltagy, 2003). Pulse proteins are relatively high in lysine and low in sulphur-containing amino acids: cysteine, methionine, and tryptophan (Leterme, Monmart, & Baudart, 1990). Combination of pulse proteins with low-lysine cereal proteins allows a nutritional adequate protein profile. The formulation of gluten-free extruded snacks made entirely from pulse legumes is also an interesting way to introduce pulse to modern consumers demanding vegan, gluten free, non-GMO and protein enriched functional foods.
Extrusion is one of the most versatile and efficient food processing technique used in preparation of starch based foods and especially to process protein-fortified extrudates (Day & Swanson, 2013). After melting, the viscous material is forced through a die where the vapour expands the material to a porous structure. Texture properties of extruded products, envisioned like solid foam, depend not only on density and cellular structure but also on the mechanical properties of the material which constitutes the cell walls of the foam. These properties depend on the morphology of the cell wall material created during extrusion under high temperature and shear. These conditions lead to many biopolymer transformations: starch melting, depolymerisation, protein denaturation, browning, and destruction of anti-nutritional components. With increasing temperature, molecular changes in proteins take place in several steps: association, dissociation, and aggregation of subunits by noncovalent (hydrophobic, ionic, H) and covalent (disulphide, isopeptide) bonds (Stanley, 1989). In the case of extrusion of soy and pea protein isolates, the increasing shear energy promotes protein unfolding resulting in protein re-association (Della Valle, Quillien, & Gueguen, 1994;Fang, Zhang, Wei, & Li, 2013). Besides, moisture content probably increases the mobility of proteins and their cross-linking (Holay & Harper, 1982).
The knowledge about the bonding forces involved in protein changes during extrusion was generally deducted from experiments on protein solubility of extrudates in selective reagents capable to disrupt specific bonds. Most of the studies on legume protein solubility have focused on high moisture extrusion (>50% w.b.) producing meat analogues from soy and pea protein isolates. Recent studies show that the meat analogue structuring is more related to thermodynamic incompatibility, which could be induced by protein-protein interactions (Tolstoguzov, 2016).
Most studies on expanded snacks, made from legume proteins extruded at lower moisture (<25% w.b.), focused mainly on the effect of extrusion variables on density, expansion and functional properties using black box modelling for optimizing. Studies on the changes of protein structure and solubility were performed after extrusion, generally in a narrow extrusion domain, and did not involve Gallant, Bouchet, Gueguen, & Melcion, 1991;Della Valle et al., 1994;Gujska & Khan, 1991). More recently, protein aggregations have been studied on pulse legume based pasta extruded at moisture content between 30 and 35% (w.b.) and low temperature (<100°C) (Laleg et al., 2017). However pasta are not expanded product. The diversity of protein sources and processing conditions make the comparison of results difficult and sometimes contradictory. Published results about the effect of protein content on sectional expansion index (SEI) showed both negative and positive impact. Paton & Spratt (1984) found that for extrusion of gluten-wheat starch blend (moisture content MC 21%, 163°C, 50 rpm), the SEI increased from 3 to 5 when adding 10% gluten. Conversely, for extrusion held at the same level of MC but higher mechanical energy (230 rpm The first objective of this work is to ascertain the effect of biopolymer transitions on the structure of extruded pea flour. The structure was studied at two levels: (1) the cell walls of foams constituted by starch-protein blends, and (2) the overall expansion. Second objective is to determine the relationship between extrusion variables, product structure, and material transformations on a mechanistic basis. A wide range of extrusion conditions was selected and resulting products were thoroughly analysed at different levels of matter organization.

Raw materials
Dehulled yellow split pea grits was obtained from Sotexpro (France). The pea (variety Karpate) was harvested in spring 2016. The pea grits were ground by a miller (Sarl Giraud, France) in order to obtain pea flour with median diameter of about 480µm.

Chemical composition (in dry basis)
Total crude protein content of pea flour was obtained using Kjeldahl procedure with a nitrogen-to-protein conversion factor of 6.25. Total starch content was determined with an enzymatic assay kit (Megazyme, Co., Ireland). Ash content was determined by incineration at 900 °C for 2h according to the French standard (NF 03-720). Total lipids were extracted according to method of with relative standard deviation lower than 5%. Fibre content was calculated by subtraction of average content (% d.b.) of ash, fat, protein and starch from total chemical content.

Extrusion trials
Pea flour was extruded using a laboratory scale co-rotating twin-screw extruder (Coperion Werner & Pfleiderer ZSK 26Mc). The screw diameter was 25.5mm. The barrel was divided into 7 sections with a total length of 740mm. During the experiments, each barrel section, except the first one, was heated separately to 40, 60, 80, 90, 90, and 90°C. The raw material was fed by a gravimetrically controlled feeder (Brabender DDW-DDSR 40). Water was added into the second barrel by a volumetric pump. The screw profile included (i) conveying elements with decreasing pitch from 36 to 24mm, (ii) followed by three 45° kneading discs (length of 12mm each) located inside the fourth barrel, (iii) three reverse elements having pitch and length of 12 mm located inside the fifth barrel, (iv) conveying elements having pitch of 24mm. This geometry assured (i) homogeneous mixing of flour and water at the inlet (ii) melting dough completely in the stagnant area generated by reverse elements and (iii) conveying the molten phase to die.
In order to obtain extrudates having a wide range of structures and biopolymer transformations,a large domain of operating parameters was applied: moisture content 18-35% w.b., screw speed 300-700 rpm, total feed rate 10-25 kg/h, extruder configuration with (A) and without (B) restrictive elements, and imposed die temperature 130-150°C. Two die geometries were used: die C consisted of a long slit die (thickness 3mm, width 15mm, length 72mm) followed by an orifice die (Ø 3mm, length 15mm), die D consisted of an orifice die mounted directly to the extruder head without slit die. Extrusion parameters are summarized in Table 1. The interval of variations of these parameters was chosen by numerical screening using a twin-screw simulation software Ludovic ® (Vergnes, Della Valle, & Delamare, 1998), in order to cover a wide range of extrusion variables, such as product temperature 115-165°C and specific mechanical energy (SME) 40-1200 kJ/kg. In addition, the knowledge of pea starch melting temperature at various moisture contents was used for setting extrusion conditions to obtain extrudates with amorphous starch mainly under low SME (< 100 kJ/kg) (Logié, Della Valle, Rolland-Sabaté, Descamps, & Soulestin, 2018).
In particular, experiments using relatively high moisture content (between 26 and 35%) were performed to produce dense material samples (low expansion) that can be considered as the constitutive material ("parietal material") of the pea flour foams (expanded samples). Die pressure and product temperature were measured just in the last section of the extruder with an accuracy of +0.25% and +0.5°C, respectively. Specific mechanical energy (SME, kJ/kg) was calculated from screw speed, measured torque and power as follows: where N and N max are the actual and maximum screw speed (N max = 1800 rpm), τ and τ empty are the actual and empty measured torque . The maximum power of extruder P max is 40 kW. Q is the total mass flow rate (kg/h).
The extrudates were dried under vacuum at 40°C overnight in order to get final moisture content less than 10% w.b. and stored in airtight plastic bags at room temperature. A part of each extrudate was ground into a fine powder using a grinder (M20, IKA Werke GmbH & Co., Germany).
The ground samples and pea flour were sieved to pass a 250 µm mesh and conditioned at a relative humidity of 59% at 20°C for two weeks.
In this work, macro structure of extrudates was characterized by density and SEI. The bulk density was obtained using glass beads displacement method. The extrudate density (ρ * ) was calculated from the mass of extrudate and its volume measured through the displacement of glass beads (Ø 1mm). The reported values (Table 2) were the averages of three repetitions (relative error 10%). The SEI was calculated as the ratio of cross-sectional area of extrudate to that of die. For each sample, diameter was measured at five random positions on each of seven pieces of extrudates. The reported SEI values were the averages of these thirty-five measurements (relative error 15%).

Melting transition
Pea flours with various moisture contents (5 to 35% wet basis) were prepared under controlled relative humidity of 11 to 97% over saturated salts at 20°C for 1-2 weeks. DSC Q10 (TA Instrument, France) was used for melting point analysis following the method of Barron et al. (2000). The samples were heated from 25 to 180°C at a heating rate of 3°C/min. Melting temperature (T m ) was defined as the offset temperature of endothermic peak. The measurements were done in duplicate (relative error 10%). The degree of starch transformation (crystallinity loss and granular disorganization) was determined by duplicate measurements of DSC gelatinization enthalpy of extrudates, in excess water, compared to untreated pea flour. Absence of residual gelatinization enthalpy means that starch is amorphous. The DSC scans were run at 3°C/min from 10 to 120°C.

Crystallinity state by WAXS
The loss of crystallinity was also examined using a wide angle x-ray scattering analysis (WAXS) (Barron et al., 2000). The analysis was performed using a Bruker D8 diffractometer (Bruker AXS Inc., USA) equipped with a copper source operating at 40 kV and 40 mA.

Water Solubility Index (WSI) and Water Absorption Index (WAI)
Water solubility index of total solids (WSI solids ) and water absorption index (WAI) of flour and ground extrudates were determined using the method of Anderson et al. (1970). Mass of 2.5 g of samples was dispersed in 30 ml of distilled water, then agitated at 30°C for 30 min and finally centrifuged at 3000 g for 10 min at 20°C. The supernatant was dried overnight in a Chopin oven at 105°C. WAI was calculated as the ratio of mass of wet pellet to dry original sample. WSI solids was determined as the percentage of mass of supernatant dry solid to dry original sample.
For determining WSI of starch (WSI starch ), the samples underwent firstly the same extraction and centrifugation procedure as described before. Thereafter, the supernatant was titrated with orcinol in an automatic spectrocolorimeter, according to the method of Tollier and Robin (1979).
The measurements of WSI and WAI were carried out in triplicate (relative error 15%).

Procedure of protein extraction
Proteins were extracted using a method adapted from Morel et al. (2000). Samples underwent two successive extractions. The first stage, aiming to extract SDS-soluble proteins, was conducted at As explained in detail later, D8 has been chosen because it has maximum total extractable proteins.

Measurement of molecular weight distribution of proteins aggregates
The size distribution of protein aggregates in the extracts was studied by Size-Exclusion High-Performance Liquid Chromatography (SE-HPLC). The SE-HPLC apparatus (Waters LC Module1 plus) was equipped with an analytical column TSK G4000-SW, 7.5 x 300 mm (Merck, France) and a guard column TSK G3000-SW, 7.5 x 75 mm (Merck, France). Apparent molecular weights were determined in the column calibrated with protein standards (Redl, Morel, Bonicel, Vergnes, & Guilbert, 1999). Once corrected for their different solid-to-solvent ratios, the integrated areas of SDSsoluble and DTE-soluble proteins of samples were expressed as percentage of the total area of the reference sample D8 giving the maximum total extractable proteins. The solubility experiments were duplicated (relative error 5%).

Color measurement
Color changes of ground samples (sieved between 250 and 500µm) were assessed by measurement of CIE Lab color space (L*, a* and b*) using a chromameter (Konica Minolta CR-400) with a standard illuminant D65 (natural daylight), and an observation angle of 2°. L* stands for lightness black-white (0, 100), a* and b* for color opponents green-red (-100, +100) and blue-yellow (-100, +100), respectively. Color measurements were done in triplicate (relative error 1%). These color changes could reflect Maillard reactions occurring between reducing sugars and free amine groups due to thermo-mechanical treatment.

Morphology of starch-protein blends
Confocal laser scanning microscopy (CLSM) was used to locate the starch/protein phases in the constitutive material of extruded samples. The extrudates were cut perpendicularly and embedded in paraffin using a method adapted from Ben Hdech et al. (1991). The slices of 10 µm were then cut using a rotary microtome. Starch was stained red with oxidation in periodic acid solution followed by coloration with Acriflavine-Schiff reagent, called as PAA staining. Proteins were stained green with 2005). Images were acquired using a confocal laser scanning microscope (Nikon A1) with attached NIS imaging system (Nikon, Germany). Samples were examined in the epifluorescence mode of the microscope, excited by a green laser beam at 561nm for Fuchsine Acid stain and at 488nm for PAA stain. The emitted light was selected by a long-pass filter (>570nm). Three-dimensional images of samples were obtained by observing 20 planes of 1µm thick, every 1µm. Each image corresponds to the projection of these 20 planes of a sample area of 318 x 318 µm 2 .

Statistical analysis
The significant difference between means of measured extrusion variables (Table 1) within a grouped samples extruded using same extruder and die configuration was determined by analysis of one-way ANOVA and Fisher's least significant different (LSD). Analysis of one-way ANOVA was also conducted to determine effect of several levels of extrusion variables on protein solubility (Table   3). All statistical analysis were performed at 5% significance level using Microsoft Excel 2010.
The minor components were ash (2.

Flour melting, extrusion and expansion
The measured melting temperature of pea flour T m was 118°C (MC = 20% w.b.), defined as the temperature at peak offset, was slightly higher than those found by Li & Ganjyal (2017)  Extrusion parameters and variables are summarized in Table 1. The selected processing conditions, using screw configuration A, resulted in a wide domain of extrusion variables: product temperature T varied between 120 and 165°C, MC between 18 and 26%, and SME between 400 and 1200 kJ/kg (Fig. 1a). The most severe extrusion conditions values corresponding to highest temperature and SME were obtained for sample C1 with low moisture content (21.4%). Conversely, samples E1-E4 were processed at lower temperature (<130°C) and low SME (<100 kJ/kg) using screw configuration B and larger MC (≥26%). Under all conditions, product temperature at the die T was higher than T m . The extrusion domain exceeded that covered in the previous studies on pea extrusion Fair correlation (R 2 =0.74) between T and SME was found, whatever the values of other variables (Fig. 1a). Indeed, higher T and SME values were obtained for lower MC. This general trend, which has not been underlined yet for legume extrusion, is common to starchy products. It is due to the influence of solid friction between flour particles and viscous dissipation in the molten phase (Colonna, Tayeb, & Mercier, 1989).
The SME effect on the proteins transformation is complex. The heat due to viscous dissipation can enhance the polymerisation whereas shear causes depolymerisation. Therefore, it is interesting to find how to control SME by processing parameters. This is illustrated in Fig. 1b, where SME was plotted as a function of the ratio of total flow rate to screw speed, Q/N, which is proportional to the

ACCEPTED MANUSCRIPT
filling ratio of screw for a given screw configuration. Good correlation (R 2 =0.86) between SME and Q/N was obtained for the screw configuration involving reverse elements, whatever MC and die geometry. The same trend was observed in the case of twin-screw extrusion of starchy products: SME increases when screw speed increases and decreases when feed rate is inc reased (Meuser & Wiedmann, 1989). Obviously, the SME depends on screw filled length, which changes with screw configuration.
In the applied domain of thermomechanical conditions, the density varied in a large interval (85-1060 kg/m 3 , Table 2). Samples extruded with the configuration B (no restrictive elements) were not expanded, likely because of low temperature and SME. They are called dense samples. Extruded samples with density lower than 200 kg/m 3 had similar aspect to directly expanded starchy snacks or foams. Higher expansion is generally preferable because it favours the obtaining of lighter and crispier products.
Density was negatively correlated with product temperature (T) (R 2 =0.77, Fig. 2a). The same trend was observed for SME (R 2 =0.80 result not shown). On the other hand, MC was correlated positively with foam density (R 2 >0.75 result not shown).
Foam density was inversely related to sectional expansion index (SEI) (R 2 =0.91) for any die geometry (Fig. 2b). This general trend that is known for starchy product extrusion has not been evidenced yet for legumes. Calculation of longitudinal expansion index using Eq. (2) Unlike starch melt, pea flour melt cannot be considered as a fully homogenous viscous fluid, this will be discussed in the later discussion about the morphology of parietal material of extrudates (section 3.4). According to SE-HPLC elution profile (Fig. 3), SDS-soluble extract was divided into six major fractions: S1 (1.2%), S2 (6.8%), S3 (32.3%), S4a (13.7%), S4b ( Legumins are hexameric proteins with subunit linked by disulphide bonds, 300-420 kDa, which contribute to S2, and Albumin 1 (cysteine-rich peptide, ~10 kDa) is part of S5.

Protein modifications
Protein solubility of dense samples (E1-E4), extruded at low temperature and SME values was close to that of the native flour (Table 2). With increasing product temperature, beyond 140°C, SDSsoluble proteins decreased from 95% to 35% (R 2 =0.83) (Fig. 4a). Conversely, DTE-soluble proteins increased from 5% to 45% with temperature (R 2 =0.75). Given the strong correlation between SME and T, the same trends were observed on variations of protein solubility with SME (result not shown). This result indicated the formation of SDS-insoluble protein aggregates via protein cross-linking by disulphide bonds, which was confirmed by the negative correlation between solubility in SDS and DTE (R 2 =0.93, result not shown). The best fitting for solubility variations with temperature was obtained using the sigmoid function (see Fig. 4a, R 2 ≥0.75): Sol. = a/(1 + exp(-k(T-T c ))) where Sol. is the solubility (in SDS, DTE…), T is the material temperature measured inside the die and  (Table A1). Onset temperature of the events governing the variation of protein solubility, T i , was found using the sigmoid model, see Appendix 3. The loss of protein solubility in SDS started at 146°C, while the onset of formation of DTE-soluble proteins occurred at 125°C. The variation of solubility with SME could be fitted with a similar sigmoid function (result not shown). By subtracting SDS and DTE-soluble proteins from the total protein content in samples (100%), the fraction of non-extracted proteins was found. It resulted from the formation of aggregates linked by other covalent bonds than S-S. This fraction could reach up to 26 % (sample C1, Table 1). Formation of these bonds started massively at a temperature close to 152°C.
The impact of material temperature on size distribution of SDS-soluble proteins of extrudates is illustrated by the variation of fractions identified by SE-HPLC (Fig. 4b). Increasing material temperature led to a marked decrease in the SDS solubility of S1, S2, S3 and S4a fractions following Eqs.
(3) (R 2 >0.85, Fig.3). The model parameters are reported in Appendix, Table A1. Conversely, ANOVA analysis showed that protein fraction of S4b and S5 did not vary significantly with temperature (Table 3)  Stanley, 1989) and gluten (Pommet, Redl, Morel, Domenek, & Guilbert, 2003). Increasing SME favoured proteins cross-linking by S-S and non S-S covalent bonds due to the enhancement of unfolding and re-association of protein aggregates under high shear force (Anderson & Ng, 2000;Camire, 1991;Della Valle et al., 1994). Although SME is directly correlated to T because of solid friction and viscous dissipation, our results suggest that it is the temperature which influenced directly these changes. Moreover, the ANOVA analysis did not show any significant effect of die geometry (configuration C and D) on protein solubility (Table 3)

A C C E P T E D M A N U S C R I P T
on the fact that protein aggregates with S-S bonds did not vary much with extrusion conditions. This discrepancy can be attributed to different methods of protein extraction and quantification. In the earlier works, neither sonication-assisted extraction procedure nor precise SE-HPLC for solubleprotein quantification was performed. Moreover, different reagents (DTT in citrate phosphate buffer vs. DTE in SDS-phosphate buffer) and extraction temperature were implemented to disrupt S-S bonds.
The use of DTT in the absence appropriate denaturant (urea, guanidine, or SDS) is less effective because it cannot reduce buried (solvent-inaccessible) disulfide bonds (Hermanson, 2013).

Total solubility, color and starch changes
Native pea flour displayed peaks at 5.6°, 9.8°, 11.3°, 15.2°, 17°, 18.2°, 21.9°, 23° and 26.3° on the WAXS diffractograms (Fig. 5a). These peaks are typical for C-type crystallinity pattern of starch Extruded starches were partially soluble with WSI 50%, i.e. lower than WSI of solids. Both indices increased with temperature (Fig. 6), and also with SME (result not shown). Both temperature and shear energy are known to increase the breakdown of starch granules and macromolecules and subsequently to raise the amount of water-soluble components (from 12 to 55%) compared to starch in native flour (11.7%). Moreover, WSI of total solids was negatively and linearly correlated with WAI 250°C] WAI of extruded corn starch increased with WSI, reaching maximum at 175°C. We found that the increase of both WSI indices with material temperature could be fitted sufficiently with sigmoid model (Fig. 6). This model allowed determination of onset temperature of starch solubility increase, close to 115°C. This value is lower than the onset temperatures of protein solubility decrease and formation of S-S bonds. The interval between onset temperature of starch and protein changes can be Extruded materials were darker as shown by the decrease of L* values from 87 to 58 and more reddish and bluish (higher a* and lower b*) with increase of T (Fig. 7a). Moreover, correlations of a* with protein solubility in DTE (R 2 =0.80) and non-extractable proteins (R 2 =0.80) showed that protein aggregations by other covalent bonds occurred at the same time as darkening and increase in redness,

ACCEPTED MANUSCRIPT
i.e. browing (Fig. 7b). Extrudate browning can be related to the development of Maillard reactions between reducing sugars and free amine groups (lysine). Reducing sugars can result from starch breakdown during extrusion (Camire, 1991). The higher intensity of thermo-mechanical treatment, the higher was the transformation of starch and proteins and the interactions between their derivatives.
Consequently, Maillard reactions can be related to the reduction of lysine availability in the extruded pea (Asp & Björck, 1989;Meuser & Wiedmann, 1989). Lysine is one of the most sensitive and most limiting amino acids in humane diets, so lysine availability is one of nutritional quality criteria for processed legumes (Asp & Björck, 1989;Hood-Niefer and Tyler, 2010;Laleg et al., 2017). Finally, extrudate color can be utilised as indicator of protein aggregation by covalent linkages others than S-S ones; in a first approach for sample screening before analysis of protein solubility using SE-HPLC.

Cell wall morphology
The morphology of starch-protein blend in the cell wall materials of expanded samples was ascertained using CSLM for two samples extruded at similar T, higher than the onset of proteins aggregation by S-S bonds: sample C12 (550 kJ/kg, 136°C) and D11 (720 kJ/kg, 140°C). Sample D11 had slightly higher starch and DTE-protein solubility, likely due to higher SME. For C12 (Fig. 8a).
CSLM micrograph shows small protein aggregates (≈10 µm, green) dispersed among some swollen   & Lourdin, 2006). These authors found that the starch-protein blends with co-continuous phases morphology of (like sample D11) had higher elongational viscosity than the blends with the morphology of particles dispersed in a matrix (sample C12). This interpretation was comforted by the fact that sample D11 was extruded with the circular die D, where, due to section contraction, elongational strains were larger than for the slit die C with prevailing shear. Clearly, this interpretation has to be confirmed by studying the rheological behaviour of pea flour envisioned like a composite material of protein aggregates with amorphous starch and by ascertaining the morphology of these materials.

ACCEPTED MANUSCRIPT
where ° is the melting temperature of anhydrous starch (in K), R is the gas constant (8.31 J/mol.K), Δ is the melting enthalpy per mole of monomer unit (or anhydroglucose) (in J/mol), V u /V 1 is the ratio of molar volume of the monomer unit to the molar volume of water (dimensionless), and 1 is the  Table A1. Fitting of measured extrudate properties

Graphical abstract
The design of high-protein expanded foods is still a challenge, likely due to the reactivity of protein components under thermomechanical treatment and its effect on extrudate structure and texture. Material temperature governs concomitant starch and protein changes during pea flour extrusion. Protein insolubilization, mainly due to S-S bonds, is triggered at T≈125°C, and contributes to expansion by morphology changes. CSLM micrographs of two samples extruded under similar conditions show different expansion (density *) and morphology of amorphous starch/protein aggregates. These differences are probably due to slight variations in specific mechanical energy (SME), moisture content (MC) or strain at the die. Highlights  Extrusion temperature, via dissipated energy, governs starch and protein changes  Protein changes lead to various amorphous starch-aggregated proteins morphologies  Morphology of starch-protein blend affects expansion through rheological properties  Onset temperature for biopolymers changes was determined from models