Impacts of the size distributions and protein contents of the native wheat powders in their structuration behaviour by wet agglomeration

– The control of the wet agglomeration processes of powders depends on the 8 contribution of the characteristics of the native powders. The objective of this work is to 9 specifically investigate the contribution of the size distribution (d 50 and span) and 10 composition (the protein contents) of the native particles, on the dimensional and structural 11 characteristics of the wet agglomerates. Different fractions of native semolina with contrasted 12 size properties (d 50 and span) and protein content are used as raw materials. The wet 13 granulation is conducted using a low shear granulator and liquid spraying condition at 14 constant dimensionless spray flux. The structure properties are evaluated by the distribution 15 of the measured values of the size, water content and compactness. We observed specific 16 effects of the span and the median diameter of native powders on the size distributions of the 17 wet agglomerates. Using small native particles (d 50 < 250 µm) improves the homogeneity of 18 the size distributions. Using slightly dispersed native particles (span < 1.0) leads to a better 19 uniformity on the size distributions. Using semolina with high protein content could lead to a 20 narrow size distribution by limiting the process of fragmentation and the formation of 21 fragments in the bed. We showed that the mechanisms involved in the agglomeration process 22 are similar whatever the size distribution and the protein content of the native powders. wet agglomeration process. The results demonstrate that the span and the median diameter of the native powder have a significant influence. As the median diameter increases, the ratio of the dough pieces and fragments in the bed increases and the ratio of nuclei and agglomerates decreases. The impact of the protein content is lower. The different structures generated by the agglomeration process result from two major modes of agglomeration: nucleation/growing and dough formation/fragmentation. We show that structures as nuclei or fragments could be individually identify because they differ 430 essentially from their hydrotextural properties. An increase in the amount of dough pieces leads to an increase in the fragments and an increase of the polydispersity of the structures. The configuration leading to the higher amount of the agglomerates population corresponds 433 to the low diameter semolina. The relative size of native particles is the major parameter to 434 modulate the formation of the fragments. In the same way, semolina with higher protein content could strengthen the structure by favouring the stickiness between the semolina particles. The process of fragmentation is limited and the ratio of fragment in the bed

The investigation of the agglomeration mechanisms also requires taking into consideration 48 the changes in the morphology or the hydro-textural characteristics of the agglomerates, such 49 as the compactness or the sphericity. Among the multiple phenomena occurring 50 simultaneously during the wet agglomeration, how to identify a priori the limiting 51 phenomenon or the preferential reactive mechanisms?

52
The wet agglomeration process requires the addition of water to generate capillary bridges, 53 which promotes the cohesion between particles. Agglomeration mechanisms result from the 54 spatial arrangement of the native particles with the binder components, promoting attractive 55 interactions and links. The contribution of liquid bridges between native particles largely 56 overtakes the physical forces and the van der Waals interactions (Hapgood et al., 2003). For 57 food powders, besides the contribution of physical phenomena, the mechanisms also depend 58 on the physicochemical reactivity of the molecules, which strengthens the adhesion forces

84
The aim of this work is to describe the wet agglomeration mechanisms by analysing the 85 experimental investigations, to understand how the initial characteristics (particle-size 86 distribution and protein content) of the native durum wheat semolina impact the 87 characteristics of the wet agglomerates. Experiments were conducted using different durum 88 wheat semolina, with different size characteristics or protein contents. The wet agglomeration 89 process was conducted by using a low shear mechanical mixer. Wet agglomerates are 90 considered as multiphase media described as a solid granular matrix that could be saturated 91 or unsaturated by two fluids: liquid and gaseous phases (Rondet, 2008;Ruiz et al., 2011). The Durum wheat semolina of industrial quality (Panzani group, France) was used as "standard 99 semolina". Four fractions of semolina with different particle size distributions were obtained 100 by sieving the standard semolina over a column of 2 metallic sieves (0.315 and 0.25 mm).

101
The fraction called "small semolina" was collected under the 0.25 mm sieve. The fraction 102 called "low span semolina" was collected between the two sieves (0.25 and 0.315 mm). The 103 fraction called "very coarse semolina" was collected over the 0.315 mm sieve. In addition, 104 the fraction called "coarse semolina" was collected over the 0.25 mm sieve, after sieving the 105 standard semolina over only one sieve at 0.25 mm. Three durum wheat semolina with 106 different protein contents (called "very low protein semolina", "low protein semolina", and 107 "high protein semolina") were also selected. These semolina were produced according to a

113
The selected semolina were characterized using standardized methods (Table 1). The water 114 content was determined according to the approved method 44-15A (AACC, 2000), by 115 weighing after oven drying at 105°C for 24 h. The particle size distributions and their 116 characteristics values (d 10 , d 50 and d 90 ) were measured by a laser particle size analyser 117 (Coulter TMLS 230, Malvern, England) at room temperature. The true density ( ) was 118 measured by using an nitrogen pycnometer. The total nitrogen content (TN) of semolina was 119 determined by the Kjeldahl method, and the crude protein content was calculated according 120 to TN x 5.7 based on the AFNOR method V 03-050 (AFNOR, 1970). All experimental 121 measurements were carried out in triplicate.

122
The selected raw materials allow studying separately the effects of selected variables (median 123 diameter, span, or protein content) to understand the effects of the physical and biochemical 124 characteristics of the native particles on the wet agglomeration mechanisms. The wet agglomeration of durum wheat semolina was carried out in a horizontal low shear 129 mixer (Sercom, France) composed by a mixing a horizontal shaft axis positioned at 6.7 cm 130 from the bottom of the tank (30 cm length, 11.5 cm width, 16.5 cm height), with 14 metal 131 rotating paddle blades (4 cm length, 2 cm width, 7.5 cm gap between 2 blades) (Fig. 1). The 132 direction of rotation of the shaft axis was inverted every 20 seconds to improve the effects of 133 mixing and homogenization of the powder. The process can be qualified as low shear mixer 134 because the mixing conditions do not lead to shear stresses as intense as those sitting within 135 high shear mixers (Rondet et al., 2012).

152
The spraying time (about 2 min) was adjusted with respect to the targeted water content of 153 the wet mass (0.45 g water/g dry matter). After the water addition step, a mixing stage at 100 154 rpm for 10 minutes was conducted to homogenize the entire wet mass. Samples of the wet 155 agglomerates were collected immediately after the end of the mixing step. for 2 min. The size distribution was obtained by weighing the mass of the products collected 165 on each sieve. The weight distribution according to size criteria was expressed as the percent 166 of total weight. Measurements were conducted in triplicate. Immediately after sieving, 167 products were sampled from the remaining on the sieves with 2, 1.25, 0.9, 0.8, 0.71 mm mesh 168 openings, and are then characterized by their water content (w j ) and compactness (ϕ j ).

170
Water content -The water content (g water / g dry matter) of wet agglomerates was

208
In the present study at unique process water content, we observe that the structures obtained 209 on the different sieves are characterized by the same trend.

210
Taking into consideration the specific shape of the particle distribution curve ( 8 called "the dough pieces". Their water content is higher than the plastic limit (w P ) of the 226 semolina. The plastic limit is defined as the water content at which the sample maintains any 227 applied deformation (Atterberg, 1911). The water content is then enough high to ensure their 228 saturation (Rondet et al., 2013).

229
The weight proportion of these different fractions is presented in Table 2 Table 2. For each experiment, these different structures were also characterized by their water 247 content and compactness as a function to their diameter (Fig. 5).

248
The impact of the span of the semolina (at similar median diameter) can be described by

253
The present results demonstrated a significant effect of the span of the semolina on the 254 weight fractions of the different structures ( Table 2). The distributions of the hydrotextural 255 properties (water content and compactness) as a function of the size are in the same range 256 (Fig. 5). Besides, the homogenization of the population of native particles around the median 257 diameter (i.e. lower span) seems to lead to a strong decrease of the population of the "small 258 particles" and of the "fragments" for the benefit of "nuclei" (Table 2)  We investigated the impact of the protein content (between 7.7 and 13.8%) of the native 296 semolina on the wet agglomeration mechanisms by comparing three raw materials (Table 1).

297
The experiments were conducted at similar mean water content ( ) and at almost  Table 2. For each experiment, the different structures are also characterized by their water 306 content and compactness as a function to their diameter (Fig. 7).

307
We observe a slight difference in the particle size distribution curve obtained with the three 308 native semolina, which have different protein content (Fig. 6). compactness for the different structures produced by the wet-agglomeration process (Fig. 8),

326
we observe that structures on each sieve follow the saturation curve (Ruiz et al., 2011). For 327 each structure, the saturation degree is equal to 1.

333
The present work demonstrates that the mechanisms leading to these agglomerated structures

12
The agglomeration mechanism of nucleation/growth, which takes place here, occurs in three contribute to build agglomerates with the same growing mechanism as ever described. This 382 paste/fragmentation mechanism could be compared to a nucleation by downsizing, by 383 difference with the "classic" nucleation, which is a bottom-up mechanism (Iveson et al., 13 dough pieces (Fig. 2). It occurs here a concomitant paste/fragmentation mechanism, which 388 generates fragments, associated to nuclei in the growing process.

389
As mentioned in Barkouti et al. (2014), to identify the major mechanisms contributing to the 390 structure layout during the growing stage, it seems interesting to combine the fluctuations of 391 water content and compactness standard deviations. Figure 10 shows that for all the

4
Values in column with the same letter were not significantly different (P<0.05).  Table 2 8 Impact of the characteristics of the native semolina on the weight fractions of the different structures produced 9 during the wet agglomeration process.

10
Weight fractions of the different structures after agglomeration

Small
Fragments Nuclei Agglomerates Dough pieces