*4.1. HSPs Characterization Results*

The results of the HSP drying tests showed that H0210 had a humidity loss of about 9.05%, while for HM200 it was 7.64%. HSPs having lower particle size release less moisture after the drying.

From the TGA results reported in Figure 2, it can be observed that, for both HSP typologies, the thermograms are characterized by a first mass drop (completed below 100 ◦C) that is correlated to the humidity loss of the HSPs. The moisture loss is greater for the H0210 sample, indicating that the HSP having higher particle size dimension releases easily the up taken water. The second mass drop for H0210 corresponds to the thermal degradation of the material and presents a similar magnitude for the two HSP typologies. The final residue is also similar in magnitude for both samples. The residue includes both inorganic compounds and carbon, normally generated when thermal degradation occurs in a nitrogen atmosphere. The superimposed derivatives of the curve show the inflection point (where the mass loss occurs) as a maximum. The main maximum peak is about the same for the two samples; however, HM200 shows an additive peak at around 198 ◦C probably indicating a major quantity of water highly linked on the surface or substances with lower thermal resistance.

**Figure 2.** *Cont*.

Hazelnut shells are composed of cellulose, hemicellulose, and lignin. However, there is a significant amount of low molecular weight compounds. In literature [11], it was observed that hazelnut shell contains about 10.6% of low molecular weight extractable substances, about 30.1% of lignin and about 49.7% of polysaccharides (cellulose and hemicellulose). From the ATR spectra reported in Figure 3, for H0210 a wide band at around 3327 cm−<sup>1</sup> was observed, attributed to the surface hydroxyl groups (-OH) mainly related to the presence of water as well as alcoholic, phenolic groups but also amino acids and carboxylic derivatives.

**Figure 3.** ATR spectra of H0210 and HM200 hazelnut shell powders.

The peak at 2920 cm−<sup>1</sup> is assigned to the asymmetric stretching band C-H; also, that at 2850 cm−<sup>1</sup> is related to the symmetrical stretching of the same bonds. These groups are also present in the structure of lignin [40]. The peak associated with the stretching of C=O (carbonyl compounds) is located at 1708 cm<sup>−</sup>1, but a shoulder is noted at 1743 cm−1. While the main peak is attributable to carboxylic acids, the second is attributable to the presence of ester groups. The presence of unsaturations and C=C bonds that occurred in the widened bands between 1606–1640 cm−<sup>1</sup> is attributable to alkenes, aromatic groups, but also amide groups (C=O stretching); while the peaks at 1400 and 1240 cm−<sup>1</sup> may be due to C−O, C−H or C−C elongation vibrations. The peak observed at 1024 cm−<sup>1</sup> is due

to C–O, present in the ethereal, alcoholic, and carboxylic groups. The band of the C–O group is more intense than that of the C=O group, and this shows that the polysaccharide component is certainly dominated in the sample. The peak at 588 cm−<sup>1</sup> is due to the folding vibration in the aromatic compounds typical of lignin, highlighting their presence.

The spectrum of HM200 was acquired in a similar way to that of H0210, but the signals are more intense. This is attributable to the lower particle size of the powder, which allows better adhesion of the sample to the crystal. The observed bands are completely similar to those of the H0210 sample, suggesting that the only difference between H0210 and HM200 is in the particle size.

The SEM micrographs (Figure 4) show, especially for H0210, the presence of irregularly shaped particles having a rough surface attributed to the external part of the hazelnut shell, which has a different morphology, depending on the filler layers. For smaller-sized samples, the amount of rough surface particles is reduced.

**Figure 4.** SEM images of H0210 (**left side**) and HM200 (**right side**).

Also cavities and reliefs are visible that correspond to the cell walls and lumens. In any case, for both HSPs, a greater variability in the filler shape and size can be observed. The morphological results are consistent to what can be found in literature and despite the different surface roughness, both HSPs can be considered as a typical particle-shaped fillers [41,42]. The "elliptical approach" was adopted to determine the diameter distribution; according to this model, the major axis of the ellipse corresponds to the length of the filler while the minor axis corresponds to the width. With this method, the length and aspect ratio are overestimated by about 10% for the fiber-shaped filler while this overestimation is practically negligible for the particulate filler [34]. Since from the SEM images the greater quantity of HSPs tends to be particle, with little presence of elongated fibers, it was preferred to adopt this elliptical model. In Figure 5 the diameter distribution curves are shown that confirm the great differences in diameters dimension between H0210 and

HM200. In particular, an average diameter of 206.7 μm and 25.8 μm was obtained for H0210 and HM200, respectively.

**Figure 5.** Diameter distributions for H0210 and HM200 HSP.
