Saffron filament samples were crushed using a pestle in an agate mortar until reaching a fine powder, whereas saffron powder samples required no pretreatment. Around 1.000 g of both saffron powder and saffron filaments was weighed to the nearest 0.001 g. The samples were then oven-dried at 103 ± 2 °C for 4 h and allowed to cool in a desiccator. Once they reached room temperature, they were weighed to the nearest 0.001 g.

Results were expressed as percentage of moisture and volatile matter content according to the following equation:(1)MVOL%=100−mfmi×100.

In this equation, m_i is mass of the sample prior to drying and m_f is the mass after drying.

The analyses were carried out following the specifications indicated in Paragraph 14 of the ISO 3632-2:2010 norm with a UVIKON 943 double-beam UV-Vis spectrophotometer (Kontron Instruments, Milan, Italy). Around 125 mg of dried saffron was transferred to a 250 mL volumetric flask and approximately 230 mL of Milli-Q water was added. The flask was covered with aluminum foil to avoid direct contact with light sources, placed in a cold-water bath (<22 °C), and the content was stirred for 1 h. The solution was diluted up to the mark with Milli-Q water, then 60 mL was filtered using rapid filtration filter paper (Extra Rapida—Perfecte 2, myCordenons, Milan, Italy). The first 40 mL was discarded and the last 20 mL was kept. Half of this volume was withdrawn and diluted up to the mark in a 100 mL volumetric flask.

The absorption characteristics of this solution were measured by recording a spectrum between 200 nm and 700 nm. Maximum absorbances at 440 nm, 257 nm and 330 nm were recorded, corresponding to the absorption of crocin, picrocrocin and safranal, respectively. The results were expressed as the “coloring power”, “bittering power” and “odorous power” of saffron using the following equations:(2)Colouring power=A1 cm1%440 nm=A440 nm×2500mw×Wd%, (3)Bittering power=A1 cm1%257 nm=A257 nm×2500mw×Wd%, (4)Odorous power=A1 cm1%330nm=A330 nm×2500mw×Wd%, where A_λ,max is the maximum absorbance at the desired wavelength, m_w is the mass of the sample in grams, and W_d(%) is the percentage weight of dry sample.

A small portion of the sample was retrieved and placed under the tip of a Nicolet 380 FTIR spectrometer (Thermo Electron Corporation, Waltham, MA, USA). A spectrum between 400 cm⁻¹ and 4000 cm⁻¹ was recorded in transmittance mode using ATR as a sampling technique and carrying out 64 scans with a resolution of 4 cm⁻¹ and performing smoothing operations (15 points).

An appropriate amount of powder or filaments was placed on a standard circular 19 mm pin stub previously covered with a graphite coating and analyzed using a TM4000PlusII Scanning Electron Microscope (Hitachi, Tokyo, Japan) coupled with an EDX microprobe. Nine different areas were selected (3.8 mm × 2.4 mm) with a 50× magnification and for each of them a microscopic image was obtained using back-scattered electrons in low vacuum conditions. Point and aerial analyses at greater magnifications were performed when necessary. Semi-quantitative elemental analysis of saffron’s main components was carried out with the EDX microprobe on all of the nine areas mentioned and mean average values were then calculated.

Principal component analysis (PCA), which is an unsupervised method used to evaluate similarities and differences between samples in terms of elemental composition, was carried out on EDX data acquired as described in the previous section. Score plots and loading plots were obtained from the starting dataset and evaluated. All analyses were carried out with the statistical package STATISTICA 7.1 (StatSoft Italia, Vigonza, Italy).

A total of 13 different metals were analyzed using an Optima 8000 Optical Emission Spectrometer (Perkin Elmer, Waltham, MA, USA): Na, Mg, Ca, K, Al, Fe, Pb, Cd, Cr, Mn, Zn, Ni and Cu.

The amount of moisture and volatile matter in saffron samples is an important parameter in defining the quality of the final product and values outside the recommended ranges are considered as signs of adulteration [13]. The percentage of moisture and volatile matter content ranged between 4.524% and 9.324% for powder samples and between 4.126% and 11.653% for saffron in filaments (Table 3).

All of the values were below the thresholds indicated in the ISO 3632 standard (10% for powder samples, 12% for saffron in filaments) and no noteworthy differences can be observed between powder and filament samples. This was expected for commercial products, which are specifically treated in order to achieve values in the desired range, whereas samples obtained directly from producers often exceed the limits and can have moisture values above 30% [13].

Based on the results of the UV-VIS spectrophotometric analysis (Figure 1), four samples (PPW-1, PFL-1, PFL-2, PFL-12) were placed in category I, two samples (CPW-2, PFL-7) in category II, and another two samples (PPW-4, PFL-4) in category III. All of the others had at least one out of the three parameters which did not meet the criteria needed for classification, therefore no category could be assigned to them and, according to the ISO 3632 standard, these are the samples of the lowest quality possible.

It is worth noting that, of the 13 samples that were denied classification in one of the three commercial categories, 9 (CPW-1, CPW-3, CPW-5, PFL-5, PFL-6, PFL-8, PFL-9, PFL-10, PFL-11) were excluded solely due to the odorous power being outside the specified range, whereas the colouring and bittering power were at least above the minimum value required for category III classification. Only four samples (CPW-4, PFL-3, PPW-2, PPW-3) were denied classification because of low colouring and/or bittering power values.

The FTIR spectra of all the samples show the same features of the reference FTIR spectrum of saffron [24]. This is shown in Figure 2, in which three FTIR spectra obtained on samples of different type and origin are represented. All of the major bands were present and correlated well with the reference spectrum in terms of relative height and position, starting from the broad band around 3350 cm⁻¹ relative to the stretching vibrations of the -OH groups, moving on to the asymmetric and symmetric stretching vibrations of -CH₂- and -CH₃- groups at 2925 cm⁻¹ and 2860 cm⁻¹, the -C=O stretching vibrations around 1720 cm⁻¹ and the C-4-OH glucose residue bending vibration at 1020 cm⁻¹ [24].

Despite the fact that it is difficult to determine saffron quality exclusively by ATR-FTIR spectroscopy [28], the experimental spectra still represent a fingerprint for this type of spice [24]. In this light, it is safe to say that all of the samples analyzed were indeed composed of saffron and no major adulterations were present. Thanks to this authentication, it was possible to proceed with the final two analyses on all of the samples, which would otherwise have been insignificant.

Microscopic investigations on saffron filaments highlighted a homogenous composition between the different samples. The samples were composed mainly of irregular fragments greater than 1 cm in length and to a lesser extent of smaller jagged particles of irregular shape, probably deriving from the main filament (an example is shown in Figure 3a). No floral contaminants such as leaves or stamens were observed in any of these samples, confirming the fact that concealing these types of adulterants is more difficult if the product has not been crushed.

Saffron powders showed a much more heterogenous morphological appearance. Some of them (CPW-2, CPW-3, CPW-5, PPW-1) were composed almost exclusively of the same irregular fragments found in filament samples, only smaller in size (less than 1 cm in length). As in the case of the filament samples, no contaminants were detected in these powders. However, others showed a limited presence of these irregularly shaped fragments. In some cases, jagged particles of non-definable size and shape were more frequent (Figure 3b). In others it was possible to notice the presence of pollen granules (Figure 3c), which are spherical/ovoidal grains with an heterogenous size distribution [29]. Finally, in some cases it was possible to observe floral contaminants, such as in sample CPW-1 (Figure 3d).

These types of contamination arise from the cultivation and processing phases of saffron, highlighting lower quality operational practices or intentional contamination [4,16]. It is often the case that manual operations are substituted by semi-automatic machinery, which certainly speeds up the process, but inevitably gives a lower quality product contaminated with pollen granules [4].

Whereas floral contamination in commercial powder samples was only sporadic, samples PPW-2 and PPW-3 showed an extensive presence of pollen granules and almost none of the typical fragments previously observed. Moreover, as highlighted in Figure 3c, these samples were characterized by the widespread presence of brighter particles associated with heavier elements. EDX analyses of both the bright particles and the pollen granules highlighted the presence of terrigenous elements such as Al, Si, Ca, K and Fe (Figure 4a–c).

Table 4 shows the elemental composition of the saffron samples analyzed, calculated as the average of nine aerial analysis (3.8 mm × 2.4 mm).

Being a fundamental plant nutrient crucial for plant growth [30], potassium was the element present in greater concentrations in all of the filament samples. Lower percentage concentrations were found in samples PFL-3 and PFL-4 mainly due to higher values of chlorine present in NaCl and KCl particles (Figure 4d–f). It is likely that this finding can be attributed to the presence of salt-rich aerosols, typical of maritime cities such as Genoa, where the samples are from.

Another important finding related to filament samples is that, despite the fact that Fe was not identified in any of the nine areas analyzed, punctual analysis revealed the presence of particles containing this element (Figure 4g–i). The elemental composition of the particles analyzed, as in the example in Figure 4i, shows a lower percentage of K when compared to the aerial analyses, but a higher concentration of Al and Si. This is likely due to impurities such as clay soil residues, dust or atmospheric particulate matter which came into contact with the sample during the cultivation or processing phases.

Saffron powders also showed K as the main constituent and lower concentrations were once again associated with a higher chlorine content (sample CPW-1). However, adulteration via addition of inorganic salts is suspected in this case, given that the origin of the sample cannot account for the higher chlorine content as in the filament samples.

Punctual analysis in all of the saffron powders revealed the presence of terrigenous particles such as the one presented in Figure 4a. In the case of commercial powders, no significant differences were observed in this regard when compared to filament samples. Instead, samples PPW-2 and PPW-3 showed a much greater presence of the aforementioned particles and this was also reflected in the higher average concentrations of Fe, Ca, Si and Al. These results are also due to the high number of pollen grains, which are rich in these elements (Figure 4j–l).

The results of the ICP-OES analysis conducted on samples PFL-1, PPW-3 and CPW-5 are reported in Table 5 and Table 6. Based on the results of the previous analyses, one sample of low (PPW-3), medium (CPW-5) and high (PFL-1) quality was chosen for this determination [27].

A clear trend can be observed between the different samples in terms of elemental concentration. Potassium was the most abundant element in all three samples and the one which showed the least variation, with values ranging between 5990 μg g⁻¹ and 6840 μg g⁻¹. Most of the other analytes (Mg, Al, Ca, Fe, Na) showed an increase in concentration moving from sample PFL-1 to CPW-5 and finally to PPW-3. The highest levels of Zn and Cu were also recorded for the latter. All of the other analytes (Cr, Mn, Pb, Ni, Cd) were found below the limit of quantification.

Considering that the elemental composition of saffron can vary greatly depending on numerous factors such as the origin of the sample [26], the values observed for samples PFL-1 and CPW-5 are comparable with literature data from other studies [25,26]. However, the results obtained for sample PPW-3 are systematically higher than the ones observed in other studies for all of the elements, with the exception of K.

Student’s paired t-test was carried out in order to compare the extraction efficiency of the two methods (p = 0.05). The results of the test (t = 0.242 < 2.069) show that it is not possible to state that there is a statistical difference between the concentrations of the metals analysed with two methods of digestion. Despite these findings, it was interesting to note that, following the open vessel digestion of sample PPW-3, an insoluble residue remained at the bottom of the container. This deposit appeared dark grey in colour with black and brownish tips, mainly in the form of small particles. The residue was recovered with a minimal amount of Milli-Q water, filtered and analysed using scanning electron microscopy (Figure 5).

The size of the particles ranged from 50 μm to 300 μm and the spectroscopic analysis revealed the presence of Si, Al, K, P and Fe, accounting for the lower concentrations obtained in ICP-OES analysis with the OV procedure, as observed by a direct comparison of the values. Moreover, through punctual analysis it was possible to obtain a more complete chemical characterization of the particles (Figure 5d–f). Based on the results obtained, the composition of the residue is in accordance with the one of silica sands [31]. First of all, this substrate is only soluble in hydrofluoric acid and could not be dissolved using a 1:10 mixture of perchloric acid and nitric acid by heating to 120 °C. In addition, silica sands are composed mainly of SiO₂, Al₂O₃ and Fe₂O₃, with the presence of other metallic oxides. So, it is possible that the PPW-3 sample was fortified with sands to increase the weight for the same number of pistils used.