Prismatic to Asbestiform Offretite from Northern Italy: Occurrence, Morphology and Crystal-Chemistry of a New Potentially Hazardous Zeolite

: A multi-methodological approach, based upon ﬁeld investigation, morphological characterization, chemical analysis and structure reﬁnement was applied to different samples of ﬁbrous offretite, a new potentially hazardous zeolite recently discovered in northern Italy. Their morphology ranges from stocky-prismatic to asbestiform. All the investigated ﬁbers may be considered as “inhalable”, and they are well within the range of the “more carcinogenic ﬁbers” regarding diameter. As regards the length, the main mode observed in the asbestiform samples is 20–25 µ m, and ~93% of the measured ﬁbers are >5 µ m and may be signiﬁcantly associated with carcinogenesis also in terms of lengths. The chemical-structural features of the investigated ﬁbers are comparable: the extra-framework cations K + , Mg 2+ and Ca 2+ are present in all samples in similar proportions, and reﬁned cell parameters are similar among the samples. Offretite occurs in 60% of the investigated sites, with an estimated amount up to 75 vol % of the associated minerals. The presence of this mineral could be of concern for risk to human health, especially if one considers the vast number of quarries and mining-related activities that are operating in the zeolite host rocks.


Introduction
In recent years, a subject of great interest has been that of inhalable mineral fibers and their potential health effects [1,2]. One of the most important factors for defining the hazardousness of mineral fibers is their size or their tendency to split into smaller fibrils that are easily inhalable [3][4][5][6][7]. The biopersistence of fibers, defined as their ability to persist in the human body to chemical, physical, and other physiological clearance mechanisms, is another important factor for inducing toxicity and carcinogenicity [8][9][10]. Accordingly, it has been shown that an inhalable fiber might be carcinogenic if it is sufficiently durable to remain chemically and physically unaltered within the lung tissue [11]. In addition to biopersistence, the presence and structural coordination of Fe in the mineral fibers were proposed to play a key role in the carcinogenic mechanisms [12][13][14]. Other properties of fibers, including trace elements content, specific surface area, interacting capability, The samples FF102 and AD13 originated from Nogare (VI) and Fittà (VR), respectively, in the Lessini Mountains (northern Italy), where a thick sequence of lava flows of the Veneto Volcanic Province extensively crops out [60,61]. These lavas range from poorly to highly vesiculated and scoriaceous basalts and basanites. Their vesicles are often filled with secondary minerals, which are dominated by zeolites of hydrothermal origin (mainly analcime, chabazite, phillipsite-harmotome, gmelinite, erionite, and offretite). Based on the different secondary mineral assemblages, multi-stage alteration processes have been described in the Lessini basalts [56].
The sample MB2287 originated from Saviore dell'Adamello (BS) and had been found in cavities of an arenaceous metasiltite, metasomatised by the Adamello intrusion. This rock, which is a fine-grained, black cornubianite belonging to the Servino Formation, is characterized by the presence of large druses filled by secondary minerals such as zeolites (chabazite, gismondine, offretite, mesolite, phillipsite) and other phases (calcite, pyrite, actinolite, muscovite). The association offretite-chabazite is already known in this area [53,55]; however, the morphology of offretite crystals in our sample is notably different concerning that described in the literature, and this is the first time that asbestiform offretite is found in northern Italy.
Morphological observations were performed at the University of Urbino Carlo Bo using an Environmental Scanning Electron Microscope (ESEM) FEI Quanta 200 FEG (FEI, Hillsboro, OR, USA), equipped with an energy-dispersive X-ray spectrometer (EDS) for microchemical analyses. Operating conditions were 30 kV accelerating voltage, 10 mm working distance, 0 • tilt angle, and variable beam diameter. The ESEM was utilized in low vacuum mode, with a specimen chamber pressure set from 0.80 to 0.90 mbar. The images were obtained using a back-scattered electron detector. For the sample MB2287, given the very small size fibers and asbestiform habit, morphometric data were collected by accurate size measurements (length and diameter) of the offretite fibers visible in several ESEM images. SEM-EDS analyses were performed on free surfaces of selected crystals and renormalized using a theoretical water content of 19.34 wt %, corresponding to the grand mean values of offretite samples previously analyzed; the final crystal-chemical formula was calculated by 18 (Si + Al) atoms per formula unit (apfu). As previously suggested [62,63], to minimize the alkali metal migration, chemical data were acquired using a low counting time and a raster scan mode to reduce the temperature increase. The reliability of the chemical analysis used to classify the offretite samples was evaluated by using the charge balance error formula (E% within ±10%; [51,64]), the Mg-content [65] and the K-content tests, similarly to erionite [66].
XRPD data were collected for all the studied samples. Crystals were selected under a binocular microscope. They were reduced in powder using an agate mortar and loaded into 0.5 mm diameter SiO 2 -glass capillaries. XRPD data were collected in transmission mode using a D8 Advance diffractometer (Bruker AXS, Karlsruhe, Germany) operating in θ/θ geometry. The instrument is fitted with focusing Göbel mirrors on the incident beam, Soller slits on both incident and (radial) diffracted beams, and a PSD VÅNTEC-1 detector. The Rietveld method was used to determine the cell parameters of offretite. However, in the case of sample AD13 that was found to consist of pure offretite, a full structure refinement was carried out using TOPAS v.4.2 [67] and modeling the peak shape by FPA (Fundamental Parameters Approach). Starting structural model of offretite was taken from [68]. We adopted the same Rietveld refinement procedure used by the present research group in several structure refinements of erionite fibers [12,69]. Owing to the occurrence of correlations between site occupancy factors and displacement parameters, the latter were constrained as follows: Absorption correction was performed following the formalism of [70], and the occurrence of preferred orientation was modeled by spherical harmonics selecting the number of terms to be used according to the procedure devised by [71]. Moreover, the refinement was performed using the ellipsoid-model of [72] describing the diffraction-vector dependent broadening of diffraction maxima. Full structural data of the samples were submitted as CIF file in the Supplementary Material.

Morphology
The FF102 sample contains stocky-prismatic crystals of offretite showing well-shaped hexagonal or pseudo-hexagonal habits with diameters of~50 µm and lengths up to 200 µm (Figure 1a,b). They occur as irregular aggregates of several individual crystals, but isolated single prisms are also observed. In some cases, the crystals termination shows smaller size and irregular morphology due to the presence of a network of fractures. The crystals of this sample have a compact aspect and a rigid behavior and do not originate any fibrous element.
In the AD13 sample, offretite occurs as very elongated, hexagonal (or pseudo-hexagonal) prismatic crystals of solid appearance, with size~50 µm concerning diameter, and up to 500 µm regarding length (Figure 1c,d). Furthermore, in the same sample, acicular to fibrous offretite crystals with diameters <1 µm and lengths up to 500 µm are also present, sometimes grouped in radiating bundles of larger size. These fibrous crystals show a rigid behavior, even if some fibers are curved and seem to display flexibility. In any case, both single prismatic and acicular crystals present a well-shaped morphology, and no fractures are recognizable.
In the MB2287 sample, offretite is found under the form of very thin fibers densely associated in parallel aggregates, which grow as a coating on the main surfaces of platy, rhombohedral chabazite crystals, creating a sandwich-like morphology (Figure 1e). In this sample, offretite is characterized by an extremely fibrous habit, consisting of fibers and fibrils with a diameter ranging from <<1 µm to about 4 µm, and variable lengths (generally about 20 µm; Figure 1f). Accordingly, it might be described as having an asbestiform habit. The coatings are always characterized by a great tendency to separate in fibers and fibrils of the thinner section, with rigid to flexible behavior.

Morphology
The FF102 sample contains stocky-prismatic crystals of offretite showing well-shaped hexagonal or pseudo-hexagonal habits with diameters of ~50 μm and lengths up to 200 μm (Figure 1a,b). They occur as irregular aggregates of several individual crystals, but isolated single prisms are also observed. In some cases, the crystals termination shows smaller size and irregular morphology due to the presence of a network of fractures. The crystals of this sample have a compact aspect and a rigid behavior and do not originate any fibrous element. In the AD13 sample, offretite occurs as very elongated, hexagonal (or pseudo-hexagonal) prismatic crystals of solid appearance, with size ~50 μm concerning diameter, and up to 500 μm regarding length (Figure 1c,d). Furthermore, in the same sample, acicular to fibrous offretite crystals with diameters <1 μm and lengths up to 500 μm are also present, sometimes grouped in radiating bundles of larger size. These fibrous crystals show a rigid behavior, even if some fibers are curved Considering the asbestiform nature of the MB2287 sample, morphometric analyses were performed on about 100 ESEM images to quantify the size (length and diameter) of all the visible fibers separated and deposited on the sampling plate, as well as those partially separated from the main crystals. The results are summarized in Figure 2. As regards the length, about 99% of the total of the measured fibers lies in the range from~2 to~30 µm (average 17.7 µm), while the remaining~1% is longer than 30 µm. With over 65% of the measured fibers, the main mode corresponds to a fiber length between~15 and~25 µm. Regarding diameter, the histogram shows the presence of a large population of fibers (~95%) characterized by a small diameter (<1.5 µm; average of 0.6 µm), and the remainder (~5%) shows a small diameter (up to 2 µm). In particular, more than 30% of the measured fibers have a very small diameter, up to 0.3 µm, with a mean value of 0.25 µm. of the measured fibers lies in the range from ~2 to ~30 μm (average 17.7 μm), while the remaining ~1% is longer than 30 μm. With over 65% of the measured fibers, the main mode corresponds to a fiber length between ~15 and ~25 μm. Regarding diameter, the histogram shows the presence of a large population of fibers (~95%) characterized by a small diameter (<1.5 μm; average of 0.6 μm), and the remainder (~5%) shows a small diameter (up to 2 μm). In particular, more than 30% of the measured fibers have a very small diameter, up to 0.3 μm, with a mean value of 0.25 μm.

Mineralogical Composition and Structural Data
Owing to the difficulty of separate pure offretite crystals, XRPD data indicate that sample FF102 contains impurities of chabazite and calcite, whereas MB2287 has large amounts of chabazite and calcite and minor quartz plus additional clay minerals ( Figure 3). Refined cell parameters (Table 2) are similar among the samples and are consistent with those reported by [51]. Addition of the present data to the unit-cell volume vs. R (R = Si/(Si + Al)) plot of [51] does not improve the poor correlation reported by those authors (Figure 4).

Mineralogical Composition and Structural Data
Owing to the difficulty of separate pure offretite crystals, XRPD data indicate that sample FF102 contains impurities of chabazite and calcite, whereas MB2287 has large amounts of chabazite and calcite and minor quartz plus additional clay minerals ( Figure 3). of the measured fibers lies in the range from ~2 to ~30 μm (average 17.7 μm), while the remaining ~1% is longer than 30 μm. With over 65% of the measured fibers, the main mode corresponds to a fiber length between ~15 and ~25 μm. Regarding diameter, the histogram shows the presence of a large population of fibers (~95%) characterized by a small diameter (<1.5 μm; average of 0.6 μm), and the remainder (~5%) shows a small diameter (up to 2 μm). In particular, more than 30% of the measured fibers have a very small diameter, up to 0.3 μm, with a mean value of 0.25 μm.

Mineralogical Composition and Structural Data
Owing to the difficulty of separate pure offretite crystals, XRPD data indicate that sample FF102 contains impurities of chabazite and calcite, whereas MB2287 has large amounts of chabazite and calcite and minor quartz plus additional clay minerals (Figure 3). Refined cell parameters (Table 2) are similar among the samples and are consistent with those reported by [51]. Addition of the present data to the unit-cell volume vs. R (R = Si/(Si + Al)) plot of [51] does not improve the poor correlation reported by those authors (Figure 4). Refined cell parameters (Table 2) are similar among the samples and are consistent with those reported by [51]. Addition of the present data to the unit-cell volume vs. R (R = Si/(Si + Al)) plot of [51] does not improve the poor correlation reported by those authors (Figure 4).   (7) 7.5868 (6) 1166.02 (15)  Rietveld refinement of samples AD13 indicates a disordered Si, Al distribution because of <Si1-O> and <Si2-O> bond distances of 1.656 and 1.651 Å, respectively. According to Jones' determinative curves [73], the Si, Al distribution at the tetrahedral sites is Si1 = Al4.03Si7.97 and Si2 = Al1.82Si4.18 corresponding to an R ratio of 0.675. This value is slightly outside the reported range 0.69-0.74 of offretite [51] and smaller than 0.724 obtained from chemical data (see below). The structure agrees with that reported by [68]. Differences are related to the minor displacement of the extraframework (EF) cations and H2O located within the gmelinite cages. As far as EF cations are referred to, the refined site scattering (s.s.) of 62.8(11) e − is moderately higher than that obtained from SEM-EDS of 55.5 e − . Similarly, the refined s.s. of the OW sites is consistent with the occurrence of 17.3(2) H2O per formula unit (pfu). This value is slightly higher than the grand mean of offretite samples of ca. 16.3 H2O pfu reported by [51]. It is worth noting that detailed scrutiny of the difference plot of the Rietveld refinement indicates, despite the adopted anisotropic peak shape approach, some difficulty to properly fit the 011, 021 and 211 reflections that could be possibly related to the occurrence of strain deriving from minor fibers curling ( Figure 5). Therefore, we may speculate that the small structural deviations mentioned above may be due to this reason. Rietveld refinement of samples AD13 indicates a disordered Si, Al distribution because of <Si1-O> and <Si2-O> bond distances of 1.656 and 1.651 Å, respectively. According to Jones' determinative curves [73], the Si, Al distribution at the tetrahedral sites is Si1 = Al 4.03 Si 7.97 and Si2 = Al 1.82 Si 4.18 corresponding to an R ratio of 0.675. This value is slightly outside the reported range 0.69-0.74 of offretite [51] and smaller than 0.724 obtained from chemical data (see below). The structure agrees with that reported by [68]. Differences are related to the minor displacement of the extraframework (EF) cations and H 2 O located within the gmelinite cages. As far as EF cations are referred to, the refined site scattering (s.s.) of 62.8(11) e − is moderately higher than that obtained from SEM-EDS of 55.5 e − . Similarly, the refined s.s. of the OW sites is consistent with the occurrence of 17.3(2) H 2 O per formula unit (pfu). This value is slightly higher than the grand mean of offretite samples of ca. 16.3 H 2 O pfu reported by [51]. It is worth noting that detailed scrutiny of the difference plot of the Rietveld refinement indicates, despite the adopted anisotropic peak shape approach, some difficulty to properly fit the 011, 021 and 211 reflections that could be possibly related to the occurrence of strain deriving from minor fibers curling ( Figure 5). Therefore, we may speculate that the small structural deviations mentioned above may be due to this reason.

Chemical Data
The point analyses acquired on the investigated samples are highly consistent, showing a variation of major elements within 2%-3%, indicating a high degree of chemical homogeneity of each sample. The average chemical composition and relative standard deviations (σ) are reported in Table 3, whereas all the collected data are shown in Figure 6.

Chemical Data
The point analyses acquired on the investigated samples are highly consistent, showing a variation of major elements within 2%-3%, indicating a high degree of chemical homogeneity of each sample. The average chemical composition and relative standard deviations (σ) are reported in Table 3, whereas all the collected data are shown in Figure 6.   The average chemical composition of stocky-prismatic offretite of sample FF102 is (K1.22Mg0.75Ca1.13)·(Al4.96Si13.04O36.00)·15.79H2O. The R ratio is in the range 0.71-0.74 (average 0.72), which is consistent with the interval 0.69-0.74 identified by [51] for offretite. The K + and Ca 2+ extraframework (EF) cations are present in a comparable amount, being in the range 1.12-1.30 apfu and 1.73-2.56 apfu, respectively. The Mg 2+ content ranges from 0.70 to 0.79 apfu and the Mg/(Ca + Na)  [51] for offretite. The K + and Ca 2+ extra-framework (EF) cations are present in a comparable amount, being in the range 1.12-1.30 apfu and 1.73-2.56 apfu, respectively. The Mg 2+ content ranges from 0.70 to 0.79 apfu and the Mg/(Ca + Na) ratio, which is considered the most significant chemical parameter for the distinction between erionite and offretite [51], is in the range 0.54-0.88 (average 0.68).
The prismatic to acicular offretite AD13 has an average chemical formula (K 1.16 Mg 0.97 Ca 1.10 )·(Al 4.95 Si 13.05 O 36.18 )·15.70H 2 O. The R ratio lies in the range 0.72-0.73, in agreement with literature data. The K + content (0.91-1.37 apfu) is similar to that of Ca 2+ (0.91-1.21 apfu), but with an average value slightly higher for K + . Moreover, the Mg 2+ content is also high, ranging from 0.88 to 1.11 apfu, while the Mg/(Ca + Na) ratio range from 0. 73 [51]. In this sample, all EF cations K + , Ca 2+ and Mg 2+ are present in comparable amounts. The K + content ranges from 0.91 to 1.06 apfu, while Ca 2+ ranges from 0.89 to 1.05 apfu. The Mg 2+ content is slightly higher than that of the previous samples, ranging from 0.96 to 1.20 apfu, and the Mg/(Ca + Na) ratio is in the range 0.91-1.27 (average 1.13). It is interesting to notice that the analyzed samples show an inverse correlation between both K and Mg and Ca and Mg content. All chemical data plot within the offretite field in the ternary compositional plot ( Figure 6).

Discussion
The main mechanisms by which an inhaled fibrous mineral particle induces pathological changes to comprise (a) physical features (diameter, length and aspect ratio); (b) chemical-mineralogical features (fibre type, chemical composition and surface reactivity); (c) the ability to generate reactive oxygen species (ROS); and (d) the biopersistence. These factors are also deeply interconnected: the fiber dimensions influence its surface reactivity, fiber structure and composition control its biopersistence, and biopersistence is associated with the fiber habit.
Concerning the physical features, the diameter seems to be the main factor controlling the ability of fibers to reach the lower respiratory tract, while macrophage phagocytosis is more dependent on fiber length [7,74]. According to [4], an elongated particle is defined as "inhalable" for humans when its diameter is <3.5 µm, while [19] considers as inhalable the fibers with a diameter-length ratio ≥1:3, a length >5 µm, and a diameter <3 µm. Fibers ≤0.25 µm in diameter and >8 µm in length seem to be the most carcinogenic [3], but the majority of fibers detected in the lung and mesothelial tissues are generally <1 µm in diameter and shorter than five µm in length [5,75,76]. Consequently, small-size fibers should not be excluded from those contributing to the induction of human MM, especially in high exposure environments. This is further supported by the fact that ultrafine fibers are particularly suitable for penetration from the proximal area to the peripheral part of the lung and migration from the lung to the pleura and the other sites within the human [6,77,78]. Therefore, it is difficult to exclude fibers of a particular dimension from those producing disease within the lung or extrapulmonary sites.
In the investigated samples, the habit of offretite varies from stocky-prismatic crystals with a solid appearance of FF102, through prismatic or acicular with the rigid mechanical behavior of AD13, to asbestiform crystals with rigid to the flexible behavior of MB2287. Offretite with prismatic and acicular crystal habits occurs more commonly, representing~84% of the total studied samples, while offretite with asbestiform habit is relatively rare and has been found in~16% of the total samples. However, it is relevant to highlight that asbestiform offretite had never been reported in northern Italy and this work represents its first finding and characterization. The stocky-prismatic and prismatic crystals have diameters of~50 µm and lengths up to 500 µm, while the acicular variety is characterized by the same lengths but very small diameters (<1 µm). The asbestiform crystals consist of fibers and fibrils with a diameter ranging from <<1 µm up to 4 µm, and a length which is in the range from~2 tõ 30 µm (average~18 µm). In particular,~95% of fibers are characterized by a small diameter (<1.5 µm, average 0.6 µm), and more than 30% of the measured fibers have a very small diameter (<0.3 µm, average 0.25 µm). According to [18] indications, all of these fibers may be considered as "inhalable," and they are well within the range of the "more carcinogenic fibers" regarding diameter [3]. The main mode observed in the offretite fibers of MB2287 is 20-25 µm, and~93% of the measured fibers are >5 µm, and therefore they may also be significantly associated with carcinogenesis when breathed in term of lengths [3,10,19]. About the physical features, particular attention also needs to be paid to reliably evaluating the toxicological effects related not only to the dimension of offretite fibers, but also to their flexibility and their tendency to split, features that may significantly modify their surface area and, consequently, their reactivity [18,79]. The various offretite samples show different mechanical behavior, having an extremely fibrous with flexible to rigid appearance or forming prismatic to acicular, brittle crystals. In the mineral fibers, the flexibility might significantly transform the shape of the particles, leading to variations of both aerodynamic properties and physical particle deposition mechanisms [80,81].
The chemical-structural features of the mineral fibers are also of critical importance to evaluate the pathological changes, being related to their durability in the human body. Bearing in mind that the interactions between fibers and biological environment occur on the fiber surface, the surface chemical composition and specific surface area are the most important elements that need to be considered. The chemical composition of the investigated offretite samples is coherent with reference data: R ranges from 0.71 to 0.74, the EF cations K + , Mg 2+ and Ca 2+ are present in all samples in comparable proportions, and Na + , Sr 2+ , Ba 2+ and Fe are absent. All compositions plot within the offretite field in the ternary diagram of Figure 6. However, observing the chemical data in greater detail, it can be observed that the prismatic to acicular crystals have similar compositions, with Mg 2+ content slightly lower than K + and Ca 2+ ; differently, the fibrous offretite is characterized by a Mg 2+ content which is slightly higher than that of the other EF cations. The crystal structure of sample AD13 shows small differences concerning that of prismatic offretite [68], possibly induced by minor curling of the fibers.
The present study indicates that, in different lithotypes of northern Italy, the zeolite offretite may crystallize with asbestiform habit, and therefore may be potentially harmful. This is even more concrete if one considers that many of these rocks are extensively quarried and widely employed for a variety of uses, most commonly as an aggregate. Crushed basalt is added to many construction jobs, including in asphalt paving, mixed with concrete, and as a rock filtering agent in many drain fields. Basalt is also used as the bedrock for railroad tracks, providing drainage and support. As a consequence, further than potential environmental exposure produced by natural weathering processes, occupational exposure during the phases of extraction and processing of offretite-containing rocks should also be taken into account, and additional environmental exposure due to the uses of such material also needs to be considered.
For the above reason, the related host-rocks should be carefully checked before their use as building or construction materials and, to this aim, the role of geoscientists is crucial in guiding safe rock extraction. The present results suggest the need for a detailed Italian mapping of natural sites characterized by the potential occurrence of fibrous minerals, with particular regard to the zeolites erionite and offretite. Furthermore, these data can be used to assess the health risks related to the exposure to mineral fibers during human activities, such as road constructions, quarry excavations and farming that may induce disturbance in the mineral fibers-bearing rocks and trigger unplanned fibers release process.
It is relevant to note that, among the fibrous zeolites, erionite is the only one that was classified as a human carcinogen [38], although other species should not be considered intrinsically safe [82,83]. Despite the lack of epidemiological information on populations exposed to natural asbestiform minerals other than asbestos and erionite, we suggest that all mineral fibers of similar size, habit, and biopersistence may carry a risk for human health.
Finally, we believe the results obtained in this research could be used: (i) to identify potentially hazardous health areas in view of the presence of zeolite minerals fibres; (ii) to provide data for the compulsory Italian mapping of natural sites that are characterized by the presence of the fibrous zeolites; and (iii) to map fibrous minerals (e.g., offretite and erionite) that are classified as non-asbestos and, therefore, not regulated by law but that could be just as hazardous to human health.