Erionite has been found worldwide in many different countries, as shown in Figure 1, yet often in publications, much of the key geological data useful in characterizing erionite (e.g., rock units, paleoenvironment, apparent mode of formation) is missing or incomplete. Notwithstanding these limitations, Italy, Turkey, and the United States are three countries where in-depth analyses into the geological occurrence and characterization of erionite has occurred (Artvinli and Bariş, 1979; Sheppard, 1996; Giordani et al., 2017). In the United States, Sheppard’s (1996) widely-cited work focused only on one geological environment (sedimentary rocks). Subsequently, a more detailed USA-focused geological review of erionite was published by Van Gosen et al. (2013) from the standpoint of an emerging national public health concern for respiratory disease. Elsewhere, global data on erionite is less abundant, but locations, where erionite has been reported include Antarctica (Vezzalini et al., 1994), Australia (England and Ostwald, 1979; Birch, 1987), Fiji (Ram et al., 2019), Finland (Lehtinen, 1976), Georgia (Batiashvili and Gvakharia, 1968), Crimea (Suprychev and Prokhorov, 1986), Scotland (Macpherson and Livingstone, 1982), Northern Ireland (Passaglia et al., 1998), Japan (Harada et al., 1967; Kawahara, 1967; Shimazu and Kawakami, 1967), Kenya (Surdam and Eugster, 1976; Bernhart Owen et al., 2019), Austria (Zirkl et al., 1962; Waltinger and Zirkl, 1974), and New Zealand (Sameshima, 1978; Irwin, 2016; Patel and Brook, 2021). Erionite has been found within rocks used for road aggregates in the United States (Van Gosen et al., 2013), and in the rock used to construct houses in Turkey (Carbone et al., 2007). Hence understanding the geological occurrence, formation processes, and geographic distribution of erionite-bearing rock is important.

As with other zeolites, erionite is usually identified within volcanic and volcanically-derived rocks (Figures 1, 4A), where the minerals typically form via diagenesis or hydrothermal alteration (Mumpton, 1979; Coombs et al., 1997). Nevertheless, erionite has also been identified in sedimentary and metamorphic rocks (Lehtinen, 1976; Sheppard, 1996; Schmieder and Jourdan, 2013). The characteristic host rocks that erionite typically occurs in are basalt and tuffs, but for this review, the volcanic rocks have been classified based on their wt% SiO₂ composition. The classification via SiO₂ is because when describing certain rock types, such as tuff, the definitions can vary according to the author. For example, the erionite in Cappadocia (Turkey) was from, per se loquendo, rhyolitic pyroclastic deposits, yet these same deposits have been referred to as both welded tuff (e.g., Wagner et al., 1985; Topal and Doyuran, 1998) and ignimbrite (Temel and Gündoğdu, 1996). Moreover, the use of the terms “felsic” (silica-rich) and “mafic” (Mg and Fe-rich) are consequently more useful from a geological standpoint (e.g., Marshak, 2019). Unfortunately, however, many erionite-related publications are not focused on geology, but rather on health and toxicity, so even rudimentary geological information is often absent. Thus, as the majority of publications reviewed did not specify host rock composition, these rocks were classified as “undifferentiated volcanics” (Figure 1), with the majority of these undifferentiated volcanics assumed to be “tuff” (Figure 4A). Nevertheless, the host rock type is important, as it can indicate possible erionite fiber size. Indeed, larger well-formed erionite crystals are frequently found within the cavities and veins of volcanic rocks, while fine-grained crystals are more often homogenously distributed within volcanoclastic or sedimentary rocks (Rakovan, 2004; Marantos et al., 2012). Indeed, most larger fibers (>1 mm) are reported from mafic rocks, typically crystallizing within vesicles (e.g., Kamb and Oke, 1960; Tschernich and Wise, 1982; Passaglia and Tagliavini, 1995). Fiber size is discussed in more detail in Section 4.1 below.

From the available literature, within mafic rocks, erionite tends to form within vesicles (e.g., Hey, 1959; Wise and Tschernich, 1976; England and Ostwald, 1979; Rychly et al., 1982; Bargar and Keith, 1984; Noh and Kim, 1986; Birch, 1987; Vezzalini et al., 1994). In contrast, for felsic rocks, most erionite was reported within the matrix of the rock (Figure 4B; e.g., Cochemé et al., 1996; Donoghue et al., 2008; Eyde and Irvin, 1979; Hay, 1964; Mumpton, 1979), but can also be found to form within vesicles (e.g., Barrows, 1980). Most reported erionite occurrences are in vesicles or disseminated within a sedimentary layer, and only very rarely was erionite reported in a vein (Figure 4C). According to Marantos et al. (2012), zeolites forming within sediments typically form the cementing matrix, crystallizing within pore spaces, and this can lead to a harder zeolitized horizon layer, relative to the underlying and overlying sedimentary layers (Davidson and Black, 1994), and therefore are more resistant to erosion. A further intriguing geological factor is the age of the host rock (Figure 5). Most erionite globally is reported from rock units formed during the Miocene epoch (23–5.3 Ma), but erionite has been reported from rocks across a range of geological timescales. However, although the timing of zeolitization must post-date the rock unit age, the exact timing or duration is unknown in most cases. For example, erionite is found within Pleistocene-age (<2.58 Ma) rocks in New Zealand (Rodgers et al., 2004), as well as Precambrian age rocks (1.90–1.87 Ga), such as at Lake Lappajärvi, Finland (Lehtinen, 1976). At Lake Lappajärvi, zeolitization probably occurred much more recently during the late Cretaceous, following the meteorite impact event at ∼77.9 Ma (Kenny et al., 2019).

Erionite (and zeolites more broadly) typically form from diagenesis or hydrothermal alteration, crystalizing from fluids present within the host rock (Giordani et al., 2017; Patel et al., 2022). When hydrothermal alteration causes zeolitization, the conditions include low pressure and temperatures (<110°C). Examples of such locations where this process has occurred include Cairns Bay, Australia (Birch, 1988) and Eastern Rhodopes, Bulgaria (Ivanova et al., 2001; Kirov et al., 2011). Another typical mechanism of zeolite formation is dissolution via diagenesis, which has occurred in a variety of locations globally, including Guanajuato, Mexico (Ortega-Guerrero and Carrasco-Núñez, 2014; Ortega-Guerrero et al., 2015), Reese River, United States (Gude and Sheppard, 1981) and Chojabaru, Japan (Shimazu and Mizota, 1972). Note that for many published studies, the erionite formation processes were not reported by the authors, which, again, limits the possibility of accurate geospatial mapping of potentially hazardous erionite-bearing units. Notwithstanding this, from the literature summarized in Supplementary Table S1, six characteristic geological settings can be recognized for erionite formation, outlined below.

(1) Hydrothermal alteration of silica-rich volcanic deposits—heated hydrothermal fluids alter the surrounding host rock and cause the precipitation of erionite and other minerals (Bargar and Beeson, 1981; Van Gosen et al., 2013). Primarily the fluids are heated from below, and typically the temperature to crystallize erionite is low at around <110°C and can be found within vesicles and fractures of volcanic rocks (Bargar and Beeson, 1981). Examples include sinters such as at Otamakokore, New Zealand (Rodgers et al., 2004), and geysers at Yellowstone, United States (Honda and Muffler, 1970; Bargar and Beeson, 1981). Typically, such host rocks are young and of Pleistocene age (∼2.58 Ma; Bargar and Beeson, 1981; Rodgers et al., 2004).

In addition, two further modes of formation are occasionally reported within the literature and include 5) diagenesis in a marine environment such as in Auckland, New Zealand (e.g., Davidson and Black, 1994), and 6) hydrothermal alteration via meteorite impact metamorphism (e.g., Lake Lappajärvi, Finland; Lehtinen, 1976; Schmieder and Jourdan, 2013).

Occasionally, there is discord within the literature about how erionite minerals (and associated zeolites) have formed. For example, two contrasting formation processes have been proposed for erionite formed within the Miocene Waitemata Group in Auckland, New Zealand (Sameshima, 1978; Davidson and Black, 1994). According to Sameshima (1978), zeolites formed within a bathyal submarine environment via hot spring activity, accompanied by hydrothermal alteration at a shallow burial depth. In contrast, Davidson and Black (1994) proposed that zeolites formed due to diagenesis within a closed hydrologic system, based on the premise that zeolitization was confined to very specific lithological layers. Thus, because zeolites appeared restricted to specific sedimentary layers, rather than being disseminated throughout the surrounding units, the hot spring theory by Sameshima (1978) was deemed to be incorrect.

A further example of conjecture within the literature concerns erionite found in southern Bulgaria, which Ivanova et al. (2001) proposed was formed from low-temperature hydrothermal solutions, heated by hot pyroclastic material (i.e., hydrothermal alteration). In contrast, Kirov et al. (2011) more recently proposed that zeolitization occurred from diagenesis within a closed system. Both hypotheses may be valid as Kirov et al. (2011) explained that temperatures may have risen due to later volcanism, causing hydrothermal fluids to alter host rocks.

The paragenetic sequence of zeolite crystallization within a rock mass can provide insights into the environment that the zeolites formed in because the sequence of zeolite mineral formation is indicative of both the fluid chemistry and host rock chemistry present within the system (Hay, 1963; Surdam and Eugster, 1976; Davidson and Black, 1994; Kirov et al., 2011; Mattioli et al., 2016). The most common zeolites forming alongside erionite are clinoptilolite, chabazite, phillipsite, analcime, and mordenite (Figure 6). Observations from Surdam and Eugster (1976) at Lake Magadi, Kenya, found that erionite was formed in environments that are silica and sodium-rich, but which are low in calcium. Indeed, the most common mineral assemblages within the High Magadi beds are erionite + analcime ± quartz ± magadiite and erionite + analcime + chabazite + quartz. Therefore, the first zeolite to form is erionite, which forms straight from trachytic glass with the addition of H₂O. Furthermore, analcime indicates a sodium-rich environment (Surdam and Eugster, 1976; Davidson and Black, 1994). However, as erionite-Ca is a mineral within the erionite series, the observations by Surdam and Eugster (1976) are specific for that area, given that to form a calcium end-member, calcium would need to be abundant within the host rock and/or fluid.

In contrast, Birsoy (2002) provided a broader definition for erionite occurrence, suggesting that erionite forms in alkaline environments. In particular, three locations specifically reported high pH (≥7) levels, including Kandovan, Iran (Ilgren et al., 2015), Tierra Blanca, Mexico (Ortega-Guerrero and Carrasco-Núñez, 2014), and Tuzgölü Basin, Turkey (Karakaya et al., 2015). Moreover, many other studies have simply referred to an alkaline environment, such as Ashio Tochigi, Japan (Matsubara et al., 1978), Olduvai Gorge, Tanzania (Hay, 1963; McHenry et al., 2020), and the studies from the United States in Durkee, Rome (Staples and Gard, 1959), Eastgate, Nevada (Papke, 1972; Sheppard, 1996), Kildeer Mountain, North Dakota (Goodman and Pierson, 2010; Saini-Eidukat and Triplett, 2014), and Reese River, Nevada (Deffeyes, 1959; Gude and Sheppard, 1981). Taken together, the reports indicate that erionite will likely form within an alkaline-rich environment. Such environments tend to be lacustrine, rather than marine, with alkaline lakes usually found in quiescent or recently extinct volcanic areas where neither water vapor nor acidic magmatic gases can reach surface waters (Stumm, 2004). The occurrence depends on peculiar climatic and geological conditions that allow evaporative concentration of the water (often evaporation much higher than water inputs and in endorheic basins), and on geochemical factors that favor chemical evolution towards an alkaline environment (Jones and Deocampo, 2003).

Such an example has been reported from Auckland, New Zealand, where Davidson and Black (1994) proposed that lithology strongly controlled zeolite paragenesis, with different units having different mineral assemblages. For the volcaniclastic sandstones of the Waitemata Group’s East Coast Bays Formation (ECBF), the paragenetic sequence was clinoptilolite + (mordenite) → chabazite + erionite. This sequence is low in Si and is associated with a closed hydrologic system such as a lacustrine environment, which is more alkaline than marine environments (which are more neutral; Davidson and Black, 1994). This indicates that the sandstone was likely sealed in-between layers of mudstone within the turbidite sequence, creating a closed hydrologic system. The crystallization of zeolites occurred as pore fluid flowed within the rock mass, liberating Si and alkali cations from the volcanic glass within the sandstone (Davidson and Black, 1994). Over time, the fluid composition changed as the alteration of the minerals continued, with the varying cation contents of the zeolite assemblages attributed to changes within the pore fluid chemistry (Davidson and Black, 1994; McHenry et al., 2020). In contrast, within conglomerate beds of the Waitemata Group, pore fluid was less restricted, and the system was an open hydrologic system while also highly permeable, allowing analcime to crystallize from Na-saturated fluids (Davidson and Black, 1994).

Most naturally occurring zeolites are non-fibrous, whereas zeolites such as clinoptilolite, edingtonite, erionite, ferrierite, gonnardite, dachiardite, kalborsite, mesolite, mordenite, natrolite, offretite, paranatrolite, scolecite and thomsonite can be fibrous (Nishido and Otsuka, 1981; Belitsky et al., 1992; Thomas and Ballantyne, 1992; Artioli and Galli, 1999; Armbruster and Gunter, 2001; Betti et al., 2022; Finocchiaro et al., 2022). In particular, epidemiological (Bariş et al., 1979; Carbone et al., 2011) and experimental data (Wagner et al., 1985; Coffin et al., 1992) show that erionite fibers have the highest carcinogenic potency among any other fibers so far studied, including fibers regulated as asbestos. Erionite fibers also have strong fibrogenic potential (Fraire et al., 1997) and biopersistence (Sanchez et al., 2009). Malignant mesothelioma (MM) is a cancer caused by a malignant transformation of the mesothelial cells which are found in the tissue lining the lungs, abdomen, and heart (Carbone et al., 2011; Carbone and Yang, 2012; Attanoos et al., 2018). Pleural MM (cancer of the tissue that lines the lungs) is the most common cancer of the three and is caused by the inhalation of fibrous material such as asbestos or erionite (Bariş et al., 1979; Robinson et al., 2005). In addition, there are also non-cancerous health issues known to be caused by inhaling erionite, such as pleural fibrosis and promoting the production of autoantibodies (Fach et al., 2003; Zebedeo et al., 2014; Ray, 2020).

Particle size is one of the most critical factors determining the toxicity of a fiber. Depending on the size and morphology, the inhaled particles can be deposited in various parts of the respiratory system with very different in situ biochemical conditions (Giordani et al., 2019). In terms of morphology, elongated particles typically pose a higher risk to human health in comparison to spherical particles, as they are more likely to be inhaled and deposited within the lung airway surfaces (Asgharian et al., 2018). For elongated particles, NIOSH has established exposure limit guidelines for asbestos and other fibrous mineral particulates that satisfy the following size requirements: length (L) ≥5 μm and a ≥3:1 aspect ratio of length to diameter (NIOSH, 1994; Beaucham et al., 2018). The World Health Organization (WHO, 1997) also specifies a diameter (w) of <3 μm for particles to be inhalable, and fiber diameter is a critical dimension as the smaller the diameter, the higher the particulate number per unit mass of dust, which will increase the inhalation potential of the fibers (WHO, 1997). The diameter also influences the ability of phagocytes to clear fibers from the respiratory tract, and the density of the fiber aids in determining its aerodynamic diameter (d_ae), which influences the depositional depth of fibers within the respiratory tract (WHO, 1986, 1984; James et al., 1994; Brown, 2015; DeWitt et al., 2016; Belluso et al., 2017; Gualtieri et al., 2017; Di Giuseppe, 2020). The aerodynamic diameter, as defined by Gonda and Abd El Khalik (1985) and modified by Di Giuseppe (2020), is: d a e = d ( 1 2 9 ( 1 ( ln ⁡ 2 β − 0.5 ) + 8 9 ( 1 ln ⁡ 2 β + 0.5 ) ) ( ρ ρ 0 ) where d = fiber diameter, β = aspect ratio (L/w), ρ = particle density, and ρ ₀ = unit density (1 g/cm³; Di Giuseppe, 2020; Gualtieri et al., 2017).

The aerodynamic diameter is critical as not only can it determine where in the respiratory tract a fiber is likely to be deposited, but it also assists in determining the inhalability of a fiber (Millage et al., 2010; Gualtieri et al., 2017; Di Giuseppe, 2020). The inhalability of fibers is important as it determines if a particle may be able to enter the body (Millage et al., 2010). In terms of erionite inhalability, there is a paucity of research, with most published work focused on asbestos and other fibers (e.g., Millage et al., 2010; Shang et al., 2015; Asgharian et al., 2018; Militello et al., 2021; Vigliaturo et al., 2021; Zanko et al., 2022). Nevertheless, from these studies, the maximum inhalable dimensions of a fibrous particle have been determined. For most particulate fibers, a d_ae < 100 μm is considered to be inhalable, and while there exists a lack of studies of ultra-large (d_ae > 100 μm) particles, such particles do not pose a significant health risk due to the limited range while airborne (Millage et al., 2010; Vigliaturo et al., 2021).

Out of the 139 locations of erionite found within the literature, only 37 included information on the morphometry of the erionite fibers, and 95 provided information on the crystal habit (Supplementary Table S1), data necessary for hazard assessment. In the literature, erionite is commonly described as being elongated; however, it is not always fibrous, which is apparent by the number of varying terms used to describe the crystal habit (Van Gosen et al., 2013; Giordani et al., 2016). Terms that have been used to describe erionite include; acicular bundles (Bargar and Keith, 1984), divergent acicular aggregates (Belitskiy and Bukin, 1968), thick fibrous minerals (de Pablo-Galán and de Chávez-García, 1996), radiating clusters (Donoghue et al., 2008), hexagonal rods (Golden et al., 1993), woolly aggregates (Matassa et al., 2015), needlelike (Metropolis, 1986), compact fibrous erionite (Passaglia et al., 1974), tiny needles (Reed, 1937), lamellae in radiating aggregates (Vezzalini et al., 1994), fibrous (Hey, 1959; England and Ostwald, 1979; Birch, 1987; Van Gosen et al., 2013), hexagonal prisms (Tschernich and Wise, 1982), radiating bundles (Passaglia and Tagliavini, 1995; Saini-Eidukat and Triplett, 2014), felted masses (Surdam and Eugster, 1976), bundles of needles (Surdam and Eugster, 1976; Karakaya et al., 2015), bundles of fibrils (Dogan et al., 2006), thick bundles which cleave into blocky rods (Mumpton, 1979), acicular crystals (Deffeyes, 1959; Matsubara et al., 1978; Van Gosen et al., 2013; Giordani et al., 2016) and woolly fibers (Deffeyes, 1959; Staples and Gard, 1959; Shimazu and Mizota, 1972). Detailed descriptions of the erionite habit can be found in Supplementary Table S1.

The fiber morphometric data that has also been reported in the literature is summarized in Figure 7, and a wide range of lengths (Figure 7A) have been reported, ranging from <5 μm (e.g., at Lessini Mounts, Italy Giordani et al., 2016) and at Tierra Blanca de Abajo, Mexico (Ortega-Guerrero and Carrasco-Núñez, 2014) and up to 15 mm at Cape Lookout, Oregon, United States (Wise and Tschernich, 1976; Van Gosen et al., 2013). Fiber diameters show less marked variability, with most <1 μm, irrespective of erionite series (Figure 7B). A bivariate scatterplot of length (L) and diameter (w) is shown in Figure 7C, with a moderately strong positive, linear relationship evident between the two variables. Thus, following the WHO (1997) guidelines, the data indicates that erionite could potentially be a hazard in at least 15 of the reported locations (i.e., L ≥ 5 μm, w < 3 μm, ≥3:1 aspect ratio). The rather limited morphometric dataset should be treated with caution (Figure 7C), but erionite-K may be the most likely erionite series to exhibit an inhalable morphometry.

Notwithstanding the above, potential issues exist in the reporting of erionite morphometric data (Figure 7) that may hinder inter-site comparisons. First, occasionally, the diameter of fiber bundles is reported by authors, as opposed to a single fibril, an example being the erionite morphometric data from Yaquina Head and Cape Lookout in Oregon, United States (Wise and Tschernich, 1976; Van Gosen et al., 2013). The same potential reporting issue occurs for Rock Island Dam, Washington, United States (Altaner and Grim, 1990), with uncertainty as to whether the authors measured either a single fibril or a bundle of fibrils. A second potential limitation in some of the erionite morphometric data is hinted at by the frequency distributions of fiber length. For example, the population distribution of fiber length from a single location may be bimodal rather than unimodal, and this may indicate fracturing due to handling, as at Lessini Mounts, Italy (Giordani et al., 2016). At that site, the primary mode for fiber length ranged from ∼40 to ∼60 μm, yet a secondary mode ranged from ∼10 to ∼25 μm. The secondary mode is likely linked to the fracturing of fibers, which may have occurred when collecting and preparing the samples, and therefore, incorrect natural fiber lengths may then be reported by authors. A compounding issue is then using the sample mean to represent a bimodal distribution (Riffenburgh, 2012), when the means and standard deviations of each mode, along with a mixing parameter, should usually apply (e.g., Ashman et al., 1994).

Biopersistence is the amount of time the mineral fibers reside within the human body following inhalation, and fibers that cannot be cleared from the respiratory tract are considered biopersistent and can therefore accumulate during chronic exposure (e.g., Bernstein et al., 2005). Erionite fibers exhibit biopersistence, and erionite-induced mesotheliomas have similar histology and long latency to those originating from asbestos, though there are still uncertainties in their respective mechanisms of carcinogenicity (Reid et al., 2021). Two key components of erionite biopersistence are 1) biodurability and 2) dissolution in surfactant or physiological fluids (Moolgavkar et al., 2001). Regarding biodurability, longer, asbestiform fibers tend to exhibit high tensile strength and elasticity (Giordani et al., 2017). In addition, in vitro acellular dissolution studies have demonstrated that while chrysotile dissolves faster than amphibole asbestos, Scholze and Conradt (1987) showed that erionite is much more biopersistent than both crocidolite and chrysotile. This is consistent with the reported mineral fiber dissolution rates reported by Gualtieri et al. (2018). For a 0.25 μm thick fiber, the calculated dissolution time of chrysotile is ∼94–177 days, very short if compared to that of amphibole fibers (49–245 years) and fibrous erionite (181 years). Thus, the biopersistence of erionite is important because the fiber can induce carcinogenicity only if it is durable enough to remain physically and chemically intact within lung tissue (Sanchez et al., 2009).

In addition to needle-like particle morphology and biopersistence, a key factor contributing to the toxicity potential of erionite has historically been deemed to be the presence of iron (e.g., Fubini and Mollo, 1995). Indeed, it is believed that the toxicity of erionite is linked to both its fibrous properties and its association with iron in natural deposits (Fach et al., 2003; Sanchez et al., 2009; Reid et al., 2021). One theory is that the toxicity of erionite is associated with iron that accumulates on the surface of the fibers and generates cytotoxic hydroxyl radicals (Fach et al., 2003; Waris and Ahsan, 2006; Pacella et al., 2018). Fraire et al. (1997) reported that erionite from Rome, Oregon, and Pine Valley, Nevada, shows contrasting effects in vivo. The sample from Rome is Fe-rich, whereas the sample from Pine Valley is Fe-poor, and results showed that Rome erionite, with Fe in some form, is more potent than Fe-poor erionite (Fraire et al., 1997). It is thought that iron in mineral fibers may be responsible for carcinogenic activity, namely via reactive oxygen species (ROS) or reactive nitrogen species (RNS) production (Roggli et al., 2010; Pacella et al., 2018). Active iron present at the surface of the fibers promotes the formation of reactive hydroxy species by the surface Fenton reaction chain (Gualtieri et al., 2018). However, the presence of iron in fibrous erionite is currently debated, and experiments by Gualtieri et al. (2018) concluded that erionite fibers may not in fact, contain structural Fe³⁺, but contain Fe³⁺ associated iron-rich impurities. Indeed, Gualtieri et al. (2016) showed that Fe found in some erionite analyses was actually coming from iron-bearing nano-particles on the surface of the erionite fibers. The fact that iron is not found in the erionite crystal structure of natural samples also has a sound geological basis. This is because, during the zeolitization process, iron typically present as Fe²⁺ in the host tuffs is leached, oxidized, and precipitated later as secondary iron-bearing phases like iron hydroxides (Gualtieri et al., 2018).

A range of techniques have been used over the decades to study erionite (Figures 2, 8). These can be generally grouped into techniques used for 1) bulk mineral analysis or 2) techniques for analyses of single mineral fibers (Dogan and Dogan, 2008). Bulk mineral analysis techniques that have typically been applied to the study of erionite include X-Ray Diffraction (XRD), Inductively Coupled Plasma Mass Spectrometry (ICP-MS), and X-ray fluorescence (XRF). Single mineral fiber analysis approaches include Polarized Light Microscopy (PLM), Phase Contrast Microscopy (PCM), Scanning Electron Microscopy with Energy Dispersive Spectroscopy (SEM-EDS), Transmission Electron Microscopy with Energy Dispersive Spectroscopy (TEM-EDS), and Electron Micro Probe Analysis (EMPA). Some example methods and case studies are discussed in the following sections. Screening level analysis of erionite, using PLM is occasionally used but requires high-dispersion refractive index liquids with the appropriate refractive index range (n = 1.460 to 1.480; Berry et al., 2019). PCM has also been used infrequently for screening of soils for erionite. An example is the work of Farcas et al. (2017), who recently reported the use of PCM analysis for detecting erionite from soils sampled in eastern Montana and western South Dakota. Farcas et al. (2017) compared the PCM analysis to PLM analysis of erionite in soils and found the PCM method to be more sensitive than PLM.

Binocular microscopy, PLM, and PCM are unreliable when discriminating amongst different fibrous minerals, such as erionite, offretite, or asbestos fibers, but may be helpful to determine if fibrous minerals are present within a sample (e.g., Berry et al., 2019). However, irrespective of the techniques utilized, Ray (2020) cautioned that erionite is more delicate to handle than asbestos minerals when preparing samples for analysis. For example, rock and soil material must be reduced to a fine powder for analysis by optical microscopy (OM), and based on the preparation of erionite samples from Pine Valley, Nevada, Ray (2020) reported that erionite is fragile and extremely susceptible to over grinding. Indeed, samples were milled to two nominal sizes, 250 μm and 75 μm, and once over-milled, bundles and fibers were destroyed and broken into non-fibrous particles (Ray, 2020). Thus, such fragments would no longer be countable by an analyst during a microscopic examination, which could misrepresent the potential toxicity of in situ erionite material. For the detection of erionite fibers in soils, the fluidized bed asbestos segregator (FBAS) preparation method is often used for both asbestos and erionite fibers (e.g., Berry et al., 2019). In particular, previous research has demonstrated that using an FBAS, even very low levels (0.002%–0.005% by weight) of fibers in soils can be readily detected when followed by TEM (Januch et al., 2013).

X-ray powder diffraction (XRD) is a convenient technique that can reveal detailed structural and chemical information about the crystallography of the material. The information XRD provides is especially advantageous as XRD can analyze the constituents of a bulk sample of heterogeneous rock. The likely minerals present within the sample can potentially be identified from the XRD diffraction patterns using online databases and software developed for XRD. The presence of erionite within bulk rock samples has been identified, which makes it a particularly valuable tool (Beaucham et al., 2018). Bulk XRD will not only provide detailed mineralogy of the composition of rock samples but also identify the different zeolites in a given specimen via their XRD pattern. However, there are fundamental issues in applying XRD on its own. For example, erionite and offretite can occur together and exhibit very similar XRD patterns, and second, low concentrations of erionite may be masked by diffractions from other minerals (Dogan and Dogan, 2008). Thus, XRD should be applied in combination with other methods, such as SEM-EDS, where individual fibers or fiber bundles can also be imaged. A further limitation of XRD is that the minimum amount of mineral needed to be present within the sample is 1%–2%, as any concentrations below this threshold will not be detected (Meeker, 2008; Eylands et al., 2009).

X-ray fluorescence (XRF) spectroscopy is a technique also used to analyze samples to determine their chemical composition. It is similar to EMPA, however, it is not as precise and is typically used for bulk rock analysis (Dogan and Dogan, 2008; Stocker et al., 2017; Oyedotun, 2018). XRF works by using X-rays to excite atoms, which causes electrons to be dislodged from the inner orbital, producing fluorescent radiation (Oyedotun, 2018). The energy of the photons emitted is distinct for the transition between specific electron orbitals within an element, and it can be measured and used to determine the abundance of the elements present within the sample being studied (Oyedotun, 2018).

As outlined above, computing the balance error (E%), the K-content, and the Mg-content is fundamentally important for accurate characterization of erionite, and ICP-MS is routinely used for this purpose (e.g., Dogan and Dogan, 2008). ICP-MS is also used to verify putative erionite detected that has been tentatively identified using other methods, such as SEM-EDS (e.g., Dogan and Dogan, 2008). ICP-MS has also been used to identify trace elements present on erionite fibers that may also play a role in fiber toxicity (e.g., Bloise et al., 2016), as well as studying the possible uptake of arsenite and arsenate (H₂AsO₄) species from aqueous solution in zeolites including erionite (Elizalde-González et al., 2001).

Given the limitations of some of the bulk analysis approaches outlined above, SEM-EDS can provide improved delineation of single minerals within a sample. SEM involves scanning an electron beam over a sample to create an image. The images of rock specimens can provide detailed information on the morphology of minerals, which is an advantageous technique when looking specifically for fibrous zeolites such as erionite (Giordani et al., 2017). Numerous studies have utilized SEM to identify erionite (Pacella et al., 2016; Giordani et al., 2017; Rinaudo and Croce, 2019). These authors used SEM primarily due to the ease at which fibers can be identified within the analyzed samples. For minerals that may look similar, EDS can be used to distinguish the minerals from one another. Additionally, SEM can image the minerals in their natural habitat for freshly fractured samples. Thus, not only can the zeolites themselves be observed, but so can the minerals that surround them. The images can provide details of the zeolite facies mineral assemblage, aiding a better understanding of the zeolites.

Energy Dispersive Spectroscopy (EDS) is a technique used in conjunction with electron microscopy (SEM or TEM). When the beam of electrons hits a sample, it generates X-rays, which are characteristic of each element (Ray, 2020). The EDS detects the X-ray energy and measures the rate at which the X-ray is emitted, producing an EDS spectrum of X-ray energy vs. counts. The spectrum gives the elemental composition of the selected sample. In this way, EDS works to quantify and identify every element within the periodic table except H, Li, and He (Newbury and Ritchie, 2013). Aside from being unable to identify every element within the periodic table, there are other areas where EDS will be imprecise. One of these areas is using EDS to measure an object where the geometry varies, as this can introduce a geometric error that can alter the quantitative results to the point that they become useless (Newbury and Ritchie, 2013). Another limitation is that during EDS analysis, the electron beam has been found to replace cations in prior mineral studies, causing the fibers to become unstable, especially if their diameter is <0.5 μm (Carbone et al., 2011; Pacella et al., 2016; Ray, 2020). This is the case for both SEM and TEM-EDS (Carbone et al., 2011; Pacella et al., 2016; Ray, 2020). Precautions to minimize the error as discussed by Pacella et al. (2016) should be taken into account. For example, when determining the chemical composition, calculating the correction factor for each oxide as a function of its particle size is useful. This will reduce size-dependent errors from arising, especially for smaller particles.

Transmission electron microscopy (TEM) accelerates a beam of electrons through a sample prepared on a small grid to observe small specimens’ morphology and structure. According to Van Gosen et al. (2013) and Ray (2020), some techniques using electron beams, such as TEM, can be less effective on zeolites, as the beam can influence the chemistry and crystal structure of the mineral. While asbestos fibers tend to be thermally stable (Martin et al., 2016), most zeolites are quite sensitive to the electron beam (Ray, 2020). Once the energy of the electron beam collides with the erionite fibers, they deform (Ray, 2020). This degradation caused by the electron beam also influences the chemistry and crystal structure. Indeed in standard TEM analyses of erionite, the selected area electron diffraction (SAED) pattern does not last long enough to be documented and measured to an appropriate standard (Van Gosen et al., 2013; Ray, 2020). As mentioned above electron beams have been found to replace some cations in prior studies of zeolites, especially during EDS analysis and the key to overcoming these problems is stabilization using cryogenic electron microscopy holders (Carbone et al., 2011; Ray, 2020). The cryogenic holders aid in stabilizing zeolite fibers during TEM, protecting them from the energy of the electron beam (Gualtieri et al., 1998; Ray, 2020). The cryogenic holder is simply the addition of cooling by liquid nitrogen. In addition, Ray (2020) has also drawn attention to the issue of over-milling with regard to TEM, which could degrade the natural morphology of the erionite mineral prior to analysis.

Similar to TEM and SEM, EMPA utilizes an electron beam to bombard a solid material and determine the sample chemistry, and can be used on geological materials in situ to acquire data which is quantitative and precise (∼1 μm; Reed, 2005). However, unlike SEM, EMPA requires a smooth polished flat surface for analysis to prevent the imperfections on the surface from interfering with the sample and electron beam interactions (Richter and Mayer, 2012). Limitations of EMPA include the fact that lighter elements cannot be detected, such as hydrogen and helium (Bennell, 2015). Furthermore, similar to EDS, the electron beam may cause the migration of cations away from the beam. Indeed, the movement of alkalis (especially Na) can also cause the concentration of Si + Al to increase, affecting the quantitative determination of the mineral composition (Kearns and Buse, 2012; Campbell et al., 2016). To decrease these effects, Campbell et al. (2016) have recommended operating protocols to determine zeolite compositions, including using a smaller defocused beam with a diameter of 20 μm, as well as prioritizing detection of elements such as Na, K, and Al with the spectrometer configuration first. The chemical data should also be obtained from several individual point analyses on each sample, to determine the homogeneity of the mineral and account for possible cation migration (Campbell et al., 2016). Additionally, as EMPA reports chemical data as oxides of the different elements, further calculations need to take place to determine the mineral formulae (Kearns and Wade, 2021). When the mineral composition has been determined, the results should be evaluated against the charge balance error formula (E%; Dogan and Dogan, 2008; Passaglia, 1970), Mg content test (Dogan and Dogan, 2008), and the K content test (Cametti et al., 2013).

Raman spectroscopy is a qualitative and quantitative technique involving shining a monochromatic laser beam on a sample (Bumbrah and Sharma, 2016), and several erionite studies have utilized the method (e.g., Croce et al., 2013). Typically, for different mineral species, spectral ranges of 4,000–100 cm⁻¹ are recorded (Rinaudo and Croce, 2019). The resulting interaction between the laser and the atoms within the specimen causes the light to scatter, and a fraction of the scattered light changes color (Bumbrah and Sharma, 2016). The changing color is due to a change in frequency caused by energy interacting with molecular vibrations. Raman spectroscopy studies the vibration of atoms to provide information on the chemical structure, phase, crystallinity, and the material’s composition, as each mineral has a unique Raman frequency (Lancelot, 2010). Minimal sample preparation is required for Raman spectroscopy, reducing the chances of sample loss and helping to ensure the original shape of the minerals remains intact (Croce et al., 2015).

In micro-Raman spectroscopy, the laser beam is focused through a microscope. This allows the diameter of the sample being analyzed to be as small as ∼200 nm, and it thus increases the precision when determining where in a sample the laser should be directed for analysis (Kattumenu et al., 2012; Piccardo et al., 2013). Micro-Raman spectroscopy has been applied to samples of erionite from different localities in Oregon and North Dakota (United States) and Cappadocia (Turkey) by Rinaudo and Croce (2019). Rinaudo and Croce (2019) also reported that the technique can be used to observe material lying on top of the mineral structure, which cannot be observed with other analytical techniques. Micro-Raman has also been applied to the study of erionite within various organs (pancreas, spleen, and liver) of mice injected with erionite (Croce et al., 2013). Indeed, Croce et al. (2013) showed that micro-Raman spectroscopy permits the recording of distinct Raman patterns of both crocidolite and erionite fibers in animal tissues and human biopsies, so it is useful in determining fiber exposure of MM patients (i.e., erionite, crocidolite, etc.).

Dogan and Dogan, (2008) provided a seminal review on putative erionite, and reported the probability of erionite being wrongly identified by several authors in the past. This was based on re-analysis, particularly using ICP-MS, to determine if the erionite balance error (E%) is ≤ 10% (Passaglia, 1970; Dogan and Dogan, 2008). The Mg²⁺ content must also not exceed 0.80 atoms per cell, and if it does, the mineral should not be characterized as erionite (e.g., Gualtieri et al., 1998; Dogan and Dogan, 2008). Indeed, Dogan and Dogan, (2008) reported ambiguous definitions, incorrect identifications, and inaccurate reporting of clinical investigations in their review. Moreover, Dogan and Dogan (2008, p. 355) concluded that “if data did not pass either the E% or Mg content test, then we propose that reference to them in the literature be disregarded.”

Given the above stringent reporting caveats stated by Dogan and Dogan, (2008), prior to 2008, the typical erionite identification issues were threefold: 1) erionite was sometimes often confused with offretite (e.g., Birch, 1988); 2) the exact erionite species was not reported (Sheppard and Gude, 1969); or 3), the erionite species was reported, but was incorrect (e.g., Dogan and Dogan, 2008). For example, Sheppard et al. (1974) utilized XRD to originally identify a zeolite overgrowth on levyne as offretite at Beech Creek, United States. Subsequent analysis by Passaglia et al. (1998) found the sample was erionite, based on XRD and EMPA, as well as other techniques. The frequent misidentifications of erionite as offretite have been due to the similar morphologies and crystal structure of these two zeolites. In some cases, despite the reporting of the chemical formula, confusion can still occur, such as the zeolite from Araules, Loire, France. This was initially identified as offretite by Pongiluppi (1976) and subsequently re-identified as erionite by Passaglia et al. (1998). However, Gualtieri et al. (1998) then determined that the mineral belonged on the erionite-offretite border, with no explicit characterization of what it could be. Finally, Dogan and Dogan (2008) classified the mineral as not being erionite because the zeolite sample had an Mg content of 0.83, which did not meet the requirement of Mg < 0.80 to be classified as erionite.