During a μ-Raman study of serpentine minerals (Compagnoni et al., 2021), the careful microscopic examination of prismatic lizardite-1T crystals from the Monte Fico quarry, the island of Elba, showed “spongy” intracrystalline microstructure, especially evident in sections perpendicular to the c axis. This observation aroused renewed interest over a mineral occurrence largely studied after its first report (Mellini and Viti, 1994).

Actually, the discovery of abundant euhedral crystals, up to 0.4×0.4×2 mm in size, had allowed several subsequent studies, overcoming the limits constituted by the usual tiny size and poor crystallinity of lizardite.

At Monte Fico, euhedral lizardite-1T crystals occur embedded within an interstitial matrix consisting of variable amounts of chrysotile and minor polygonal serpentine. Viti and Mellini (1997) estimated the chemical compositions (Mg2.79Fe0.092+Fe0.073+Al0.07)Σ=3.02(Si1.9lAl0.09)Σ=2.00O5(OH)3.90,(Mg2.80Fe0.112+Al0.06)Σ=2.97Si2.00O5(OH)4, for lizardite and chrysotile, respectively. Based upon Mössbauer and Fourier transform IR data, Fuchs et al. (1998) proposed for lizardite the improved formula (Mg2.74Fe0.102+Fe0.053+Al0.11)Σ=3.00(Si1.94Al0.05)Σ=2.00O5.05(OH)3.95. The Monte Fico lizardite has been extensively studied from many points of view. Among these, Gregorkiewitz et al. (1996) performed neutron Rietveld refinement at different temperatures. Monte Fico lizardite was used by Rinaudo et al. (2003) as Raman reference material and by Auzende et al. (2004) during high-pressure (HP) Raman investigation. Hilairet et al. (2006) determined the serpentine P–V equation of state. Hirose et al. (2006) studied its mechanical behaviour and microstructural evolution during dehydration reaction, and subsequently Viti and Hirose (2009) performed high-temperature torsion experiments. It has been used in thermogravimetric study by Viti (2010) and in high-temperature Raman and FTIR study of phase transformations by Trittschack et al. (2012). Amiguet et al. (2012) studied its rheological behaviour. Lacinska et al. (2016) used it for study of carbon capture by mineralization. Gailhanou et al. (2018) determined its thermodynamic properties by calorimetry.

Therefore, after so much work, it was amazing to find now, just by optical microscopy, unexpected microstructural features, possibly capable of exposing bias in previous characterizations.

Fresh serpentinites are exposed in two large disused quarries located in Monte Fico (Viti and Mellini, 1997). The serpentinites, fractured by tectonics, consist of subrounded blocks several metres thick, separated by joints filled with polygonal serpentine or macrocrystalline lizardite veins. The blocks, usually massive, preserve well the original peridotite structure, but the original high-temperature minerals were pseudomorphically replaced during low-temperature serpentinization, which mostly occurred on the ocean floor (e.g. Viti and Mellini, 1998). In addition to the previous joint-filling veins, several other kinds of vein occur: chrysotile veins; antigorite veins (Viti and Mellini, 1996); lizardite–chrysotile veins; and more complex veins, where chrysotile, polygonal serpentine and lizardite are present in variable relative amounts.

The most interesting veins are those containing euhedral prismatic lizardite. These veins, which are a few metres long and a few decimetres thick, mostly consist of prismatic lizardite-1T embedded in a fibrous matrix of chrysotile and minor polygonal serpentine, whose amount varies from sample to sample. In the field, prismatic lizardite defines an evident mineral lineation, which locally appears slightly deformed. Following the interpretation of O'Hanley (1991), this assemblage was due to dynamic serpentinite recrystallization assisted by deformation and shearing along active joints.

From one of the samples studied by Viti and Mellini (1997), two thin sections were obtained, parallel (MM91A) and perpendicular (MM91B) to the elongation of lizardite crystals. Perpendicularly to lineation, most of the prismatic lizardite crystals show a stumpy or roundish shape and a greenish colour (Fig. 1a). Conversely, parallel to mineral lineation, most of the lizardite crystals have a long prismatic habit (Fig. 1b) and evident pleochroism (Fig. 1b, left). However, in spite of the clear mineral lineation defined by the preferred orientation of lizardite prisms, radial aggregates of prismatic crystals also occur in both sections (Fig. 1c). This indicates that some lizardite crystals locally grew in a more complex way, with a three-dimensional radial arrangement, possibly indicative of static conditions. In summary, lizardite has a prismatic habit, with crystals elongated along the crystallographic c axis (Fig. 1a and b). The lizardite size ranges from 1 to 3 mm in length by 0.3 mm across, with an aspect ratio of up to 10. Most crystals show a spaced (001) cleavage system perpendicular to prism elongation (Fig. 1b). Lizardite is optically uniaxial with a negative sign and low birefringence (ca. 0.008). Prismatic crystals show parallel extinction and are pleochroic with ε being yellowish (parallel to crystal elongation, i.e. parallel to [001]) and ω being light greenish (perpendicular to crystal elongation). The elongation sign is negative (length fast).

The lizardite crystals are embedded in a light beige to colourless interstitial matrix of fibrous serpentine (Fig. 1a, left, and 1b, left), recognized to consist of chrysotile and minor polygonal serpentine on the grounds of TEM observations (Viti and Mellini, 1997). The amount of interstitial fibrous serpentine varies from place to place and from sample to sample. Post-crystalline deformation is evident, tested by local bending and stretching of lizardite crystals and kinking of the interstitial chrysotile (Fig. 1d).

Observed under the polarizing microscope at high magnification, lizardite reveals an internal microstructure, with “fibrous” (Fig. 1e, left) or “spongy/cribriform” appearance (Fig. 1e, right), for crystals seen parallel or perpendicularly to prism elongation, respectively. This microstructure is mostly evident in sections perpendicular to the lizardite elongation, where the spongy appearance derives from lighter domains (5–10 µm wide), surrounded by darker micrometric rims (Fig. 2a). The microstructure is due to a fine intermixture of phases with different “reliefs”. Under crossed polarizers, the interference figures of sections cut perpendicularly to the lizardite prism elongation give a centred black cross, typical of negative uniaxial crystals cut ⊥ to ε. Furthermore, the lack in the interference figure of the isochromes indicates a low birefringence mineral. Both features are compatible with lizardite-1T, which is the volumetrically prevailing phase. The domain extensions were evaluated processing the microscopic images with the MultiSpec program (Purdue Research Foundation); first, the three available bands (red, green and blue) were imported and, afterwards, 11 areas were selected to characterize the two classes by means of the Gaussian maximum likelihood function. This evaluation, performed on several sections, indicates that lizardite (87 %) prevails over “voids” (13 %). As the accompanying TEM observations show that intermixed chrysotile (Ctl) and polygonal serpentine (PS) fibres do not fill the entire void volume, we suppose a still greater lizardite content, possibly on the order of 90 % lizardite.

Attempts to discriminate the chemical compositions by SEM–energy dispersive spectroscopy (EDS) were unsuccessful, as similar values were obtained everywhere. This differs from what was observed by Viti and Mellini (1997) when comparing lizardite and the intercrystalline, interstitial chrysotile and polygonal serpentine, which develop at a definitely larger scale. Therefore, we tried further characterization by μ-Raman spectroscopy, whose laser beam with a 100× objective has a spatial resolution of 1 µm, comparable with the average size of the two intermixed phases (Fig. 2).

The optical examination of the lizardite thin sections has evidenced an intracrystalline microstructure within the prismatic crystals. The microstructure appears as an intermixture of phases with different reliefs. Basal sections of the most abundant phase produce sharp interference figures, typical of optically negative uniaxial crystals, cut perpendicularly to the optic axis. Obviously, these are the properties expected for lizardite. The minor phase forms micrometric regions with chemical composition very close to lizardite. Although hypothetically distinguishable from lizardite by careful SEM-EDS analysis, the close chemical compositions and possible three-dimensional mutual contamination effects precluded this possibility.

Owing to the size of the intracrystalline microstructure, contamination effects may also be present in the μ-Raman spectra, having a lateral spatial resolution of 1 µm. Sections cut perpendicularly to the crystal elongation (i.e. ⊥-to-ε optic axis) show the two phases as separate lighter and darker regions (Fig. 2). Whereas the major phase is univocally determined as lizardite on the basis of μ-Raman data, the darker domains produce poorly resolved μ-Raman spectra, interpreted as the sum of lizardite and chrysotile (with possible polygonal serpentine too).

TEM observations confirm the presence of chrysotile (plus polygonal serpentine), with further details that reveal a still more complex microstructure. Basically, the optically darker domains consist of veins hosting a network of chrysotile and polygonal serpentine fibres, oriented in three directions rotated by n60∘ relative to each other. A further generation of chrysotile and polygonal serpentine occurs as fibres grown perpendicularly to the previous ones (and therefore parallel to the optic axis). Somewhere, the second-generation polygonal serpentine appears to have grown at the expense of the first-generation chrysotile. When particularly developed, it is recognized as 30-sector polygonal serpentine (Fig. 6a).

Chrysotile and lizardite show tight crystallographic relationships, such as those evident from Fig. 4, suggesting that the two minerals grew in a concerted way. In other words, the lizardite crystals grew up with a skeletal habit and the chrysotile fibres filled the voids, growing epitactically on the lizardite crystals. The possible comparison of these fibres with the lizardite scrolls observed by Wicks and Chatfield (2005) seems unlikely, due to the important differences existing between the two samples. In particular, whereas lizardite scrolls seem triggered by an elastic stress recovery acting on very thin TO-layer packets of lizardite, in the Monte Fico lizardite {100} and {010} facets act as nucleation sites for chrysotile.

In general, skeletal growth of minerals is promoted by fast crystal growth, made possible in a fluid-rich environment. Such conditions occurred in the Monte Fico veins, as suggested by the huge size of lizardite prisms. These conditions would be common for the formation of the present intracrystalline lizardite–chrysotile–polygonal serpentine microstructure and for the formation of the larger-scale intercrystalline lizardite–chrysotile–polygonal serpentine microstructure previously described by Viti and Mellini (1997).

Although definitely pervasive, the minor phases (chrysotile and polygonal serpentine) characterizing the intracrystalline microstructure (probably consisting of more than 90 % lizardite) apparently behave as background noise, unable to induce any evident features in the different characterizations undergone by the Monte Fico lizardite.