Serpentinite, asbestos and the geology of the Rechnitz Window

In the pure magnesium system, the hydration of olivine is described by the model reaction 2 Mg₂SiO₄ + 3 H₂O = Mg₃Si₂O₅(OH)₄ + Mg(OH)₂, that is, forsterite plus water to serpentine plus brucite (Frost and Beard 2007, reaction 13; Klein et al. 2009, reaction R13). If additional SiO₂ is available, for instance from the simultaneous hydration of orthopyroxene, the reaction proceeds via 3 Mg₂SiO₄ + 2 SiO₂ + 4 H₂O = 2 Mg₃Si₂O₅(OH)₄ and serpentine forms without brucite (Frost and Beard 2007, reaction 8; cf. Klein et al. 2009, reaction R12). Brucite is therefore not an obligatory phase but one that depends on the silica supply and is often cryptic.

The governing quantity is the silica activity a_SiO₂, the chemical driving force of silica relative to quartz. It controls which mineral assemblage is stable: as a_SiO₂ rises, the equilibrium shifts from brucite (Mg₃Si₂O₅(OH)₄ + H₂O = 3 Mg(OH)₂ + 2 SiO₂, Frost and Beard 2007, reaction 9) through serpentine toward talc (Mg₃Si₂O₅(OH)₄ + 2 SiO₂ = Mg₃Si₄O₁₀(OH)₂ + H₂O, reaction 7). Serpentinites are among the most silica-poor rocks of the Earth's crust: in an equilibrium of olivine, serpentine and brucite, a_SiO₂ is around 10^−2.5, and across the serpentinite system as a whole it lies between about 10^−2.5 and 10⁻⁵ (Frost and Beard 2007).

The redox budget hangs on this. The Fe²⁺ bound in olivine is oxidized to Fe³⁺ and fixed predominantly in secondary magnetite, and alongside that in the serpentine itself; the oxygen required is supplied by the decomposition of water (2 H₂O = 2 H₂ + O₂), which releases H₂ (Frost and Beard 2007; Klein et al. 2009). The oxidation can be written via the iron serpentine end-member component: 2 Fe₃Si₂O₅(OH)₄ + O₂ = 2 Fe₃O₄ + 4 SiO₂ + 4 H₂O, equivalently in its hydrogen-producing form Fe₃Si₂O₅(OH)₄ = Fe₃O₄ + 2 SiO₂ + H₂O + H₂ (Frost and Beard 2007, reaction 23; Klein et al. 2009). The hydrogen thus generated makes serpentinites the most strongly reduced rocks of the Earth's crust, reducing enough that partially serpentinized peridotites can carry metallic iron-nickel alloys such as awaruite (FeNi₃) (Frost and Beard 2007).

The less obvious conclusion: a_SiO₂ controls not only the mineral assemblage but also the distribution of iron and hence the hydrogen yield. At high silica activity, magnetite is not stable; the iron remains bound in iron-rich serpentine and brucite, and less H₂ forms. Only at low a_SiO₂ does magnetite crystallize alongside magnesium-rich serpentine and brucite, which drives H₂ production (Frost and Beard 2007). In the mesh textures of abyssal serpentinites, accordingly, a substantial part of the iron is trivalent: about 30 to 50 percent in the serpentine-brucite rims (Klein et al. 2009).

At Rechnitz: for the serpentinites of Glashütten and Rumpersdorf in the Kőszeg-Rechnitz Window, Demény et al. (2007) demonstrate a strong ¹⁸O enrichment of up to 16.2 ‰ and interpret it as the result of low-temperature serpentinization or of interaction with ¹⁸O-rich fluids, in contrast to the mantle-like δ¹⁸O values (5.9 to 6.8 ‰) of the serpentinite of Bienenhütte in the Bernstein Window.

All three serpentine polymorphs are trioctahedral sheet silicates of the 1:1 or TO type: one tetrahedral [SiO₄] sheet is linked to exactly one brucite-like, trioctahedral [Mg(O,OH)₆] octahedral sheet (Wicks and O'Hanley 1988). The key to the three structural forms is a geometric incompatibility of the two sheets: the lateral dimension of an ideal magnesium-occupied octahedral sheet (b ≈ 9.43 Å) is larger than that of an ideal silicon-occupied tetrahedral sheet (b ≈ 9.1 Å). This misfit is considerable and generates a strain that each of the three structures resolves differently (Wicks and O'Hanley 1988). The same substance, the same strain, three solutions: a platy, a tube-shaped fibrous and a wavy mineral.

Lizardite stays planar. It bridges the misfit within the flat sheet, chiefly through a ditrigonal rotation of the tetrahedra, which according to Wicks and O'Hanley (1988) is the essential mechanism, together with limited coupled substitution of Al³⁺ and Fe³⁺ for Mg and Si, for example after (Mg₆₋ₓAlₓ)(Si₄₋ₓAlₓ)O₁₀(OH)₈; at an Al content of x ≈ 0.6 the lateral areas of the two sheets exactly match (Wicks and O'Hanley 1988; cf. Evans 2004). Chrysotile, by contrast, resolves the misfit through curvature: the sheet rolls up into concentric or spiral cylinders, with the larger octahedral sheet on the outside, on the convex side of the arc (Wicks and O'Hanley 1988). Strain-free would be a single radius of curvature of about 88 Å (Whittaker 1957); real fibres have outer radii of around 110 to 135 Å (Yada 1971, cited in Wicks and O'Hanley 1988). It is precisely this curved layer structure, says Evans (2004), that "turns a sheet silicate into a fibrous mineral".

Antigorite takes a third path: a modulated structure built from an alternating wave in which the tetrahedral sheet runs continuously but reverses its polarity, that is, the direction of curvature, at the inflection points; this reversal repeats periodically and forms a superstructure (Wicks and O'Hanley 1988). Its wavelength A is not fixed but usually lies at about 33.7 to 43.1 Å (Wicks and O'Hanley 1988); the periodicity parameter m denotes the number of tetrahedra per superperiod (Kunze 1961, cited in Wicks and O'Hanley 1988). At every curvature reversal, 3 Mg and 6 (OH) are missing per unit cell relative to Si, which is why the formula of antigorite deviates from the ideal serpentine formula and comes out somewhat more silicon-rich and magnesium-poor (Wicks and O'Hanley 1988). The less obvious conclusion: chrysotile's fibrous, health-relevant habit is a crystallographic consequence of strain release, not a matter of chemical composition.

At Rechnitz: for the fully serpentinized ultramafic bodies in the windows of Bernstein, Rechnitz and Eisenberg, Koller (1985) demonstrates by X-ray diffraction lizardite and chrysotile together as the essential mineral content, chrysotile among other things in mesh texture. In the same rocks, then, the planar and the rolled-up resolution of the same misfit occur side by side; asbestos, according to Koller, is relatively often present on the open fracture surfaces.

Which of the serpentine phases belongs in the phase diagram of the system MgO-SiO₂-H₂O (MSH) at crustal conditions is the core question of Evans (2004). His finding: chrysotile is nowhere the most stable phase. Lizardite is marginally more stable at low temperatures, yet the difference in Gibbs energy in the range 300 to 400 °C amounts to no more than about 2 kJ; above about 300 °C, antigorite + brucite is more stable than both (Evans 2004). This gap is small compared with the natural scatter of the Gibbs energy of lizardite itself, which on account of surface energy, polytypism and crystal defects reaches several hundred joules per mole; minimally strained chrysotile with a radius of curvature of about 90 Å can even be more stable than lizardite, while strongly strained chrysotile (≈ 200 Å) cannot (Evans 2004).

What is decisive, therefore, is not thermodynamics but kinetics. Chrysotile grows preferentially in isotropically strained, fluid-filled voids and pores as well as in veins, usually only after the active hydration of the surroundings has ended; Evans models the chrysotile veins in lizardite- or antigorite-bearing rock as nucleation of weakly strained chrysotile followed by kinetically favoured accretion of higher-energy layers under mild fluid supersaturation (Evans 2004). Lizardite, by contrast, forms by replacement under the expansion pressure of hydration. Because chrysotile is promoted by fluid and not by shear stress, the two behave like a strain-antistrain pair that can grow side by side in the same partially or fully serpentinized rock (Evans 2004).

From this follows the less obvious conclusion: it is not temperature and pressure but the circumstances of growth that decide whether chrysotile or lizardite forms (Evans 2004). Where fibrous chrysotile, the white asbestos, occurs, it marks the pathways along which fluids found open space, not a particular pressure-temperature position. The asbestos content of a serpentinite thus follows its fluid and fracture history, not the equilibrium phase diagram.

At Rechnitz: Koller (1985) finds the asbestos of the Rechnitz serpentinites above all on the open fracture surfaces, precisely in the fluid-conducting openings that, according to Evans, favour the growth of chrysotile. The Eo-Alpine metamorphism of the Rechnitz series at 330 to 370 °C lies in the transitional range in which lizardite is progressively replaced by antigorite (Schwartz et al. 2013).

Amphiboles are chain silicates, and in contrast to the single chains of the pyroxenes they are double-chain silicates: the silicate framework consists of a double chain of corner-linked tetrahedra joined to a band of edge-linked octahedra, both extended along the c-axis (Hawthorne and Oberti 2007). Per formula unit this double chain comprises eight tetrahedra, T₈O₂₂, that is, two single chains of the repeat-unit type Si₄O₁₁ that grow together via the bridging oxygens O(5), O(6) and O(7) into the double chain (Hawthorne and Oberti 2007). Hawthorne and Oberti (2007) write the general amphibole formula as A B₂ C₅ T₈ O₂₂ W₂ with four cation sites: A sits at the centre of a large cavity between the back-to-back double chains and takes up Na, K, Ca, Li or a vacancy (□); the B cations (Na, Li, Ca, Mn²⁺, Fe²⁺, Mg) occupy the M(4) site at the edge of the octahedral band; the C cations (Mg, Fe²⁺, Mn²⁺, Al, Fe³⁺, Mn³⁺, Ti⁴⁺, Li) fill the three octahedrally coordinated sites M(1), M(2) and M(3) in the interior of the band; T stands for the tetrahedral cations Si, Al and Ti⁴⁺, and W for the anions (OH), F, Cl or O²⁻ on the O(3) position (Hawthorne and Oberti 2007).

Tremolite and actinolite are, in this scheme, a complete solid-solution series of the calcic amphiboles. The tremolite end-member □Ca₂Mg₅Si₈O₂₂(OH)₂ has an empty A site, Ca₂ on B = M(4) and Mg₅ on C = M(1)–M(3) (Hawthorne and Oberti 2007). From there the coupled isovalent substitution Fe²⁺ ⇌ Mg on the C sites, that is the Fe²⁺-Mg distribution across the octahedral sites M(1)–M(3) (Hawthorne and Oberti 2007), leads continuously to the iron-rich end-member □Ca₂Fe²⁺₅Si₈O₂₂(OH)₂. Tremolite and actinolite thus differ solely in the Mg/Fe²⁺ ratio of the octahedral sites, not in the structural principle of the double chain.

From this follows the point decisive for the asbestos question. An amphibole that grew asbestiform, formed in thin, flexible, individually separable fibres, and a cleavage fragment, produced by the breaking of a massive or prismatic crystal along its cleavage, can carry the same chemical composition, because the composition resides in the occupancy of the A, B, C and T sites, not in the habit. Both can at the same time fall into the same elongate particle geometry. Nevertheless they formed in crystal-chemically different ways: the one as directed fibre growth, the other as a fragment. It is precisely this decoupling of composition and geometry on the one hand from genesis on the other that makes the identification and counting of amphibole fibres analytically tricky. The counting-related and regulatory side of this distinction we address on the standards page, section 5.

At Rechnitz: Koller (1985) documents in the serpentinites of the Rechnitz series tremolite above all in tectonically reworked zones and on the open fracture surfaces, and he assigns to the youngest, greenschist-facies metamorphic event dark-blue alkali amphiboles of riebeckitic to Mg-riebeckitic composition. In the local sequence, then, the calcic amphibole tremolite, whose actinolitic counterpart has been detected in Burgenland material, and the sodium-bearing riebeckite-magnesioriebeckite occur side by side.

That a fibre remains in the tissue is, according to Hume and Rimstidt (1992), not a question of equilibrium solubility but of dissolution kinetics. The fluids in lung tissue contain very low magnesium and silicon concentrations and are markedly undersaturated with respect to chrysotile; thermodynamically, then, chrysotile ought to dissolve, and its persistence is "simply a result of a slow dissolution rate" (Hume and Rimstidt 1992). The governing reaction at pH < 9 is Mg₃Si₂O₅(OH)₄ + 6 H⁺ = 3 Mg²⁺ + 2 H₄SiO₄ + H₂O (Hume and Rimstidt 1992). It proceeds in two steps: first the brucite-like magnesium hydroxide sheet of chrysotile is leached out rapidly, leaving behind a silicate framework whose slower dissolution then controls the overall rate. The fibre lifetime, that is, the biodurability, therefore depends on the rate of silica release (Hume and Rimstidt 1992). For the undersaturation range prevailing in lung tissue, this rate is independent of pH and amounts to 5.9 (± 3.0) × 10⁻¹⁰ mol m⁻² s⁻¹ at 37 °C (Hume and Rimstidt 1992).

It is exactly here that the kinetics connects with the crystal structure. The sheet that gives way first and fast is the same outer, brucite-like Mg(OH) octahedral sheet of the TO sheet that, in the rolled chrysotile fibre, lies on the convex outer side of the arc and is thus exposed to the fluid (cf. the TO-sheet section above; Wicks and O'Hanley 1988). The amphiboles offer the fluid no such freely accessible hydroxide sheet: in the amphibole double chain the magnesium and iron cations sit on the octahedral sites M(1), M(2) and M(3) in the interior of the octahedral band, enveloped by the silicate framework of the corner-linked double chain T₈O₂₂ (cf. the double-chain section above; Hawthorne and Oberti 2007). Structurally it follows that the kinetically decisive first step, the rapid leaching of an exposed Mg hydroxide sheet, lacks its surface of attack in the amphiboles; the attackable part is here built into the more resistant silicate framework. This statement is a structural inference from the two structural plans already established, not a measured amphibole dissolution rate: Hume and Rimstidt (1992) note explicitly that for the iron-rich amphiboles such as crocidolite no rate data are available, so a quantitative rate for amphibole asbestos remains open here. How far the greater durability explains the differing harmfulness we set out on the standards page.

The less obvious conclusion: biodurability is the geochemical reason why the same hazard class splits apart along mineralogy. The leachable Mg sheet makes chrysotile less biopersistent than the framework-built silicates; as an order of magnitude, Hume and Rimstidt (1992) give a modelled lifetime of 9 (± 4.5) months for a chrysotile fibre of 1 µm diameter, against 438 years for an equally large fibre of amorphous silica and 1.7 × 10⁶ years for a quartz grain of the same size. This is why the local amphibole finding, tremolite and actinolite, carries its own weight; the epidemiological side of this weighting is on the standards page, section 5.

At Rechnitz: in the Rechnitz rocks both kinds of asbestos occur, the chrysotile of the serpentine (Koller 1985; see the TO-sheet section above) and the amphibole asbestos tremolite or actinolite, whose actinolitic member is detected in one of our own samples (Koller 1985; see Own sampling). By the biodurability logic, the amphibole asbestos is the more durable of the two; the health weighting of this difference is on the standards page.

Each of Koller's (1985) three stages carries its own diagnostic mineral assemblage, and together they trace a path from oceanic crust into the subduction zone and back. The oceanic metamorphism is marked by high-temperature hornblende phases, namely barroisite, pargasite and magnesiohornblende, accompanied by metasomatic changes (Na, Ca influx) and intense oxidation; Koller (1985) brackets it at 750 to below 400 °C at a pressure of at most 1 kilobar and interprets the widespread metamorphic traces as evidence of a mid-ocean ridge with relatively low spreading rates. It is the imprint the rock acquired on the sea floor itself.

The Eo-Alpine high-pressure low-temperature stage is the tectonic memory of the burial. Its index minerals are Mg-rich pumpellyite, ferroglaucophane, an alkali pyroxene of composition around Ac₆₃Jd₂₁, lawsonite (preserved only as pseudomorphs) and stilpnomelane; from this paragenesis Koller (1985) derives 330 to 370 °C at 6 to 8 kilobar, and a K-Ar dating on zoned crossite-riebeckite mixed crystals gave 65 ± 6 million years. These conditions require, according to Koller (1985), a Cretaceous burial of the Penninic oceanic crust in a subduction zone to depths of at least 15 to 25 kilometres; the low geothermal gradient corresponds exactly to that of a subduction zone. The youngest, greenschist-facies regional metamorphism overprints this high-pressure fabric at 390 to 430 °C and less than 3 kilobar (K-Ar 19 to 22 million years); its index minerals are actinolite, an alkali pyroxene with a small jadeite component (Ac₈₅Jd_<5) and dark-blue alkali amphiboles of riebeckitic to magnesioriebeckitic composition, with the metamorphic grade increasing from north to south (Koller 1985).

The protolith was ocean floor. By their composition the ultramafic rocks are peridotites with mostly harzburgitic, subordinately lherzolitic chemistries, that is, low CaO and Al₂O₃ with high MgO contents (Koller 1985); the associated metabasites carry an N-type MORB signature (Koller 1985). Combining the magmatic relicts, the geochemistry and the isotopy, Demény et al. (2007) assign this sequence to an ophiolite sequence of N-MORB basalts and harzburgite bodies of the suboceanic mantle (drawing on Melcher et al. 2002 and Meisel et al. 1997) and note that the geochemical compositions have outlasted the later metamorphism. Geodynamically this fits into a closed picture: Penninic oceanic lithosphere, whose subduction began, according to Demény et al. (2007), in the Cretaceous, buried to blueschist depths and exhumed again during the Middle Miocene crustal extension (Demény et al. 2007). The less obvious conclusion: the asbestos-bearing serpentinite is a piece of ocean-floor mantle that was hydrated near the sea floor, carried to subduction depths and subsequently uplifted. Its mineralogy, in particular the succession of amphibole generations from the high-temperature oceanic hornblende through the high-pressure alkali amphibole to the greenschist-facies actinolite and riebeckite, is the tape recording of this path from ocean through subduction to exhumation.

At Rechnitz: the four windows of Möltern, Bernstein, Rechnitz and Eisenberg expose one and the same subducted and re-exhumed Penninic oceanic lithosphere; because their ultramafic core is hydrated mantle harzburgite and their amphibole generations have preserved the oceanic-subductive path, asbestos-bearing serpentinite is geologically expectable across the entire belt, not a local special case.

Why the isotopy reveals anything at all about serpentinization rests on the temperature dependence of the equilibrium fractionation. The oxygen isotope fractionation between two coexisting phases, written as Δ_A−B = 1000 ln α_A−B ≈ δ¹⁸O_A − δ¹⁸O_B, decreases with rising temperature and runs, in the high-temperature range, approximately proportional to 1/T² (Wenner and Taylor 1971; Saccocia et al. 2009). A measured fractionation between two minerals that grew at the same time is thus a geothermometer. For serpentinites the only broadly applicable mineral pair is serpentine-magnetite, because during serpentinization magnetite very commonly forms from the released iron (Wenner and Taylor 1971; cf. the redox section above). Wenner and Taylor (1971) define α_serp−mag = R_serp/R_mag with R = (¹⁸O/¹⁶O) and Δ_serp−mag = 1000 ln α_serp−mag ≈ δ_serp − δ_mag.

Their finding is diagnostically sharp. "Normal" lizardite-chrysotile serpentinites show very large serpentine-magnetite fractionations of Δ = 12.9 to 15.1 ‰, whereas nine antigorites show markedly smaller values of 8.7 to 4.8 ‰ (Wenner and Taylor 1971). Since Δ falls with temperature, this proves that antigorite formed at higher temperatures than the lizardite-chrysotile serpentinites, in agreement with the phase diagram from the multisystem section above. Translated into temperatures, Wenner and Taylor (1971) estimate for continental lizardite-chrysotile 85 to 115 °C, for oceanic lizardite or chrysotile 130 °C and 185 °C, for oceanic antigorite 235 °C and for continental antigorites 220 to 460 °C; the low fractionation values and the predominance of lizardite and chrysotile thus characterize both abyssal and continental serpentinization as a low-temperature process. The experimental gap in this field-based calibration has been closed by Saccocia et al. (2009): from partial-exchange experiments at 250 to 450 °C and 50 MPa they derive for the system serpentine-water 1000 ln α_serp−water(¹⁸O) = 3.49 × 10⁶/T² − 9.48 (T in kelvin), with the fractionation falling from 3.1 ± 1.2 ‰ at 250 °C to −3.1 ± 1.3 ‰ at 450 °C. This allows a measured δ¹⁸O_serp to be translated, at known fluid composition, directly into a formation temperature, and conversely a temperature into an expected serpentine phase.

The δ¹⁸O values themselves are a fingerprint of the fluid history. Unaltered mantle peridotite lies at about 5.5 ‰ (Mattey et al. 1994, cited in Demény et al. 2007); markedly higher δ¹⁸O values require either serpentinization at low temperature, where the large mineral-water fractionation makes the rock ¹⁸O-rich, or interaction with ¹⁸O-rich, for example sediment-equilibrated, fluids (Demény et al. 2007). The hydrogen carries complementary information: δD reveals the source of the fluid, because seawater lies near 0 ‰, meteoric water strongly negative and mantle-derived or metamorphic fluids in between. The two systems are differently temperature-sensitive. Saccocia et al. (2009) calibrate the serpentine-water D/H fractionation over 25 to 450 °C as 1000 ln α_serp−water(D/H) = 3.436 × 10⁶/T² − 34.736 × 10³/T + 21.67 and show that serpentine becomes increasingly depleted in deuterium relative to the coexisting water toward lower temperatures (from −20 ± 2 ‰ at 450 °C to −32 ± 15 ‰ at 250 °C).

It is on exactly this basis that the Rechnitz data set of Demény et al. (2007) reads. The δ¹⁸O values of the silicates scatter widely, from 5.9 to 16.4 ‰, and systematically by locality: rodingite, chlorite-rich blackwall, serpentinized harzburgite and tremolite vein of Bienenhütte in the Bernstein Window preserve mantle-near values of 5.9 to 6.8 ‰, while the serpentinite of Glashütten and Rumpersdorf in the Kőszeg-Rechnitz Window as well as the silicates of the ophicalcites carry a strong ¹⁸O enrichment up to 16.2 ‰ (Demény et al. 2007). The hydrogen isotopy, by contrast, is almost homogeneous: δD_mean = −63 ± 7 ‰ (n = 7), only two serpentinites reach about −106 ‰ and are ascribed to a limited, late-metamorphic infiltration of meteoric water that, according to Demény et al. (2007), remained local, since even samples taken directly on slickenside surfaces show no deviating δD values. The decisive point is what is missing: the strong D enrichment that a high-temperature interaction with seawater would leave behind was not observed. Demény et al. (2007) interpret this to mean that the originally D-rich, ¹⁸O-poor oceanic serpentinite signature was overprinted by sediment-derived fluids during the Alpine metamorphism (negative δD and positive δ¹⁸O shift), and they link the H isotopy to the subduction-related dewatering of the Penninic oceanic crust and a mantle metasomatism caused by it, whose reconstructed fluid composition of about −40 ‰ corresponds to that of mantle-derived amphibole megacrysts (Demény et al. 2005, cited in Demény et al. 2007).

The less obvious conclusion closes the circle back to the mineralogy at the surface. The serpentine-magnetite thermometer ties a measured δ¹⁸O to the serpentinization temperature and hence to the expectable serpentine phase (multisystem logic above), and the locality-by-locality contrast of the δ¹⁸O values reads directly the fluid history of these particular quarry rocks: high values at Glashütten and Rumpersdorf as a mark of low temperature or sediment-equilibrated fluids, mantle-buffered values at Bienenhütte. The asbestos content follows, as shown in the multisystem section, the fluid and fracture history; the isotopy makes this fluid history measurable for each outcrop individually.

At Rechnitz: the isotopy separates the quarry rocks by their fluid history. The serpentinite of Glashütten and Rumpersdorf records, with δ¹⁸O up to 16.2 ‰, a low-temperature or sediment-influenced serpentinization, while the rock of the Bernstein Window at Bienenhütte has preserved, at 5.9 to 6.8 ‰, a mantle-near oxygen signature; the δD values, almost uniform across the whole belt at around −63 ‰, and the absence of the D enrichment typical of high-temperature seawater point at the same time to the subduction-related dewatering as the common background. Even within one belt, then, the fluid histories differ from outcrop to outcrop.