Abstract
Double seismic zones (DSZs) are a feature of some subducting slabs, where intermediate-depth earthquakes (~70–300 km) align along two separate planes. The upper seismic plane is generally attributed to dehydration embrittlement, whereas mechanisms forming the lower seismic plane are still debated. Thermal conductivity of slab minerals is expected to control the temperature evolution of subducting slabs, and therefore their seismicity. However, effects of the potential anisotropic thermal conductivity of layered serpentine minerals with crystal preferred orientation on slab’s thermal evolution remain poorly understood. Here we measure the lattice thermal conductivity of antigorite, a hydrous serpentine mineral, along its crystallographic b- and c-axis at relevant high pressure-temperature conditions of subduction. We find that antigorite’s thermal conductivity along the c-axis is ~3–4 folds smaller than the b-axis. Our numerical models further reveal that when the low-thermal-conductivity c-axis is aligned normal to the slab dip, antigorite’s strongly anisotropic thermal conductivity enables heating at the top portion of the slab, facilitating dehydration embrittlement that causes the seismicity in the upper plane of DSZs. Potentially, the antigorite’s thermal insulating effect also hinders the dissipation of frictional heat inside shear zones, promoting thermal runaway along serpentinized faults that could trigger intermediate-depth earthquakes.
Similar content being viewed by others
Introduction
Slab subduction is the main driving force of plate tectonics, which introduces complex chemical heterogeneities in Earth’s interior1. During subduction, the rocks within the slab are subjected to increasing pressure (P) and temperature (T) conditions, which alter their physical and chemical properties with impacts on the thermochemical evolution of the slab, and the geodynamics of the subduction zone and surrounding mantle. In particular, the temperature profile of the subduction zone is key to influence the stability of the minerals and rocks within it, as well as its geological processes, e.g., arc volcanism and magmatism2 and earthquakes3. Therefore, quantifying the temperature evolution of a slab along subduction is fundamental to understand a large variety of geophysical, geodynamical, and geochemical phenomena around a subduction zone.
Interestingly, in some subduction zones, earthquakes occur at intermediate-depth (~70–300 km) along two distinct planes separated by ~20–40 km, known as double seismic zones (DSZs), see refs. 4,5,6 for example features of DSZs, where the separation of two seismic planes correlates with the age of a slab, i.e., with its temperature and thickness at the trench. Since its first observation in northeast Japan7, several mechanisms have been proposed to explain the origin of DSZs, including dehydration embrittlement6,8,9,10,11,12,13, fluid-related embrittlement14, transformation faulting15, plastic shear instability16,17, grain size reduction18, and thermal runaway12,19,20,21,22. Of particular importance are the dehydration embrittlement and thermal runaway, as they are considered the two major mechanisms for intermediate-depth seismicity12,19. Dehydration embrittlement is triggered by the presence of free water H2O in the interstitial pores of the rock, which reduces the differential stress required for brittle fracture. Most of the water released at depth originates from the breakdown of antigorite23,24, a high-pressure polymorph of serpentine minerals that contains large amounts of water (~12 wt%) and serves as the major hydrous mineral in subduction zones. Antigorite, however, does not survive at the high T condition in Earth’s interior, and breaks down at ~700–1000 K23,24. This temperature is typically reached inside the slab at about 70–300 km depth, depending on the P-T path of the slab24,25. As a result, the intermediate-depth seismicity at the upper plane of a DSZ has often been attributed to the temperature-induced breakdown and dehydration of antigorite, see, e.g., refs. 3,6,26. The mechanism that causes seismicity at the lower plane, however, remains debated.
Thermal conductivity of minerals within a subduction zone plays key roles in controlling the heat flow and the temperature profile in the region. Thus, precise determination of such critical property at relevant P-T conditions would offer crucial insights into the thermal structure and dynamics of the subduction zone, as well as the fate of antigorite, with potential implications for the formation and distribution of intermediate-depth earthquakes. Lattice thermal conductivity or diffusivity of mantle minerals under extreme P-T conditions have been studied for decades. Precise measurements, however, are challenging due to the difficulty and limitation of previous conventional experimental techniques. As a result, very few experimental data under relevant mantle conditions were available and the measurement accuracy was insufficient. Recent successful combinations of time-resolved optical techniques with diamond anvil cells (DACs) have enabled precise measurements of minerals’ thermal conductivity under high P-T conditions, offering novel insights to the complex thermal evolution and dynamics in Earth’s deep interior27,28,29,30.
Prior studies on the thermal conductivity of antigorite have been limited to polycrystalline samples at relatively low pressure regime31,32. Given its layered crystal structure, antigorite is characterized by a strong anisotropy in its elastic constants along different crystal orientations33,34,35,36. Since the lattice heat transport scales with the elastic constants37, antigorite’s thermal conductivity is expected to be anisotropic as well. Such hypothesis, however, has not been experimentally tested. On the other hand, serpentinites are featured by a weak rheology and, upon shear deformation during slab subduction, serpentine crystals tend to develop a crystal preferred orientation (CPO), in which the crystal layers are parallel to the shear plane, i.e., antigorite’s [001] direction orients perpendicularly to slab’s dip direction38,39. As discussed above, the breakdown depth of antigorite during subduction depends on slab’s temperature evolution, which could be critically controlled by antigorite’s thermal conductivity. Moreover, previous thermomechanical modelings, e.g., refs. 6,9, have revealed the important roles that the slab age and convergence velocity play on antigorite’s dehydration profile within a subducting slab, while the antigorite’s thermal conductivity has often been assumed as a constant (e.g., ~3–4 W m−1 K−1), regardless of its crystal orientation and P-T conditions. The potential thermal conductivity anisotropy and CPO of antigorite during subduction would result in distinct thermal conductivity along a direction perpendicular and parallel to the slab’s dip direction. Understanding the effects of P, T, and CPO on antigorite’s thermal conductivity along subduction is thus critically needed, as it would significantly advance numerical modeling of the temperature profile in subduction zones.
In this work, we experimentally show a strong thermal conductivity anisotropy in antigorite under relevant P-T conditions along subduction. Our numerical simulations on the two-dimensional (2D) heat diffusion and temperature profile within a subducting slab indicate that antigorite’s anisotropic thermal conductivity with CPO promotes dehydration embrittlement and potentially thermal runway, which could play key roles in triggering intermediate-depth earthquakes.
Results
Lattice thermal conductivity of antigorite at high pressure and room temperature
We used ultrafast time-domain thermoreflectance (TDTR) coupled with DAC to precisely measure the lattice thermal conductivity of single-crystalline natural antigorite to ~13 GPa at room temperature. The TDTR is an ultrafast pump-probe spectroscopy for high-precision thermal conductivity measurements at high pressures and wide range of temperatures, see, e.g., refs. 27,30,40 and Methods for details. Figure 1 shows the effect of pressure on the thermal conductivity of single-crystal antigorite along the [010] (in-plane b-axis, Λ010, black circles) and [001] (cross-plane c-axis, Λ001, red squares) orientation at room temperature. At ambient conditions, the Λ010 is ~4.6 W m−1 K−1, ~4 times larger than the Λ001 (~1.1 W m−1 K−1). Upon compression, Λ010 and Λ001 both increase with pressure and reach 10.0 (±1.9) and 2.7 (±0.4) W m−1 K−1, respectively, around 7 GPa, where the difference between them achieves a maximum value of 7.3 W m−1 K−1. In other words, the thermal conductivity anisotropy (Λ010/Λ001) at ambient conditions is ~4.3 and becomes less anisotropic as the anisotropy progressively decreases to ~2.5 at P ~ 4–5 GPa; afterwards the anisotropy slightly increases to ~3.7 at 7 GPa. Since thermal conductivity is an ensemble contribution of heat transport that involves heat capacity, elastic constant, and phonon mean-free-path of all the available phonon modes37, such trend with pressure can be primarily accounted for by the pressure evolution of the elastic constant along the in-plane and cross-plane direction, see, e.g., ref. 36. Note that the Λ010 at 7 GPa is larger than the thermal conductivity of olivine (~5.1–8.3 W m−1 K−1)41,42, the major mineral phase in the upper mantle. Interestingly, at P > 7 GPa the difference between Λ010 and Λ001 reduces progressively: the Λ010 decreases considerably with pressure, while the Λ001 reaches a plateau of ~3.0–3.5 W m−1 K−1 between 8 and 13 GPa, making them approach with each other. The occurrence of the dramatic change in the trend of Λ010 with pressure around 7 GPa coincides with a displacive phase transition observed in antigorite24,43,44, where pressure induces the distortion of Si-O tetrahedral and octahedral sheets45.
In Fig. 1 we also plotted literature results for polycrystalline antigorite at ambient pressure31 (open blue circle) and to 8.5 GPa32 (open green stars), respectively, which are bracketed by our b- and c-axis data at P < 4 GPa. Nevertheless, the data by ref. 32. (open green stars) appear to be nearly a constant, rather than a typically increasing trend upon compression, and become comparable to our c-axis data at P ~ 4–9 GPa. We are not aware of the detailed experimental conditions in Osako’s study, and thus could only propose that their polycrystalline sample may have been progressively oriented preferentially to near the c-axis under their compression conditions. By contrast, to check the potential variation of crystal orientation during our compression conditions, we have re-characterized the orientation from the quenched samples out of the DAC using electron backscattered diffraction or X-ray diffraction (in situ measurements of crystal orientation and thermal conductivity at high pressures are currently not available). We found that the orientation remained essentially the same, suggesting that the minor variation, if there is, of crystal orientation is expected to have minor effects on the uncertainty of our antigorite’s thermal conductivity.
Note that we have also measured the thermal conductivity along the a-axis, Λ100, at ambient conditions, which is ~4.5 W m−1 K−1 and close to that of the b-axis. Since thermal conductivity scales with the square of sound velocity37, such behavior can be understood by the similar elastic constant and sound velocity along the b- and a-axis, even at high pressures34,46. Therefore, even though we did not measure the pressure dependence of Λ100, we expect it to be similar to Λ010 and much larger than the Λ001. As we show later in the sections of numerical modeling and discussions, what critically impacts slab’s thermal evolution is the thermal insulating effect by the low value of Λ001, and the strong thermal conductivity anisotropy. Lack of data for Λ100(P, T) does not influence our conclusions of this study.
Thermal conductivity at simultaneous high pressure and temperature conditions
To quantify the combined effects of pressure and temperature on the thermal conductivity of antigorite, we further performed simultaneous high P-T measurements. Figure 2 and Table 1 summarize the P-T conditions for each measurement run and its thermal conductivity along a specific crystal orientation, which allow us to track how the antigorite’s thermal conductivity changes along slab subduction. Typical P-T profiles at slab surface and Moho are plotted as red and blue curve, respectively, in Fig. 2 from ref. 3. To quantify the temperature dependence of thermal conductivity, we compared our data at a given pressure and room temperature (Fig. 1) with those at a similar pressure and high temperatures (Fig. 2). For instance, we considered Λ001 at 5 GPa and room temperature (Fig. 1), as well as measurement run 4 and 7 (Fig. 2). For simplicity, if we assume the Λ of antigorite can be phenomenologically modeled as Λ(T) = Λ0Tn, where Λ0 is a normalization constant at room temperature, then the temperature exponent n is determined by fitting a linear regression slope in the lnΛ-lnT plot. We found that n ~ –0.55 (±0.01) along the c-axis and n ~ –0.58(±0.06) along the b-axis. Considering our data uncertainty of ~15%, such dependences are reasonably in line with the typical temperature dependence of T–0.5 for minerals27,40,47,48. The combined P-T dependences indicate that during slab subduction, antigorite’s thermal conductivity anisotropy (Λ010/Λ001) remains at ~3–4 until ~230 km depth. Such critical finding is employed to model the thermal evolution of a sinking slab containing antigorite with CPO, see the numerical modeling section below.
Numerically modeling the thermal evolution of a sinking slab
Our experiments reveal that antigorite’s thermal conductivity is strongly anisotropic even at high P-T conditions (\({\Lambda }^{010} \sim 3{\Lambda }^{001}\)). Potentially, such thermal conductivity anisotropy could induce temperature anomalies within a sinking slab, thus influencing the dehydration depth of antigorite, as well as promoting the onset of thermal runway. Therefore, investigating the large-scale effects of antigorite’s thermal conductivity anisotropy would provide critical insights to the mechanisms responsible for the intermediate-depth earthquakes. For this purpose, we further performed 2D numerical modeling to compute the temperature evolution of a sinking slab containing antigorite (see Methods section). The full description of the model is reported in Supplementary Information: physical and numerical model (Note S1–S2, Fig. S4–S8, Table S1), thermodynamic parameters (Note S3–S5, Fig. S9, Tables S2–S3), model limitations (Note S6), and code benchmark (Note S7, Figs. S10–S11, Table S7). In total, we run 10 different models divided in three sets (Table S4). In the first set, used as a reference, we assumed that the hydrous layer was made entirely of dry olivine (model 1) or wet olivine (model 2). In the other two sets, we considered the [001] direction of antigorite (i.e., the c-axis) as the “insulating direction”. We tested two ideal end-member configurations: vertical insulator with antigorite’s [001] direction parallel to the slab dip (models 3–6, Fig. 3a), and horizontal insulator with antigorite’s [001] direction perpendicular to the slab dip (models 7–10, Fig. 3b), see Methods section for details. The assumption of perfect alignment of single-crystalline antigorite along the slab dip angle and the potential effects of grain boundary are fully discussed in the Supplementary Information Note S8, Fig. S13, and Table S8, and Note S9, respectively. In set 2 and 3, antigorite fraction \({{{{{{\rm{\varphi }}}}}}}_{{{{{{\rm{Atg}}}}}}}\) was increased to simulate different degrees of serpentinization in the slab: \(0.1\) (models 3,7); \(0.3\) (models 4,8); \(0.5\) (models 5, 9); \(1.0\) (models 6,10). Figure 4 illustrates two examples of 2D temperature fields from model 1 (Fig. 4a) and model 10 (Fig. 4b) and their temperature difference (Fig. 4c).
We monitored the temperature \({{{{{\rm{T}}}}}}\) (Fig. 4) and the heat flux \({{{{{\rm{Q}}}}}}\) (Fig. S12) in three specific planes from the slab surface (0 km): \(7\) km (external side of the hydrous layer), \(9\) km (inside the hydrous layer) and \(1\)1 km (internal side of the hydrous layer). In the horizontal insulator configuration (Fig. 3b), the thermal energy coming from the hot ambient mantle is not efficiently transported toward the cold core of the slab, accumulating at the interface between the oceanic crust and the insulating layer (e.g., Fig. 4c). Consequently, the temperature at the base of the crust (7 k\({{{{{\rm{m}}}}}}\)) in the horizontal insulator configuration (\({{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{001}\), Fig. 5c teal dashed lines) is higher than the temperature in the vertical insulator configuration (\({{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{010}\), Fig. 5c purple dotted lines). Moreover, in the horizontal insulator configuration (Fig. 3b), the heat flow toward the cold core of the slab is hampered for the whole thickness of the hydrous layer, and thus its internal temperature is higher than the vertical insulator configuration: \({{{{{{\rm{T}}}}}}}_{9{{{{{\rm{km}}}}}}}^{010} \, < \, {{{{{{\rm{T}}}}}}}_{9{{{{{\rm{km}}}}}}}^{001}\) (Fig. 5b). The insulation effect caused by the hydrous layer with antigorite in the horizontal insulator configuration is shown in Fig. 5a, where the temperature at \(11\) km from the slab surface is lower than the vertical insulator configuration: \({{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{010} \, > \, {{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{001}\). The temperature differences between the two configurations (models 7–10 vs. models 3–6) range between \(+20\) \({{{{{\rm{K}}}}}} \, < \, ({{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{7-10}-{{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{3-6}) \, < \,+230\) \({{{{{\rm{K}}}}}}\) (Fig. 5c), \(+10\) \({{{{{\rm{K}}}}}} \, < \,({{{{{{\rm{T}}}}}}}_{9{{{{{\rm{km}}}}}}}^{7-10}-{{{{{{\rm{T}}}}}}}_{9{{{{{\rm{km}}}}}}}^{3-6}) \, < \,+100{{{{{\rm{K}}}}}}\) (Fig. 5b), and \(-3{{{{{\rm{K}}}}}} \, < \, ({{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{7-10}-{{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{3-6}) \, < \, -60\) \({{{{{\rm{K}}}}}}\) (Fig. 5a).
We compared these results also with the two reference models: 1 (dry olivine, \({\Lambda }^{{{{{{\rm{DryOl}}}}}}}\)) and 2 (wet olivine, \({\Lambda }^{{{{{{\rm{WetOl}}}}}}}\)). The thermal conductivity of wet olivine containing ~7000 wt. ppm water is 30–40% lower than its dry counterpart41, and thus the hydrous layer in model 2 acts as a thermal insulator compared to model 1 (solid green lines in Fig. 5)\(:\) \(({{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{{{{{{\rm{WetOl}}}}}}}-{{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{{{{{{\rm{DryOl}}}}}}}) \, < \, 80\) \({{{{{\rm{K}}}}}}\) (Fig. 5c) and \(({{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{{{{{{\rm{WetOl}}}}}}}-{{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{{{{{{\rm{DryOl}}}}}}}) \, > \, -30\) \({{{{{\rm{K}}}}}}\) (Fig. 5a). The vertical insulator configuration (models 3–6) does not produce a significant temperature difference compared to the dry reference even in the extreme case of complete serpentinization (Fig. 5): \(-6\) \({{{{{\rm{K}}}}}} \, < \,\) \({{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{3-6}-{{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{{{{{{\rm{DryOl}}}}}}} \, < \, -60\) \({{{{{\rm{K}}}}}}\); and \(+1.5\) \({{{{{\rm{K}}}}}} < \) \({{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{3-6}-{{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{{{{{{\rm{DryOl}}}}}}} \, < \,+ 11\) \({{{{{\rm{K}}}}}}\). In the horizontal insulator configuration, however, the temperature difference increases with the degree of serpentinization, and reaches its maximum in the extreme case of complete serpentinization:\(+15\) \({{{{{\rm{K}}}}}} \, < \,({{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{7-10}-{{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}^{{{{{{\rm{DryOl}}}}}}}) \, < \,+170\) \({{{{{\rm{K}}}}}}\) at the external side (Fig. 5c) and \(-3\) \({{{{{\rm{K}}}}}} \, < \,({{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{7-10}-{{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}^{{{{{{\rm{DryOl}}}}}}}) \, < \, -65\) \({{{{{\rm{K}}}}}}\) at the internal side (Fig. 5a). Figure 5 indicates that a serpentinization degree of 50% (model 9) causes a temperature difference similar to the wet olivine case (model 2), where the oceanic crust is \(\sim 80\) \({{{{{\rm{K}}}}}}\) hotter than the dry reference (model 1). Given the limited water storage capacity of olivine in the upper mantle (only about few hundreds wt. ppm49) with relatively high thermal conductivity, antigorite with [001] direction would be a key mineral phase to act as a thermal blanket in the shallow upper mantle.
We further evaluated the effect of the thermal insulating layer on antigorite’s breakdown depth (Fig. 6), which depends on the slab’s internal temperature25. Along subduction the breakdown temperature of antigorite lays between \(700 \, < \,{{{{{\rm{T}}}}}} \, < \, 1000\) \({{{{{\rm{K}}}}}}\)23,24, and therefore we computed the maximum depth of the \(700\) \({{{{{\rm{K}}}}}}\) and \(1000\) \({{{{{\rm{K}}}}}}\) isotherms (\({{{{{{\rm{D}}}}}}}^{700}\) and \({{{{{{\rm{D}}}}}}}^{1000}\)) inside the slab (Table S5). When the antigorite acts as a horizontal insulator (teal colors in Fig. 6), the temperature of the external side of the hydrous layer (\({{{{{{\rm{T}}}}}}}_{7{{{{{\rm{km}}}}}}}\)– teal squares and diamonds in Fig. 6) is higher, and hence the two isotherms shift to shallower depths compared to the dry reference (model 1): \(-1 \, < \, \triangle {{{{{{\rm{D}}}}}}}_{7{{{{{\rm{km}}}}}}}^{700}\le -\)10 km and \({-1 \, < \, \triangle{{{{{{\rm{D}}}}}}}_{7{{{{{\rm{km}}}}}}}^{1000} \, < \, -}\)23 km. Inside the hydrous layer the trend is the same (\({{{{{{\rm{T}}}}}}}_{9{{{{{\rm{km}}}}}}}\) - down- and up-pointing triangles in Fig. 6): \(-1 \, < \, \triangle {{{{{{\rm{D}}}}}}}_{9{{{{{\rm{km}}}}}}}^{700} \, < \, -1\)0 km and \(0 \, < \, \triangle{{{{{{\rm{D}}}}}}}_{9{{{{{\rm{km}}}}}}}^{1000} \, < \, -\)3 km. On the other hand, the temperature on the internal side of the hydrous layer (\({{{{{{\rm{T}}}}}}}_{11{{{{{\rm{km}}}}}}}\), five-pointed and six-pointed stars in Fig. 6) is lower, and the two isotherms survive at greater depths: \(+1 \, < \, \triangle {{{{{{\rm{D}}}}}}}_{11{{{{{\rm{km}}}}}}}^{700} \, < \,+1\)3 k\({{{{{\rm{m}}}}}}\) and \(0 \, < \, \triangle{{{{{{\rm{D}}}}}}}_{11{{{{{\rm{km}}}}}}}^{1000} \, < \,+2\) km. Consequently, the breakdown depth of antigorite at the external side of the insulating layer is ~ 15 km shallower, while that at the internal side of the insulating layer is only mildly affected.
Discussion
The hydrothermal circulation active at the slow-spreading ridges can reach the bottom of the oceanic crust, and thus produce a few-km thick layer of serpentinized rocks within the lithospheric mantle50,51. Moreover, additional serpentinization occurs at the trench-rise system, where the bending of the oceanic lithosphere creates a network of normal faults along which seawater percolates down to ~20 km from the slab surface52,53,54,55. Potentially, the unbending of the slab in the mantle could generate a pressure gradient sufficient to pump seawater toward the inner portion of the slab, and to push the serpentinization front down to ~40 km from the slab surface56. The occurrence of this serpentinization process, however, is still under debate. In certain regions (e.g., Lesser Antilles), the seawater alteration of the oceanic lithosphere can produce a 3-km-thick layer of hydrous rocks containing ~50–70 vol% of serpentine57,58. Antigorite has a highly anisotropic phyllosilicate structure33, which is very compressible in the cross-plane direction (the [001] direction), and relatively stiff in the in-plane direction34,36. Therefore, antigorite is characterized by a weak rheology and, during shear deformation, it likely forms a CPO, in which the [001] direction tends to align perpendicularly to the shear plane/foliation38,39,59. In other words, the serpentinized rocks of the slab accommodate most of the shear deformation developed during subduction60 by orienting antigorite’s [001] direction perpendicularly to slab’s dip38,39. This configuration might create a layer of rocks, where the seismic wave that propagates normal to the plane of slab subduction is slower than in other directions61.
The combination of antigorite’s CPO and strong thermal conductivity anisotropy as revealed by our present results can play a key role in triggering intermediate-depth earthquakes. In particular, the insulating effect of oriented antigorite can promote dehydration embrittlement in the upper plane of a DSZ and potentially thermal runaway within the serpentinized faults (Fig. 7). Dehydration embrittlement is typically expected to be responsible for the upper seismic plane of a DSZ, because the hydrous minerals, e.g., antigorite, contained in the outer ~10 km of the slab are largely exposed to the rising temperature9,10. As shown in our thermal evolution models, the presence of a 3-km-thick thermally-insulating layer of serpentinites (\({{{{{{\rm{\varphi }}}}}}}_{{{{{{\rm{Atg}}}}}}}\) > 50%, models 9 and 10, which is achieved in old oceanic lithosphere with slow-spreading ridges, e.g., Lesser Antilles57,58) efficiently traps the heat at the base of the oceanic crust (~7 km from the slab surface), thus causing a temperature increase of ~50–150 K and an ~7–15 km upward shift of the dehydration front (Table S5). In other words, the antigorite’s thermal insulating effect facilitates dehydration embrittlement in the outer 10 km of the slab, while hindering the same mechanism in the inner regions of the slab by preserving it in colder conditions. Such effect is expected to enhance the separation of the two seismic planes of DSZs, and should be taken into account to better understand the features of global DSZs4.
Moreover, the temperature differences caused by antigorite’s thermal insulating effect may contribute to local seismic anomalies. Using the scaling parameter reported by ref. 62, temperature differences of +170 K and –65 K between model 10 and model 1 would lead to a seismically detectable perturbation of compressional velocity Vp of about –8.2\(\times\)10−2 km s−1 and +3.1\(\times\)10−2 km s−1, respectively (i.e., about –1% and +0.4% of the Vp in the upper mantle63). In the case of 50% serpentinization (relevant to, e.g., Lesser Antilles57,58), temperature differences of +80 K and –25 K would lead to a Vp perturbation of about –3.8\(\times\)10−2 km s−1 and +1.2\(\times\)10−2 km s−1 (about –0.5% and +0.15%, respectively, of the Vp in the upper mantle), which may be difficult to clearly observe by seismology. In addition, the uncertainties in inferring the thermal structure from seismic interpretation are estimated to be about ±100 K in the upper mantle63, which is comparable to the thermal anomaly produced in our models. Seismic perturbations, however, can also be attributed to several factors other than temperature, including (1) the intrinsic density of a given mineral phase (antigorite is less dense than olivine), (2) the composite elastic properties of the mineral assemblage, and (3) the presence of fluid/melts, etc. Thus, it would be challenging to attribute univocally a seismic or temperature anomaly to a single factor.
On the other hand, the seismicity in the lower plane of a DSZ is more enigmatic and under debate. Depending on the variable temperature profile and water content, etc., in a subduction zone, the lower seismic plane could be caused by either also the dehydration of hydrous minerals9,13, or other rupture mechanisms. Another plausible trigger for intermediate-depth earthquakes is the thermal runaway12,19,20,21,22, a ductile deformation mechanism in shear zones, where the intense frictional heating induces weakening of rocks and causes the formation of a self-localizing slip planes. To produce such slip planes, the rate of heat production inside the shear zone has to be larger than the rate of heat dissipation from the shear zone19. When this condition occurs, the net heat confined inside the shear zone leads to a temperature increase, which promotes ductile deformation and generates weak slip planes. The severe frictional heat sometimes can lead to the formation of lenses of molten rock (e.g., pseudotachylites20,64).
Although we have not tested this mechanism in our models, we propose that antigorite’s anisotropic thermal insulating effect can promote the onset of thermal runaway. The ideal setting to trigger thermal runaway are the pre-existing shear zones22, in particular the highly serpentinized normal faults produced in trench-rise system. It has been suggested that mineral transformations, such as dehydration of antigorite, in the fault zones would assist the fault slip that generates heat, leading to self-localizing thermal runaway (see ref. 12 and references therein). During shear deformation, the antigorite crystals hosted within the fault zones should orient themselves with the [001] direction perpendicular to the slip plane. In this configuration, the heat produced by friction can flow easily along the slip plane (due to the high \({\Lambda }^{010}\)), but rarely escapes from the shear zone60 due to the thermal blanketing effect by the surrounding antigorite’s low \({\Lambda }^{001}\). Thermal runaway is also aided by the weak mechanical resistance of antigorite crystals along the basal planes17. Moreover, to achieve runaway deformation regime, the background temperature of the shear zone has to be lower than a critical temperature \({{{{{{\rm{T}}}}}}}_{{{{{{\rm{c}}}}}}}\), below which the thermal relaxation generates unstable ductile deformation19. Typically, the critical temperature of a mantle rock is <700 K65. Our models show that the thermal insulating effect of oriented antigorite keeps the slab interior in colder temperatures, which may be a necessary condition for the onset of thermal runaway. Potentially, the antigorite-assisted thermal runaway could be responsible for the lower seismic plane of a DSZ12. To better our understanding of the thermal runaway mechanism, future numerical modelings on the effect of anisotropic heat diffusion inside shear zones are required to test such scenario.
Finally, the anisotropic thermal conductivity of antigorite can also be relevant in the serpentinized forearc regions66. The surface heat flow on top of the forearc is typically very low (~30–40 mW m−2) and, in old subduction zones, is uniform from the trench to the volcanic arc. Such anomalous constant value may be associated with the presence of insulating antigorite at depth. Future corner flow models67 should include antigorite’s anisotropic thermal conductivity to compute the rheology of a serpentinized mantle wedge.
In conclusion, we have experimentally demonstrated a strong thermal conductivity anisotropy of antigorite under high P-T conditions along subduction, where the conductivity along the [001] is ~3–4 times lower than that along the [010]. Our 2D numerical simulations show that such anisotropy combined with its CPO during subduction can significantly influence the thermal evolution of a sinking slab with high degrees of serpentinization (\(\ge\)50%). In particular, when antigorite is oriented with the [001] direction perpendicular to the slab dip, it acts as an effective thermal insulator that hinders heat flowing toward the slab interior. This effect increases the temperature at the interface between the lower crust and the lithospheric mantle, thus promoting dehydration embrittlement in the outer ~10 km of the slab. Potentially, the anisotropic insulation by oriented antigorite can promote thermal runaway within the serpentinized faults inside the slab, triggering intermediate-depth earthquakes. Future experimental studies on the thermal conductivity of relevant phyllosilicates in the sediments and crust (e.g., muscovite, biotite, chlorite, phengite, and talc, etc.) along specific crystal orientation and P-T conditions of slab subduction will offer more comprehensive datasets for the thermal properties of slab’s minerals. Combined with detailed thermal evolution modeling, these would substantially advance our understanding of important geological, geophysical, and geodynamical features in a subduction zone, including the thermal state, flow dynamics, water transportation, magmatism, and seismic structure of the region.
Methods
Sample characterization and preparation
We collected natural antigorite samples from Baibao River (Hualien, Taiwan) and used energy-dispersive X-ray spectroscopy to determine their average chemical composition—(Mg2.80Fe0.05)Si2.08O5(OH)3.77 with a density of 2.58 ± 0.03 g cm−3. We also performed X-ray diffraction and Raman spectroscopy measurements (Supplementary Fig. S1) to confirm antigorite’s crystal structure and vibrational spectrum, respectively, which are both in good agreement with RRUFF standard database for antigorite68. We oriented the antigorite crystals to two principal directions along the in-plane b-axis [010] and cross-plane c-axis [001] by either electron backscattered diffraction or single-crystal X-ray diffraction.
For high pressure thermal conductivity measurements at room temperature, we first manually polished the crystals down to a thickness of ~30 μm, and then deposited an ~90-nm-thick aluminum (Al) film on the crystals. We loaded the crystal sample along with a few ruby balls into a symmetric piston-cylinder diamond anvil cell (DAC) equipped with a pair of 500 μm culets and a stainless-steel gasket pre-indented to ~80 μm thick. We compressed the crystal by loading silicone oil as the pressure medium, see Supplementary Fig. S2 for the schematic illustration of our sample geometry and experimental set-up. We calibrated the pressure within the sample chamber by the shift of the ruby R1 fluorescence peak69 with an uncertainty of ~0.1–0.2 GPa.
To perform simultaneous high P-T thermal conductivity measurements, we used an externally-heated BX-90 DAC (EHDAC), in which a toroidal resistive heater was placed next to the sample chamber, providing a spatially-homogeneous high T environment within the small sample chamber. We compressed the sample by high-pressure gas loading of Ar as the pressure medium. During our high P-T measurements, we also inserted a gas membrane in the EHDAC, allowing us to in situ control the pressure during heating. The pressure uncertainty is ~0.1–0.5 GPa due to the broadening of ruby peak upon heating. Details of the EHDAC assemblage and pressure uncertainty during high P-T measurements were described in refs. 27,70.
Thermal conductivity measurements
We measured the lattice thermal conductivity of single-crystal antigorite at high pressure and a wide range of temperature conditions by time-domain thermoreflectance (TDTR). TDTR is a well-developed ultrafast optical pump-probe method that enables precise measurement of the thermal conductivity of a material29,71. In our TDTR setup, we split the output of a mode-locked Ti: sapphire laser into a pump beam, which heated the Al film coated on the antigorite sample, and a probe beam, which monitored the changes in optical reflectivity of the Al film caused by temperature variations. The small intensity changes of the reflected probe beam, synchronous with the 8.7 MHz modulation frequency, were measured by a silicon photodiode detector coupled with a lock-in amplifier. Details of the TDTR were described elsewhere, e.g., ref. 71.
We determined the thermal conductivity of the sample by comparing the ratio of thermoreflectance in-phase signal Vin to out-of-phase Vout (-Vin/Vout) with calculations based on a bidirectional heat-flow thermal model, see, e.g., refs. 70,72 and references therein. The thermal model requires inputs of several parameters, including laser spot size (7.6 μm), and thickness, thermal conductivity, and volumetric heat capacity of each layer (i.e., silicone oil or Ar, Al film, and antigorite). All input parameters are pre-determined or in situ measured during the measurements (see, e.g., refs. 72,73,74,75 for details), while antigorite’s thermal conductivity is the only significant unknown and free parameter to be determined. The volumetric heat capacity of antigorite is a key parameter when determining its thermal conductivity. We calculated the P-T dependent volumetric heat capacity of antigorite based on the results by ref. 32 combined with its equation of state44. Note that the uncertainty of our thermal conductivity data majorly arises from analysis uncertainty, instead of measurement uncertainty. We estimated the data uncertainty by evaluating the total uncertainties propagated by all the input parameters used in the thermal model, see Supplementary Figs. S3, S4 as well as relevant references therein. We found the error is ~10% at P < 5 GPa and <20% at 13 GPa.
Numerical modeling
We designed a 2D model of slab subduction to investigate the effects of the strong anisotropy of antigorite’s thermal conductivity on the heat propagation within the slab. The full description of the physical and numerical model of the slab, its governing equations, the limitations, and the code benchmark are reported in the Supplementary Information (Note S1–S7, Figs. S4–S10, Tables S1–S4, S7). We wrote a MATLAB code to compute the heat diffusion equation \({{{{{\rm{\rho }}}}}}{{{{{\rm{Cp}}}}}}(\partial {{{{{\rm{T}}}}}}/\partial {{{{{\rm{t}}}}}})=- \nabla \cdot (- \Lambda \nabla {{{{{\rm{T}}}}}})\) with the finite difference method76 (Note S2). The symbols represent: \(\Lambda ({{{{{\rm{W}}}}}}\,{{{{{{\rm{m}}}}}}}^{-1}\,{{{{{{\rm{K}}}}}}}^{-1})\) thermal conductivity (Note S3), \({{{{{\rm{\rho }}}}}}\left({{{{{\rm{kg}}}}}}{{{{{{\rm{m}}}}}}}^{-3}\right)\) density (Note S4), \({{{{{\rm{Cp}}}}}}({{{{{\rm{J}}}}}}{{{{{{\rm{kg}}}}}}}^{-1}{{{{{{\rm{K}}}}}}}^{-1})\) heat capacity (Note S5), \({{{{{\rm{T}}}}}}({{{{{\rm{K}}}}}})\) temperature, \({{{{{\rm{t}}}}}}({{{{{\rm{s}}}}}})\) time, and \(\partial {{{{{\rm{T}}}}}}/\partial {{{{{\rm{t}}}}}}({{{{{\rm{K}}}}}}{{{{{{\rm{s}}}}}}}^{-1})\) temperature evolution over time. In 2D, the parameter \(\nabla {{{{{\rm{T}}}}}}({{{{{\rm{K}}}}}}{{{{{{\rm{m}}}}}}}^{-1})\) indicates the temperature gradient along the vertical (\(\partial {{{{{\rm{T}}}}}}/\partial {{{{{\rm{y}}}}}}\)), and the horizontal direction (\(\partial {{{{{\rm{T}}}}}}/\partial {{{{{\rm{x}}}}}}\)) (Note S2). We simplified the slab as a \(600\times 12\)0 k\({{{{{\rm{m}}}}}}\) rectangle, and we discretized the domain into a mesh of square cells \(\triangle {{{{{\rm{x}}}}}}=\triangle {{{{{\rm{y}}}}}}=500{{{{{\rm{m}}}}}}\) (i.e., \(242\times 1202\) nodes), see Figs. S5–S7. In the finite difference method76 (Note S2), \(\partial {{{{{\rm{T}}}}}}\) was approximated as the temperature difference between two consecutive nodes (\(\triangle {{{{{\rm{T}}}}}}={{{{{{\rm{T}}}}}}}_{{{{{{\rm{i}}}}}}}-{{{{{{\rm{T}}}}}}}_{{{{{{\rm{i}}}}}}+1}\)), and the temperature gradient \(\nabla {{{{{\rm{T}}}}}}\) was computed considering the vertical (\(\partial {{{{{\rm{y}}}}}} \sim \triangle {{{{{\rm{y}}}}}}\)) and horizontal (\(\partial {{{{{\rm{x}}}}}} \sim \triangle {{{{{\rm{x}}}}}}\)) grid spacing, i.e., \({\nabla {{{{{\rm{T}}}}}}}_{{{{{{\rm{V}}}}}}}=\triangle {{{{{\rm{T}}}}}}/\triangle {{{{{\rm{y}}}}}}\) and \({\nabla {{{{{\rm{T}}}}}}}_{{{{{{\rm{H}}}}}}}=\triangle {{{{{\rm{T}}}}}}/\triangle {{{{{\rm{x}}}}}}\). For the calculation we set the time step to \(\partial {{{{{\rm{t}}}}}} \sim \triangle {{{{{\rm{t}}}}}}=\) \(2000\) \({{{{{\rm{yrs}}}}}}\). At the four edges of the slab we added an extra stencil of nodes (Fig. S6), in which we set isothermal boundary conditions (Note S2). The slab thickness (\(12\)0 km) was computed using the analytical solution \(2.32\sqrt{{{{{{\rm{\kappa }}}}}}{{{{{\rm{t}}}}}}}\)77, assuming a thermal diffusivity of \({{{{{\rm{\kappa }}}}}}=1 \times {10}^{-6}({{{{{{\rm{m}}}}}}}^{2}{{{{{{\rm{s}}}}}}}^{-1})\) and a slab age of \(80\) My\({{{{{\rm{rs}}}}}}\)78 (Note S1). The slab was composed of three layers (Fig. S6): (1) a \(7\)-\({{{{{\rm{km}}}}}}\)-thick crust79, (2) a \(3\)-\({{{{{\rm{km}}}}}}\)-thick layer of hydrous lithosphere80, and (3) a \(110\)-\({{{{{\rm{km}}}}}}\)-thick dry lithosphere. For simplicity, we assumed the lithospheric slab and the oceanic crust to be composed entirely of olivine, while the hydrous layer was made of an assemblage of antigorite and olivine with variable fractions of \({{{{{{\rm{\varphi }}}}}}}_{{{{{{\rm{Atg}}}}}}}\) and \({{{{{{\rm{\varphi }}}}}}}_{{{{{{\rm{Ol}}}}}}}=1-{{{{{{\rm{\varphi }}}}}}}_{{{{{{\rm{Atg}}}}}}}\)(Note S3). The initial temperature profile of the slab at the trench was computed with the half-space cooling equation77 (Note S1, Equation 2, Fig. S5). In the simulation, the slabs subducted vertically (i.e., dip angle \({{{{{{\rm{\theta }}}}}}}_{{{{{{\rm{dip}}}}}}}=90^\circ\)), with a constant velocity (\({{{{{{\rm{v}}}}}}}_{{{{{{\rm{sink}}}}}}}=5{{{{{\rm{cm}}}}}}\) \({{{{{{\rm{yr}}}}}}}^{-1}\)) (Note S1, Fig. S7). Subduction occurred from \(12\)0 k\({{{{{\rm{m}}}}}}\) depth to avoid the interactions between the slab and the overriding plate81 down to \(23\)0 k\({{{{{\rm{m}}}}}}\), i.e., the typical breakdown depth of antigorite3. During subduction, the increasing P-T conditions were taken from the adiabatic profile of the mantle82 (Note S1, Fig. S8), while the P-T-phase evolution of each parameter was computed with the equations extrapolated from literature (Note S3–S5, Fig. S9): (A) olivine ΛDryOl, ΛWetOl41,47, ρOl83 and CpOl83; (B) antigorite ΛAtg (this study), ρAtg32,44, CpAtg32. The aggregate thermal conductivity of the hydrous layer was computed with the geometric average of olivine and antigorite contributions84 (Note S3): ΛHydLayer\(=\)(ΛOl)φOl\(*\)(ΛAtg)φAtg. We chose to use geometric averaging because it provides the best correlation between the computed and the measured thermal conductivity of a random grainy medium85. To investigate the anisotropy of antigorite’s thermal conductivity we distinguished the horizontal \({\Lambda }_{{{{{{\rm{H}}}}}}}\) and vertical \({\Lambda }_{{{{{{\rm{V}}}}}}}\) components along the corresponding axis (Note S2). The limitations of the model (Note S6) include: (1) the 2D geometry, which produces a colder slab compared to the 3D geometry; (2) the simplified petrology, composed only by olivine and antigorite; (3) the P-T independent mineral assemblage; (4) the absence of self-consistent gravity-driven subduction; (5) the lack of additional heating source (e.g., radioactive contribution); (6) the limited domain size (the surrounding mantle is only 500 m thick); (7) the lack of interactions between the slab and the overriding plate; (8) the absence of interactions between the slab and the mantle wedge (corner flow).
Data availability
The main data generated in this study are available in https://zenodo.org/records/11104058.
Code availability
The MATLAB script used to perform numerical modeling is available in https://zenodo.org/records/11104058.
References
Crameri, F. et al. A transdisciplinary and community-driven database to unravel subduction zone initiation. Nat. Commun. 11, 3750 (2020).
Pearce, J. A. & Peate, D. Tectonic implications of the composition of volcanic arc magmas. Annu. Rev. Earth Planet Sci. 23, 251 (1995).
Wei, S. S., Wiens, D. A., van Keken, P. E. & Cai, C. Slab temperature controls on the Tonga double seismic zone and slab mantle dehydration. Sci. Adv. 3, e1601755 (2017).
Brudzinski, M. R., Thurber, C. H., Hacker, B. R. & Engdahl, E. R. Global prevalence of double benioff zones. Science 316, 1472–1474 (2007).
Seno, T. & Yamanaka, Y. Double seismic zones, compressional deep trench-outer rise events, and superplume. Geophys. Monogr. 96, 347 (1996).
Yamasaki, T. & Seno, T. Double seismic zone and dehydration embrittlement of the subducting slab. J. Geophys. Res. 108, B42212 (2003).
Umino, N. & Hasegawa, A. On the Two-Layered Structure of Deep Seismic Plane in Northeastern Japan Arc. J. Seismol. Soc. Jpn. 27, 125–139 (1975).
Kirby, S. H., Stein, S., Okal, E. A. & Rubie, D. C. Metastable mantle phase transformations and deep earthquakes in subducting oceanic lithosphere. Rev. Geophys. 34, 261–306 (1996).
Peacock, S. M. Are the lower planes of double seismic zones caused by serpentine dehydration in subducting oceanic mantle? Geology 29, 299–302 (2001).
Wang, J., Zhao, D. & Yao, Z. Seismic anisotropy evidence for dehydration embrittlement triggering intermediate-depth earthquakes. Sci. Rep. 7, 2613 (2017).
Jung, H., Green, H. W. & Dobrzhinetskaya, L. F. Intermediate-depth earthquake faulting by dehydration embrittlement with negative volume change. Nature 428, 545 (2004).
Ferrand, T. P. Seismicity and mineral destabilizations in the subducting mantle up to 6 GPa, 200 km depth. Lithos. 334–335, 205–230 (2019).
Kim, D., Jung, H. & Lee, J. Impact of chlorite dehydration on intermediate-depth earthquakes in subducting slabs. Commun. Earth Environ. 4, 491 (2023).
van Keken, P. E., Kita, S. & Nakajima, J. Thermal structure and intermediate-depth seismicity in the Tohoku-Hokkaido subduction zones. Solid Earth 3, 355–364 (2012).
Kao, H. & Rau, R.-J. Detailed structures of the subducted Philippine Sea plate beneath northeast Taiwan’ A new type of double seismic zone. J. Geophys. Res. Solid Earth 104, 1015–1033 (1999).
Hobbs, B. & Ord, A. Plastic instabilities: implications for the origin of intermediate and deep focus earthquakes. J. Geophys. Res. Solid Earth 93, 10521 (1988).
Reynard, B., Nakajima, J. & Kawakatsu, H. Earthquakes and plastic deformation of anhydrous slab mantle in double Wadati-Benioff zones. Geophys. Res. Lett. 37, L24309 (2010).
Thielmann, M., Rozel, A., Kaus, B. J. P. & Ricard, Y. Intermediate-depth earthquake generation and shear zone formation caused by grain size reduction and shear heating. Geology 43, 791–794 (2015).
Thielmann, M. Grain size assisted thermal runaway as a nucleation mechanism for continental mantle earthquakes: Impact of complex rheologies. Tectonophysics 746, 611–623 (2018).
John, T. et al. Generation of intermediate-depth earthquakes by self-localizing thermal runaway. Nat. Geosci. 2, 137–140 (2009).
Ogawa, M. Shear instability in a viscoelastic material as the cause of deep focus earthquakes. J. Geophys. Res. 92, 13801–13810 (1987).
Kelemen, P. B. & Hirth, G. A periodic shear-heating mechanism for intermediate-depth earthquakes in the mantle. Nature 446, 787–790 (2007).
Wunder, B. & Schreyer, W. Antigorite: high-pressure stability in the system MgO-SiO2-H2O (MSH). Lithos 41, 213–227 (1997).
Reynard, B. Serpentine in active subduction zones. Lithos 178, 171–185 (2013).
Ohtani, E. The role of water in Earth’s mantle. Natl Sci. Rev. 7, 224–232 (2020).
Ferrand, T. P. et al. Dehydration-driven stress transfer triggers intermediate-depth earthquakes. Nat. Commun. 8, 15247 (2017).
Hsieh, W.-P. et al. Low thermal conductivity of hydrous phase D leads to a self‐preservation effect within a subducting slab. J. Geophys. Res. Solid Earth 127, e2022JB024556 (2022).
Okuda, Y. et al. Thermal conductivity of Fe-bearing post-perovskite in the Earth’s lowermost mantle. Earth Planet Sci. Lett. 547, 116466 (2020).
Zhou, Y., Dong, Z. Y., Hsieh, W. P., Goncharov, A. F. & Chen, X. J. Thermal conductivity of materials under pressure. Nat. Rev. Phys. 4, 319 (2022).
Hsieh, W.-P. et al. Low thermal conductivity of iron-silicon alloys at Earth’s core conditions with implications for the geodynamo. Nat. Commun. 11, 3332 (2020).
Horai, K. Thermal conductivity of rock-forming minerals. J. Geophys Res 76, 1278 (1971).
Osako, M., Yoneda, A. & Ito, E. Thermal diffusivity, thermal conductivity and heat capacity of serpentine (antigorite) under high pressure. Phys. Earth Planet. Inter. 183, 229–233 (2010).
Capitani, G. & Mellini, M. The modulated crystal structure of antigorite: The m = 17 polysome. Am. Mineral. 89, 147–158 (2004).
Marquardt, H., Speziale, S., Koch-Müller, M., Marquardt, K. & Capitani, G. C. Structural insights and elasticity of single-crystal antigorite from high-pressure Raman and Brillouin spectroscopy measured in the (010) plane. Am. Mineral. 100, 1932–1939 (2015).
Mookherjee, M. & Stixrude, L. Structure and elasticity of serpentine at high-pressure. Earth Planet Sci. Lett. 279, 11–19 (2009).
Satta, N. et al. Single-crystal elasticity of antigorite at high pressures and seismic detection of serpentinized slabs. Geophys. Res. Lett. 49, e2022GL099411 (2022).
Ashcroft, N. W. & Mermin, N. D. Cengage Learning, Solid State Physics (1976).
Katayama, I., Hirauchi, K. I., Michibayashi, K. & Ando, J. I. Trench-parallel anisotropy produced by serpentine deformation in the hydrated mantle wedge. Nature 461, 1114–1117 (2009).
Padrón-Navarta, J. A., Tommasi, A., Garrido, C. J. & López Sánchez-Vizcaíno, V. Plastic deformation and development of antigorite crystal preferred orientation in high-pressure serpentinites. Earth Planet Sci. Lett. 349–350, 75–86 (2012).
Hsieh, W.-P., Deschamps, F., Okuchi, T. & Lin, J.-F. Effects of iron on the lattice thermal conductivity of Earth’s deep mantle and implications for mantle dynamics. Proc. Natl Acad. Sci. USA 115, 4099–4104 (2018).
Chang, Y.-Y., Hsieh, W.-P., Tan, E. & Chen, J. Hydration-reduced lattice thermal conductivity of olivine in Earth’ s upper mantle. Proc. Natl Acad. Sci. USA 114, 4078 (2017).
Osako, M., Ito, E. & Yoneda, A. Simultaneous measurements of thermal conductivity and thermal diffusivity for garnet and olivine under high pressure. Phys. Earth Planet. Inter. 144, 311–320 (2004).
Yang, C., Inoue, T., Yamada, A., Kikegawa, T. & Ando, Jichi Equation of state and phase transition of antigorite under high pressure and high temperature. Phys. Earth Planet. Inter. 228, 56–62 (2014).
Hilairet, N., Daniel, I. & Reynard, B. Equation of state of antigorite, stability field of serpentines, and seismicity in subduction zones. Geophys. Res. Lett. 33, L02302 (2006).
Capitani, G. C. & Stixrude, L. A first-principle investigation of antigorite up to 30 GPa: structural behavior under compression. Am. Mineral. 97, 1177–1186 (2012).
Bezacier, L., Reynard, B., Cardon, H., Montagnac, G. & Bass, J. D. High-pressure elasticity of serpentine and seismic properties of the hydrated mantle wedge. J. Geophys. Res. Solid Earth 118, 527–535 (2013).
Xu, Y. et al. Thermal diffusivity and conductivity of olivine, wadsleyite and ringwoodite to 20 GPa and 1373 K. Phys. Earth Planet. Inter. 144, 321–336 (2004).
Zhang, Y., Yoshino, T., Yoneda, A. & Osako, M. Effect of iron content on thermal conductivity of olivine with implications for cooling history of rocky planets. Earth Planet Sci. Lett. 519, 109–119 (2019).
Bolfan-Casanova, N. Water in the Earth. Miner. Mag. 69, 229 (2005).
Ranero, C. R. & Sallarès, V. Geophysical evidence for hydration of the crust and mantle of the Nazca plate during bending at the north Chile trench. Geology 32, 549–552 (2004).
Contreras-Reyes, E. et al. Deep seismic structure of the Tonga subduction zone: Implications for mantle hydration, tectonic erosion, and arc magmatism. J. Geophys. Res. Solid Earth 116, B10103 (2011).
Green, H. W., Chen, W. P. & Brudzinski, M. R. Seismic evidence of negligible water carried below 400-km depth in subducting lithosphere. Nature 467, 828–831 (2010).
Grevemeyer, I., Ranero, C. R., Flueh, E. R., Kläschen, D. & Bialas, J. Passive and active seismological study of bending-related faulting and mantle serpentinization at the Middle America trench. Earth Planet Sci. Lett. 258, 528–542 (2007).
Ranero, C. R., Morgan, J. P., McIntosh, K. D. & Reichert, C. Flexural faulting and mantle serpentinization at the Middle American. Nature 425, 367–373 (2003).
Faccenda, M., Gerya, T. V. & Burlini, L. Deep slab hydration induced by bending-related variations in tectonic pressure. Nat. Geosci. 2, 790–793 (2009).
Faccenda, M., Gerya, T. V., Mancktelow, N. S. & Moresi, L. Fluid flow during slab unbending and dehydration: Implications for intermediate-depth seismicity, slab weakening and deep water recycling. Geochem. Geophys. Geosyst. 13, Q01010 (2012).
Allen, R. W., Collier, J. S. & Henstock, T. J. The role of crustal accretion variations in determining slab hydration at an Atlantic subduction zone. J. Geophys. Res. Solid Earth 127, e2022JB024349 (2022).
Cooper, G. F. et al. Variable water input controls evolution of the Lesser Antilles volcanic arc. Nature 582, 525–529 (2020).
Jung, H. Crystal preferred orientations of olivine, orthopyroxene, serpentine, chlorite, and amphibole, and implications for seismic anisotropy in subduction zones: a review. Geosci. J. 21, 985–1011 (2017).
Hirth, G. & Guillot, S. Rheology and tectonic significance of serpentinite. Elements 9, 107–113 (2013).
Kern, H., Liu, B. & Popp, T. Relationship between anisotropy of P and S wave velocities and anisotropy of attenuation in serpentinite and amphibolite. J. Geophys. Res. 102, 3051–3065 (1997).
Deal, M. M., Nolet, G. & Van Der Hilst, R. D. Slab temperature and thickness from seismic tomography 1. Method and application to Tonga. J. Geophys. Res. Solid Earth 104, 28789–28802 (1999).
Cammarano, F., Goes, S., Vacher, P. & Giardini, D. Inferring upper-mantle temperatures from seismic velocities. Phys. Earth Planet. Inter. 138, 197–222 (2003).
Ueda, T., Obata, M., Di Toro, G., Kanagawa, K. & Ozawa, K. Mantle earthquakes frozen in mylonitized ultramafic pseudotachylytes of spinel-lherzolite facies. Geology 36, 607–610 (2008).
Hobbs, B. E., Ord, A. & Teyssier, C. Earthquakes in the ductile regime? Pure Appl. Geophys. 124, 309–336 (1986).
Hyndman, R. D. & Peacock, S. M. Serpentinization of the forearc mantle. Earth Planet Sci. Lett. 212, 417–432 (2003).
van Keken, P. E., Kiefer, B. & Peacock, S. M. High-resolution models of subduction zones: Implications for mineral dehydration reactions and the transport of water into the deep mantle. Geochem. Geophys. Geosyst. 3, 1056 (2002).
Antigorite R070228. https://rruff.info/Antigorite.
Dewaele, A., Loubeyre, P. & Mezouar, M. Equations of state of six metals above 94 GPa. Phys. Rev. B 70, 094112 (2004).
Hsieh, W.-P. High-pressure thermal conductivity and compressional velocity of NaCl in B1 and B2 phase. Sci. Rep. 11, 21321 (2021).
Cahill, D. G. et al. Nanoscale thermal transport. II. 2003-2012. Appl. Phys. Rev. 1, 011305 (2014).
Hsieh, W.-P., Chen, B., Li, J., Keblinski, P. & Cahill, D. G. Pressure tuning of the thermal conductivity of the layered muscovite crystal. Phys. Rev. B 80, 180302 (2009).
Hsieh, W.-P., Tsao, Y. & Lin, C. Thermal conductivity of helium and argon at high pressure and high temperature. Materials 15, 6681 (2022).
Hsieh, W.-P. Thermal conductivity of methanol-ethanol mixture and silicone oil at high pressures. J. Appl Phys. 117, 235901 (2015).
Chen, B., Hsieh, W., Cahill, D. G., Trinkle, D. R. & Li, J. Thermal conductivity of compressed H 2 O to 22 GPa: a test of the Leibfried-Schlomann equation. Phys. Rev. B 83, 132301 (2011).
Gerya, T. Introduction to Numerical Geodynamic Modelling (Cambridge University Press, 2019).
Turcotte, D. L. & Schubert, G. Geodynamics (Cambridge University Press, 2014). https://doi.org/10.1017/CBO9780511843877.
Stein, C. A. & Stein, S. A model for the global variation in oceanic depth and heat flow with lithospheric age. Nature 359, 123–129 (1992).
Grevemeyer, I., Ranero, C. R. & Ivandic, M. Structure of oceanic crust and serpentinization at subduction trenches. Geosphere 14, 395–418 (2018).
John, T., Scambelluri, M., Frische, M., Barnes, J. D. & Bach, W. Dehydration of subducting serpentinite: Implications for halogen mobility in subduction zones and the deep halogen cycle. Earth Planet Sci. Lett. 308, 65–76 (2011).
Eberle, M. A., Grasset, O. & Sotin, C. A numerical study of the interaction between the mantle wedge, subducting slab, and overriding plate. Phys. Earth Planet. Inter. 134, 191–202 (2002).
Katsura, T. A revised adiabatic temperature profile for the mantle. J. Geophys. Res. Solid Earth 127, e2021JB023562 (2022).
Su, C., Liu, Y., Song, W., Fan, D. & Wang, Z. Thermodynamic properties of San Carlos olivine at high temperature and high pressure. Acta Geochimic 37, 171–179 (2018).
Grose, C. J. & Afonso, J. C. New constraints on the thermal conductivity of the upper mantle from numerical models of radiation transport. Geochem. Geophys. Geosyst. 20, 2378–2394 (2019).
Fuchs, S., Schütz, F., Förster, H. J. & Förster, A. Evaluation of common mixing models for calculating bulk thermal conductivity of sedimentary rocks: Correction charts and new conversion equations. Geothermics 47, 40–52 (2013).
Acknowledgements
This work was partially supported by the Academia Sinica and the National Science and Technology Council (NSTC) of Taiwan, Republic of China, under Contract IA-111-M02 (to W.P.H.) and 110-2628-M-001-001-MY3 (to W.P.H.). Y.H.C. acknowledges support by the NSTC 105-2922-I-008-212. E.M. acknowledges the support of the Helmholtz Young Investigators Group CLEAR (VH-NG-1325) and DFG project - BL 1690/1-1. The numerical model was written in MATLAB using the license 139702 (GFZ_network_concurrent). E.M. would like to thank N. Satta, and G. Criniti for the fruitful discussion, M. Jarema for the support in optimizing the MATLAB script, and S. Sarkar for English proof reading.
Author information
Authors and Affiliations
Contributions
W.P.H. conceived, designed, and supervised the project. Y.H.C., Y.C.T., and W.P.H. conducted experiments and analyzed data. E.M. performed numerical modeling. All authors wrote, reviewed, and commented on the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Haemyeong Jung, Nicolas Riel, and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Chien, YH., Marzotto, E., Tsao, YC. et al. Anisotropic thermal conductivity of antigorite along slab subduction impacts seismicity of intermediate-depth earthquakes. Nat Commun 15, 5198 (2024). https://doi.org/10.1038/s41467-024-49418-3
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-024-49418-3
- Springer Nature Limited