Metastable Wedge Research Articles

Elastic Anomalies Across the α‐β Phase Transition in Orthopyroxene: Implication for the Metastable Wedge in the Cold Subduction Slab

AbstractSingle‐crystal elasticity of both α‐ and β‐orthopyroxene was determined up to 20 GPa and 300 K by Brillouin scattering. Using the derived full elastic moduli (Cij), we investigated the contribution of the metastable pyroxene to the seismically observed 3%–5% low‐velocity anomalies along the subducting slab in the top transition zone. Our modeled results show that a harzburgite wedge with a 1000‐K colder geotherm and metastable α‐orthopyroxene and olivine displays compressional (VP) and shear‐wave (VS) velocities 3.0%–3.6(6)% and 2.0%–2.8(6)% lower than the surrounding mantle at 410–460 km depth, respectively. At deeper depth up to 520 km, VP and VS of this metastable wedge with β‐orthopyroxene and olivine are 3.6%–4.4(6)% and 2.8%–4.3(6)% lower than the pyrolitic mantle, respectively. The presence of both metastable orthopyroxene and olivine instead of metastable olivine alone helps better explain the origin of the low‐velocity anomalies within the subduction slab in the top transition zone.

A Metastable Fo-III Wedge in Cold Slabs Subducted to the Lower Part of the Mantle Transition Zone: A Hypothesis Based on First-Principles Simulations

The metastable olivine (Ol) wedge hypothesis assumes that Ol may exist as a metastable phase at the P conditions of the mantle transition zone (MTZ) and even deeper regions due to inhibition of the phase transitions from Ol to wadsleyite and ringwoodite caused by low T in the cold subducting slabs. It is commonly invoked to account for the stagnation of the descending slabs, deep focus earthquakes and other geophysical observations. In the last few years, several new structures with the forsterite (Fo) composition, namely Fo-II, Fo-III and Fo-IV, were either experimentally observed or theoretically predicted at very low T conditions. They may have important impacts on the metastable Ol wedge hypothesis. By performing first-principles calculations, we have systematically examined their crystallographic characteristics, elastic properties and dynamic stabilities from 0 to 100 GPa, and identified the Fo-III phase as the most likely metastable phase to occur in the cold slabs subducted to the depths equivalent to the lower part of the MTZ (below the ~600 km depth) and even the lower mantle. As disclosed by our theoretical simulations, the Fo-III phase is a post-spinel phase (space group Cmc21), has all cations in sixfold coordination at P < ~60 GPa, and shows dynamic stability for the entire P range from 0 to 100 GPa. Further, our static enthalpy calculations have suggested that the Fo-III phase may directly form from the Fo material at ~22 GPa (0 K), and our high-T phase relation calculations have located the Fo/Fo-III phase boundary at ~23.75 GPa (room T) with an averaged Clapeyron slope of ~−1.1 MPa/K for the T interval from 300 to 1800 K. All these calculated phase transition pressures are likely overestimated by ~3 GPa because of the GGA method used in this study. The discrepancy between our predicted phase transition P and the experimental observation (~58 GPa at 300 K) can be explained by slow reaction rate and short experimental durations. Taking into account the P-T conditions in the cold downgoing slabs, we therefore propose that the Fo-III phase, rather than the Ol, highly possibly occurs as the metastable phase in the cold slabs subducted to the P conditions of the lower part of the MTZ (below the ~600 km depth) and even the lower mantle. In addition, our calculation has showed that the Fo-III phase has higher bulk seismic velocity, and thus may make important contributions to the high seismic speeds observed in the cold slabs stagnated near the upper mantle-lower mantle boundary. Future seismic studies may discriminate the effects of the Fo-III phase and the low T. Surprisingly, the Fo-III phase will speed up, rather than slow down, the subducting process of the cold slabs, if it metastably forms from the Ol. In general, the Fo-III phase has a higher density than the warm MTZ, but has a lower density than the lower mantle, as suggested by our calculations.

An implosive component to the source of the deep Sea of Okhotsk earthquake of 24 May 2013: Evidence from radial modes and CMT inversion

We study the spectral amplitudes of the first two Earth radial modes, 0S0 and 1S0, excited by the Sea of Okhotsk earthquake of 24 May 2013, the largest deep event ever recorded, in the search for an isotropic component to its source. In contrast to the case of the 1994 Bolivian earthquake, we detect an implosive component MI=-1.1×1027 dyn*cm, equivalent to 3% of the full scalar moment, but 14% of the lone deviatoric component exciting the Earth’s radial modes. An independent moment tensor inversion, using the standard GlobalCMT algorithm but after relaxing its zero-trace constraint, similarly yields an implosive isotropic component, albeit with a larger amplitude, while it fails to document one in the case of the 1994 Bolivian deep earthquake. An implosive component to the source is expected in the model of transformational faulting in which deep earthquake rupture nucleates and grows upon transformation of metastable olivine to ringwoodite in the cold subducting slab. This interpretation is supported by quantitative estimates (0.9–4 m) of the thickness of the transformed shear zone, which scale favorably, relative to earthquake fault length, with the upper end of the range of laboratory results reported for ices, germanates and silicates. The resulting extent of the transformation in the metastable wedge is consistent with the local geometry of the deep slab, as recently determined by rupture modeling and aftershock distribution. Our results are in contrast to those for the two runner-up largest deep earthquakes, the 1994 Bolivian and 1970 Colombian shocks, for which a similar isotropic component could not be detected. We attribute this difference to variability in the ratio of isotropic to deviatoric components, which combined with the smaller size of the 1970 and 1994 events, would make any putative implosive component fall below detection levels, especially in the case of the 1970 Colombian earthquake for which only analog narrow-band records were available.

Strength‐limiting mechanisms in high‐confinement brittle‐like failure: Adiabatic transformational faulting

AbstractBrittle‐like failure under high confinement impacts many geophysical phenomena, including earthquakes at depths beyond a few tens of kilometers. We reanalyze recent observations of high‐confinement brittle‐like failure of nominally thermodynamically stable ice and antigorite and show that the brittle‐like failure only occurs when the strain rate within an incipient adiabatic instability accelerates to the point that the attendant increase in localized temperature approaches a phase transition temperature, initiating transformational faulting. We test a closed‐form model for this process based on independently measureable parameters by comparing model predictions to observations of brittle‐like failure of ice and antigorite under high confinement. Model results do not constrain the mechanism of transformational faulting. However, the similarity of brittle‐like failure characteristics across a variety of geologic materials suggests a similarity of process. We propose that localized heating due to an adiabatic instability initiates dynamic recrystallization near the phase transformational temperature. Transformational superplasticity during recrystallization concentrates plastic strain ahead of the transformed region, causing dynamic recrystallization there and propagating slip in an autocatalytic manner. The newly recrystallized grains rapidly harden as strain accumulates, limiting the stress drop. As the recrystallized grains harden and the stress increases, localized heating reinitiates the adiabatic instability in the still warm region, concentrating plastic strain due to thermal softening and initiating another wave of dynamic recrystallization and partial stress drop. This process, termed adiabatic transformational faulting, could explain how deep earthquakes can propagate beyond the metastable wedge since the propagation of slip is due to dynamic recrystallization rather than to the primary phase transformation.

Evidence for a metastable olivine wedge inside the subducted Mariana slab

An anomalous seismic velocity structure is revealed inside the subducted Pacific plate at northern Mariana near the deepest earthquake foci. We observe anomalous later phases in the P wave coda for a small cluster of deep earthquakes and an anomalous differential P wave slowness between a few pairs of the earthquakes belonging to the cluster. We interpret these observations with a 5% low velocity wedge which is nearly 25 km wide at the depth of 590 km and is truncated at 630 km. The wedge is bounded by two steeply dipping (60°–70°) transition layers with the thickness of about 4 km. The highly localized seismic anomaly inside the slab represents a wedge of metastable olivine (α) embedded in dry ringwoodite (γ). Deep earthquakes occur across only one of the boundaries of the metastable wedge, with some of them outside the wedge. The mechanism of deep earthquakes is discussed based on the spatial pattern of seismicity relative to the wedge.

Stresses along the metastable wedge of olivine in a subducting slab: possible explanation for the Tonga double seismic layer

A numerical calculation of the stresses associated with changes in volume during phase transitions of olivine in a descending slab results in a double layer of high shear stress along the metastable olivine wedge in the depth range 350–460 km. Stress in the upper layer is in-plane tensional and stress in the lower layer is down-dip compressional. The modeled stress field agrees with observations of stress in the Tonga double seismic zone. High shear stress also exists in the slab at depths below the metastable wedge. This stress distribution involves down-dip compression and trench-parallel tension, which agrees with about half of the focal mechanisms in the Tonga slab at depths of 460–690 km. The model supports the idea that at least two possible stress release mechanisms for deep earthquakes may act in the Tonga subducting slab. One, transformational faulting, is restricted to the metastable wedge while the other one acts below the metastable wedge.

2-D viscoelastic FEM simulation on stress state in the deep part of a subducted slab

The characteristics of the stress fields in deep subducting slabs are studied using viscoelastic plain strain finite element method. When introducing the new rheology structure given by Karato, et al into our computation, there emerge two regions with great shear stress just below the olivine-spinel phase transition zone, which encompass the low viscosity zone below the lower tip of the metastable wedge. Further, the directions of the main compressional stress of these two regions are all along the dip direction of the slab. These are in accordance with the seismic observations that there are two deep seismic zones in a slab and the directions of the main compressional stress in these two seismic zones are along the dip direction of the slab. Smaller effective viscosity probably caused by smaller grain size in the phase transformation zone does not have great influence on the stress state in the deep part of the slab. There is the maximum of shear stress at the transition region from olivine to spinel and the direction of the main compressional stress in this region is roughly perpendicular to the trend of the phase boundary no matter whether there exists metastable wedge, which nevertheless do not correspond to some well-known seismic observations.

Seismological constraints on the mechanism of deep earthquakes: temperature dependence of deep earthquake source properties

Seismological observations are of vital importance for understanding the mechanism of deep earthquakes. Most of the seismological observables for earthquakes deeper than 300 km are similar to shallow earthquakes. Deep earthquakes clearly represent shear failure on a planar surface as shown by their double couple mechanisms. Deep earthquake aftershock sequences show temporal decay rates and magnitude–frequency relations ( b-values) that are similar to shallow earthquakes, with the aftershocks occurring preferentially along the mainshock fault planes. In addition, deep earthquake rupture velocities are similar to those of shallow earthquakes. However, there are also some observations that are clearly distinctive relative to shallow earthquakes. Stress drops of deep earthquakes show a large variation but are larger, on average, than shallow earthquakes. Different deep seismic zones show very different b-values, in contrast to shallow earthquakes, which show similar b-values worldwide. In addition, deep earthquakes show fewer aftershocks than shallow earthquakes. A variety of observations suggest that deep earthquakes are highly sensitive to the temperature of the slab. Both deep earthquake b-values and the rate of deep earthquake aftershock occurrence are inversely correlated with the temperature of the deep slabs, suggesting that these factors are temperature-controlled to an extent much greater than with shallow earthquakes. Large deep earthquakes in warm slabs show slower rupture velocities, larger stress drops, and lower seismic efficiencies than similar earthquakes in cold slabs. The width of deep seismic zones is also probably temperature controlled, but deep earthquake rupture can propagate outside the normal limits of Benioff zone seismicity. Simple thermal models for the Tonga slab near the 1994 deep earthquake suggest that the temperature at the rupture termination point was at least 200°C warmer than the temperature that limits smaller earthquakes in the slab. These observations can be used to evaluate physical models for deep earthquakes, including brittle slip along fluid-weakened faults, transformational faulting, and thermal (and perhaps melt lubricated) shear instabilities. It is difficult to explain the large fault widths of some deep earthquakes using a fluid weakening model, since hydrated materials are expected in only a narrow depth zone at the top of the slab, and the fault planes do not have the expected orientations for reactivated faults. The lateral extent of the largest deep earthquakes cast doubt on the transformational faulting model, in which events should be confined within a narrow metastable wedge. Seismological studies have also failed to find evidence for the existence of metastable olivine in slabs. The temperature dependence of deep earthquakes argues in favor of a temperature-activated phenomenon, such as thermal shear instabilities and perhaps fault zone melting.

Avalanche Stratification - Experimental Tests of the “Metastable Wedge” and “Continuous Flow” Models

ABSTRACTWhen an initially homogeneous binary mixture of granular media such as fine and coarse sand is poured near the closed edge of a “quasi-two-dimensional” Hele-Shaw cell consisting of two vertical transparent plates held a narrow distance apart, the mixture spontaneously forms alternating segregated layers. Experimental measurements of this stratification effect are reported in order to determine which model, one which suggests that segregation only occurs when the granular material contained within a metastable heap between the critical and maximum angle of repose avalanches down the free surface, or one for which the segregation results from smaller particles becoming trapped in the top surface and being removed from the moving layer during continuous flow. The result reported here indicate that the Metastable Wedge model provides a natural explanation for the initial mixed zone which precedes the formation of the layers, while the Continuous Flow model explains the observed upward moving kink of segregated material for higher granular flux rates, and that both mechansims are necessary in order to understand the observed pairing of segregated layersfor intermediate flow rates and cell separations.

Variable conductivity: Effects on the thermal structure of subducting slabs

Understanding the thermal structure of the lithosphere in subduction zone environments is a key element in the realization of the source of deep focus earthquakes and other geodynamic processes. Model predictions of the thermal structure of subducting slabs are dependent upon a priori knowledge of how thermal conductivity varies within the slab. Using recent results for the variation of thermal conductivity as a function of temperature, pressure, and gross mineralogy, we investigate how models that utilize a realistic model for variable conductivity compare with canonical models for constant conductivity. Results suggest that jumps in thermal conductivity across the phase boundaries that arise due to transformations to β and γ‐spinel phases lead to old, fast, steeply dipping slabs being considerably warmer than previously thought. Although these results do not rule out the possibility of a metastable wedge source of deep focus earthquakes, they do imply the likelihood of metastable olivine being found at depth is less than once thought.

Velocity anomalies and spatial distributions of physical properties in horizontally lying slabs beneath the Northwestern Pacific region

We calculated velocity anomalies by constructing thermal structural models of horizontally lying slabs beneath Kuriles, Central Japan, Northern Izu–Bonin and Central Izu–Bonin arcs. The values of V p and V s anomalies become maximum in the depth range of 100 to 200 km within the slabs, reaching 5 to 6.4% for V p anomalies and 7 to 8.5% for V s anomalies. The calculated V s anomalies are larger than the V p anomalies, and the ratio of the former to the latter is 1.2 to 1.5 within the slab. This suggests that V s anomalies are more sensitive to temperature variation than V p anomalies, indicating that the existence of the horizontally lying slabs is detectable in more detail by using S-wave. We also calculated spatial distributions of physical properties in the horizontally lying slabs beneath the four regions. In the calculation, we assumed that the material of the slabs consists of olivine (Mg 0.89Fe 0.11) 2SiO 4. The estimated values of physical properties within the slabs are larger than those in the surrounding upper mantle, except for Poisson's ratio, while the values within the metastable wedge of olivine become smaller. This is most remarkable within the slab beneath the Central Izu–Bonin arc due to the lowest temperature among the four regions. The horizontally lying slab above the 660 km seismic discontinuity beneath the Central Izu–Bonin arc is denser up to 60 kg/m 3 and the slab within the metastable wedge of olivine is lighter up to −120 kg/m 3. The values of the physical properties, except for Poisson's ratio, decrease with increasing the horizontal distances of the horizontally lying portions of the slabs. This is due to thermal assimilation, which is largely dependent on moving velocity of the slabs.

Phase diagram for avalanche stratification of granular media

When a binary mixture of granular materials is poured into a quasi-two-dimensional Hele-Shaw cell, alternating stratified layers of large and small particles are formed along the top surface. This effect is studied as the plate separation of the cell and the flow rate at which the granular mixture is poured are systematically varied. A nontrivial ``phase diagram'' is found, with pairing of the stratification layers occurring for a finite range of plate separation and flow rate values. Numerical simulations based upon a model involving periodic avalanching of a metastable wedge of granular material above the critical angle of repose reproduce both the observed stratification pattern and the flow rate dependence.

Comment on “Seismological evidence for the existence of anisotropic zone in the metastable wedge inside the subducting Izu‐Bonin Slab” by Takashi Iidaka and Kazushige Obara

Comment on “Seismological evidence for the existence of anisotropic zone in the metastable wedge inside the subducting Izu‐Bonin Slab” by Takashi Iidaka and Kazushige Obara

Reply [to “Comment on ‘Seismological evidence for the existence of anisotropic zone in the metastable wedge inside the subducting Izu‐Bonin Slab’ by Takashi Iidaka and Kazushige Obara”

Reply [to “Comment on ‘Seismological evidence for the existence of anisotropic zone in the metastable wedge inside the subducting Izu‐Bonin Slab’ by Takashi Iidaka and Kazushige Obara”

Seismological evidence for the existence of anisotropic zone in the metastable wedge inside the subducting Izu‐Bonin Slab

The mechanism of deep earthquakes has been a puzzle since their discovery almost 70 years ago. Experimental observations at high‐pressure have reported faulting caused by spinel‐filled anti‐cracks in metastable olivine and suggesting that the anticrack mechanism is responsible for deep earthquakes. We compared the shear‐wave splitting of earthquakes from the upper and lower planes of the double seismic zone to detect anisotropy caused by the faultings inside the metastable wedge. We detected the anisotropic zone inside the metastable wedge at the depth of about 340–390 km, which corresponds to the depth range of a second peak on the histogram of earthquakes. The observed polarization direction, east‐west, is almost parallel to the direction of subduction, which can be explained by a view that the shear‐wave splitting is caused by the en echelon arrays of spinel‐filled anti‐cracks and anticrack‐associated faultings.

Stress fields associated with metastable phase transitions in descending slabs and deep-focus earthquakes

Inside subducting slabs the interaction of different metastable phase transformations (α-olivine → β-spinel, β-spinel → γ-spinel, γ-spinel → perovskite + magnesiowüstite) associated with the latent heat release and absorption for these transitions can result in a complex thermal slab structure. Thermokinetic coupling processes cause thermal anomalies and buoyancy contrasts between slab and mantle that strongly influence the stress field within subducting plates. We have described the thermal field within the downgoing slab in a Lagrangian framework by a self-consistent, high-resolution, two-dimensional thermokinetic coupling model, which also incorporates the effects of nucleation site saturation. Using an adaptive method, the threshold for the nucleation site saturation is found to be less than 0.1% transformation degree in the cold slab interior. This can significantly shift the metastable wedge of α-olivine deeper, as suggested from previous thermokinetic slab models. However, this effect is compensated by the latent heat release. During the α-olivine → β-spinel and β-spinel → γ-spinel phase transitions, latent heat release produce very sharp phase boundaries in the slab. For fast slabs the phase boundaries reveal significant metastable perturbations from the equilibrium state in the cold interior. However, the strong thermal interaction between both ongoing phase transitions results in sharp overlapping phase boundaries with dramatic consequences for the slab stress field. We have investigated the detailed physical properties and the state of stress in the deeper portion of a subducting plate by using up-to-date temperature field and physical properties determined from recent high P-T experiments. We find that the slab is denser up to 100 kg m −3 than the surrounding mantle. Remarkable denser portions up to 260 kg m −3 are located just below the uplifted α → β and β → γ equilibrium phase transitions, while lighter portions up to −210 kg m −3 can be found just above the depressed γ-spinel to perovskite + magnesiowüstite transition and for the metastable wedge of α-olivine. These density differences effectively act as a body force and produce a significant stress field, which we calculated with a two-dimensional finite-element code. The results show that the buoyancy-induced forces produce maximum shear stress up to 23 MPa along the metastable wedge and deeper portion just above the depressed last phase transition involving γ-spinel. For the latter, the dominant state of stress is down-dip compression. The calculated P-T dependent state of stress is very similar to the depth distribution of deep-focus

Formation of a metastable olivine wedge in a descending slab

We present a thermal model of a descending slab in which the transformation of olivine to spinel is controlled by pressure‐ and temperature‐dependent reaction kinetics. Two different formulations of the kinetics are considered with the main discriminant being the temperature range over which olivine converts to spinel at pressures of about 15 GPa (about 500°–515°C and 560°–650°C). We use a finite element method to solve the coupled heat conduction (perpendicular and parallel to the dip of the slab) and kinetics equations, and we include the latent heat of the phase transformation. Latent heat release together with heat conduction parallel to the dip of the slab reduces significantly the length of the metastable olivine wedge and results in a very thin (≲5 km) two‐phase region. We employ the thermal parameter υτ sin δ (υ is the velocity of the descending slab, τ is the age of the slab, and δ is the dip of the slab to the horizontal) to interpret the results for the length of the metastable wedge. For values of the thermal parameter smaller than about 4000 and 7000 km, depending on the model of the kinetics, no metastable olivine wedge exists (the critical value of 7000 km for the thermal parameter corresponds to the kinetics model with the lowest transition temperature range). The length of the metastable olivine wedge is also found to be very sensitive to the model of the kinetics and to the effects of adiabatic heating. If the occurrence of deep earthquakes is related to the transformation of metastable olivine to spinel, then data on earthquake depth versus thermal parameter require that the onset of the reaction takes place at temperatures of about 550°–575°C. In this case the slab thermal parameter must be larger than 10,000 km for the metastable olivine wedge to extend down to 660 km depth. But deep earthquakes occur near 660 km depth in slabs with thermal parameter as small as about 5000 km (South America, for example). Either some deep earthquakes are unrelated to olivine metastability or our knowledge of olivine‐spinel reaction kinetics is incomplete.

Transformation shear instability and the seismogenic zone for deep earthquakes

We use a numerical model for olivine‐spinel transformation to study deep earthquake nucleation and to delineate the seismogenic region within a subducting slab. The model includes laboratory‐derived flow laws, latent heat release, and phase transformation kinetics. We calculate deformation, transformation state, grain growth, and rheology for several paths within a subducting slab. Strain rate perturbations are imposed to define the necessary conditions for instability. Strain rate perturbations decay for ξ < a critical value ξc, and thus the coldest, interior portion of the metastable wedge deforms stably. For ξ≥ξc, strain rate perturbations grow, shear strength decreases with strain, and the system is potentially unstable. The instability condition is mapped to delineate the seismogenic zone within a subducting slab. The model seismogenic zone is bounded by ξc, and, at larger percent transformations, by coarsening of spinel grains and saturation of the transformation weakening effect. The model predicts a narrow seismogenic region along the outer edges of the metastable wedge and thus a “double seismic” zone, consistent with some seismic observations. Several simplifying assumptions are required due to lack of thermo‐kinetic data and incomplete knowledge of constituent mineralogy and rheology. However, the model provides a quantitative definition of the instability condition and a framework for testing the hypothesis of transformation‐induced instability.

Two-dimensional thermo-kinetic model for the olivine-spinel phase transition in subducting slabs

We have investigated the effects of the latent-heat release on the kinetics of the olivine-spinel phase transition to clarify the role of the thermo-kinetic coupling process for the structure of the metastable olivine-wedge in subducting slabs. We have laid out the mathematical formulation of a two-dimensional time-dimensional time-dependent model consisting of the kinetic equations, which are cast as a system of four nonlinear ordinary differential equations (ODE) at each spatial grid point and the time-dependent partial differential equation (PDE) for the temperature, which is coupled to the kinetics by virtue of latent-heat release. This set of ODE-PDE system has been solved by the differential-algebraic method. The structure of the kinetic phase boundary is strongly determined by thermo-kinetic coupling effects during the phase transition. For slow, warm slabs a very narrow phase boundary is obtained near the typical depth for equilibrium phase transformations. From laboratory data we obtain a small latent-heat release (< 10 kJ mol −1), which results in a small heating up of the slab (around 50°). Hence thermo-kinetic coupling effects will not significantly influence the structure of the phase boundary in this regime. For fast, cold slabs narrow regions with metastable olivine may be pushed down to a depth of about 600 km while the thermo-kinetic coupling due to the latent-heat release drastically reduces the depth and the width of the region where olivine and spinel coexist in the cold slab interior. Below the metastable wedge the latent-heat results in a significant and localized heating of the cold slab interior (around 150°), because in this regime the heat release is three times higher. The depth of the metastable wedge in the subducting slab is found to be very sensitive to certain thermodynamic parameters such as the activation energy for growth and the internal slab heating caused by the phase transformation. We propose that deep or intermediate earthquakes occur due to a thermal runaway-effect caused by shear instabilities while these effects are enhanced by the latent-heat release associated with the olivine-spinel transformation. The correlation between fast subducting velocity and the concentration of deep-focus earthquakes at around 600 km depth, as shown for the Tonga-Kermadec trench, can be predicted by this 2-D thermo-kinetic model.

The metastable olivine wedge in fast subducting slabs: Constraints from thermo-kinetic coupling

The effects of a strong non-linear thermo-kinetic coupling in the olivine-spinel transition are investigated with a 2D model for fast-subducting slabs. The thermo-kinetic coupling problem is solved in the Lagrangian framework by the differential algebraic method and by using experimental input parameters. Narrow regions with metastable olivine are found in fast and cold slabs down to the 660 km discontinuity, while the thermo-kinetic coupling reduces the depth and width of the zone where olivine and spinel coexist. In contrast to recent kinetic models, both latent heat release and heat diffusion place important constraints on the structure of the metastable wedge. Both processes trigger the ongoing phase transformation in the cold slab interior. A thermal instability over a short timescale (10 4–10 5 yr) can occur due to a positive thermo-kinetic feedback. This effect is enhanced at greater depths. The depth of the metastable wedge is found to be extremely sensitive to certain geophysical and thermodynamic parameters such as the lithospheric basal temperature, slab width and the activation energy for diffusion. Variations in these parameters within the experimental uncertainties can cause significant changes in the depth of the metastable wedge. Hence predictions about the exact position of the metastable olivine wedge are rather difficult. We propose that deep focus earthquakes may occur due to a thermal runaway caused by shear heating, while the thermal instability is enhanced by the latent heat release associated with the olivine-spinel phase transition.