Kinetic hindrance of phase transformations in dense MgSiO3 polymorphs under shock compression Authors: Sota Takagi, Donghoon Kim, Yingwei Fei, Sally June Tracy Published: 2026 ABSTRACT Equilibrium phase diagrams define the pressure-temperature stability fields of MgSiO3 polymorphs, yet whether these phase transformations can proceed under shock compression remains poorly constrained. Here, we present in situ synchrotron x-ray diffraction (XRD) measurements of presynthesized dense MgSiO3 polymorphs—bridgmanite and akimotoite—subjected to laser-driven shock compression. Bridgmanite preserves its perovskite structure up to at least 274 gigapascals, far beyond the equilibrium perovskite-postperovskite boundary, whereas akimotoite retains its ilmenite structure up to 87 to 116 gigapascals and loses crystallinity at 344 to 401 gigapascals. These observations provide direct evidence that neither the perovskite-postperovskite transition from bridgmanite nor the ilmenite-perovskite transition from akimotoite was observed under the shock conditions investigated. Our results indicate that kinetic limitations dominate MgSiO3 transformations on nanosecond timescales, such that equilibrium phase relations cannot be directly applied at extreme shock. Placed in the context of recent in situ XRD studies across a range of silicate minerals, these results suggest an important role for kinetic accessibility relating to crystal topology—particularly the three-dimensional connectivity and stiffness of the polyhedral network—rather than thermodynamic stability alone, in the structural response to shock, with implications for mineral physics and modeling planetary impact processes. INTRODUCTION Phase transitions of silicates under extreme pressure-temperature (P-T) conditions govern the interior structure, dynamics, and evolution of terrestrial planets. Bridgmanite, the perovskite-type (Pv) polymorph of MgSiO3, constitutes the dominant mineral phase in Earth’s lower mantle and exerts primary control over its physical and chemical properties (1, 2). In the lowermost mantle, bridgmanite transforms into the postperovskite (pPv) phase at ∼125 GPa and ∼2500 K, with a positive Clapeyron slope (dP/dT) (3). Akimotoite, the ilmenite-type polymorph of MgSiO3, is stable under transition-zone to uppermost lower-mantle conditions (∼18 to 25 GPa) (4) and may persist metastably in subducting slabs (5) and shock-metamorphosed materials (6). Extensive high-pressure and high-temperature (high P-T) studies using diamond anvil cells and multianvil apparatus have constrained the equations of state (EOS) and phase relations of these dense MgSiO3 polymorphs (7–14). Static compression techniques probe mineral behavior under quasiequilibrium conditions, whereas dynamic compression is essential for understanding material responses during rapid, high-energy processes such as planetary collisions, giant impacts, and hypervelocity meteorite bombardment. Shock experiments on MgSiO3 polymorphs, including enstatite (15–19), glass (15, 17, 20), and presynthesized bridgmanite (20–23), have extended the accessible P-T range to terapascal pressures and constrained EOS, melting relations, and thermodynamic parameters relevant to super-Earth interiors and extreme impact events. Despite these advances, the crystal structures and phase stability of MgSiO3 minerals under shock compression remain poorly constrained. Traditional shock experiments rely on velocimetry or postshock recovery, which provide essential thermodynamic constraints but cannot directly identify crystal structures during shock loading. The absence of direct structural measurements introduces substantial uncertainty in phase identification. For example, in recent pulsed-power shock experiments on MgSiO3 bridgmanite, the solid Hugoniot was interpreted to represent pPv because the inferred Hugoniot P-T conditions exceeded the Pv-pPv boundary (22). However, no structural measurements were obtained, and the density and elastic-property contrasts between Pv and pPv diminish at ultrahigh pressures, making phase identification based solely on bulk properties increasingly uncertain. This limitation underscores the need for in situ x-ray diffraction (XRD) to directly determine crystal structures under dynamic compression. Recent advances in in situ XRD have begun to address this gap. Notably, in situ XRD studies of MgSiO3 and (Mg,Fe)SiO3 glass (24) and enstatite (25) revealed amorphization, melting, or transformation into disordered or metastable structures during shock compression, with no evidence of crystallization into stable high-pressure phases. These observations suggest that silicate materials under dynamic compression may favor kinetically accessible metastable or amorphous states rather than equilibrium crystal structures. Whether transformations predicted by equilibrium phase diagrams can operate during shock compression therefore remains a fundamental open question. More broadly, it is unclear how kinetic constraints and crystal structure influence nonequilibrium transformation pathways under ultrafast compression. Here, we address this problem using in situ XRD measurements of presynthesized bridgmanite and akimotoite subjected to laser-driven shock compression. RESULTS Laser-driven shock experiments combined with in situ XRD and velocimetry diagnostics were conducted at the Dynamic Compression Sector (DCS) of the Advanced Photon Source (APS), Argonne National Laboratory. The experimental configuration is shown in Fig. 1A. Presynthesized bridgmanite (MgSiO3) and akimotoite (MgSiO3 and Mg0.92Fe0.08SiO3) were sandwiched between a polyimide ablator and a LiF window. Synchrotron x-ray pulses probed their crystal structures under shock loading, while simultaneous velocity measurements constrained the shock pressures. Shock pressures were determined from interface velocity histories for shots #477, #45, and #471 (Fig. 1B), while those for the remaining shots were inferred from estimated polyimide pressure with impedance matching (IM) method (see Materials and Methods). The XRD images were collected during compression for all shots except for the 339(25)-GPa bridgmanite shot, for which diffraction was recorded during release. Table 1 summarizes the experimental conditions, estimated pressures, and XRD results for all shots. Experimental configuration and diagnostics for laser-driven shock in situ XRD. (A) Experimental configuration of the laser-shock in situ XRD measurements. UV, ultraviolet. (B) The interface velocity histories for shots #477, #45, and #471. Table 1.Shot conditions and the results of XRD (lattice parameter, unit-cell volume, and density) and estimated pressure.RunSampleLaser conditionsLattice parameters (Å)V (Å3)ρXRD (g/cm3)X-ray timing relative to breakout (ns)Observed phasePeak Psample (GPa)2023-R1 shot #57Bdm63.6 J–250 μm–10 nsa = 4.42(3), b = 4.44(5), c = 6.37(9)125(2)5.3(1)+1.50Released bdm339(25)b 144 (probed)c2023-R1 shot #55Bdm47.7 J–250 μm–10 nsa = 4.16(5), b = 4.38(9), c = 6.1(2)111(5)6.0(2)−0.36Compressed bdm274(21)b2022-R3 shot #477Bdm79.2 J–500 μm–10 nsa = 4.35(2), b = 4.51(4), c = 6.39(8)125(2)5.32(9)−1.30Compressed bdm127(8)2023-R1 shot #45Bdm56.4 J–500 μm–10 nsa = 4.445(5), b = 4.524(9), c = 6.41(2)129(5)5.17(2)−4.13Compressed bdm101(9)BdmAmbienta = 4.795(5), b = 4.914(9), c = 6.90(2)163(6)4.103(1)2023-R1 shot #56Aki79.5 J–250 μm–10 nsQ1 = 2.933(4) Å−1, Q2 = 3.771(4) Å−1—5.90–6.08a—Amorphous344(25)–401(30)bAkiAmbienta = 4.723(1), c = 13.541(7)262.4(2)3.814(2)2022-R3 shot #471Aki (Fe0.08)55.0 J–500 μm–10 nsa = 4.48(2), c = 12.74(9)221.4(3)4.52(5)−0.15Compressed aki87(8)–116(10)Aki (Fe0.08)Ambienta = 4.733(1), c = 13.591(7)263.3(2)3.800(2)aThe density on the akimotoite Hugoniot (pressure-density) curve for a corresponding pressure (344 to 401 GPa) is shown.bThe pressures for shot #55, #57, and #56 were estimated from polyimide pressures.cThe pressure at XRD probe timing, which was estimated from XRD density on release isentrope. Shot conditions and the results of XRD (lattice parameter, unit-cell volume, and density) and estimated pressure. The density on the akimotoite Hugoniot (pressure-density) curve for a corresponding pressure (344 to 401 GPa) is shown. The pressures for shot #55, #57, and #56 were estimated from polyimide pressures. The pressure at XRD probe timing, which was estimated from XRD density on release isentrope. Figure 2 shows one-dimensional (1D) XRD patterns as a function of shock pressure for bridgmanite and akimotoite. Most patterns include contributions from ambient, compressed, and release regions of the sample due to transmission geometry (Fig. 1A) and possible spatial pressure gradients (fig. S1). For bridgmanite, all patterns can be indexed as compressed bridgmanite, together with ambient or released bridgmanite. The compressed peaks consistently appear at lower d-spacings than their ambient counterparts, indicating lattice contraction. Peak positions shift systematically with increasing pressure from 101(9) to 274(21) GPa, consistent with progressive lattice compression. The 339(25)-GPa pattern deviates from this trend because the XRD pattern was acquired during release. Comparisons between ambient and shocked 2D diffraction images (figs. S2 to S4) reveal no detectable pPv reflections: Peaks in the d-spacing range overlapping the expected pPv 022 position (∼2.3 to 2.5 Å) appear at identical pixel positions in the ambient image (figs. S3 and S4), confirming their assignment to uncompressed Pv 112. To quantify this constraint, we estimated the detection limit of pPv in the 1D XRD pattern assuming random grain orientation; under this assumption, the presence of textured pPv cannot be excluded. The minimum detectable pPv volume fraction was estimated from the noise level at the expected pPv 020 position in the 274(21)-GPa shot, yielding a 3σ upper limit of ∼16 vol % (fig. S5). Collectively, these diffraction data indicate that bridgmanite preserves its Pv structure as the dominant phase to at least 274(21) GPa under shock compression, with any pPv abundance below the ∼16 vol % detection limit. Pressure dependence of XRD patterns. Bridgmanite (left) and akimotoite (right). Diffraction peaks from ambient or released states observed in the patterns during shock were marked with an asterisk (*). Minor impurity observed in the ambient state was marked with a black triangle. a.u., arbitrary units. Akimotoite (Mg0.92Fe0.08SiO3) preserves its ilmenite structure at 87 to 116 GPa and shows no evidence for transformation to bridgmanite or pPv. In contrast, at 344 to 401 GPa (MgSiO3), all sharp diffraction peaks disappear and are replaced by two broad diffuse features observed around ∼1.67 and ∼2.14 Å, corresponding to Q-values (Q = 2π/d) of 2.933 and 3.771 Å−1. The broad features coincide with the first and second sharp diffraction peaks of compressed MgSiO3 glass and liquid (fig. S6), consistent with a dense disordered state with predominantly sixfold Si–O coordination (24–26). Lattice parameters and unit-cell volumes under shock compression conditions were extracted by peak fitting of the XRD patterns (Fig. 3). For bridgmanite, both volume and lattice parameters decrease continuously with increasing pressure, reaching 111 Å3 (V/V0 = 0.68) at 274(21) GPa. The effective linear axial compressibilities of the a, b, and c axes are 4.7 × 10−4, 3.7 × 10−4, and 4.1 × 10−4 GPa−1, respectively, which are lower by a factor of ∼1.2 to 1.7 than those reported from static compression experiments below 60 GPa (7). Both the c/a and b/a ratios increase from 1.025 and 1.439 at ambient conditions to 1.053 and 1.466 at 274(21) GPa, following trends observed under static compression (7). For akimotoite, the unit-cell volume decreases to 221.4 Å3 (V/V0 = 0.84) at 87 to 116 GPa. The effective linear axial compressibilities for a and c axes are 5.2 × 10−4 and 6.2 × 10−4 GPa−1, respectively. The c axis is 1.2 times more compressible than the a axis, consistent with static data, although the absolute compressibility values are approximately half of those reported below 25 GPa (27, 28). The c/a ratio decreases modestly from 2.872 to 2.844 at 87 to 116 GPa, a smaller reduction than observed in low-pressure static data. The akimotoite data point lies slightly above the Hugoniot curve in Fig. 3A, consistent with transmission-geometry XRD probing a spatially averaged unit-cell volume, while the reported pressure corresponds to the peak shock-front value. Pressure dependence of the unit-cell volume, lattice parameters, and ratio of lattice parameters under laser-shock conditions. Bridgmanite (left) and akimotoite (right). Black lines in the lattice parameters and their ratio are the trend lines for the data in this study. Figure 4A shows densities as a function of pressure. Densities for shots with crystalline diffraction peaks at peak compressed state—shots #45, #477, and #55 (Bdm); and #471 (Aki)—were derived from XRD unit-cell volumes, whereas those for shot #57 (Bdm), probed during release, and #56 (Aki), which showed no crystalline peaks, were inferred from the Hugoniots. Because of thermal pressure contributions and pressure distribution in the samples, all measurements lie above the 300-K isotherm. The bridgmanite results are consistent with previously reported Hugoniot data (21, 22). The akimotoite data point at 87 to 116 GPa also aligns with an akimotoite Hugoniot calculated from the isentrope using reported thermodynamic parameters (28). Shock compression simultaneously increases pressure and temperature, and both variables influence the structural response of silicates on ultrafast timescales. Figure 4B shows the corresponding pressure-temperature diagram. Hugoniot temperatures were estimated using the Mie-Grüneisen-Debye model with thermodynamic parameters for bridgmanite (13) and akimotoite (28). For bridgmanite, the 101-GPa shot approaches the Pv-pPv transition boundary reported from static high P-T experiments (29). The 127(8)- and 274(21)-GPa shots and the peak conditions of the 339(25)-GPa bridgmanite shot exceed the boundary. The released state at the XRD timing of the 339(25)-GPa shot is plotted near the Pv-pPv coexistence region. Despite these conditions exceeding the Pv-pPv equilibrium boundary, the XRD results demonstrate that the Pv structure is dominant in shocked bridgmanite. For akimotoite, the conditions of the 87- to 116-GPa shot, at which compressed akimotoite is observed in XRD, lie far beyond the akimotoite-bridgmanite phase boundary reported from static high P-T experiments (3, 30, 31). The conditions of the 344- to 401-GPa akimotoite shot, where XRD showed loss of crystallinity, slightly exceed the lowest estimates of the MgSiO3 melting curve (16). The persistence of the crystal structures of both bridgmanite and akimotoite in the solid region demonstrates that phase transformations are kinetically inhibited under nanosecond shock loading. Pressure-density diagram and pressure-temperature diagram. (A) Pressure-density diagram for bridgmanite and akimotoite with the Hugoniot data (21, 22). The Hugoniot and isotherm curves were drawn with the parameters from the literature (22, 27, 28, 42). The densities of the bridgmanite 339(25)-GPa shot and akimotoite 344- to 401-GPa shot were determined from the Hugoniot because the 339(25)-GPa shot was probed during release and the 344- to 401-GPa shot showed no crystalline diffraction peaks, so neither yielded a peak-state unit-cell volume. (B) Pressure-temperature diagram with melting curve (16, 22) and Pv-pPv phase transition boundary (29). DISCUSSION The absence of the Pv-pPv transition under shock compression highlights a fundamental difference in transformation behavior between static and shock loading. Under static high-pressure and high-temperature conditions, the Pv-pPv transition may proceed on timescales of 101 to 102 s, and quantitative kinetic constraints have been established by Langrand et al. (32), who modeled the transformation using Avrami kinetics and reported exponents of n ≈ 1.5. These values are inconsistent with a purely martensitic shear transformation and instead suggest a reconstructive transformation character, although the mechanistic details remain debated (32–35). From the structural perspective, the Pv-pPv transition requires cooperative rotations and rearrangements of SiO6 octahedra to form the layered pPv structure (34, 35), involving a substantial kinetic barrier associated with large-scale structural reorganization, rather than a simple lattice shear (35). Shock compression, by contrast, operates on shorter timescales, raising the question of whether kinetic barriers established under static conditions have any relevance to the dynamic regime. In Fig. 5, we plot our shock-compression data alongside characteristic transformation time, τ (in seconds), extrapolated from the static kinetic parameters of Langrand et al. (32), to evaluate whether kinetic inhibition is consistent with the observed absence of pPv. Characteristic transformation times were estimated using activation parameters from Langrand et al. (32): Q0 = 437(186) kJ/mol and V* = 16.1(6.3) cm3/mol for the shear-transformation model; Q0 = 426(180) kJ/mol and V* = 14.4(6.4) cm3/mol for the nucleation-and-growth model. The prefactor lnk was fixed to its mean value, as variations in lnk primarily shift the absolute timescale without altering the pressure-temperature dependence of the transformation kinetics. For each model, uncertainty contours span a low-barrier case (green dashed: Q0 − 1σ, V* + 1σ) and a high-barrier case (blue dashed: Q0 + 1σ, V* − 1σ). This comparison is qualitative rather than quantitative: The many-orders-of-magnitude extrapolation from static experimental timescales (101 to 102 s) to nanosecond shock conditions (∼1 to 10 ns) precludes a precise prediction. As shown in Fig. 5, at ∼127 GPa, the calculated transformation timescales exceed the nanosecond (10−9 s) duration of shock compression for all considered parameter sets, consistent with the absence of pPv. For the low-barrier and mean parameter combinations, however, both models consistently predict that the Pv → pPv transition should be kinetically accessible within the shock duration for the 274-GPa shot. Only the most extreme high-barrier case places the 274-GPa data point near the boundary of the accessible regime, and even this is marginal given the inherent uncertainty in the shock duration estimate. The absence of detectable pPv (<16 vol %; fig. S5) is therefore inconsistent with this prediction, indicating a fundamental mechanistic difference between shock and static loading that cannot be attributed to parameter uncertainty alone. In addition, fundamental differences between static and shock compression—including nonhydrostatic stress states, extreme heating rates, and high strain rates—may suppress the cooperative rotations and rearrangements of SiO6 octahedra required for the Pv-pPv transition. While the microscopic origin of this kinetic inhibition remains unresolved, our results demonstrate that transformation kinetics inferred from static compression experiments cannot be directly extrapolated to nanosecond shock conditions. Characteristic transformation times, τ (in seconds), for the Pv-pPv transition as a function of pressure and temperature (P-T), extrapolated from static-experiment kinetic parameters. (A) Shear-transformation model (mean activation energy Q0 = 437 kJ/mol, activation volume V* = 16.1 cm3/mol, and lnk = −17.7). (B) Nucleation-and-growth model (mean activation energy Q0 = 426 kJ/mol, activation volume V* = 14.4 cm3/mol, and lnk = −17.5). Background colors show log10(τ) for the mean parameters. Dashed contour lines mark log10(τ) = −9, −8, −7, and −6 for the low-barrier (green: Q0 − 1σ, V* + 1σ) and high-barrier (blue: Q0 + 1σ, V* − 1σ) end-members. The solid black curve is the equilibrium Pv-pPv phase boundary. Symbols with error bars indicate estimated P-T conditions for three bridgmanite shots from this study. Static-experiment kinetic parameters (32). For the 339(25)-GPa shot, diffraction was obtained during release at ∼144 GPa, corresponding to conditions near the Pv-pPv boundary. No structural information at peak compression is available for this shot. Two scenarios are conceivable: Bridgmanite may have remained metastably in the Pv structure throughout the event, or pPv may have transiently formed during peak compression. If pPv formed at peak compression, it either reverted to Pv before the XRD frame or persisted below the detection limit of ∼16 vol % at the probe time (fig. S5). The available data do not permit distinction between these two scenarios for the 339(25)-GPa shot. The τ contour lines in Fig. 5 qualitatively illustrate how kinetic accessibility may vary with shock duration across the range relevant to natural planetary impact. Extrapolating the shear-transformation model along the bridgmanite Hugoniot, the expected Pv-pPv transition threshold decreases from ∼143 to >287 GPa at nanosecond timescales to ∼135 to 237 GPa at microsecond (log10τ = −6) and ∼118 to 184 GPa at second (log10τ = 0), spanning fast-to-slow kinetic parameter sets. The second-timescale estimates, approaching the model calibration range, carry relatively smaller uncertainty than nanosecond extrapolations. These results suggest that the kinetic inhibition documented here at nanosecond timescales may be reduced at the microsecond-to-second durations characteristic of natural planetary impacts, although direct experimental validation at intermediate timescales remains lacking. We further note that these conclusions pertain specifically to shock loading; quasi-isentropic compression experiments, which access comparable pressures along lower-temperature paths, may reveal different structural outcomes even at nanosecond timescales. The kinetic suppression of phase transitions in akimotoite and bridgmanite observed at nanosecond timescales provides context for understanding mineral behavior during planetary impacts, although direct application to natural impact conditions requires explicit consideration of the timescale difference. Asteroidal and planetary collisions in the solar system typically occur at velocities of ∼5 to 20 km/s (36, 37). For impacts at these velocities, the duration of peak shock loading scales with impactor size, ranging from microseconds (decimeter-scale fragments) to seconds (planet-scale events)—many orders of magnitude longer than our ∼10-ns laser-shock experiments. Under these conditions, silicate materials can experience peak shock pressures of hundreds of gigapascals. Numerical simulations of such impacts generally use EOS models to describe pressure-density-energy relations, and progress has been made in refining these models (38, 39). These approaches usually assume structural phase transitions according to equilibrium phase diagrams (40). Our results suggest that equilibrium phase relations may not always be realized under ultrashort dynamic compression. When a phase transition is assumed to occur during shock compression, both the energetic cost associated with the transition and the difference in physical properties between low- and high-pressure phases can influence the shock temperature evolution. For example, molecular dynamics simulations estimate that the enthalpy change associated with the Pv-pPv transition is on the order of ∼7.5 kJ/mol at 125 GPa and 2000 K (35). Hugoniot temperature calculations using parameters for bridgmanite (13) and pPv (22) yield a temperature difference of ∼240 K at ∼500 GPa (Fig. 4B). Bridgmanite remains stable as a solid phase up to pressures of ∼500 GPa (22). Therefore, in the pressure regime relevant to many impact scenarios, shock compression occurs predominantly within the solid state. Under such conditions, whether or not structural phase transitions are realized may appreciably affect shock-induced temperature increases. The extent to which the nanosecond-scale kinetic inhibition documented here persists at microsecond-to-second impact timescales remains an open question; as illustrated in Fig. 5 and discussed above, the kinetic model suggests that the timescale gap may reduce the effective kinetic barrier in natural impact events. Incorporating structure- and timescale-dependent kinetic models into planetary impact simulations therefore represents a potentially important direction for future work. At 344 to 401 GPa, the shock conditions of the MgSiO3 akimotoite shot exceed the lower bound of the MgSiO3 melting curve (16) while remaining below the higher bound (22). The disappearance of all sharp diffraction peaks and the appearance of broad diffuse scattering can therefore be interpreted as either solid-state amorphization or shock melting, and neither possibility can be definitively excluded from the present data. We note that the two akimotoite shots in this study differ in composition (Mg0.92Fe0.08SiO3 at 87 to 116 GPa; MgSiO3 at 344 to 401 GPa); accordingly, no composition-dependent behavior is claimed in the following discussion. Disentangling the effects of composition and pressure would require additional experiments using controlled compositions across comparable pressure ranges, which is beyond the scope of the present study. At nanosecond timescales, the suppression of cooperative atomic rearrangements implies that the response of silicate minerals to shock compression depends not only on thermodynamic stability but also on crystal-structure–dependent deformation pathways. Bridgmanite exhibits lattice compression trends similar to those observed in static experiments (7–14), indicating that high strain rates (>108 s−1) do not substantially alter its elastic behavior and that shock heating may have relaxed nonhydrostatic stresses. In contrast, akimotoite displays departures from static compression behavior in its c/a ratio (28), with a smaller pressure-induced change than observed under static conditions, similar to behavior reported for shock-compressed stishovite (41). These contrasting responses likely reflect differences in crystal structure: The 3D, corner-sharing SiO6 network of bridgmanite accommodates compression more uniformly than the layered topology of ilmenite-structured akimotoite, thereby reducing the propensity for localized structural collapse under shock loading. The retention of crystalline structure in bridgmanite to at least 274(21) GPa is directly paralleled by the behavior of stishovite (SiO2) under shock compression. Schoelmerich et al. (41) reported no transition to thermodynamically stable or metastable high-pressure phases in stishovite shocked to 336 GPa (41). Bridgmanite and stishovite share a common structural feature—3D, corner-sharing SiO6 octahedral networks—and both have high bulk moduli [e.g., ∼255 GPa (42) and ∼307 GPa (41), respectively]. That study suggested that this pronounced kinetic hindrance, even for the displacive stishovite-to-CaCl2 transition, reflects the high initial density and low compressibility of stishovite, as well as nonhydrostatic shock conditions. More broadly, 3D corner-sharing SiO6 octahedral networks can distribute stress through cooperative polyhedral distortions, allowing the crystal structure to be preserved even when reconstructive transformations are kinetically inhibited. Akimotoite retains its ilmenite structure at 87 to 116 GPa—far above its equilibrium boundary with bridgmanite at ∼20 to 30 GPa (30)—not through framework rigidity but through the sluggish kinetics of the ilmenite-to-bridgmanite transition (4, 30). In contrast to bridgmanite’s 3D corner-sharing SiO6 network, akimotoite adopts a layered structure composed of edge-sharing SiO6 octahedra arranged in the ab plane, which lacks equivalent pathways for stress redistribution; if the loss of crystallinity at 344 to 401 GPa reflects solid-state amorphization, this structural distinction may account for the susceptibility of akimotoite to collapse under extreme shock conditions. Kinetic hindrance to equilibrium transformation is, in fact, a general feature of silicates under nanosecond-timescale shock, and pressure-induced amorphization in silicates is commonly interpreted as a metastable collapse of the crystal structure when phase transformation pathways are kinetically inhibited (43–45). SiO2 polymorphs with SiO4 tetrahedra—quartz and coesite—transform to a metastable d-NiAs–type structure rather than their equilibrium high-pressure structures (46, 47). Mg silicates with chain and isolated SiO4 tetrahedra—forsterite and enstatite—pass through transient metastable crystalline states before collapsing into disordered states (25, 48). Open framework silicates such as feldspars undergo direct shock-induced amorphization (49–51). Across a wide range of silicate structures—from SiO2 polymorphs and chain Mg silicates to open-framework silicates—no phase attains its equilibrium high-pressure structure, and most lose long-range order. In marked contrast, bridgmanite and stishovite, both comprising dense 3D networks of corner-sharing SiO6 octahedra with high bulk moduli, retain crystallinity to extreme pressures. These observations suggest that crystal structure—particularly the 3D connectivity and stiffness of the polyhedral network— plays an important role in the structural response of silicates to shock rather than strain rate or deviatoric stress alone. In summary, in situ XRD measurements demonstrate that phase transformations in MgSiO3 are strongly suppressed under nanosecond shock compression. Bridgmanite remains crystalline and does not undergo the Pv-pPv transition to at least 274(21) GPa, indicating that kinetic barriers can overwhelm equilibrium thermodynamic driving forces at ultrafast timescales. Akimotoite retains its ilmenite structure to 87 to 116 GPa; at 344 to 401 GPa, akimotoite loses crystallinity under conditions consistent with either solid-state amorphization or melting. The bridgmanite results, together with the retention of structure in stishovite to 336 GPa (41), suggest that dense silicates built on 3D polyhedral networks are strongly resistant to shock-induced structural transformation under nanosecond-timescale kinetic constraints, whereas silicates with less-connected structures are more susceptible. Consequently, these results indicate that ultrafast compression can fundamentally alter transformation pathways in silicate minerals, providing refined constraints for modeling planetary impact processes. MATERIALS AND METHODS Target preparation Bridgmanite samples used in this study were obtained from materials synthesized in previous shock experiments (22), where full details of the synthesis procedure are described. The initial density, measured using the Archimedes method, was 4.10(4) g/cm3. Unit-cell parameters determined by XRD were a = 4.795(5) Å, b = 4.914(9) Å, and c = 6.90(2) Å, corresponding to a unit-cell volume of 163(6) Å3 and a crystallographic density of 4.10(1) g/cm3. Akimotoite samples were produced as a secondary phase during bridgmanite synthesis, forming in regions of the sample assembly that experienced lower temperatures due to pressure-temperature gradients near the capsule edge. Two compositions, MgSiO3 and (Mg0.92Fe0.08)SiO3, were recovered. Orthopyroxene was first synthesized from powder mixtures of MgO (Alfa Aesar, purity 99.995%), SiO2 (Alfa Aesar, purity 99.99%), and FeO (Sigma-Aldrich, purity 99.6%) at 1.5 GPa and 1100°C for 48 hours using a piston-cylinder apparatus. The sintered orthopyroxene was subsequently transformed at ∼25 GPa and 1700°C in a multianvil press. Akimotoite-dominant regions were isolated and used for laser-shock experiments. XRD yielded unit-cell parameters of a = 4.733(1) Å and c = 13.591(7) Å, corresponding to a unit-cell volume of 263.3(2) Å3 and a crystallographic density of 3.800(2) g/cm3. Both bridgmanite and akimotoite samples were double polished to thicknesses of ∼35 to 85 μm. Each sample was glued on one side to a ∼55- or 85-μm-thick polyimide, which served as the laser ablator, and on the opposite side to a ∼500-μm-thick single-crystal (100) LiF window for velocity measurements. Laser-shock in situ XRD measurements Laser-shock experiments were conducted at the DCS of the APS, Argonne National Laboratory. The in situ XRD and velocimetry system at DCS was used for these measurements (52). Shock compression was generated using a 100-J ultraviolet Nd:glass laser with a wavelength of 351 nm (53). The laser pulse duration was ∼10 ns with spot diameters of 250 or 500 μm and pulse energies of ∼30 to 80 J. Synchrotron x-rays with a peak energy of 23.6 keV and a bandwidth of ∼1.5%, produced using the first harmonic of the U17 undulator and a multilayer monochromator, were used to probe the shocked samples. Transmitted diffraction patterns were recorded using a Rayonix SX165 detector, with diffraction from the LiF window masked during analysis. 2D diffraction images were azimuthally integrated to obtain 1D XRD profiles using the pyFAI package (54). Interface velocities between the sample and LiF window were measured using a line-imaging velocity interferometer system for any reflector (VISAR), operated with velocity-per-fringe constant of 1.625 and 0.913 km/s for two channels. The angle between the drive laser and x-ray beam was 45°, and the target normal was tilted by 38° relative to the x-ray direction. Single pulses of the drive laser, synchrotron x-rays, and VISAR probe laser were synchronized and delivered to each target. Lattice parameters were calculated using the UnitCellJ software (55) by fitting reflections 110, 112, 220, and 312 for bridgmanite and 10−2, 104, 113, 11−6, and 12−4 for akimotoite, which are free from overlap with reflections from the ambient and released states. Shock pressure estimation For shots with adequate VISAR fringes, shock pressures were extracted by IM at the sample-LiF interface using the measured interface velocity, which is the most direct method. For shots lacking sufficient fringe contrast, pressures were estimated by laser-intensity–calibrated IM: The polyimide shock pressure (Ppolyimide) was first estimated from the measured laser intensity (IL) using the relation Ppolyimide = 3188(188) × [IL (PW/cm2)/0.351]0.804 (see the Supplementary Materials) (14), and IM at the polyimide-sample interface was then applied. Shock wave transit-time analysis was performed as a supplemental cross-check; details are given in the Supplementary Materials. For bridgmanite shots #45 and #477, interface velocities from VISAR fringe shifts were used with the bridgmanite Hugoniot, Us = 7.922(139) + 1.249(15)up (22), and the LiF Hugoniot, Us = 5.201(25) + 1.323(9)up (56), yielding 101(9) GPa and 127(8) GPa, respectively. Here, Us and up are the shock wave velocity and the particle velocity, respectively. For bridgmanite shots #55 and #57, VISAR fringe contrast was insufficient; pressures were estimated by IM with Ppolyimide using the bridgmanite Hugoniot, yielding 274(21) GPa and 339(25) GPa, respectively. For akimotoite shot #471, VISAR fringe shifts were available. Because Hugoniot data for akimotoite are lacking, IM was performed using the polyimide Hugoniot, Us = 1.79(11) + 1.43(2)up (57), with akimotoite’s shock impedance bracketed by enstatite (ρ0 = 3.219 g/cm3) (25) and bridgmanite (ρ0 = 4.10 g/cm3) (22), which bound the density of akimotoite (ρ0 = 3.80 g/cm3), yielding 87 to 116 GPa. For akimotoite shot #56, VISAR data were insufficient. Ppolyimide was therefore estimated as above, and the same enstatite-bridgmanite bracketing was applied, yielding 344 to 401 GPa.