Carbon dioxide emissions from dolomite decarbonation play an essential role in the weakening of carbonate faults by lowering the effective normal stress, which is thermally activated at temperatures above 600–700 °C. However, the mechanochemical effect of low-crystalline ultrafine fault gouge on the decarbonation and slip behavior of dolomite-bearing faults remains unclear. In this study, we obtained a series of artificial dolomite fault gouges with systematically varying particle sizes and dolomite crystallinities using a high-energy ball mill. The laboratory-scale pulverization of dolomite yielded MgO at temperatures below 50 °C, indicating that mechanical decarbonation without significant heating occurred due to the collapse of the crystalline structure, as revealed by X-ray diffraction and solid-state nuclear magnetic resonance results. Furthermore, the onset temperature of thermal decarbonation decreased to ∼400 °C. Numerical modeling reproduced this two-stage decarbonation, where the pore pressure increased due to low-temperature thermal decarbonation, leading to slip weakening on the fault plane even at 400–500 °C; i.e., 200–300 °C lower than previously reported temperatures. Thus, the presence of small amounts of low-crystalline dolomite in a fault plane may lead to a severely reduced shear strength due to thermal decomposition at ∼400 °C with a small slip weakening distance.

An understanding of fault weakening during earthquake slip is essential to comprehend the facilitation of rupture propagation. Various mechanisms, such as flash heating (De Paola et al., 2011b), thermal pressurization (Miller et al., 2004), frictional melting (Di Toro et al., 2006), thermal decomposition (Han et al., 2007; Carpenter et al., 2015), mechanochemical changes (Hirono et al., 2013), and crystal plasticity of nanoparticles (De Paola et al., 2015; Spagnuolo et al., 2015; Violay et al., 2019), have been suggested for the reduction of frictional strength. Carbonate-bearing faults, wherein the thermal decomposition of carbonate and the resultant CO2 emissions simultaneously affect the frictional strength, merit a quantitative investigation based on mass and energy conservation laws related to pore pressure and temperature, respectively.

In carbonate-bearing-faults, CO2 has been recognized as a geochemical signature of earthquake propagation (De Paola et al., 2011a; Rowe et al., 2012; Violay et al., 2013). Moreover, the relicts of decarbonation, such as degassing bubbles, vesicles, and amorphous oxide minerals, have been observed as evidence of significant heating due to coseismic slip (De Paola et al., 2011b; Collettini et al., 2013; Fondriest et al., 2013; Siman-Tov et al., 2013; Smith et al., 2013; Delle Piane et al., 2017; Ohl et al., 2020). While the decarbonation of carbonate is known to be thermally activated at ∼700 °C (Rodriguez-Navarro et al., 2012), the mechanical deformation-induced CO2 emission has also been reported in calcite (Kristóf-Makó and Juhász, 1999).

Mechanical deformation processes, such as pulverization, occur extensively in slip zones (Brantut et al., 2010; Delle Piane et al., 2017; Fondriest et al., 2017), and mechanochemical changes can affect the shear strength of faults (Hirono et al., 2013). In carbonate faults, nanograins or amorphous materials are also observed in fault planes (Siman-Tov et al., 2013; Spagnuolo et al., 2015; Delle Piane et al., 2018; Ohl et al., 2020). However, the mechanochemical effect on decarbonation behavior is not fully understood from an earthquake propagation perspective.

In this study, we investigated pulverized ultrafine dolomite powder, which mimics the properties of natural and artificial fault gouges, using high-energy ball mills and spectroscopy techniques for a systematic exploration of the mechanochemical effects on the strength of dolomite faults. The effects on slip behavior in dolomite faults were then quantified based on a series of thermomechanical numerical simulations.

A commercial dolomite powder (SD200, SungSin, Republic of Korea) was mechanically pulverized using a high-energy ball mill (Emax, Retsch GmbH, Germany). Ultrafine dolomite particles with varying sizes and crystallinities were obtained by grinding powders for up to 480 min. During grinding, thermal heating was suppressed using a water-cooling system. The maximum temperature on the outer surface of the grinding jaw was maintained below 50 °C. The characteristics of ground dolomite were analyzed using laser diffraction particle size analysis, X-ray diffraction (XRD) spectrometry, Brunauer–Emmett–Teller specific surface area analysis, scanning electron microscopy (SEM), 25Mg solid-state nuclear magnetic resonance (NMR) spectroscopy, thermogravimetry (TG), and differential thermal analysis (DTA).

To understand the effects that low-temperature (i.e., ∼400 °C) decarbonation has on the slip behavior of dolomite faults, we performed series of one-dimensional (1-D) finite element modeling (Sulem and Famin, 2009) with a 0.005 m shear zone under a seismic slip of 0.5 m/s. The depth, normal stress, and ambient temperature of the domain, considering mature carbonate faults (Rice, 2006), were 7 km, 180 MPa, and 210 °C, respectively. The phyllosilicate content of carbonate faults declines at depths >7 km (Chen et al., 2015), indicating that the frictional properties of carbonate are the main factors controlling seismic behaviors at this depth.

We tested whether low-crystalline dolomite enhances CO2 emissions and shear strength weakening at low temperatures (i.e., ∼400 °C) compared with high-crystalline dolomite that exhibits decarbonation at high temperatures (i.e., ∼700 °C). A detailed description of the experimental and numerical methods is provided in the Supplemental Material1. In short, the conservation equations of mass and energy for pore pressure (Equation S7 in the Supplemental Material) and temperature (Equation S8) were coupled via a second-order Runge-Kutta time integration (So et al., 2013).

Formation of Ultrafine Particles with Low Crystallinity

The SEM images (Figs. 1A and 1B), laser diffraction particle size (Fig. 1C), and specific surface area analysis (Fig. 1D) show that the microsized dolomite particles (mean size of ∼10 μm) were pulverized into ultrafine powder (mean size of ∼0.2 μm) aggregates after grinding for 120 min. The sharp peak in the XRD pattern (Fig. 1D) gradually diminished due to the loss of crystallinity, especially for the first 120 min of grinding. The height of the primary peak (104) decreased to ∼15% compared with that observed prior to grinding (Fig. 1E), indicating the coexistence of amorphous and crystalline phases in the pulverized dolomite. The similar overall changes in the crystallinity and particle size indicated that particle size reduction and amorphization occurred simultaneously when mechanical energy was applied.

Mechanical Decarbonation of Dolomite

The 25Mg solid-state NMR spectra showed two peaks, corresponding to Mg in dolomite [CaMg(CO3)2] at 8 ppm and periclase (MgO) at 25 ppm (Fig. 2A; Pallister et al., 2009). During grinding, the intensity of the dolomite peak gradually decreased at the expense of the periclase peak up to 120 min, indicating that dolomite partially decomposed during grinding. The loss of crystallinity and its decarbonation were found to be correlated, indicating simultaneous amorphization and decarbonation; i.e., mechanical decarbonation. Although thermal decarbonation may also partly occur due to the high local flash temperature during impact, the temperature of the grinding jaw was maintained below 50 °C during grinding. The observation that the NMR spectra clearly showed a MgO peak, which was not observed in the XRD pattern, indicates that MgO had an amorphous phase. The MgO peak observed in the as-received dolomite would have formed due to grinding of the commercial dolomite by the manufacturer. In addition, CaO and/or Ca(OH)2 can also form via mechanical decarbonation; however, this cannot be proven by the 25Mg NMR spectrum.

The simultaneous occurrence of amorphization and decarbonation indicated that the level of mechanical decarbonization was related to the degree of amorphization. In this study, the formation of MgO via mechanical decarbonation proceeded to ∼30% (Fig. 2B). However, mechanical decarbonation in a fault depends on the magnitude of the mechanical energy applied to the fault zone during slip (Aretusini et al., 2017; Kaneki et al., 2020). Thus, the occurrence of mechanical decarbonization depends on the fault.

Effect of Crystallinity on Thermal Decarbonation

The mechanochemical effects on the thermal decarbonation of dolomite were determined via TG and DTA (Figs. 2C and 2D). Thermal decarbonation can occur due to thermal heating during slip, unless the strain rate is low. Based on the endothermic reaction accompanying a weight loss of ∼47 wt% at ∼780 °C, the occurrence of single-stage thermal decarbonation in dolomite was confirmed. Upon grinding, additional low-temperature decarbonation was observed from 400 °C to 700 °C, which slightly overlapped with the high-temperature decarbonation at 700 °C. Upon grinding, the degree of low-temperature decarbonation increased, whereas that of high-temperature decarbonation decreased, indicating that high-temperature decarbonation shifted to low-temperature decarbonation with the loss of crystallinity.

Low-temperature decarbonation occurred over a wider temperature range (400–700 °C) than high-temperature decarbonation (700–800 °C). These results also indicate that two-stage thermal decarbonation was induced by loss of crystallinity. Therefore, low- and high-temperature decarbonation are related to dolomite with low and high crystallinity, respectively. The physicochemical reactions of low-crystalline materials generally occur within a wider temperature range than those of high-crystalline materials because the binding energy between atoms in disordered structure varies depending on the bond length or angle. The low-temperature decarbonation possibly resulted from the lowered decarbonation temperature of the MgCO3 layer in low-crystalline dolomite, while the decarbonation temperature of the CaCO3 layer rarely changed in low-crystalline dolomite (Kristóf and Juhász, 1993). Further, the large surface area of ultrafine particles may lower the decarbonation temperature.

Effect of Low-Temperature Decarbonation on Fault Weakening

The activation energy (Ea) of high-crystalline dolomite varies significantly between 97 and 333 kJ/mol depending on the particle size and presence of impurities (DeAngelis et al., 2007; Gunasekaran and Anbalagan, 2007). Thus, we determined that two values of Ea, i.e., 320 and 220 kJ/mol, can explain the high- and low-temperature decarbonation, respectively, observed in our laboratory experiments (Fig. S4). Temperature evolution during seismic slip strongly depends on the fraction of low-crystalline dolomite in the shear zone (Fig. 3A). The high-temperature decarbonation at ∼700 °C is clearly reproduced for the case of pure crystalline dolomite (black line). As the fraction of low-crystalline dolomite increases due to pulverization, the fraction of the phase with Ea = 220 kJ/mol also increases by up to 50%, indicating a composition of 100% low-crystalline dolomite. The temperature evolution in the shear zone shows two-stage increases of temperature and pore pressure, which are consistent with the experimental data shown in Figures 2C and 2D. In the first stage, the concentration of the low-crystalline phase does not affect the temperature increase because frictional heating dominates the thermal evolution (time ≤ 0.5 s). With further slip, the system reaches a quasi-equilibrium between frictional heating and the endothermic CO2 emission from low-crystalline dolomite.

Figure 3B shows that the time and temperature at which the pore pressure in the center of the shear zone reaches ∼170 MPa are controlled by the fraction of low-crystalline dolomite. The shear zone with pure crystalline dolomite rapidly achieves a pore pressure of ∼170 MPa in a single stage after ∼2.5 s (black line). In contrast, in the presence of higher contents of low-crystalline dolomite, more time is required to reach the ∼170 MPa pore pressure in two different stages. For instance, at a concentration of 16.6% (gray line), the pore pressure reached ∼170 MPa after 9 s.

Figure 3C displays the relationship between the temperature and shear strength (Equation S9). Crystalline dolomite displays a single strength drop at 650–700 °C, which is indicative of high-temperature fault weakening (black line). When low-crystalline dolomite is included, an additional abrupt drop in the shear strength is observed at 400–450 °C (blue-shaded zone) prior to high-temperature weakening at 650–700 °C (red-shaded zone). These results indicate that the presence of low-crystalline dolomite induces low-temperature thermal decomposition and fault plane weakening at ∼400 °C.

The 1-D nature of our model may have prohibited off-fault fracturing (Okubo et al., 2019) during earthquake slip, inducing a permeability increase around the slip zone and resisting the buildup of pore pressure along the fault plane. We suggest that a two-dimensional model is required to account for dynamic rupture propagation due to lateral heterogeneity in the low-crystalline dolomite, as well as the off-fault fracturing.

Multistep Decarbonation of Pulverized Dolomite and Fault Weakening

Dolomite along a fault plane is pulverized during seismic slip, affecting the occurrence of both mechanical and thermal decarbonation at a temperature lower than the typical thermal decarbonation temperature of dolomite (Fig. 4). The emission of CO2 from mechanical decarbonation results in fault weakening, which is termed fault weakening via “mechanical pressurization” (Hirono et al., 2013). The pressurization induced by thermal decomposition of pulverized dolomite can be categorized into “low-temperature pressurization” and “high-temperature pressurization” depending on the decomposition temperature, i.e., ∼400 °C and ∼700 °C, respectively. The numerical simulation results show that the presence of a small amount of low-crystalline dolomite possibly leads to significant fault weakening due to low-temperature pressurization at temperatures ∼200–300 °C lower than those previously reported. In the stage of high-temperature thermal decarbonation, the grain-size-sensitive flow (De Paola et al., 2015) activated by high temperature and small grain size may act as additional fault weakening mechanisms. The low-crystalline dolomite nanoparticles could be obliterated due to recrystallization in the presence of fluids during the interseismic period.

While the presence of CaO and MgO in carbonate fault gouge has been considered as evidence of flash heating (>∼600 °C) (De Paola et al., 2011b, 2015; Fondriest et al., 2013), the latter can also occur via mechanical decarbonation; thus, the presence of CaO and MgO cannot be definitively regarded as evidence of a high-temperature environment. A previous study (Pittarello et al., 2008) suggested that >97% of the elastic strain energy is dissipated by significant frictional heating >1000 °C. However, under the regime of lower-temperature pressurization (300–500 °C), the ratio of heat dissipation to surface energy might be lower. The vigorous pressurization induced by thermal decomposition at lower temperature may lead to a larger fraction of the energy being available for creation of new surface fractures. While mechanical decarbonation has been reported in calcite (Kristóf-Makó and Juhász, 1999), the mechanochemical effects of lowering the thermal decarbonation temperature of calcite need further investigation.

We thank the editor, C. Clark, and reviewers S. Aretusini, M. Violay, and an anonymous reviewer for their helpful reviews. This study was supported by the National Research Foundation (NRF) of Korea, grant NRF-2019R1F1A1061301 awarded to H. Kim, and grant NRF-2019R1A6A1A03033167 awarded to B. So. H. Kim was also supported by the “Human Resources Program in Energy Technology” from the Korea Institute of Energy Technology Evaluation and Planning, and granted financial resources from the Ministry of Trade, Industry and Energy, Republic of Korea (grant 20194010201730).

1Supplemental Material. Experimental and numerical methods. Please visit to access the supplemental material, and contact with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.