Spontaneous complexity in the dynamics of slow laboratory earthquakes

Spontaneous complexity in the dynamics of slow laboratory earthquakes | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Spontaneous complexity in the dynamics of slow laboratory earthquakes Giacomo Pozzi, Giuseppe Volpe, Jacopo Taddeucci, Massimo Cocco, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6097579/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 07 Oct, 2025 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Abstract Rock deformation experiments play a key role in our understanding of earthquake physics and friction constitutive laws. These laws commonly describe the response of analogue laboratory faults as a simple and homogeneous system, without accounting for the spatial-temporal evolution of structures in the sample. However, increasing experimental evidence suggests that slip instability is closely tied to heterogeneity, complex rheologies, and inhomogeneous boundary conditions. To address this, we designed a novel transparent setup to observe real-time deformation, track the spatial-temporal evolution of shear fabric, and document unstable slip in experimental faults. Our video documentation reveals that the progressive development of fault fabrics results in heterogeneous but not random stress redistribution. We show that stress and structural heterogeneities play a key role in the nucleation, propagation, and arrest of slip instabilities, raising questions about the robustness of scaling laboratory frictional laws to nature. Earth and environmental sciences/Solid Earth sciences/Seismology Earth and environmental sciences/Solid Earth sciences/Tectonics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 1. Introduction Faults accommodate deformation through a spectrum of slip behaviours ranging from earthquakes to aseismic slip and creep. This variety includes the newly discovered phenomena of low frequency and slow earthquakes 1 , whose underlying mechanisms are still debated 2–4 . In this context, laboratory experiments have played a key role in interpreting the inferred fault slip behaviours in terms of dynamic frictional instabilities 5–8 . The dependence of friction on sliding velocity, for example, laid the foundation for the rate and state theory 6–8 (RSF), which provides a widely employed physical framework in earthquake dynamics. The fine tuning of constitutive RSF parameters in numerical simulations and laboratory experiments reproduces the full spectrum of slip behaviour 9–11 and even determines whether unstable events repeat regularly in time or evolve into a deterministic chaos 12 . These results are obtained without considering complex fault dynamics, which implies that second-order variations of friction can alone lead to strong bifurcations of the dynamic system. In fact, experimental results are commonly portrayed with the relatively simple dynamics of a single degree of freedom spring-slider. This interpretative framework built on the frictional properties of rocks relies on the response of the whole experimental sample, treated as homogeneous, without distinguishing processes occurring at smaller scales. However, several studies show that even minor spatial heterogeneities can have a strong impact on bulk behaviour. Heterogeneities may be found in the form of fault roughness 13–18 , frictional parameters 19–21 , lithology 22–27 , and shear fabrics 28,29 . Fabrics are of particular interest since these characteristic structures found in experimentally deformed samples are also found in natural faults 30–34 . A growing body of evidence suggests that fault zone structures and shear fabrics are central to earthquake dynamics, yet they remain poorly understood even in the laboratory, hidden within sample assemblies and unveiled only after their entire deformation history is completed. Here, we report on novel deformation experiments to investigate in real time the spontaneous evolution of fault fabrics and their effect on slow laboratory earthquakes. A transparent assembly and a high-speed optical camera was used to capture the evolution of the experimental fault zone in tandem with its macroscopic mechanical behaviour. We report on a large variety of rupture patterns, suggesting a complexity of dynamic processes in space and time modulated by fault fabric and applied loading rates. 2. A direct view on fault processes We performed rock deformation experiments on the biaxial apparatus BRAVA 35 using a modified double-direct shear (DDS) configuration, where two simulated fault zones are deformed under controlled conditions (Fig. 1 a). In all experiments we maintained a constant normal load of 25 kN (nominal normal stress of 10 MPa) but varied the forcing velocity of the central slider, here called loading point velocity (LPV). Details on the setup and experimental procedure are included in the Methods section. Differently from a common DDS setup, the central slider (Fig. 1 a) is not made in metal but by transparent Plexiglas (polymethyl methacrylate). This assembly allows recording the evolution of the mechanical properties of the material (Fig. 1 b) while simultaneously visualising fault zone deformation (Fig. 1 c). We studied fault zones composed of montmorillonite clay powder because: i) it is a relevant rock-forming mineral typically found in subduction zones associated with shallow slow slip and tectonic fault tremor 36,37 ; ii) it is frictionally unstable at laboratory conditions comparable to the natural settings 38 ; and iii) it has adhesive properties 39 that ensure strong coupling at the layer boundary and thus shear deformation internal to the fault zone. On one side of the assembly, a high speed camera films the light of a uniform light source reflected by the sample-block interface (Fig. 1 a), hereafter called for simplicity “fault surface”. We use the reflectivity as a proxy for the fault zone stress as well as a marker of the structures developing in the sample layer 38 . This is possible thanks to the phenomenon of frustrated total internal reflection, the same that makes a fingerprint appear when pressing a thumb against polished glass. Here, the quantity of light reflected is inversely proportional to the number of clay platelets pressed against the fault surface. Therefore, the luminosity of the fault surface is function of the poroelastic state of the fault zone, where porosity is function of the state of stress 40 . Tests to calibrate the relation between luminosity and fault zone stress state yield a yet qualitative but strong, inverse and nonlinear proportionality 38 (Supplementary Section SM1). In the following sections we first document the development of fault structures and then investigate how unstable slow slip starts and propagates along the fault zone. 3. Evolution of fault zone structures Our experiments provide remarkable videos on the development of fault zone structures with shear displacement. For this documentation, we shear the initially homogeneous layer of montmorillonite powder applying a constant normal load of 25 kN and a constant LPV = 10 µm/s to the central slider until the macroscopic friction (apparent friction) achieves a constant value (Fig. 1 b). This practice, so-called run-in, is common in rock-deformation experiments and is used to achieve steady state shear strength, i.e., when the fault is considered at macroscopic mechanical equilibrium with the imposed boundary conditions. All of our experiments exhibit consistent behaviour during the run-in, developing a specific structure that forms in a sequential pattern (Fig. 1 c). During shear loading to peak friction conditions (Fig. 1 b, frames I-III), there is a progressive increase of stress (darkening) from the top to the bottom of the fault prior to formation of fault shear fabric structures (Fig. 1 c, frames I-III). Only after peak friction, a first structure forms with a local decrease of stress (increase of luminosity, Fig. 1 c, white arrow). This feature corresponds to the termination of oblique zones of localized shear, called Riedel 41 shear bands (R-shears). They are arcuate in the XY plane (Fig. 1 c), with convexity oriented towards the loading point. Progressively, new shear bands form with a rather regular spacing (~ 5 mm, Fig. 1 c III-VIII), populating the experimental fault in the direction of the loading point (positive Y direction). Towards the end of the experiment, the luminosity of the fault is no longer homogeneous but shows persistent local stress minima marking the terminations of the R-shears (Fig. 2 a). A remarkable observation is that, after 250 s of shearing (displacement of 2.5 mm), friction has already achieved steady-state conditions (Fig. 1 b), but the shear fabric continues to evolve with progressive population of the fault zone with R-shears (frames VI to VIII). The geometry and sequential development of the shear fabric elements is clear in both the real-time images (Fig. 1 ) and the deformed samples recovered after the experiments (Fig. 2 b and Supplementary Fig. S2). The R-shears form at ~ 30° to the shear plane and cut the entire sample layer from one boundary to the other (Fig. 2 b). They dip roughly in the same direction of slip, representing shear localization features that produce “normal” kinematic motion (Fig. 2 b). Our images also show a subsidiary set of R-shears that are rarely identified in previous studies. These features form at the lateral edges of the fault zone (Figs. 1 and 2 c), joining the first R-shear set at sharp angle (Fig. 1 c, black arrow). Formation of this set is accompanied by the lateral extrusion of material from the assembly. The R-shears bound low-strain regions referred to as “micro-lithons” 42 (ML, Fig. 2 ) and they correspond to so-called spectator regions. In XY view, they are characterised by an upper (closer to the loading point) bright and a lower dark area (Fig. 1 c) suggesting a gradient in stress distribution between the R-shears (Fig. 1 c and 2 c). In addition to R-shears, two boundary shear zones form as a consequence of shear localisation: one in proximity of the central slider (upper B-shear, B U ) and one in proximity of the lateral forcing block (lower B-shear, B L ). These present a well-developed S-C fabric 43 (close-up of Fig., 2b) and represent the high-strain zones accommodating shear motion. An important feature is that strain is partitioned between both B-shears after their development, with B L being mostly active close to the older R-shears and B U close to the younger R-shears. This observation and further information on the development of the fault structures are detailed in Supplementary Section SM3. 4. Slip instabilities We now explore the influence of shear fabric on the genesis (nucleation), propagation, and arrest of slip instabilities. In this set of experiments, we repeat the procedure described above and, after the run-in, we vary the LPV between 0.01 and 10 µm/s and observe unstable slip behaviour occurring in stick slip cycles. These slow laboratory earthquake events occur only for LPVs below 7.5 µm/s, finding agreement with previous works on slip stability and forcing velocity dependence 38,44 . The cycles are composed of a “stick” stage, characterised by a nearly linear increase of friction in time, and a “slow slip” stage, characterised by spontaneous slip accompanied by a quasi-dynamic drop in friction (Fig. 3 a). During slow slip events, the luminosity changes only subtly across the fault surface (variations < 1% of the sensitivity range). In fact, the position of the main fault structures as well as the distribution of stress (luminosity) do not vary substantially throughout the stick-slip series (see Supplementary Section SM3). Nevertheless, the slight luminosity change recorded during the friction drops shows negative correlation with the friction (balloon in Fig. 3 a) and clearly illuminates the passage of a rupture front across the fault. Hereby we define “rupture” as the phenomenological description of the spatial-temporal variation of mechanical attributes (e.g., stress and luminosity) in response of the propagation of an instability. A quick visualisation of the ruptures nucleating and propagating along the shear direction (Y) is shown in Fig. 3 c, which depicts the stressing rates (time derivative of luminosity) as a function of time (Fig. 3 c). We can immediately see that a plethora of slip phenomena emerges, modulated by the LPV. Nearly periodic stick-slip is observed at LPV < 2.5 µm/s with a transition to more complex, aperiodic events, and eventually aseismic slip for higher LPV. All of the slow slip events present some differences in rupture patterns (Supplementary Section SM4), but two main groups with specific characteristics emerge (Fig. 3 a). The first, “regular” slow slip events, is observed mainly at LPVs ranging between 0.25 and 2.5 µm/s. The second, “irregular” slow slip events, is observed at LPV < 0.25 µm/s. These two categories are of particular interest and are described in detail in the following sections. At higher LPVs, the slip behaviour quickly transits to chaotic stick-slip. Chaotic events are irregularly spaced in time, present variable friction drops, overlap in time, and are characterised by complicated rupture patterns (compare regular stick-slip with chaotic stick slip in Fig. 3 c). Together with each major rupture propagating in the direction of shear (dark blue traces), numerous smaller events nucleate at other locations and interfere with each other (light red-blue patterns), causing a chaotic stressing rate across the experimental fault (Fig. 3 c). The clear fluctuations of friction visible during regular ruptures (t 10 µm/s chaotic stick-slip quenches to stable sliding. Surprisingly, despite the measured constant friction, luminosity changes reveal that ruptures still continue to nucleate and propagate in the fault zone (Fig. 3 c). These ruptures present small stressing rates, (lighter colours, Fig. 3 c), random propagation patterns, and irregular timing. These features, occurring in concert, masks the existence of instabilities from the mechanical time series to produce what macroscopically appears as aseismic shear motion. 4.1. Regular slow slip Regular slow slip events starts at the boundary of the uppermost micro-lithon (Fig. 4 c,d), close to the subsidiary R-shear next to R9 (circle in Fig. 4 g), where stress heterogeneities are larger. From this region, the rupture front propagates in the X (Fig. 4 f) and Y (Fig. 4 e) directions. Ruptures propagate as wrinkle-like pulses 38 with dimension smaller than the separation between the Riedel shears (< 5 mm, Fig. 4 e,f). The width of the wrinkle appears to be proportional to the propagation velocity of the pulse 45 . We observed a range of rupture speeds depending on propagation direction and shear fabric. For instance, in Fig. 4 b the rupture migrates at ~ 110 mm/s in the horizontal direction (Fig. 4 f), remaining mostly confined within the uppermost micro-lithon (Figs. 5 a, frames I-VII, and 5c), above R9. Most of the macroscopic friction drop occurs during this stage (from black to white arrow, Fig. 4 b). Note that the slow rupture velocity (7 mm/s), reported in Fig. 4 d up to this point, is only apparent because the averaging of apparent stress change in the X direction does not capture motion in the Y direction. When the rupture encounters the rightmost boundary (Figs. 4 f, white arrow, and 5a, frame VII), it overcomes the bounding Riedel and propagates in the shear direction, relieving stress across the entire fault surface. In the meantime, additional minor ruptures propagate within the uppermost micro-lithon (Fig. 4 f, blue stripes past the white arrow), contributing to a further local stress drop. Rupture propagation is faster in the shear (Y) direction (v ~ 225 mm/s) than along the X-direction. Our measurements show a range of interesting spatiotemporal complexity including momentary phases of arrest and deceleration at each local stress maximum within the micro-lithons (Fig. 4 e, dashed lines). The final rupture arrest occurs when the front approaches the lowermost boundary of the sample. All of these so-called regular rupture events propagate across the entire fault surface. 4.2. Complex ruptures and irregular slow slip Irregular slow slip events occur at LPVs = 100 nm/s. These differ from regular slow slip events by one or more features such as: two-stepped friction drops, partial ruptures (confined ruptures), and/or complex propagation patterns. In three consecutive events occurring at low LPV (100 nm/s, Fig. 6 a,c), the first two nucleate at the topmost micro-lithon and do not rupture the entire fault (Fig. 6 c), leading to an increase of stress in the unruptured lower half of the sample. This region is then ruptured by a third event nucleating inside of it (Fig. 6 d, black arrow). Unlike regular slow slip events, the rupture slowly expands (~ 14 mm/s, Fig. 6 f) from the nucleation site (Fig. 7 , frames I-IV) and then accelerates to a much faster propagation speed. We observe that the nucleation stage is composed of a small rupture that induces the nucleation of another one in the neighbouring micro-lithon across R3 (Fig. 7 , white arrow, frames V-VI). Notably, the event behaves as a wrinkle-like rupture even during the nucleation phase since the stressing rate is restricted to a thin region close to the rupture front (Fig. 6 e,f). After the nucleation stage, the rupture expands at faster rates (> 50 mm/s, Fig. 6 e,f) towards the sides of the fault and reaches the topmost micro-lithon (Fig. 7 , frames VI-IX), then rapidly propagates across it from left to right (frames IX-X) and continues to the bottom left (frames X-XII). This rupture shows a clockwise rupture pattern (Fig. 7 d), with faster propagation at the boundaries of the fault zone. Again, arrest coincides with a decrease of rupture front speed (Fig. 6 d). The rupture event clearly correlates with the first half of the slip event, which is modulated in two steps of roughly equal friction drop (starting at the white and orange arrows in Fig. 6 b). Only the first step corresponds to the variation in luminosity recorded at the fault surface. This effect is probably linked to the diachronous rupture of each side of the double-direct shear configuration. This is a common feature of irregular slow slip events observed at low LPVs (< 0.5 µm/s). The description of irregular slow slip events applies as well to the main events recorded during chaotic slow slip (Fig. 3 b,c). 5. Discussion Our experiments qualitatively reveal the state of stress, the development of self-organized structures, and how these control the nucleation, propagation, and arrest of slip instabilities. Here we provide an explanation of this self-organized shear fabric. 5.1. Fabric development and spontaneous redistribution of stress on-fault At the beginning of each experiment, we force the microstructural reorganisation of the fault structure from the initial homogeneity (uniaxially pressed powder) to an optimal shear fabric (localised shear zones). This configuration is reproducible across experiments and consists of the formation of two types of shear bands (Fig. 8 a): i) two boundary (B-) shear zones forming at the fault surface (B U ) and close to the lateral block (B L ,), and ii) several Riedel shears (R1-R9), which together bound sigmoidal low strain domains, the micro-lithons (ML). The micro-lithons are consistently ten in number and with rather regular spacing, suggesting that there is a geometrical control between sample dimension (layer thickness, contact area), experimental conditions (run-in velocity and normal load), and optimal stress distribution 31,41,46 . The shear strain is accommodated mainly by both B-shears and is partitioned between the two through the activity of R-shears (Fig. 8 a). The formation of each R-shear relieves stress from the micro-lithon it bounds, with the largest stress decrease recorded in its proximity. Immediately after the formation of the R-shear, stress increases inside the micro-lithon close to the previous R-shear (Supplementary Section SM3) producing the small-scale fluctuations of stress in the Y direction (Fig. 2 a). This effect recalls the way stress is relieved at asperities when material is plucked-out in response to adhesive wear 47,48 , transferring the strain from the sliding surface to a new, deeper slip zone. In our case, the lower B-shear. In the fault’s final configuration, the top of the R-shears separates the least stressed regions from neighbouring loci of stress concentration (Fig. 8 a). The formation of R-shears seems thus a necessary mechanism that redistributes stress across the fault while boundary shear zones creep to accommodate stable sliding. This effect is extremely important in producing heterogeneities that determine the initiation and mode of fault instability 49,50 . Given the impossibility to form further R-shears towards the loading point, the last formed micro-lithon is the most stressed region of the fault and presents large stress heterogeneities at its boundaries (Fig. 8 a), especially in correspondence of a low stress embayment. This is the area mostly affected by the loading conditions and that becomes inherently unstable. Lastly, it is important to point out an important feature. Although a fault zone structure fully populated by R-shears can be reasonably seen as a final state of mechanical equilibrium, the analysis of both time series and footages does not confirm this interpretation. The bulk friction has in fact already achieved steady state conditions (Fig. 1 b) after the formation of the fifth R-shear, as well as the average luminosity. This means that the continuous evolution of the inner structure of the fault zone throughout the experiment, which is responsible of the spatial reorganisation of stress, leads to a first-order mechanical equilibrium (e.g., macroscopic friction, stress), characterized by the contribution of all points of the fault (i.e., bulk response). This is what is commonly observed with the single observation point (load cell) of the experimental apparatus. However, second order variations may still evolve with strain within the sample, spontaneously following the fabric development, at a scale smaller than the bulk fault, without requiring further energy dissipation. This has important implications on the rate-dependence of friction, sliding stability, and their relationship with strain and strain rate 51–53 . 5.2. Nucleation of slip instabilities As much as all slow slip events display rather variable rupture patterns (Supplementary Section SM4), most initiate at the highest stress heterogeneity (regular slow slip events, Fig. 8 b), located at the bounding R-shear of the uppermost micro-lithon (Figs. 3 , 4 , 5 ). Irregular events (Fig. 8 b) may also originate at smaller stress heterogeneities inside the fault (Fig. 6 g). The ruptures thus start in regions where sharp local stress gradients are observed, but the absolute entity of the local stress seems subordinate to the intrinsic instability of the area and memory effects. This is clear since the overall stress state of the fault remains remarkably unchanged across several creep-slip cycles (Supplementary Section SM3) while minor redistribution of stress can disrupt the previous rupture patterns and favour nucleation in other regions (Figs. 6 c). For this reason, irregular slip events may emerge as a consequence of incomplete ruptures or overlapping chaotic rupture patterns (Fig. 3 c). In our experiments, most events display no clear distinction between a nucleation and a dynamic phase sensu Ohnaka and Shen 54 . Instead, different rupture speed are recorded depending on the direction of the rupture front (slower in the X direction, Fig. 4 e,f), while a fictious nucleation stage may emerge due to a specific choice of directional averaging of the fault luminosity (e.g., Fig. 4 d). This highlights the importance of the observation point on the fault (e.g., camera or other sensors), here privileged by the planar view. A distinct, slow nucleation phase is clearly observed only when the instability initiates in more central areas of the fault (irregular events, Figs. 8 b). Given the general lack of a distinguishable, systematic nucleation phase, a reliable measure of the nucleation size for use in numerical models 55 is unattainable. Neither can nucleation size be calculated theoretically since the material is rate-strengthening 38 and the stress distribution is heterogeneous. Nevertheless, since we observe partitioning of strain between the different shear zones, it is reasonable to assume that slow slip events tend to nucleate away from the B-shear zones, for example, inside the R-shears (Fig. 8 c). In support of this hypothesis, we recorded the formation of plumose structures along R-shears formed in over-consolidated clay experiments (Supplementary Section SM5). These structures indicate the energetic formation of R-shear through fracture-like mechanisms, potentially associated with the nucleation of slip instabilities. The beginning of slow slip instabilities in our experiments is phenomenologically related to the formation and migration of dilatant waves (increased luminosity) from a source region of stress heterogeneity, such as those forming at elastic bi-material interfaces 56,57 and widely known as Schallamach waves 58 . These are self-healing detachment pulses that start and propagate in a similar manner to what observed in our experiments. The similarity implies that the instability grows as the local loss of adhesion, which manifests in clay as region with growing porosity (increased luminosity) and possible detachment of platelets from the central slider. To corroborate this mechanistic interpretation, we highlight the existence of different regimes at different loading point velocities (Fig. 3 ), as observed in analogue tests on bi-material interfaces 56,58 . Finally, it was demonstrated that slow slip self-healing pulses can emerge as function of stiffness contrast and finite size of the setup at elastic-frictional bi-material interfaces 59 . In our experiments it is difficult to adapt existing models for a number of reasons: i) deformation occurs in a thin boundary shear zone (< 100 µm), which is not a proper bi-material interface, ii) an important fraction of deformation is accommodated by creep, and iii) instabilities form independently of the stiffness of the central slider 38 . In this light, the intrinsic elastic properties (bending) of the clay platelets 60 , the preferred frictional sliding along the S-C foliation 38 , and the sub-mm stress heterogeneities on the fault zone may play a far more important role in nucleation than the bulk properties of the sample and experimental setup. Our observations suggest that slip instabilities nucleate within the gouge layer as a result of stress heterogeneity determined by shear fabric evolution. In other words, nucleation of dynamic instabilities is a volumetric process driven by shearing and sensitive to tiny variations of the loading rate. This implies that the macroscopic description of dynamic slip instabilities as retrieved from the single observation point on the load cell in classic rock deformation experiments does not resolve the complex spatial-temporal evolution of sliding at smaller wavelengths. Thus, the macroscopic slip of the whole sample is the result of strain partitioning within the volume. 5.3. Propagation, arrest, and quenching of slow slip Beside nucleation, propagation, and arrest, the rupture pattern is strongly influenced by stress distribution and fault structures. All ruptures tend to be initially confined within the micro-lithons where they originate, with the high-stressed areas acting as barriers 17 . These barriers are overcome when the macroscopic rupture breaks perpendicular to the ML boundary and especially when it joins the low-stressed fault edges (Fig. 5 frames VI-VII). The same effect of the fault edge on rupture propagation is reported in analogue experiments in plastic faults 61 . We observe that each jump across the R-shears can be met with some latency and causes a progressive deceleration of the rupture edge (Fig. 4 e). This effect is observed especially in the region where the earliest R-shears formed, the same area of the fault where some partial ruptures stop at low LPVs (Fig. 6 c). This effect is probably related to the region of strain partitioning between the upper and lower boundary (B-)shear zones (Fig. 8 a, Supplementary Section SM3). Here, especially at low LPVs, the rupture may “sink” into the R-shears, dissipating the available elastic energy into distributed permanent deformation. With increasing LPV the bulk frictional instability of the experimental fault is quenched towards stable sliding (LPV > 7.5 µm/s, Fig. 3 c). This, however, is not related to the disappearance of slow slip instabilities. Slow, pulse-like ruptures continue to nucleate especially at the large stress heterogeneity where regular events originate (Fig. 3 c) and, at the same time, propagate while other subsidiary ruptures still sweep across the fault and nucleate in other locations. The bulk stability recorded in the time series (apparent friction) is then the result of combined effect of: i) interfering rupture patterns, ii) recurrence time smaller than that required for a complete rupture, iii) inhomogeneous redistribution of stress across cycles, and iv) different nucleation sites. We argue that the inherent instability of the material is also related to the sample size 14,59 , which signifies that the embryonal phases of an earthquake (small scale processes) is severely affected by the role of heterogeneities across different scales. For example, the fault patch acting as an asperity in nature might vary substantially depending on larger scale geological features 62 while microstructures explored here maintain the same dimension independently of the setting. Our observations suggest that macroscopic slip associated with the friction drop is due to strain partitioning between B- and R-shear within the fault zone characterized by spatial-temporal evolution defining the so-called rupture patterns. Rupture involves the evolution of fabric over the entire gouge layer and is affected by small-scale heterogeneities and fluctuations of the loading velocity. While this does not preclude the interpretation of macroscopic dynamic slip instabilities, it complicates the interpretation of mechanical processes acting at the micro scale. 5.4. Concluding remarks Our experiments allowed to image the dynamics of slow slip in laboratory clay faults. The experimental fault evolves with slip to form a reproducible inner structure composed of localised shear zones and low strain domains. This structure is in continuous evolution with strain, despite the apparent early achievement of bulk mechanical equilibrium (steady state conditions). This structure represents the self-organisation and redistribution of stress across the fault zone (i.e., the gouge layer), which has important effects on the genesis and propagation of unstable events. At low forcing velocities, slow slip events nucleate due to adhesion loss at local stress heterogeneities, which are not necessarily the largest inside the fault zone. The nucleation stage does not show a characteristic, recognisable length scale, and it is associated with stress heterogeneities in the gouge volume. The slip instability inferred from the mechanical data is indeed associated with the spatial-temporal evolution of a strain pulse. Slip instabilities propagate as self-healing dilatant pulses causing only minor redistribution of stress across the fault zone but do introduce memory effects in the regions they swipe through. As a consequence, each slow slip event is different and can follow complex rupture patterns, especially at both low (LPV 2.5 µm/s) loading rates. Even when the fault is macroscopically stable (steady-state apparent friction) at high loading rates, slip pulses continue to nucleate and propagate, but become chaotic and of smaller entity. The dynamics of slow slip appear thus complex even in experimental small faults, questioning whether constitutive laws for earthquake nucleation and frictional instability may represent processes across all scales, which would require proper addressing. 6. Methods Rock deformation experiments were carried out in BRAVA 35 , a bi-triaxial rock deformation apparatus (Fig. M1 a) hosted at the HPHT laboratory at the Istituto Nazionale di Geofisica e Vulcanologia (Rome, Italy). The experimental setup consists of a double direct shear sample holder, composed of two lateral forcing blocks compressing two parallel sample layers against a transparent Plexiglas central slider (Fig. M1 b). The sample has an initial thickness of 5 mm and is compressed on a nominal squared area of contact of 50 by 50 mm (Fig. M1 b). The samples were humidified beforehand at 100% room humidity for 12 h, allowing a thin film of water to be adsorbed on the surface of the platelets. The contact surface between the forcing blocks and the sample is flat. Normal stress and shear motion are controlled by two orthogonal hydraulic pistons (Fig. M1 a) and are transferred to the sample by the lateral blocks and the central slider (Fig. M1 b), respectively. On one side, a shielded led lamp provides with a homogeneous white light source. On the other, a high speed camera (NAC Memrecam) films the light reflected by the sample-block interface (Fig. M1 c). Our video measurements have a temporal resolution of 0.02 to 0.0067 s and a spatial resolution of ~ 50 µm and ~ 130 µm in the shear (S.D.) and horizontal (H.D.) directions, respectively, determined by the angle of view (Fig. 1 b). Video analysis was carried with a custom MATLAB© code performing simple operations on raster frames (.tif) extracted from the videos, including static and rolling background removal. In the static background removal (Figs. 4 c-d, 5 a, 6 c-d, 7 a) the background image corresponds to the first frame of interest of the video and is systematically subtracted from the following frames. The resulting video displays changes in the luminosity of the fault surface (quantified in arbitrary units, a.u.) with respect to the initial frame. The change in reflectivity obtained with static background removal qualitatively shows the stress change on the fault surface. The rolling back- ground removal (Figs. 3 c, 4 f-e, 6 f-e) is operated by subtracting to each frame the frame preceding by a fixed separation of time. The optimal separation for visualization has been selected by trial and error. The rolling back-ground removal qualitatively indicates the stress change rate on the fault surface, which is the discrete derivative of the stress change. Linear pixel-wise averaging in the X and Y directions were used on selected areas to image the propagation of rupture fronts. This analysis could not provide quantitative results due to the strong non-linearity between luminosity, stress, and poroelastic state of the material (Supplementary Section SM1). Nevertheless, our analysis show robust inverse correlation between luminosity and stress, corroborating the results and supporting the discussion. Declarations Author contributions  Conceptualization: G.P. and G.V. Data curation, formal analysis, investigation, methodology: G.P., J.T., and G.V. Writing—original draft: G.P. Writing—review and editing: G.V., J.T., M.C., C.M., C.C. Resources and funding acquisition: G.P, M.C. Visualization: G.P.; Software: G.P., J.T. Project administration: G.P. Acknowledgements - Competing interests The authors declare no competing interests. Data availability Raw data files are available on Zenodo (files will be uploaded and the statement updated upon finalisation of the manuscript). References Bürgmann, R. (2018). The geophysics, geology and mechanics of slow fault slip. Earth and Planetary Science Letters, 495, 112–134. https://doi.org/10.1016/j.epsl.2018.04.062 Peng, Z., & Gomberg, J. (2010). An integrated perspective of the continuum between earthquakes and slow-slip phenomena. In Nature Geoscience (Vol. 3, Issue 9, pp. 599–607). Nature Publishing Group. https://doi.org/10.1038/ngeo940 Kirkpatrick, J. D., Fagereng, Å., & Shelly, D. R. (2021). Geological constraints on the mechanisms of slow earthquakes. Nature Reviews Earth & Environment 2021 2:4, 2(4), 285–301. https://doi.org/10.1038/s43017-021-00148-w Ide, S., & Beroza, G. C. (2023). Slow earthquake scaling reconsidered as a boundary between distinct modes of rupture propagation. Proceedings of the National Academy of Sciences of the United States of America, 120(32). https://doi.org/10.1073/pnas.2222102120 Brace, W. F., & Byerlee, J. D. (1966). Stick-slip as a mechanism for earthquakes. Science, 153(3739), 990–992. https://doi.org/10.1126/science.153.3739.990 Dieterich, J. H. (1972). Time-dependent friction in rocks. Journal of Geophysical Research, 77(20), 3690–3697. https://doi.org/10.1029/jb077i020p03690 Ruina, A. (1983). Slip instability and state variable friction laws. Journal of Geophysical Research, 88(B12), 10359–10370. https://doi.org/10.1029/JB088iB12p10359 Marone, C. (1998). Laboratory-Derived Friction Laws and Their Application To Seismic Faulting. Annual Review of Earth and Planetary Sciences, 26(1), 643–696. https://doi.org/10.1146/annurev.earth.26.1.643 Leeman, J. R., Saffer, D. M., Scuderi, M. M., & Marone, C. (2016). Laboratory observations of slow earthquakes and the spectrum of tectonic fault slip modes. Nature Communications, 7, 1–6. https://doi.org/10.1038/ncomms11104 Scuderi, M. M., Collettini, C., Viti, C., Tinti, E., & Marone, C. (2017). Evolution of shear fabric in granular fault gouge from stable sliding to stick slip and implications for fault slip mode. Geology, 45(8), 731–734. https://doi.org/10.1130/G39033.1 Im, K., Saffer, D., Marone, C., & Avouac, J. P. (2020). Slip-rate-dependent friction as a universal mechanism for slow slip events. Nature Geoscience, 13(10), 705–710. https://doi.org/10.1038/s41561-020-0627-9 Barbot, S. (2019). Slow-slip, slow earthquakes, period-two cycles, full and partial ruptures, and deterministic chaos in a single asperity fault. Tectonophysics, 768(March), 228171. https://doi.org/10.1016/j.tecto.2019.228171 Selvadurai, P. A., & Glaser, S. D. (2015). Laboratory-developed contact models controlling instability on frictional faults. Journal of Geophysical Research: Solid Earth, 120(6), 4208–4236. https://doi.org/10.1002/2014JB011690 Harbord, C. W. A., Nielsen, S. B., De Paola, N., & Holdsworth, R. E. (2017). Earthquake nucleation on rough faults. Geology, 45(10), 931–934. https://doi.org/10.1130/G39181.1 Passelègue, F. X., Almakari, M., Dublanchet, P., Barras, F., Fortin, J., & Violay, M. (2020). Initial effective stress controls the nature of earthquakes. Nature Communications, 11(1), 1–8. https://doi.org/10.1038/s41467-020-18937-0 Gounon, A., Latour, S., Letort, J., & El Arem, S. (2022). Rupture Nucleation on a Periodically Heterogeneous Interface. Geophysical Research Letters, 49(20), 1–9. https://doi.org/10.1029/2021GL096816 Cebry, S. B. L., Sorhaindo, K., & McLaskey, G. C. (2023). Laboratory Earthquake Rupture Interactions With a High Normal Stress Bump. Journal of Geophysical Research: Solid Earth, 128(11), 1–24. https://doi.org/10.1029/2023JB027297 Xu, S., Fukuyama, E., Yamashita, F., Kawakata, H., Mizoguchi, K., & Takizawa, S. (2023). Fault strength and rupture process controlled by fault surface topography. Nature Geoscience, 16(1), 94–100. https://doi.org/10.1038/s41561-022-01093-z Corbi, F., Funiciello, F., Moroni, M., Van Dinther, Y., Mai, P. M., Dalguer, L. A., & Faccenna, C. (2013). The seismic cycle at subduction thrusts: 1. Insights from laboratory models. Journal of Geophysical Research: Solid Earth, 118(4), 1483–1501. https://doi.org/10.1029/2012JB009481 Rubino, V., Lapusta, N., & Rosakis, A. J. (2022). Intermittent lab earthquakes in dynamically weakening fault gouge. Nature, 606(7916), 922–929. https://doi.org/10.1038/s41586-022-04749-3 Song, J. Y., & McLaskey, G. C. (2024). Laboratory Earthquake Ruptures Contained by Velocity Strengthening Fault Patches. Journal of Geophysical Research: Solid Earth, 129(4), 1–16. https://doi.org/10.1029/2023JB028509 Niemeijer, A. R., & Spiers, C. J. (2005). Influence of phyllosilicates on fault strength in the brittle-ductile transition: Insights from rock analogue experiments. Geological Society Special Publication, 245, 303–327. https://doi.org/10.1144/GSL.SP.2005.245.01.15 Tembe, S., Lockner, D. A., & Wong, T.-F. (2010). Effect of clay content and mineralogy on frictional sliding behavior of simulated gouges: Binary and ternary mixtures of quartz, illite, and montmorillonite. Journal of Geophysical Research, 115(B3), B03416. https://doi.org/10.1029/2009JB006383 Tesei, T., Collettini, C., Barchi, M. R., Carpenter, B. M., & Di Stefano, G. (2014). Heterogeneous strength and fault zone complexity of carbonate-bearing thrusts with possible implications for seismicity. Earth and Planetary Science Letters, 408, 307–318. https://doi.org/10.1016/j.epsl.2014.10.021 Bedford, J. D., Faulkner, D. R., & Lapusta, N. (2022). Fault rock heterogeneity can produce fault weakness and reduce fault stability. Nature Communications, 13(1), 1–7. https://doi.org/10.1038/s41467-022-27998-2 Arts, J. P. B., Niemeijer, A. R., Drury, M. R., Willingshofer, E., & Matenco, L. C. (2024). The frictional strength and stability of spatially heterogeneous fault gouges. Earth and Planetary Science Letters, 628(January). https://doi.org/10.1016/j.epsl.2024.118586 Ruggieri, R., Scuderi, M. M., Trippetta, F., Tinti, E., Brignoli, M., Mantica, S., Petroselli, S., Osculati, L., Volontè, G., & Collettini, C. (2021). The role of shale content and pore-water saturation on frictional properties of simulated carbonate faults. Tectonophysics, 807. https://doi.org/10.1016/j.tecto.2021.228811 Scuderi, M. M., Tinti, E., Cocco, M., & Collettini, C. (2020). The Role of Shear Fabric in Controlling Breakdown Processes During Laboratory Slow-Slip Events. Journal of Geophysical Research: Solid Earth, 125(11), 1–23. https://doi.org/10.1029/2020JB020405 Pozzi, G., Scuderi, M. M., Tinti, E., Nazzari, M., & Collettini, C. (2022). The Role of Fault Rock Fabric in the Dynamics of Laboratory Faults. Journal of Geophysical Research: Solid Earth, 127(6). https://doi.org/10.1029/2021JB023779 Rutter, E. H., Maddock, R. H., Hall, S. H., & White, S. H. (1986). Comparative microstructures of natural and experimentally produced clay-bearing fault gouges. Pure and Applied Geophysics PAGEOPH, 124(1–2), 3–30. https://doi.org/10.1007/BF00875717 Logan, J. M., Dengo, C. A., Higgs, N. G., & Wang, Z. Z. (1992). Fabrics of experimental fault zones: Their development and relationship to mechanical behavior BT - Fault Mechanics and Transport Properties of Rocks. In Fault Mechanics and Transport Properties of Rocks: Vol. Internatio (Issue 2). Chester, F. M., & Logan, J. M. (1987). Composite planar fabric of gouge from the Punchbowl fault, California. Journal of Structural Geology, 9(5–6), 621-IN6. Fagereng, Å., & Sibson, R. H. (2010). Mélange rheology and seismic style. Geology, 38(8), 751–754. https://doi.org/10.1130/G30868.1 Collettini, C., Niemeijer, A., Viti, C., Smith, S. A. F., & Marone, C. (2011). Fault structure, frictional properties and mixed-mode fault slip behavior. Earth and Planetary Science Letters, 311(3–4), 316–327. https://doi.org/10.1016/j.epsl.2011.09.020 Collettini, C., Di Stefano, G., Carpenter, B., Scarlato, P., Tesei, T., Mollo, S., Trippetta, F., Marone, C., Romeo, G., & Chiaraluce, L. (2014). A novel and versatile apparatus for brittle rock deformation. International Journal of Rock Mechanics and Mining Sciences, 66, 114–123. https://doi.org/10.1016/j.ijrmms.2013.12.005 Vrolijk, P. (1990). On the mechanical role of smectite in subduction zones. Geology, 18(8), 703–707. https://doi.org/10.1130/0091-7613(1990)0182.3.CO;2 Saffer, D. M., & Marone, C. (2003). Comparison of smectite- and illite-rich gouge frictional properties: application to the updip limit of the seismogenic zone along subduction megathrusts. Earth and Planetary Science Letters, 215(1–2), 219–235. https://doi.org/10.1016/S0012-821X(03)00424-2 Volpe, G., Collettini, C., Taddeucci, J., Marone, C., & Pozzi, G. (2024). Frictional instabilities in clay illuminate the origin of slow earthquakes. Science Advances, 10(26), 1–11. https://doi.org/10.1126/sciadv.adn0869 Miller, C. M. (2011). Adhesion and the Surface Energy Components of Natural Minerals and Aggregates. Texas A&M University. Mondol, N. H., Bjørlykke, K., & Jahren, J. (2008). Experimental compaction of clays: Relationship between permeability and petrophysical properties in mudstones. Petroleum Geoscience, 14(4), 319–337. https://doi.org/10.1144/1354-079308-773 Tchalenko, J. S. (1968). The evolution of kink-bands and the development of compression textures in sheared clays. Tectonophysics, 6(2), 159–174. https://doi.org/10.1016/0040-1951(68)90017-6 Haines, S. H., Kaproth, B., Marone, C., Saffer, D., & Van der Pluijm, B. (2013). Shear zones in clay-rich fault gouge: A laboratory study of fabric development and evolution. Journal of Structural Geology, 51, 206–225. https://doi.org/10.1016/j.jsg.2013.01.002 Volpe, G., Pozzi, G., & Collettini, C. (2022). Y-B-P-R or S-C-C ′ ? Suggestion for the nomenclature of experimental brittle fault fabric in phyllosilicate-granular mixtures. Journal of Structural Geology, 165(July), 104743. https://doi.org/10.1016/j.jsg.2022.104743 Leeman, J. R., Marone, C., & Saffer, D. M. (2018). Frictional Mechanics of Slow Earthquakes. Journal of Geophysical Research: Solid Earth, 123(9), 7931– 7949. https://doi.org/10.1029/2018JB015768 Nielsen, S., & Madariaga, R. (2003). On the self-healing fracture mode. Bulletin of the Seismological Society of America, 93(6), 2375–2388. https://doi.org/10.1785/0120020090 Atmaoui, N., Kukowski, N., Stöckhert, B., & König, D. (2006). Initiation and development of pull-apart basins with Riedel shear mechanism: Insights from scaled clay experiments. International Journal of Earth Sciences, 95(2), 225–238. https://doi.org/10.1007/s00531-005-0030-1 Tesei, T., Carpenter, B. M., Giorgetti, C., Scuderi, M. M., Sagy, A., Scarlato, P., & Collettini, C. (2017). Friction and scale-dependent deformation processes of large experimental carbonate faults. Journal of Structural Geology, 100, 12–23. https://doi.org/10.1016/j.jsg.2017.05.008 Scholz, C. H. (2019) The Mechanics of Earthquakes and Faulting. 3rd ed. Cambridge University Press. Mildon, Z. K., Roberts, G. P., Faure Walker, J. P., & Toda, S. (2019). Coulomb pre-stress and fault bends are ignored yet vital factors for earthquake triggering and hazard. Nature Communications, 10(1), 1–9. https://doi.org/10.1038/s41467-019-10520-6 Ben-zion, Y., & Dresen, G. (2022). A Synthesis of Fracture, Friction and Damage Processes in Earthquake Rupture Zones. Pure and Applied Geophysics, 179(12), 4323–4339. https://doi.org/10.1007/s00024-022-03168-9 Beeler, N. M., Tullis, T. E., Blanpied, M. L., & Weeks, J. D. (1996). Frictional behavior of large displacement experimental faults. Journal of Geophysical Research: Solid Earth, 101(B4), 8697–8715. https://doi.org/10.1029/96JB00411 Mair, K., & Marone, C. (1999). Friction of simulated fault gouge for a wide range of velocities and normal stresses. Journal of Geophysical Research: Solid Earth, 104(B12), 28899–28914. https://doi.org/10.1029/1999JB900279 Noël, C., Giorgetti, C., Collettini, C., & Marone, C. (2024). The effect of shear strain and shear localization on fault healing. Geophysical Journal International, 236(3), 1206–1215. https://doi.org/10.1093/gji/ggad486 Ohnaka, M., & Shen, L. (1999). Scaling of the shear rupture process from nucleation to dynamic propagation: Implications of geometric irregularity of the rupturing surfaces. Journal of Geophysical Research: Solid Earth, 104(B1), 817–844. https://doi.org/10.1029/1998jb900007 Rubin, A. M., & Ampuero, J. P. (2005). Earthquake nucleation on (aging) rate and state faults. Journal of Geophysical Research: Solid Earth, 110(11), 1–24. https://doi.org/10.1029/2005JB003686 Yamaguchi, T., Ohmata, S., & Doi, M. (2009). Regular to chaotic transition of stick-slip motion in sliding friction of an adhesive gel-sheet. Journal of Physics Condensed Matter, 21(20). https://doi.org/10.1088/0953-8984/21/20/205105 Shlomai, H., & Fineberg, J. (2016). The structure of slip-pulses and supershear ruptures driving slip in bimaterial friction. Nature Communications, 7, 1–7. https://doi.org/10.1038/ncomms11787 Barquins, M. (1985). Sliding friction of rubber and Schallamach waves - A review. Materials Science and Engineering, 73(C), 45–63. https://doi.org/10.1016/0025-5416(85)90295-2 Brener, E. A., Weikamp, M., Spatschek, R., Bar-Sinai, Y., & Bouchbinder, E. (2016). Dynamic instabilities of frictional sliding at a bimaterial interface. Journal of the Mechanics and Physics of Solids, 89, 149–173. https://doi.org/10.1016/j.jmps.2016.01.009 Barsoum, M. W., Zhao, X., Shanazarov, S., Romanchuk, A., Koumlis, S., Pagano, S. J., Lamberson, L., & Tucker, G. J. (2019). Ripplocations: A universal deformation mechanism in layered solids. Physical Review Materials, 3(1), 1–9. https://doi.org/10.1103/PhysRevMaterials.3.013602 Gvirtzman, S., & Fineberg, J. (2023). The Initiation of Frictional Motion—The Nucleation Dynamics of Frictional Ruptures. Journal of Geophysical Research: Solid Earth, 128(2), 1–18. https://doi.org/10.1029/2022JB025483 Brodsky, E. E., Kirkpatrick, J. D., & Candela, T. (2016). Constraints from fault roughness on the scale-dependent strength of rocks. Geology, 44(1), 19–22. https://doi.org/10.1130/G37206.1 Additional Declarations There is NO Competing Interest. Supplementary Files FigureM1.png Figure M1 – Experimental apparatus, setup, and sample assembly: a) side view of the apparatus BRAVA with mounted sample; b) top view of the experimental setup in scale; c) ensembled double direct shear sample holder. S.D., H.D., and T.D are the shear, horizontal, and thickness directions, respectively. All figures maintain the original proportions. SV1Riedels232831Luminosity.gif Supplementary Video SV1 SV2Riedels232831RBG05.gif Supplementary Video SV2 SV3SlowSlip250402404SBG.gif Supplementary Video SV3 SV4SlowSlip250402404RBG0.02.gif Supplementary Video SV4 SV5SlowSlip100221223SBG.gif Supplementary Video SV5 SV6SlowSlip100221223RBG0.02.gif Supplementary Video SV6 Supplementarymaterials.docx Cite Share Download PDF Status: Published Journal Publication published 07 Oct, 2025 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6097579","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":425170654,"identity":"46b4339a-5f5b-454a-b1a9-51f64aee92bc","order_by":0,"name":"Giacomo Pozzi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIiWNgGAWjYDACZmROQoUEmD5AgpYzUC349SADxjYoA58WeXfmhx8+5tgwmLO3P/7wcJ5FnnxEAuPhD3i0GB5mM5acuS2NwbLnQIJB4jaJYsMbCfgdZtjMYMbMu+0wg8GNhAMJQC2JG3sOENLC/g2qJbHhQOIcIrTIM/PAbElmbEhskEicz96AX4sBM08xyC88lj3HmBkSjkkkbmBvbDhwBp8t/cc3fvi4zUYOFGIff9TUJc5vZj78oQKfLVAn8BggRBgb8GgA2gKTNsAQGQWjYBSMglEABQCTXlGKmThjBgAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0001-7832-9429","institution":"Istituto Nazionale di Geofisica e Vulcanologia","correspondingAuthor":true,"prefix":"","firstName":"Giacomo","middleName":"","lastName":"Pozzi","suffix":""},{"id":425170655,"identity":"5982024b-0a0d-4fd2-9554-b4ba485c1638","order_by":1,"name":"Giuseppe Volpe","email":"","orcid":"https://orcid.org/0000-0003-0165-0446","institution":"La Sapienza University of Rome","correspondingAuthor":false,"prefix":"","firstName":"Giuseppe","middleName":"","lastName":"Volpe","suffix":""},{"id":425170656,"identity":"ad1734cb-c1f9-4c3d-9f62-d5b6fd415e24","order_by":2,"name":"Jacopo Taddeucci","email":"","orcid":"https://orcid.org/0000-0002-0516-3699","institution":"Istituto Nazionale di Geofisica e Vulcanologia","correspondingAuthor":false,"prefix":"","firstName":"Jacopo","middleName":"","lastName":"Taddeucci","suffix":""},{"id":425170657,"identity":"1bf8a1d4-581a-4bc8-b209-3737abfbce25","order_by":3,"name":"Massimo Cocco","email":"","orcid":"https://orcid.org/0000-0001-6798-4225","institution":"Istituto Nazionale di Geofisica e Vulcanologia","correspondingAuthor":false,"prefix":"","firstName":"Massimo","middleName":"","lastName":"Cocco","suffix":""},{"id":425170658,"identity":"4e481613-5d1d-4abd-8ca8-539d63d51570","order_by":4,"name":"Chris Marone","email":"","orcid":"","institution":"La Sapienza Universit`a di Roma","correspondingAuthor":false,"prefix":"","firstName":"Chris","middleName":"","lastName":"Marone","suffix":""},{"id":425170659,"identity":"ee7b0bd1-4010-44e8-b31a-b22e206df32f","order_by":5,"name":"Cristiano Collettini","email":"","orcid":"https://orcid.org/0000-0002-4828-2516","institution":"Sapienza Università di Roma","correspondingAuthor":false,"prefix":"","firstName":"Cristiano","middleName":"","lastName":"Collettini","suffix":""}],"badges":[],"createdAt":"2025-02-24 14:05:44","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6097579/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6097579/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41467-025-63984-0","type":"published","date":"2025-10-07T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":78189498,"identity":"803264bd-238e-4296-9812-5ff983477cc9","added_by":"auto","created_at":"2025-03-10 19:42:30","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":2012307,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eExperimental setup, shear strength evolution, and fault structure development:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) Double direct shear arrangement with two fault zones, transparent central block, and camera to capture images of the inner fault surface (XY plane). b) Friction and average fault surface luminosity during shear deformation at constant LPV = 10 µm/s. c) Snapshots of the fault surface luminosity showing spatiotemporal evolution of luminosity during shear. Temporal sequence of snapshots is given in b). Riedel R1 shears begin to form after the peak stress (grey box in panel b). They form first at the bottom (white arrow) and then progressively above as slip accumulates. Higher-stressed areas appear darker. Black arrow indicates subsidiary R-shears forming through lateral extrusion of the sample. Dashed region in VIII: area analysed in Figs. 2-3. Video version of panel c) in Supplementary Videos SV1-2.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/57fe6480857b2a36fb479ba2.png"},{"id":78189499,"identity":"b3913167-9057-4f5f-97fa-2d7f3d939fd1","added_by":"auto","created_at":"2025-03-10 19:42:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":7711534,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eStructural elements of the experimental fault zone:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) Luminosity profile along the dashed region of Fig. 1c (frame VIII), correlated with a sketch of the kinematic cross section (YZ) of the deformed sample (Y = 0 is at the right of Panel a). Dashed lines R1-R9 indicate the position and sequence of formation of the R-shears, sketched below. Lower sub-panels below show a schematic YZ section interpreted from microstructural analysis (see Panel b) with the following structural elements: upper (B\u003c/em\u003e\u003csub\u003e\u003cem\u003eU\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e) and lower (B\u003c/em\u003e\u003csub\u003e\u003cem\u003eL\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e) B-shears, R-shears (R), and micro-lithons (ML). Shaded areas above R-shears show relaxed regions of low stress. b) Electron back-scattered image of a YZ sample cross-section showing the upper and lower B-shears, R-shears, and micro-lithons. The close-up in the red box shows the S-C fabric\u003c/em\u003e\u003csup\u003e\u003cem\u003e43 \u003c/em\u003e\u003c/sup\u003e\u003cem\u003econtained in B\u003c/em\u003e\u003csub\u003e\u003cem\u003eU\u003c/em\u003e\u003c/sub\u003e\u003cem\u003e.\u003c/em\u003e \u003cem\u003ec) Sketch of the structural elements of the deformed sample (XYZ coordinate system is defined in Fig. 1).\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/11ec31d1a8b568b1e87b70cc.png"},{"id":78189578,"identity":"efe83477-ea27-4288-836e-0819fa62f24b","added_by":"auto","created_at":"2025-03-10 19:42:48","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":3118212,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eUnstable sliding in laboratory slow slip:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) Evolution of friction (black, close-up in b) and average luminosity (blue) as a function of time for a range of load point velocities (LPV). Note that the average luminosity is anticorrelated with friction (balloon) but appears nearly invariant throughout the stick-slip series when compared to the variations during run-in (Fig. 1b). c) Qualitative stressing rate (inverse of luminosity change rate, dL/dt with dt = 20 ms) in the shear direction restricted to the stripe of Fig. 1c, frame VIII. Red circles indicate the approximative time and location of nucleation.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/55f5721c51f90358596e7baf.png"},{"id":77969994,"identity":"c0d50d5c-d529-4b1f-9c53-c3b8ab2705bd","added_by":"auto","created_at":"2025-03-07 10:40:45","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":4744932,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eRegular slow slip:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) The evolution of friction and average luminosity across three slow slip cycles, with b) close-up of the third event. Note that luminosity increases during the entire friction drop. c) Qualitative stress change (inverse of luminosity change) along the shear direction with time, with d) close-up of the third event. e-f) Qualitative stressing rate (inverse of luminosity change rate, dL/dt with dt = 20 ms) in the e) shear and f) horizontal directions. g) Snapshot of the luminosity at t = 360 s. Dashed regions show locations of stressing rate calculations for panels e) and f). Arrows: black arrow, start of the instability; white, change of main rupture direction. Yellow circle: nucleation site. Video version of panels d-f) in Supplementary Videos SV3-4.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/be0f4b8a076a2456b5b5092b.png"},{"id":77971335,"identity":"8c360492-5f67-4bac-8a19-9c71cc857205","added_by":"auto","created_at":"2025-03-07 10:56:45","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":10904300,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eRegular rupture:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) Qualitative stress change (inverse of luminosity change) across the experimental fault surface with respect to a reference frame a t = 0, not shown. Same event as Fig 4b. Each frame is separated by ∆t = 0.1 s (frame I is at t = 0.1), reported with Roman numerals (panel b). Dotted lines indicate the rupture front position in the previous time frame. b) Friction curve showing the timing of each frame in a). c) Luminosity with rupture fronts of panel a dotted in white and nucleation site circled in yellow. d) Rupture progression (dark to light) with nucleation site in yellow.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/429f28e7e1666597e5006039.png"},{"id":77970460,"identity":"08ed32e9-4ae8-4215-a8c3-419c72dc442f","added_by":"auto","created_at":"2025-03-07 10:48:45","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":5152292,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eIrregular slow slip:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) The evolution of friction and average luminosity across three irregular slow slip cycles, with b) close-up of the third event. Note that luminosity increases only during the first part of the friction drop. c) Qualitative stress change (inverse of luminosity change) along the shear direction with time, with d) close-up of the third event. e-f) Qualitative stressing rate (inverse of luminosity change rate, dL/dt with dt = 20 ms) in the e) shear and f) horizontal directions. g) Snapshot of the luminosity at t = 130 s. Dashed regions show locations of stressing rate calculations for panels e) and f). Arrows: black, start of the third instability; white, the rupture reaches the topmost stressed microlithon; orange, second part of the friction drop not matched by changes in luminosity. Yellow circle: nucleation site. Video version of panels d-f) in Supplementary Videos SV5-6.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/f2fe0993480ad62bff73b8d0.png"},{"id":77969118,"identity":"f564fd4f-7c57-458a-8c61-554c290f1d14","added_by":"auto","created_at":"2025-03-07 10:32:46","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":11076367,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eComplex rupture:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) Qualitative stress change (inverse of luminosity change) across the experimental fault surface with respect to a reference frame a t = 0, not shown. Same event as Fig 6b. Each frame is separated by ∆t = 0.1 s (frame I is at t = 0.1), reported with Roman numerals (panel b). Dotted lines indicate the rupture front position in the previous time frame. b) Friction curve showing the timing of each frame in a). c) Luminosity with rupture fronts of panel a dotted in white and nucleation site circled in yellow. d) Rupture progression (dark to light) with nucleation site in yellow.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/21614910137c67c4dc3bfd77.png"},{"id":77970002,"identity":"9f04456d-e713-46ba-ab5a-246b2fbe6950","added_by":"auto","created_at":"2025-03-07 10:40:46","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":2182704,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eSchematic of fault structures and slip modes: \u003c/strong\u003e\u003c/em\u003e\u003cem\u003eall panels use the labels of structures coherently with previous figures.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/385f39a00e62d18eca5dac2f.png"},{"id":93009983,"identity":"2656372c-84a2-49ad-a5ec-a85207c17d38","added_by":"auto","created_at":"2025-10-08 07:09:35","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":46288337,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/3aac8389-0314-4223-ba7a-2034db6e6a45.pdf"},{"id":77969106,"identity":"81354892-7c23-464b-ab64-97046d12bf35","added_by":"auto","created_at":"2025-03-07 10:32:45","extension":"png","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":106273,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eFigure M1 – Experimental apparatus, setup, and sample assembly:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e a) side view of the apparatus BRAVA with mounted sample; b) top view of the experimental setup in scale; c) ensembled double direct shear sample holder. S.D., H.D., and T.D are the shear, horizontal, and thickness directions, respectively. All figures maintain the original proportions.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"FigureM1.png","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/2dc8069d795cba4eb1bdff23.png"},{"id":77969116,"identity":"b1e7da66-5e84-4745-9c9b-b69792eb5cd8","added_by":"auto","created_at":"2025-03-07 10:32:45","extension":"gif","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":15824920,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video SV1\u003c/p\u003e","description":"","filename":"SV1Riedels232831Luminosity.gif","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/be54a73970e169f1f74a6ffc.gif"},{"id":77970010,"identity":"12561bef-4902-44c5-8f34-d9c6facd9d81","added_by":"auto","created_at":"2025-03-07 10:40:46","extension":"gif","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":58886665,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video SV2\u003c/p\u003e","description":"","filename":"SV2Riedels232831RBG05.gif","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/4c920aa79beb770d20f4c0ee.gif"},{"id":77969128,"identity":"d6020e53-e998-4d24-a357-f1931705f704","added_by":"auto","created_at":"2025-03-07 10:32:46","extension":"gif","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":33457349,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video SV3\u003c/p\u003e","description":"","filename":"SV3SlowSlip250402404SBG.gif","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/5fc7589b877911c5db9c09c2.gif"},{"id":77969121,"identity":"cf40989b-6a83-4500-a5b1-24a4ebd9608b","added_by":"auto","created_at":"2025-03-07 10:32:46","extension":"gif","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":32870645,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video SV4\u003c/p\u003e","description":"","filename":"SV4SlowSlip250402404RBG0.02.gif","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/778c3484a05dc853c4e72f13.gif"},{"id":77970461,"identity":"220e0219-7abc-4fd5-b97f-4c1e8f00a40c","added_by":"auto","created_at":"2025-03-07 10:48:46","extension":"gif","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":33164678,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video SV5\u003c/p\u003e","description":"","filename":"SV5SlowSlip100221223SBG.gif","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/4003eb4b51d1078c2f219b7c.gif"},{"id":77969151,"identity":"94f5dc49-6a7f-45ee-994c-94a42d9b346c","added_by":"auto","created_at":"2025-03-07 10:32:47","extension":"gif","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":65177762,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Video SV6\u003c/p\u003e","description":"","filename":"SV6SlowSlip100221223RBG0.02.gif","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/bbf5f50877364b3949c4e114.gif"},{"id":77969108,"identity":"153e7ff1-6f2e-4cdd-8217-b7be6e12f919","added_by":"auto","created_at":"2025-03-07 10:32:45","extension":"docx","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":8877717,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"Supplementarymaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-6097579/v1/3f2979f5cff2f3bd7d9bf523.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Spontaneous complexity in the dynamics of slow laboratory earthquakes","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eFaults accommodate deformation through a spectrum of slip behaviours ranging from earthquakes to aseismic slip and creep. This variety includes the newly discovered phenomena of low frequency and slow earthquakes\u003csup\u003e1\u003c/sup\u003e, whose underlying mechanisms are still debated\u003csup\u003e2\u0026ndash;4\u003c/sup\u003e. In this context, laboratory experiments have played a key role in interpreting the inferred fault slip behaviours in terms of dynamic frictional instabilities\u003csup\u003e5\u0026ndash;8\u003c/sup\u003e. The dependence of friction on sliding velocity, for example, laid the foundation for the rate and state theory\u003csup\u003e6\u0026ndash;8\u003c/sup\u003e (RSF), which provides a widely employed physical framework in earthquake dynamics. The fine tuning of constitutive RSF parameters in numerical simulations and laboratory experiments reproduces the full spectrum of slip behaviour\u003csup\u003e9\u0026ndash;11\u003c/sup\u003e and even determines whether unstable events repeat regularly in time or evolve into a deterministic chaos\u003csup\u003e12\u003c/sup\u003e. These results are obtained without considering complex fault dynamics, which implies that second-order variations of friction can alone lead to strong bifurcations of the dynamic system. In fact, experimental results are commonly portrayed with the relatively simple dynamics of a single degree of freedom spring-slider.\u003c/p\u003e \u003cp\u003eThis interpretative framework built on the frictional properties of rocks relies on the response of the whole experimental sample, treated as homogeneous, without distinguishing processes occurring at smaller scales. However, several studies show that even minor spatial heterogeneities can have a strong impact on bulk behaviour. Heterogeneities may be found in the form of fault roughness\u003csup\u003e13\u0026ndash;18\u003c/sup\u003e, frictional parameters\u003csup\u003e19\u0026ndash;21\u003c/sup\u003e, lithology\u003csup\u003e22\u0026ndash;27\u003c/sup\u003e, and shear fabrics\u003csup\u003e28,29\u003c/sup\u003e. Fabrics are of particular interest since these characteristic structures found in experimentally deformed samples are also found in natural faults\u003csup\u003e30\u0026ndash;34\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eA growing body of evidence suggests that fault zone structures and shear fabrics are central to earthquake dynamics, yet they remain poorly understood even in the laboratory, hidden within sample assemblies and unveiled only after their entire deformation history is completed. Here, we report on novel deformation experiments to investigate in real time the spontaneous evolution of fault fabrics and their effect on slow laboratory earthquakes. A transparent assembly and a high-speed optical camera was used to capture the evolution of the experimental fault zone in tandem with its macroscopic mechanical behaviour. We report on a large variety of rupture patterns, suggesting a complexity of dynamic processes in space and time modulated by fault fabric and applied loading rates.\u003c/p\u003e"},{"header":"2. A direct view on fault processes","content":"\u003cp\u003eWe performed rock deformation experiments on the biaxial apparatus BRAVA\u003csup\u003e35\u003c/sup\u003e using a modified double-direct shear (DDS) configuration, where two simulated fault zones are deformed under controlled conditions (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). In all experiments we maintained a constant normal load of 25 kN (nominal normal stress of 10 MPa) but varied the forcing velocity of the central slider, here called loading point velocity (LPV). Details on the setup and experimental procedure are included in the Methods section. Differently from a common DDS setup, the central slider (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea) is not made in metal but by transparent Plexiglas (polymethyl methacrylate). This assembly allows recording the evolution of the mechanical properties of the material (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb) while simultaneously visualising fault zone deformation (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec). We studied fault zones composed of montmorillonite clay powder because: i) it is a relevant rock-forming mineral typically found in subduction zones associated with shallow slow slip and tectonic fault tremor\u003csup\u003e36,37\u003c/sup\u003e; ii) it is frictionally unstable at laboratory conditions comparable to the natural settings\u003csup\u003e38\u003c/sup\u003e; and iii) it has adhesive properties\u003csup\u003e39\u003c/sup\u003e that ensure strong coupling at the layer boundary and thus shear deformation internal to the fault zone.\u003c/p\u003e \u003cp\u003eOn one side of the assembly, a high speed camera films the light of a uniform light source reflected by the sample-block interface (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea), hereafter called for simplicity \u0026ldquo;fault surface\u0026rdquo;. We use the reflectivity as a proxy for the fault zone stress as well as a marker of the structures developing in the sample layer\u003csup\u003e38\u003c/sup\u003e. This is possible thanks to the phenomenon of frustrated total internal reflection, the same that makes a fingerprint appear when pressing a thumb against polished glass. Here, the quantity of light reflected is inversely proportional to the number of clay platelets pressed against the fault surface. Therefore, the luminosity of the fault surface is function of the poroelastic state of the fault zone, where porosity is function of the state of stress\u003csup\u003e40\u003c/sup\u003e. Tests to calibrate the relation between luminosity and fault zone stress state yield a yet qualitative but strong, inverse and nonlinear proportionality\u003csup\u003e38\u003c/sup\u003e (Supplementary Section SM1). In the following sections we first document the development of fault structures and then investigate how unstable slow slip starts and propagates along the fault zone.\u003c/p\u003e"},{"header":"3. Evolution of fault zone structures","content":"\u003cp\u003eOur experiments provide remarkable videos on the development of fault zone structures with shear displacement. For this documentation, we shear the initially homogeneous layer of montmorillonite powder applying a constant normal load of 25 kN and a constant LPV\u0026thinsp;=\u0026thinsp;10 \u0026micro;m/s to the central slider until the macroscopic friction (apparent friction) achieves a constant value (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb). This practice, so-called run-in, is common in rock-deformation experiments and is used to achieve steady state shear strength, i.e., when the fault is considered at macroscopic mechanical equilibrium with the imposed boundary conditions. All of our experiments exhibit consistent behaviour during the run-in, developing a specific structure that forms in a sequential pattern (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec).\u003c/p\u003e \u003cp\u003eDuring shear loading to peak friction conditions (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb, frames I-III), there is a progressive increase of stress (darkening) from the top to the bottom of the fault prior to formation of fault shear fabric structures (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec, frames I-III). Only after peak friction, a first structure forms with a local decrease of stress (increase of luminosity, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec, white arrow). This feature corresponds to the termination of oblique zones of localized shear, called Riedel\u003csup\u003e41\u003c/sup\u003e shear bands (R-shears). They are arcuate in the XY plane (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec), with convexity oriented towards the loading point. Progressively, new shear bands form with a rather regular spacing (~\u0026thinsp;5 mm, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec III-VIII), populating the experimental fault in the direction of the loading point (positive Y direction). Towards the end of the experiment, the luminosity of the fault is no longer homogeneous but shows persistent local stress minima marking the terminations of the R-shears (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). A remarkable observation is that, after 250 s of shearing (displacement of 2.5 mm), friction has already achieved steady-state conditions (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb), but the shear fabric continues to evolve with progressive population of the fault zone with R-shears (frames VI to VIII).\u003c/p\u003e \u003cp\u003eThe geometry and sequential development of the shear fabric elements is clear in both the real-time images (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) and the deformed samples recovered after the experiments (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb and Supplementary Fig. S2). The R-shears form at ~\u0026thinsp;30\u0026deg; to the shear plane and cut the entire sample layer from one boundary to the other (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). They dip roughly in the same direction of slip, representing shear localization features that produce \u0026ldquo;normal\u0026rdquo; kinematic motion (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). Our images also show a subsidiary set of R-shears that are rarely identified in previous studies. These features form at the lateral edges of the fault zone (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec), joining the first R-shear set at sharp angle (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec, black arrow). Formation of this set is accompanied by the lateral extrusion of material from the assembly.\u003c/p\u003e \u003cp\u003eThe R-shears bound low-strain regions referred to as \u0026ldquo;micro-lithons\u0026rdquo;\u003csup\u003e42\u003c/sup\u003e (ML, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) and they correspond to so-called spectator regions. In XY view, they are characterised by an upper (closer to the loading point) bright and a lower dark area (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec) suggesting a gradient in stress distribution between the R-shears (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec). In addition to R-shears, two boundary shear zones form as a consequence of shear localisation: one in proximity of the central slider (upper B-shear, \u003cem\u003eB\u003c/em\u003e\u003csub\u003e\u003cem\u003eU\u003c/em\u003e\u003c/sub\u003e) and one in proximity of the lateral forcing block (lower B-shear, \u003cem\u003eB\u003c/em\u003e\u003csub\u003e\u003cem\u003eL\u003c/em\u003e\u003c/sub\u003e). These present a well-developed S-C fabric\u003csup\u003e43\u003c/sup\u003e (close-up of Fig., 2b) and represent the high-strain zones accommodating shear motion. An important feature is that strain is partitioned between both B-shears after their development, with \u003cem\u003eB\u003c/em\u003e\u003csub\u003e\u003cem\u003eL\u003c/em\u003e\u003c/sub\u003e being mostly active close to the older R-shears and \u003cem\u003eB\u003c/em\u003e\u003csub\u003e\u003cem\u003eU\u003c/em\u003e\u003c/sub\u003e close to the younger R-shears. This observation and further information on the development of the fault structures are detailed in Supplementary Section SM3.\u003c/p\u003e"},{"header":"4. Slip instabilities","content":"\u003cp\u003eWe now explore the influence of shear fabric on the genesis (nucleation), propagation, and arrest of slip instabilities. In this set of experiments, we repeat the procedure described above and, after the run-in, we vary the LPV between 0.01 and 10 \u0026micro;m/s and observe unstable slip behaviour occurring in stick slip cycles. These slow laboratory earthquake events occur only for LPVs below 7.5 \u0026micro;m/s, finding agreement with previous works on slip stability and forcing velocity dependence\u003csup\u003e38,44\u003c/sup\u003e. The cycles are composed of a \u0026ldquo;stick\u0026rdquo; stage, characterised by a nearly linear increase of friction in time, and a \u0026ldquo;slow slip\u0026rdquo; stage, characterised by spontaneous slip accompanied by a quasi-dynamic drop in friction (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). During slow slip events, the luminosity changes only subtly across the fault surface (variations\u0026thinsp;\u0026lt;\u0026thinsp;1% of the sensitivity range). In fact, the position of the main fault structures as well as the distribution of stress (luminosity) do not vary substantially throughout the stick-slip series (see Supplementary Section SM3). Nevertheless, the slight luminosity change recorded during the friction drops shows negative correlation with the friction (balloon in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea) and clearly illuminates the passage of a rupture front across the fault. Hereby we define \u0026ldquo;rupture\u0026rdquo; as the phenomenological description of the spatial-temporal variation of mechanical attributes (e.g., stress and luminosity) in response of the propagation of an instability. A quick visualisation of the ruptures nucleating and propagating along the shear direction (Y) is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec, which depicts the stressing rates (time derivative of luminosity) as a function of time (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). We can immediately see that a plethora of slip phenomena emerges, modulated by the LPV.\u003c/p\u003e \u003cp\u003eNearly periodic stick-slip is observed at LPV\u0026thinsp;\u0026lt;\u0026thinsp;2.5 \u0026micro;m/s with a transition to more complex, aperiodic events, and eventually aseismic slip for higher LPV. All of the slow slip events present some differences in rupture patterns (Supplementary Section SM4), but two main groups with specific characteristics emerge (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). The first, \u0026ldquo;regular\u0026rdquo; slow slip events, is observed mainly at LPVs ranging between 0.25 and 2.5 \u0026micro;m/s. The second, \u0026ldquo;irregular\u0026rdquo; slow slip events, is observed at LPV\u0026thinsp;\u0026lt;\u0026thinsp;0.25 \u0026micro;m/s. These two categories are of particular interest and are described in detail in the following sections. At higher LPVs, the slip behaviour quickly transits to chaotic stick-slip. Chaotic events are irregularly spaced in time, present variable friction drops, overlap in time, and are characterised by complicated rupture patterns (compare regular stick-slip with chaotic stick slip in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). Together with each major rupture propagating in the direction of shear (dark blue traces), numerous smaller events nucleate at other locations and interfere with each other (light red-blue patterns), causing a chaotic stressing rate across the experimental fault (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). The clear fluctuations of friction visible during regular ruptures (t\u0026thinsp;\u0026lt;\u0026thinsp;613 s, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec) disappear when multiple instances nucleate and propagate in short time intervals (interval 616\u0026ndash;617 s) or synchronously (interval 626\u0026ndash;629 s). At LPV\u0026thinsp;\u0026gt;\u0026thinsp;10 \u0026micro;m/s chaotic stick-slip quenches to stable sliding. Surprisingly, despite the measured constant friction, luminosity changes reveal that ruptures still continue to nucleate and propagate in the fault zone (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). These ruptures present small stressing rates, (lighter colours, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec), random propagation patterns, and irregular timing. These features, occurring in concert, masks the existence of instabilities from the mechanical time series to produce what macroscopically appears as aseismic shear motion.\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e4.1. Regular slow slip\u003c/h2\u003e \u003cp\u003eRegular slow slip events starts at the boundary of the uppermost micro-lithon (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec,d), close to the subsidiary R-shear next to R9 (circle in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eg), where stress heterogeneities are larger. From this region, the rupture front propagates in the X (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef) and Y (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee) directions. Ruptures propagate as wrinkle-like pulses\u003csup\u003e38\u003c/sup\u003e with dimension smaller than the separation between the Riedel shears (\u0026lt;\u0026thinsp;5 mm, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee,f). The width of the wrinkle appears to be proportional to the propagation velocity of the pulse\u003csup\u003e45\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eWe observed a range of rupture speeds depending on propagation direction and shear fabric. For instance, in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb the rupture migrates at ~\u0026thinsp;110 mm/s in the horizontal direction (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef), remaining mostly confined within the uppermost micro-lithon (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea, frames I-VII, and 5c), above R9. Most of the macroscopic friction drop occurs during this stage (from black to white arrow, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb). Note that the slow rupture velocity (7 mm/s), reported in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed up to this point, is only apparent because the averaging of apparent stress change in the X direction does not capture motion in the Y direction. When the rupture encounters the rightmost boundary (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef, white arrow, and 5a, frame VII), it overcomes the bounding Riedel and propagates in the shear direction, relieving stress across the entire fault surface. In the meantime, additional minor ruptures propagate within the uppermost micro-lithon (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef, blue stripes past the white arrow), contributing to a further local stress drop. Rupture propagation is faster in the shear (Y) direction (v\u0026thinsp;~\u0026thinsp;225 mm/s) than along the X-direction. Our measurements show a range of interesting spatiotemporal complexity including momentary phases of arrest and deceleration at each local stress maximum within the micro-lithons (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee, dashed lines). The final rupture arrest occurs when the front approaches the lowermost boundary of the sample. All of these so-called regular rupture events propagate across the entire fault surface.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e4.2. Complex ruptures and irregular slow slip\u003c/h2\u003e \u003cp\u003eIrregular slow slip events occur at LPVs\u0026thinsp;=\u0026thinsp;100 nm/s. These differ from regular slow slip events by one or more features such as: two-stepped friction drops, partial ruptures (confined ruptures), and/or complex propagation patterns. In three consecutive events occurring at low LPV (100 nm/s, Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea,c), the first two nucleate at the topmost micro-lithon and do not rupture the entire fault (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec), leading to an increase of stress in the unruptured lower half of the sample. This region is then ruptured by a third event nucleating inside of it (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ed, black arrow). Unlike regular slow slip events, the rupture slowly expands (~\u0026thinsp;14 mm/s, Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef) from the nucleation site (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e, frames I-IV) and then accelerates to a much faster propagation speed.\u003c/p\u003e \u003cp\u003eWe observe that the nucleation stage is composed of a small rupture that induces the nucleation of another one in the neighbouring micro-lithon across R3 (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e, white arrow, frames V-VI). Notably, the event behaves as a wrinkle-like rupture even during the nucleation phase since the stressing rate is restricted to a thin region close to the rupture front (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ee,f). After the nucleation stage, the rupture expands at faster rates (\u0026gt;\u0026thinsp;50 mm/s, Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ee,f) towards the sides of the fault and reaches the topmost micro-lithon (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e, frames VI-IX), then rapidly propagates across it from left to right (frames IX-X) and continues to the bottom left (frames X-XII). This rupture shows a clockwise rupture pattern (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ed), with faster propagation at the boundaries of the fault zone. Again, arrest coincides with a decrease of rupture front speed (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ed). The rupture event clearly correlates with the first half of the slip event, which is modulated in two steps of roughly equal friction drop (starting at the white and orange arrows in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eb). Only the first step corresponds to the variation in luminosity recorded at the fault surface. This effect is probably linked to the diachronous rupture of each side of the double-direct shear configuration. This is a common feature of irregular slow slip events observed at low LPVs (\u0026lt;\u0026thinsp;0.5 \u0026micro;m/s). The description of irregular slow slip events applies as well to the main events recorded during chaotic slow slip (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb,c).\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eOur experiments qualitatively reveal the state of stress, the development of self-organized structures, and how these control the nucleation, propagation, and arrest of slip instabilities. Here we provide an explanation of this self-organized shear fabric.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e5.1. Fabric development and spontaneous redistribution of stress on-fault\u003c/h2\u003e \u003cp\u003eAt the beginning of each experiment, we force the microstructural reorganisation of the fault structure from the initial homogeneity (uniaxially pressed powder) to an optimal shear fabric (localised shear zones). This configuration is reproducible across experiments and consists of the formation of two types of shear bands (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea): i) two boundary (B-) shear zones forming at the fault surface (B\u003csub\u003eU\u003c/sub\u003e) and close to the lateral block (B\u003csub\u003eL\u003c/sub\u003e,), and ii) several Riedel shears (R1-R9), which together bound sigmoidal low strain domains, the micro-lithons (ML). The micro-lithons are consistently ten in number and with rather regular spacing, suggesting that there is a geometrical control between sample dimension (layer thickness, contact area), experimental conditions (run-in velocity and normal load), and optimal stress distribution\u003csup\u003e31,41,46\u003c/sup\u003e. The shear strain is accommodated mainly by both B-shears and is partitioned between the two through the activity of R-shears (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eThe formation of each R-shear relieves stress from the micro-lithon it bounds, with the largest stress decrease recorded in its proximity. Immediately after the formation of the R-shear, stress increases inside the micro-lithon close to the previous R-shear (Supplementary Section SM3) producing the small-scale fluctuations of stress in the Y direction (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). This effect recalls the way stress is relieved at asperities when material is plucked-out in response to adhesive wear\u003csup\u003e47,48\u003c/sup\u003e, transferring the strain from the sliding surface to a new, deeper slip zone. In our case, the lower B-shear.\u003c/p\u003e \u003cp\u003eIn the fault’s final configuration, the top of the R-shears separates the least stressed regions from neighbouring loci of stress concentration (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea). The formation of R-shears seems thus a necessary mechanism that redistributes stress across the fault while boundary shear zones creep to accommodate stable sliding. This effect is extremely important in producing heterogeneities that determine the initiation and mode of fault instability\u003csup\u003e49,50\u003c/sup\u003e. Given the impossibility to form further R-shears towards the loading point, the last formed micro-lithon is the most stressed region of the fault and presents large stress heterogeneities at its boundaries (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea), especially in correspondence of a low stress embayment. This is the area mostly affected by the loading conditions and that becomes inherently unstable.\u003c/p\u003e \u003cp\u003eLastly, it is important to point out an important feature. Although a fault zone structure fully populated by R-shears can be reasonably seen as a final state of mechanical equilibrium, the analysis of both time series and footages does not confirm this interpretation. The bulk friction has in fact already achieved steady state conditions (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb) after the formation of the fifth R-shear, as well as the average luminosity. This means that the continuous evolution of the inner structure of the fault zone throughout the experiment, which is responsible of the spatial reorganisation of stress, leads to a first-order mechanical equilibrium (e.g., macroscopic friction, stress), characterized by the contribution of all points of the fault (i.e., bulk response). This is what is commonly observed with the single observation point (load cell) of the experimental apparatus. However, second order variations may still evolve with strain within the sample, spontaneously following the fabric development, at a scale smaller than the bulk fault, without requiring further energy dissipation. This has important implications on the rate-dependence of friction, sliding stability, and their relationship with strain and strain rate\u003csup\u003e51–53\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e5.2. Nucleation of slip instabilities\u003c/h2\u003e \u003cp\u003eAs much as all slow slip events display rather variable rupture patterns (Supplementary Section SM4), most initiate at the highest stress heterogeneity (regular slow slip events, Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eb), located at the bounding R-shear of the uppermost micro-lithon (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Irregular events (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eb) may also originate at smaller stress heterogeneities inside the fault (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eg). The ruptures thus start in regions where sharp local stress gradients are observed, but the absolute entity of the local stress seems subordinate to the intrinsic instability of the area and memory effects. This is clear since the overall stress state of the fault remains remarkably unchanged across several creep-slip cycles (Supplementary Section SM3) while minor redistribution of stress can disrupt the previous rupture patterns and favour nucleation in other regions (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec). For this reason, irregular slip events may emerge as a consequence of incomplete ruptures or overlapping chaotic rupture patterns (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec).\u003c/p\u003e \u003cp\u003eIn our experiments, most events display no clear distinction between a nucleation and a dynamic phase \u003cem\u003esensu\u003c/em\u003e Ohnaka and Shen\u003csup\u003e54\u003c/sup\u003e. Instead, different rupture speed are recorded depending on the direction of the rupture front (slower in the X direction, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee,f), while a fictious nucleation stage may emerge due to a specific choice of directional averaging of the fault luminosity (e.g., Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed). This highlights the importance of the observation point on the fault (e.g., camera or other sensors), here privileged by the planar view. A distinct, slow nucleation phase is clearly observed only when the instability initiates in more central areas of the fault (irregular events, Figs.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eb). Given the general lack of a distinguishable, systematic nucleation phase, a reliable measure of the nucleation size for use in numerical models\u003csup\u003e55\u003c/sup\u003e is unattainable. Neither can nucleation size be calculated theoretically since the material is rate-strengthening\u003csup\u003e38\u003c/sup\u003e and the stress distribution is heterogeneous. Nevertheless, since we observe partitioning of strain between the different shear zones, it is reasonable to assume that slow slip events tend to nucleate away from the B-shear zones, for example, inside the R-shears (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ec). In support of this hypothesis, we recorded the formation of plumose structures along R-shears formed in over-consolidated clay experiments (Supplementary Section SM5). These structures indicate the energetic formation of R-shear through fracture-like mechanisms, potentially associated with the nucleation of slip instabilities.\u003c/p\u003e \u003cp\u003eThe beginning of slow slip instabilities in our experiments is phenomenologically related to the formation and migration of dilatant waves (increased luminosity) from a source region of stress heterogeneity, such as those forming at elastic bi-material interfaces\u003csup\u003e56,57\u003c/sup\u003e and widely known as Schallamach waves\u003csup\u003e58\u003c/sup\u003e. These are self-healing detachment pulses that start and propagate in a similar manner to what observed in our experiments. The similarity implies that the instability grows as the local loss of adhesion, which manifests in clay as region with growing porosity (increased luminosity) and possible detachment of platelets from the central slider. To corroborate this mechanistic interpretation, we highlight the existence of different regimes at different loading point velocities (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), as observed in analogue tests on bi-material interfaces\u003csup\u003e56,58\u003c/sup\u003e. Finally, it was demonstrated that slow slip self-healing pulses can emerge as function of stiffness contrast and finite size of the setup at elastic-frictional bi-material interfaces\u003csup\u003e59\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn our experiments it is difficult to adapt existing models for a number of reasons: i) deformation occurs in a thin boundary shear zone (\u0026lt; 100 µm), which is not a proper bi-material interface, ii) an important fraction of deformation is accommodated by creep, and iii) instabilities form independently of the stiffness of the central slider\u003csup\u003e38\u003c/sup\u003e. In this light, the intrinsic elastic properties (bending) of the clay platelets\u003csup\u003e60\u003c/sup\u003e, the preferred frictional sliding along the S-C foliation\u003csup\u003e38\u003c/sup\u003e, and the sub-mm stress heterogeneities on the fault zone may play a far more important role in nucleation than the bulk properties of the sample and experimental setup.\u003c/p\u003e \u003cp\u003eOur observations suggest that slip instabilities nucleate within the gouge layer as a result of stress heterogeneity determined by shear fabric evolution. In other words, nucleation of dynamic instabilities is a volumetric process driven by shearing and sensitive to tiny variations of the loading rate. This implies that the macroscopic description of dynamic slip instabilities as retrieved from the single observation point on the load cell in classic rock deformation experiments does not resolve the complex spatial-temporal evolution of sliding at smaller wavelengths. Thus, the macroscopic slip of the whole sample is the result of strain partitioning within the volume.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e5.3. Propagation, arrest, and quenching of slow slip\u003c/h2\u003e \u003cp\u003eBeside nucleation, propagation, and arrest, the rupture pattern is strongly influenced by stress distribution and fault structures. All ruptures tend to be initially confined within the micro-lithons where they originate, with the high-stressed areas acting as barriers\u003csup\u003e17\u003c/sup\u003e. These barriers are overcome when the macroscopic rupture breaks perpendicular to the ML boundary and especially when it joins the low-stressed fault edges (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e frames VI-VII). The same effect of the fault edge on rupture propagation is reported in analogue experiments in plastic faults\u003csup\u003e61\u003c/sup\u003e. We observe that each jump across the R-shears can be met with some latency and causes a progressive deceleration of the rupture edge (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee). This effect is observed especially in the region where the earliest R-shears formed, the same area of the fault where some partial ruptures stop at low LPVs (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec). This effect is probably related to the region of strain partitioning between the upper and lower boundary (B-)shear zones (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea, Supplementary Section SM3). Here, especially at low LPVs, the rupture may “sink” into the R-shears, dissipating the available elastic energy into distributed permanent deformation.\u003c/p\u003e \u003cp\u003eWith increasing LPV the bulk frictional instability of the experimental fault is quenched towards stable sliding (LPV \u0026gt; 7.5 µm/s, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). This, however, is not related to the disappearance of slow slip instabilities. Slow, pulse-like ruptures continue to nucleate especially at the large stress heterogeneity where regular events originate (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec) and, at the same time, propagate while other subsidiary ruptures still sweep across the fault and nucleate in other locations. The bulk stability recorded in the time series (apparent friction) is then the result of combined effect of: i) interfering rupture patterns, ii) recurrence time smaller than that required for a complete rupture, iii) inhomogeneous redistribution of stress across cycles, and iv) different nucleation sites. We argue that the inherent instability of the material is also related to the sample size\u003csup\u003e14,59\u003c/sup\u003e, which signifies that the embryonal phases of an earthquake (small scale processes) is severely affected by the role of heterogeneities across different scales. For example, the fault patch acting as an asperity in nature might vary substantially depending on larger scale geological features\u003csup\u003e62\u003c/sup\u003e while microstructures explored here maintain the same dimension independently of the setting.\u003c/p\u003e \u003cp\u003eOur observations suggest that macroscopic slip associated with the friction drop is due to strain partitioning between B- and R-shear within the fault zone characterized by spatial-temporal evolution defining the so-called rupture patterns. Rupture involves the evolution of fabric over the entire gouge layer and is affected by small-scale heterogeneities and fluctuations of the loading velocity. While this does not preclude the interpretation of macroscopic dynamic slip instabilities, it complicates the interpretation of mechanical processes acting at the micro scale.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e5.4. Concluding remarks\u003c/h2\u003e \u003cp\u003eOur experiments allowed to image the dynamics of slow slip in laboratory clay faults. The experimental fault evolves with slip to form a reproducible inner structure composed of localised shear zones and low strain domains. This structure is in continuous evolution with strain, despite the apparent early achievement of bulk mechanical equilibrium (steady state conditions). This structure represents the self-organisation and redistribution of stress across the fault zone (i.e., the gouge layer), which has important effects on the genesis and propagation of unstable events. At low forcing velocities, slow slip events nucleate due to adhesion loss at local stress heterogeneities, which are not necessarily the largest inside the fault zone. The nucleation stage does not show a characteristic, recognisable length scale, and it is associated with stress heterogeneities in the gouge volume. The slip instability inferred from the mechanical data is indeed associated with the spatial-temporal evolution of a strain pulse. Slip instabilities propagate as self-healing dilatant pulses causing only minor redistribution of stress across the fault zone but do introduce memory effects in the regions they swipe through. As a consequence, each slow slip event is different and can follow complex rupture patterns, especially at both low (LPV \u0026lt; 0.25 µm/s) and high (LPV \u0026gt; 2.5 µm/s) loading rates. Even when the fault is macroscopically stable (steady-state apparent friction) at high loading rates, slip pulses continue to nucleate and propagate, but become chaotic and of smaller entity. The dynamics of slow slip appear thus complex even in experimental small faults, questioning whether constitutive laws for earthquake nucleation and frictional instability may represent processes across all scales, which would require proper addressing.\u003c/p\u003e \u003c/div\u003e"},{"header":"6. Methods","content":"\u003cp\u003eRock deformation experiments were carried out in BRAVA\u003csup\u003e35\u003c/sup\u003e, a bi-triaxial rock deformation apparatus (Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003eM1\u003c/span\u003ea) hosted at the HPHT laboratory at the Istituto Nazionale di Geofisica e Vulcanologia (Rome, Italy). The experimental setup consists of a double direct shear sample holder, composed of two lateral forcing blocks compressing two parallel sample layers against a transparent Plexiglas central slider (Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003eM1\u003c/span\u003eb). The sample has an initial thickness of 5 mm and is compressed on a nominal squared area of contact of 50 by 50 mm (Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003eM1\u003c/span\u003eb). The samples were humidified beforehand at 100% room humidity for 12 h, allowing a thin film of water to be adsorbed on the surface of the platelets. The contact surface between the forcing blocks and the sample is flat. Normal stress and shear motion are controlled by two orthogonal hydraulic pistons (Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003eM1\u003c/span\u003ea) and are transferred to the sample by the lateral blocks and the central slider (Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003eM1\u003c/span\u003eb), respectively. On one side, a shielded led lamp provides with a homogeneous white light source. On the other, a high speed camera (NAC Memrecam) films the light reflected by the sample-block interface (Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003eM1\u003c/span\u003ec). Our video measurements have a temporal resolution of 0.02 to 0.0067 s and a spatial resolution of ~ 50 µm and ~ 130 µm in the shear (S.D.) and horizontal (H.D.) directions, respectively, determined by the angle of view (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb).\u003c/p\u003e\u003cp\u003eVideo analysis was carried with a custom MATLAB© code performing simple operations on raster frames (.tif) extracted from the videos, including static and rolling background removal. In the static background removal (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec-d, \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea, \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec-d,\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ea) the background image corresponds to the first frame of interest of the video and is systematically subtracted from the following frames. The resulting video displays changes in the luminosity of the fault surface (quantified in arbitrary units, a.u.) with respect to the initial frame. The change in reflectivity obtained with static background removal qualitatively shows the stress change on the fault surface. The rolling back- ground removal (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef-e, \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef-e) is operated by subtracting to each frame the frame preceding by a fixed separation of time. The optimal separation for visualization has been selected by trial and error. The rolling back-ground removal qualitatively indicates the stress change rate on the fault surface, which is the discrete derivative of the stress change. Linear pixel-wise averaging in the X and Y directions were used on selected areas to image the propagation of rupture fronts. This analysis could not provide quantitative results due to the strong non-linearity between luminosity, stress, and poroelastic state of the material (Supplementary Section SM1). Nevertheless, our analysis show robust inverse correlation between luminosity and stress, corroborating the results and supporting the discussion.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003c/em\u003eConceptualization: G.P. and G.V. Data curation, formal analysis, investigation, methodology: G.P., J.T., and G.V. Writing\u0026mdash;original draft: G.P. Writing\u0026mdash;review and editing: G.V., J.T., M.C., C.M., C.C. Resources and funding acquisition: G.P, M.C. Visualization: G.P.; Software: G.P., J.T. Project administration: G.P.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e-\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRaw data files are available on Zenodo \u003cem\u003e(files will be uploaded and the statement updated upon finalisation of the manuscript).\u003c/em\u003e\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eB\u0026uuml;rgmann, R. (2018). The geophysics, geology and mechanics of slow fault slip. Earth and Planetary Science Letters, 495, 112\u0026ndash;134. https://doi.org/10.1016/j.epsl.2018.04.062\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeng, Z., \u0026amp; Gomberg, J. (2010). An integrated perspective of the continuum between earthquakes and slow-slip phenomena. In Nature Geoscience (Vol. 3, Issue 9, pp. 599\u0026ndash;607). Nature Publishing Group. https://doi.org/10.1038/ngeo940\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKirkpatrick, J. D., Fagereng, \u0026Aring;., \u0026amp; Shelly, D. R. (2021). Geological constraints on the mechanisms of slow earthquakes. Nature Reviews Earth \u0026amp; Environment 2021 2:4, 2(4), 285\u0026ndash;301. https://doi.org/10.1038/s43017-021-00148-w\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIde, S., \u0026amp; Beroza, G. C. (2023). Slow earthquake scaling reconsidered as a boundary between distinct modes of rupture propagation. Proceedings of the National Academy of Sciences of the United States of America, 120(32). https://doi.org/10.1073/pnas.2222102120\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrace, W. F., \u0026amp; Byerlee, J. D. (1966). Stick-slip as a mechanism for earthquakes. Science, 153(3739), 990\u0026ndash;992. https://doi.org/10.1126/science.153.3739.990\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDieterich, J. H. (1972). Time-dependent friction in rocks. Journal of Geophysical Research, 77(20), 3690\u0026ndash;3697. https://doi.org/10.1029/jb077i020p03690\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuina, A. (1983). Slip instability and state variable friction laws. Journal of Geophysical Research, 88(B12), 10359\u0026ndash;10370. https://doi.org/10.1029/JB088iB12p10359\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarone, C. (1998). Laboratory-Derived Friction Laws and Their Application To Seismic Faulting. Annual Review of Earth and Planetary Sciences, 26(1), 643\u0026ndash;696. https://doi.org/10.1146/annurev.earth.26.1.643\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeeman, J. R., Saffer, D. M., Scuderi, M. M., \u0026amp; Marone, C. (2016). Laboratory observations of slow earthquakes and the spectrum of tectonic fault slip modes. Nature Communications, 7, 1\u0026ndash;6. https://doi.org/10.1038/ncomms11104\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScuderi, M. M., Collettini, C., Viti, C., Tinti, E., \u0026amp; Marone, C. (2017). Evolution of shear fabric in granular fault gouge from stable sliding to stick slip and implications for fault slip mode. Geology, 45(8), 731\u0026ndash;734. https://doi.org/10.1130/G39033.1\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIm, K., Saffer, D., Marone, C., \u0026amp; Avouac, J. P. (2020). Slip-rate-dependent friction as a universal mechanism for slow slip events. Nature Geoscience, 13(10), 705\u0026ndash;710. https://doi.org/10.1038/s41561-020-0627-9\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarbot, S. (2019). Slow-slip, slow earthquakes, period-two cycles, full and partial ruptures, and deterministic chaos in a single asperity fault. Tectonophysics, 768(March), 228171. https://doi.org/10.1016/j.tecto.2019.228171\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSelvadurai, P. A., \u0026amp; Glaser, S. D. (2015). Laboratory-developed contact models controlling instability on frictional faults. Journal of Geophysical Research: Solid Earth, 120(6), 4208\u0026ndash;4236. https://doi.org/10.1002/2014JB011690\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarbord, C. W. A., Nielsen, S. B., De Paola, N., \u0026amp; Holdsworth, R. E. (2017). Earthquake nucleation on rough faults. Geology, 45(10), 931\u0026ndash;934. https://doi.org/10.1130/G39181.1\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePassel\u0026egrave;gue, F. X., Almakari, M., Dublanchet, P., Barras, F., Fortin, J., \u0026amp; Violay, M. (2020). Initial effective stress controls the nature of earthquakes. Nature Communications, 11(1), 1\u0026ndash;8. https://doi.org/10.1038/s41467-020-18937-0\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGounon, A., Latour, S., Letort, J., \u0026amp; El Arem, S. (2022). Rupture Nucleation on a Periodically Heterogeneous Interface. Geophysical Research Letters, 49(20), 1\u0026ndash;9. https://doi.org/10.1029/2021GL096816\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCebry, S. B. L., Sorhaindo, K., \u0026amp; McLaskey, G. C. (2023). Laboratory Earthquake Rupture Interactions With a High Normal Stress Bump. Journal of Geophysical Research: Solid Earth, 128(11), 1\u0026ndash;24. https://doi.org/10.1029/2023JB027297\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, S., Fukuyama, E., Yamashita, F., Kawakata, H., Mizoguchi, K., \u0026amp; Takizawa, S. (2023). Fault strength and rupture process controlled by fault surface topography. Nature Geoscience, 16(1), 94\u0026ndash;100. https://doi.org/10.1038/s41561-022-01093-z\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCorbi, F., Funiciello, F., Moroni, M., Van Dinther, Y., Mai, P. M., Dalguer, L. A., \u0026amp; Faccenna, C. (2013). The seismic cycle at subduction thrusts: 1. Insights from laboratory models. Journal of Geophysical Research: Solid Earth, 118(4), 1483\u0026ndash;1501. https://doi.org/10.1029/2012JB009481\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRubino, V., Lapusta, N., \u0026amp; Rosakis, A. J. (2022). Intermittent lab earthquakes in dynamically weakening fault gouge. Nature, 606(7916), 922\u0026ndash;929. https://doi.org/10.1038/s41586-022-04749-3\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong, J. Y., \u0026amp; McLaskey, G. C. (2024). Laboratory Earthquake Ruptures Contained by Velocity Strengthening Fault Patches. Journal of Geophysical Research: Solid Earth, 129(4), 1\u0026ndash;16. https://doi.org/10.1029/2023JB028509\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNiemeijer, A. R., \u0026amp; Spiers, C. J. (2005). Influence of phyllosilicates on fault strength in the brittle-ductile transition: Insights from rock analogue experiments. Geological Society Special Publication, 245, 303\u0026ndash;327. https://doi.org/10.1144/GSL.SP.2005.245.01.15\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTembe, S., Lockner, D. A., \u0026amp; Wong, T.-F. (2010). Effect of clay content and mineralogy on frictional sliding behavior of simulated gouges: Binary and ternary mixtures of quartz, illite, and montmorillonite. Journal of Geophysical Research, 115(B3), B03416. https://doi.org/10.1029/2009JB006383\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTesei, T., Collettini, C., Barchi, M. R., Carpenter, B. M., \u0026amp; Di Stefano, G. (2014). Heterogeneous strength and fault zone complexity of carbonate-bearing thrusts with possible implications for seismicity. Earth and Planetary Science Letters, 408, 307\u0026ndash;318. https://doi.org/10.1016/j.epsl.2014.10.021\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBedford, J. D., Faulkner, D. R., \u0026amp; Lapusta, N. (2022). Fault rock heterogeneity can produce fault weakness and reduce fault stability. Nature Communications, 13(1), 1\u0026ndash;7. https://doi.org/10.1038/s41467-022-27998-2\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArts, J. P. B., Niemeijer, A. R., Drury, M. R., Willingshofer, E., \u0026amp; Matenco, L. C. (2024). The frictional strength and stability of spatially heterogeneous fault gouges. Earth and Planetary Science Letters, 628(January). https://doi.org/10.1016/j.epsl.2024.118586\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuggieri, R., Scuderi, M. M., Trippetta, F., Tinti, E., Brignoli, M., Mantica, S., Petroselli, S., Osculati, L., Volont\u0026egrave;, G., \u0026amp; Collettini, C. (2021). The role of shale content and pore-water saturation on frictional properties of simulated carbonate faults. Tectonophysics, 807. https://doi.org/10.1016/j.tecto.2021.228811\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScuderi, M. M., Tinti, E., Cocco, M., \u0026amp; Collettini, C. (2020). The Role of Shear Fabric in Controlling Breakdown Processes During Laboratory Slow-Slip Events. Journal of Geophysical Research: Solid Earth, 125(11), 1\u0026ndash;23. https://doi.org/10.1029/2020JB020405\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePozzi, G., Scuderi, M. M., Tinti, E., Nazzari, M., \u0026amp; Collettini, C. (2022). The Role of Fault Rock Fabric in the Dynamics of Laboratory Faults. Journal of Geophysical Research: Solid Earth, 127(6). https://doi.org/10.1029/2021JB023779\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRutter, E. H., Maddock, R. H., Hall, S. H., \u0026amp; White, S. H. (1986). Comparative microstructures of natural and experimentally produced clay-bearing fault gouges. Pure and Applied Geophysics PAGEOPH, 124(1\u0026ndash;2), 3\u0026ndash;30. https://doi.org/10.1007/BF00875717\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLogan, J. M., Dengo, C. A., Higgs, N. G., \u0026amp; Wang, Z. Z. (1992). Fabrics of experimental fault zones: Their development and relationship to mechanical behavior BT - Fault Mechanics and Transport Properties of Rocks. In Fault Mechanics and Transport Properties of Rocks: Vol. Internatio (Issue 2).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChester, F. M., \u0026amp; Logan, J. M. (1987). Composite planar fabric of gouge from the Punchbowl fault, California. Journal of Structural Geology, 9(5\u0026ndash;6), 621-IN6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFagereng, \u0026Aring;., \u0026amp; Sibson, R. H. (2010). M\u0026eacute;lange rheology and seismic style. Geology, 38(8), 751\u0026ndash;754. https://doi.org/10.1130/G30868.1\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollettini, C., Niemeijer, A., Viti, C., Smith, S. A. F., \u0026amp; Marone, C. (2011). Fault structure, frictional properties and mixed-mode fault slip behavior. Earth and Planetary Science Letters, 311(3\u0026ndash;4), 316\u0026ndash;327. https://doi.org/10.1016/j.epsl.2011.09.020\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollettini, C., Di Stefano, G., Carpenter, B., Scarlato, P., Tesei, T., Mollo, S., Trippetta, F., Marone, C., Romeo, G., \u0026amp; Chiaraluce, L. (2014). A novel and versatile apparatus for brittle rock deformation. International Journal of Rock Mechanics and Mining Sciences, 66, 114\u0026ndash;123. https://doi.org/10.1016/j.ijrmms.2013.12.005\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVrolijk, P. (1990). On the mechanical role of smectite in subduction zones. Geology, 18(8), 703\u0026ndash;707. https://doi.org/10.1130/0091-7613(1990)018\u0026lt;0703:OTMROS\u0026gt;2.3.CO;2\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaffer, D. M., \u0026amp; Marone, C. (2003). Comparison of smectite- and illite-rich gouge frictional properties: application to the updip limit of the seismogenic zone along subduction megathrusts. Earth and Planetary Science Letters, 215(1\u0026ndash;2), 219\u0026ndash;235. https://doi.org/10.1016/S0012-821X(03)00424-2\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVolpe, G., Collettini, C., Taddeucci, J., Marone, C., \u0026amp; Pozzi, G. (2024). Frictional instabilities in clay illuminate the origin of slow earthquakes. Science Advances, 10(26), 1\u0026ndash;11. https://doi.org/10.1126/sciadv.adn0869\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiller, C. M. (2011). Adhesion and the Surface Energy Components of Natural Minerals and Aggregates. Texas A\u0026amp;M University.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMondol, N. H., Bj\u0026oslash;rlykke, K., \u0026amp; Jahren, J. (2008). Experimental compaction of clays: Relationship between permeability and petrophysical properties in mudstones. Petroleum Geoscience, 14(4), 319\u0026ndash;337. https://doi.org/10.1144/1354-079308-773\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTchalenko, J. S. (1968). The evolution of kink-bands and the development of compression textures in sheared clays. Tectonophysics, 6(2), 159\u0026ndash;174. https://doi.org/10.1016/0040-1951(68)90017-6\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHaines, S. H., Kaproth, B., Marone, C., Saffer, D., \u0026amp; Van der Pluijm, B. (2013). Shear zones in clay-rich fault gouge: A laboratory study of fabric development and evolution. Journal of Structural Geology, 51, 206\u0026ndash;225. https://doi.org/10.1016/j.jsg.2013.01.002\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVolpe, G., Pozzi, G., \u0026amp; Collettini, C. (2022). Y-B-P-R or S-C-C \u0026prime; ? Suggestion for the nomenclature of experimental brittle fault fabric in phyllosilicate-granular mixtures. Journal of Structural Geology, 165(July), 104743. https://doi.org/10.1016/j.jsg.2022.104743\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeeman, J. R., Marone, C., \u0026amp; Saffer, D. M. (2018). Frictional Mechanics of Slow Earthquakes. Journal of Geophysical Research: Solid Earth, 123(9), 7931\u0026ndash; 7949. https://doi.org/10.1029/2018JB015768\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNielsen, S., \u0026amp; Madariaga, R. (2003). On the self-healing fracture mode. Bulletin of the Seismological Society of America, 93(6), 2375\u0026ndash;2388. https://doi.org/10.1785/0120020090\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAtmaoui, N., Kukowski, N., St\u0026ouml;ckhert, B., \u0026amp; K\u0026ouml;nig, D. (2006). Initiation and development of pull-apart basins with Riedel shear mechanism: Insights from scaled clay experiments. International Journal of Earth Sciences, 95(2), 225\u0026ndash;238. https://doi.org/10.1007/s00531-005-0030-1\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTesei, T., Carpenter, B. M., Giorgetti, C., Scuderi, M. M., Sagy, A., Scarlato, P., \u0026amp; Collettini, C. (2017). Friction and scale-dependent deformation processes of large experimental carbonate faults. Journal of Structural Geology, 100, 12\u0026ndash;23. https://doi.org/10.1016/j.jsg.2017.05.008\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScholz, C. H. (2019) The Mechanics of Earthquakes and Faulting. 3rd ed. Cambridge University Press.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMildon, Z. K., Roberts, G. P., Faure Walker, J. P., \u0026amp; Toda, S. (2019). Coulomb pre-stress and fault bends are ignored yet vital factors for earthquake triggering and hazard. Nature Communications, 10(1), 1\u0026ndash;9. https://doi.org/10.1038/s41467-019-10520-6\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBen-zion, Y., \u0026amp; Dresen, G. (2022). A Synthesis of Fracture, Friction and Damage Processes in Earthquake Rupture Zones. Pure and Applied Geophysics, 179(12), 4323\u0026ndash;4339. https://doi.org/10.1007/s00024-022-03168-9\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeeler, N. M., Tullis, T. E., Blanpied, M. L., \u0026amp; Weeks, J. D. (1996). Frictional behavior of large displacement experimental faults. Journal of Geophysical Research: Solid Earth, 101(B4), 8697\u0026ndash;8715. https://doi.org/10.1029/96JB00411\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMair, K., \u0026amp; Marone, C. (1999). Friction of simulated fault gouge for a wide range of velocities and normal stresses. Journal of Geophysical Research: Solid Earth, 104(B12), 28899\u0026ndash;28914. https://doi.org/10.1029/1999JB900279\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNo\u0026euml;l, C., Giorgetti, C., Collettini, C., \u0026amp; Marone, C. (2024). The effect of shear strain and shear localization on fault healing. Geophysical Journal International, 236(3), 1206\u0026ndash;1215. https://doi.org/10.1093/gji/ggad486\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOhnaka, M., \u0026amp; Shen, L. (1999). Scaling of the shear rupture process from nucleation to dynamic propagation: Implications of geometric irregularity of the rupturing surfaces. Journal of Geophysical Research: Solid Earth, 104(B1), 817\u0026ndash;844. https://doi.org/10.1029/1998jb900007\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRubin, A. M., \u0026amp; Ampuero, J. P. (2005). Earthquake nucleation on (aging) rate and state faults. Journal of Geophysical Research: Solid Earth, 110(11), 1\u0026ndash;24. https://doi.org/10.1029/2005JB003686\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYamaguchi, T., Ohmata, S., \u0026amp; Doi, M. (2009). Regular to chaotic transition of stick-slip motion in sliding friction of an adhesive gel-sheet. Journal of Physics Condensed Matter, 21(20). https://doi.org/10.1088/0953-8984/21/20/205105\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShlomai, H., \u0026amp; Fineberg, J. (2016). The structure of slip-pulses and supershear ruptures driving slip in bimaterial friction. Nature Communications, 7, 1\u0026ndash;7. https://doi.org/10.1038/ncomms11787\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarquins, M. (1985). Sliding friction of rubber and Schallamach waves - A review. Materials Science and Engineering, 73(C), 45\u0026ndash;63. https://doi.org/10.1016/0025-5416(85)90295-2\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrener, E. A., Weikamp, M., Spatschek, R., Bar-Sinai, Y., \u0026amp; Bouchbinder, E. (2016). Dynamic instabilities of frictional sliding at a bimaterial interface. Journal of the Mechanics and Physics of Solids, 89, 149\u0026ndash;173. https://doi.org/10.1016/j.jmps.2016.01.009\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarsoum, M. W., Zhao, X., Shanazarov, S., Romanchuk, A., Koumlis, S., Pagano, S. J., Lamberson, L., \u0026amp; Tucker, G. J. (2019). Ripplocations: A universal deformation mechanism in layered solids. Physical Review Materials, 3(1), 1\u0026ndash;9. https://doi.org/10.1103/PhysRevMaterials.3.013602\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGvirtzman, S., \u0026amp; Fineberg, J. (2023). The Initiation of Frictional Motion\u0026mdash;The Nucleation Dynamics of Frictional Ruptures. Journal of Geophysical Research: Solid Earth, 128(2), 1\u0026ndash;18. https://doi.org/10.1029/2022JB025483\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrodsky, E. E., Kirkpatrick, J. D., \u0026amp; Candela, T. (2016). Constraints from fault roughness on the scale-dependent strength of rocks. Geology, 44(1), 19\u0026ndash;22. https://doi.org/10.1130/G37206.1\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6097579/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6097579/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRock deformation experiments play a key role in our understanding of earthquake physics and friction constitutive laws. These laws commonly describe the response of analogue laboratory faults as a simple and homogeneous system, without accounting for the spatial-temporal evolution of structures in the sample. However, increasing experimental evidence suggests that slip instability is closely tied to heterogeneity, complex rheologies, and inhomogeneous boundary conditions. To address this, we designed a novel transparent setup to observe real-time deformation, track the spatial-temporal evolution of shear fabric, and document unstable slip in experimental faults. Our video documentation reveals that the progressive development of fault fabrics results in heterogeneous but not random stress redistribution. We show that stress and structural heterogeneities play a key role in the nucleation, propagation, and arrest of slip instabilities, raising questions about the robustness of scaling laboratory frictional laws to nature.\u003c/p\u003e","manuscriptTitle":"Spontaneous complexity in the dynamics of slow laboratory earthquakes","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-07 10:32:40","doi":"10.21203/rs.3.rs-6097579/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"72e25b5f-c6b9-4b70-bd96-bf65fe0f8a2f","owner":[],"postedDate":"March 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":45315362,"name":"Earth and environmental sciences/Solid Earth sciences/Seismology"},{"id":45315363,"name":"Earth and environmental sciences/Solid Earth sciences/Tectonics"}],"tags":[],"updatedAt":"2025-10-08T07:08:49+00:00","versionOfRecord":{"articleIdentity":"rs-6097579","link":"https://doi.org/10.1038/s41467-025-63984-0","journal":{"identity":"nature-communications","isVorOnly":false,"title":"Nature Communications"},"publishedOn":"2025-10-07 04:00:00","publishedOnDateReadable":"October 7th, 2025"},"versionCreatedAt":"2025-03-07 10:32:40","video":"","vorDoi":"10.1038/s41467-025-63984-0","vorDoiUrl":"https://doi.org/10.1038/s41467-025-63984-0","workflowStages":[]},"version":"v1","identity":"rs-6097579","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6097579","identity":"rs-6097579","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}