Frictional resistance of faults during accelerating and decelerating

LETTERS PUBLISHED ONLINE: 20 SEPTEMBER 2009 | DOI: 10.1038/NGEO637

Frictional resistance of faults during accelerating and decelerating earthquake slip Hiroki Sone *† and Toshihiko Shimamoto * The dynamic friction of faults during earthquake slip is a critical control on earthquake ruptures in the crust. Extrapolation of brittle crack theories to natural earthquakes1–3 has led to the commonly held view that fault friction reduces during rapid earthquake slip, a process known as slip-weakening1 . High-velocity gouge experiments4,5 and recent analyses of thermal pressurization6,7 and frictional melting8 support such a notion; however, these studies dealt with constant rates of slip along faults. Here we present the results of experiments aimed at understanding the frictional behaviour of fault zone materials under variable slip rates—conditions that are more representative of natural earthquakes. Our results show that faults undergo a sequence of strengthening, weakening and healing during acceleration and deceleration of slip. Such a sequence may be explained by the extrapolation of rate-andstate frictional behaviour9,10 at low slip velocities to more realistic slip rates, but involving different physical mechanisms and a different scale. The initial strengthening should impose a barrier for rupture growth into large earthquakes. The healing on decelerating fault motion may lead to pulse-like earthquake ruptures11–14 and static stress drops that are low in comparison with the dynamic stress changes15 . Kinematic slip histories of earthquake ruptures inverted from seismograms suggest that fault ruptures propagate in a ‘pulselike’ mode11 , in which the rise time of fault slip is much shorter than the total duration of the event. Numerical and theoretical studies12–14 have suggested that a strong velocityweakening fault would heal itself, shortly after the passage of the propagating rupture front, to yield such ‘pulse-like’ behaviour. Moreover, the recovery of fault shear traction at the final stage of seismic slip seems to be a common phenomenon found from elasto-dynamic analyses of coseismic fault slip16,17 . Although frictional shear strengths of crustal materials are now understood to weaken markedly above sub-seismic slip rates4,5,8,18,19 , the continuous frictional response to continuously changing slip velocities is not documented yet. We conducted high-velocity frictional experiments (at constant and changing velocities) under 0.56 MPa normal stress using natural fault gouge samples collected from the Chelungpu fault, Taiwan, to investigate the fault frictional behaviour during accelerating and decelerating fault motion (Fig. 1). Figure 2 shows results from constant-velocity experiments with slip rates ranging from 0.1 to 2.1 m s−1 . Peak friction between 0.68 and 1.05 is reached normally within 0.30 m displacement after the onset of fault motion, consistent with Byerlee friction at low pressures20 . Peak friction is followed by a nearly exponential decay with slip towards a steady-state value between 0.06 and 0.48

a

Teflon

Gouge

0.56 MPa

c

b

G

HR

HR

SZ

Figure 1 | Configuration of the simulated fault. a, Schematic representation. b,c, Microscopic image (b) and structural sketch (c) of the fault within the region indicated by the red rectangle in a after 50-m slip displacement at 1.03 m s−1 equivalent slip rate. HR: host rock, G: gouge sample, SZ: slip zone. The scale bar is 1 mm.

(Fig. 2a). As in a previous study4 , the post-peak ‘slip-weakening’ can be described by: µ(d) = µss + (µp − µss )exp[ln(0.05)d/Dc ]

(1)

where d is the fault displacement after the peak friction. The parameters µp and µss represent the peak and the steady-state coefficient of friction, respectively, and Dc is the weakening distance required to achieve 95% of the weakening from µp to µss . Peak friction can be determined from experimental results and it tends to increase slightly from 0.1 to 0 .5 m s−1 and decreases on further increases in slip rate (Fig. 2b). Friction experiments21 with the same material in a direct shear apparatus at much slower slip rates (