Hales discontinuity

The impedance increase at depths of 60 km - 90 km is called the Hales discontinuity (the H). The impedance I is defined as the product of density and seismic velocity: I = $\rho \cdot v$, where $\rho$ = density and $v$ = P- or S-wave velocity.
This impedance increase was first detected in long-range seismic refraction profiles, such as the Early-Rise experiments in continental North America (Green and Hales, 1968; Hales, 1969). More recently, the H was detected by ScS reverberation studies in continental regions (Australia) and island arc regions (Revenaugh and Jordan, 1991b), by array studies (Simpson et al., 1974), and also by surface wave studies in oceanic regions (Woods et al., 1991).
The discontinuity is explained by a phase transition from spinel (sp) $\rightarrow$ garnet (Hales, 1969). This transition has been observed in laboratory experiments in rocks of pyrolite composition at upper mantle temperatures and pressures (Green and Ringwood, 1967), and both minerals (gt and sp) are common in xenoliths originating from different depths (Nixon, 1987). More recent studies place the phase transition at depths of 45 km - 55 km (Jenkins and Newton, 1979; Webb and Wood, 1986), but temperature and mantle composition have a strong influence on the depth of this transition.
The Clapeyron slope $\frac{dp}{dT}$ of the spinel $\rightarrow$ garnet transition has been determined to be positive for temperatures below 900$^{\circ}$C, but the structure of the phase transition is quite complicated (Wood and Yuen, 1983; Jenkins and Newton, 1979). The Clapeyron slope is flattening for temperatures higher than 900$^{\circ}$C, in good agreement to seismological detections of this discontinuity and might become negative for very high temperatures.

The P-velocity jump across this discontinuity has been found to be $\Delta$v$_p \sim$ 3.2% (Green and Hales, 1968; Hales, 1969), and a reflection coefficient of 3.5% has been found (Revenaugh and Jordan, 1991b). If the density change, $\Delta\rho$, and the shear velocity change, $\Delta$v$_s$, associated with the transition follow Birch's law (Birch, 1952; Anderson et al., 1968) then the reflection coefficient and $\Delta$v$_p$ imply a minimum of $\Delta\rho$ $\ge$ 3.2% and $\Delta$v$_s$ $\ge$ 3.8% (Revenaugh and Jordan, 1991b). These values are in good agreement with laboratory studies for the transition, which find density and velocity increases of up to 3% (Green and Liebermann, 1976; Webb and Wood, 1986).
Divergent from the phase transition explanation, Hirn et al.(1975) and other authors (Fuchs, 1983; Forsyth, 1977) quote an increase in amount of preferred orientation of olivine as the mechanism producing the velocity jump across the H. The lattice preferred orientation (LPO) of olivine as source for the discontinuity in these models fail to explain the high S-velocity contrasts found by Revenaugh and Jordan (1991b) and the correlation of the discontinuity depth with tectonics. On the other hand the olivine-orientation model calls for a large variation of the reflection coefficient with azimuth which cannot be found in the data. Therefore, this model cannot explain this discontinuity and the spinel $\rightarrow$ garnet phase transition is most likely the explanation for the Hales discontinuity.