THE PLASTIC LAYER OF THE EARTH’S MANTLE
by Don L. Anderson
SCIENTIFIC AMERICAN (July 1962)
Earthquake waves indicate that at a depth between 37 and 155 miles the stuff of the
earth is less rigid than above and below it. Such a layer would have an important bearing
on tectonic processes.
Earth scientists have often pointed out that physical conditions inside our own planet
are less well understood than those in stars light-years away. Even more paradoxical is
the fact that the region within a few hundred miles of the surface presents more problems
and gives rise to more technical controversy than the region below. One long-standing item
of debate is the zone called the low-velocity layer.
In 1926 the seismologist Beno Gutenberg suggested that earthquake waves slow down when
they travel through a zone roughly 100 to 200 kilometers (60 to 120 miles) below the
surface. He attributed the effect to a decrease in the rigidity of the material in the
zone compared with that above and below it. Most authorities considered his evidence to be
dubious at best and for 30 years they largely ignored his proposal. Recently a mass of
data has accumulated that strongly supports the concept of a low-velocity, low-rigidity
layer. Its existence has important implications for all theories concerned with structural
changes in and near the earth's surface.
The idea that the earth becomes plastic-if not, indeed, liquid-at moderate depths goes
back to the earliest days of geology. Volcanic lava flows pointed to a molten interior not
too far below the surface. Observations on the rate of increase of temperature in deep
mines indicated that if the temperature continues to increase at the same rate, rocks
should be molten at depths of less than 100 kilometers. The enormous cracks and folds
found in the earth's crust suggested upheavals in a mobile substratum. All this agreed
with prevailing views of the origin of the solar system, which held that the earth and
other planets had been torn loose from the sun and had had time to solidify only at the
surface.
One of the most compelling arguments for some degree of fluidity in the interior came
from the principle of isostatic equilibrium. As long ago as 1854 gravity measurements led
geologists to suspect that the earth's crust floats on a denser material. Like other
floating bodies, the crust seeks an equilibrium, riding deeper where it is heavier and
rising higher where it is lighter. Subsequent studies, of both the strength of gravity and
the propagation of earthquake waves, confirmed the notion, indicating that mountains have
deep roots that support them just as the submerged portion of an iceberg supports the part
above water, whereas plains resemble ice floes, having smooth upper and lower surfaces.
Moreover, when the load on a part of the crust changes suddenly (on the geological time
scale), the surface can be observed to respond by rising or sinking to restore
equilibrium. For example, land covered by ice during the last glaciation is still rising
at the rate of about a meter per century. Obviously this behavior implies that the
material under the crust can flow, if only slowly.
On the other hand, several facts appeared to rule out the idea of widespread fluid
material anywhere near the surface. From the tidal distortions of the solid earth in
response to the pulls of the sun and moon, Lord Kelvin calculated that the earth is more
rigid than steel. Studies of earthquake waves indicated that at depths down to thousands
of kilometers the earth transmits not only compression waves (P waves) but also
transverse, or shear, waves (S waves). Shear waves, which oscillate at right angles to
their direction of motion, cannot propagate through liquids because liquids have no shear
strength. When liquids are subjected to shearing forces, they simply flow. Finally,
seismologists discovered that earthquakes originate as deep as 700 kilometers below the
surface. Since an earthquake represents the abrupt yielding of rock to accumulated stress,
it characterizes brittle, not plastic, material.
The answer to this apparent contradiction is suggested by the properties of
noncrystalline materials such as glass and pitch, which behave like solids in the short
run and like fluids over longer periods. They transmit shear waves and can support loads
for a short time, but under a steady, long-lasting force they are plastic; that is, they
flow and change their shape permanently. Under conditions of high temperature and high
pressure the rock under the crust could also behave plastically. It would respond like a
rigid solid to the relatively short-lived stresses that build up to cause earthquakes and
the even briefer stresses involved in earthquake waves, while flowing slowly to adjust to
the long-term stresses caused by changes in the weight of overlying material. Some
geologists believe that the plastic substance under the crust is a glassy basalt. Recent
evidence suggests, however, that it is crystalline. At high temperature even a crystalline
material can flow easily, because melting at the boundaries of individual crystal grains
allows them to slide over one another.
In 1909 the Yugoslav seismologist Andrija Mohorovicic proposed that at some distance
below- the surface there is a discontinuity where the velocity of earthquake waves jumps
from about seven kilometers per second to eight. Subsequent measurements placed the
Mohorovicic discontinuity, or Moho, at an average depth of 35 kilometers below the surface
of the continents and only about five kilometers below the ocean floor. Under high
mountains the Moho is as deep as 65 kilometers. Geologists saw in the Moho the lower
boundary of the rigid, floating crust. The material between the Moll and the presumably
liquid core of the earth they named the mantle. Yet the fact that seismic waves travel
faster below the Moho than they do above it implies a greater rigidity at the top of the
mantle than in the crust. It now seems clear that the Moho marks a change in chemical
composition or crystal structure rather than an abrupt transition from strong to weak
material.
The first seismic evidence for this transition was not forthcoming until Gutenberg
announced the low-velocity zone. Actually what he had found was a decrease in the
amplitude of compressional waves reaching the surface at a distance between 100 and 1,000
kilometers from an earthquake. At 1,000 kilometers the amplitudes were only a hundredth as
great as they were at 100 kilometers. Beyond 1,000 kilometers the amplitudes increased
sharply.
Gutenberg explained the effect by assuming a subsurface layer in which the earthquake
waves travel slower than they do in the regions above or below. A wave entering this layer
obliquely from above would be refracted downward, away from the surface, as light is bent
downward when it passes from air to water. On leaving the bottom of the layer the wave
would be refracted upward again (see illustration on
page 5). The result is that the wave would arrive at the surface farther away from its
source than it would if there had been no decrease in velocity. Hence a gap would appear
between the last "ray" that had missed the low-velocity layer and the first one
to enter it. As the illustration shows, the gap, or shadow zone, is greatest for an
earthquake originating just above the top of the layer Those coming from deeper levels
evince no gap. From the extent of the shadow zone for different earthquakes, Gutenberg
calculated that the layer is centered at a depth of about 150 kilometers, and that between
100 and 200 kilometers the velocity is some 6 per cent less than it is just under the
Moho. Such a decrease in velocity means that the rock within the layer must be
substantially less rigid than the material above and below it. The velocity does not reach
the value it had at the ~ base of the crust until some 250 or 300' kilometers below the
surface.
If the low-velocity layer were perfectly uniform, and if the waves really traveled as
rays, the shadow zone at the surface would he completely "black." No waves at
all would emerge within its limits. Actually the layer
is full of inhomogeneities, and seismic waves do not travel strictly along classical
ray paths. Like all waves, they bend around corners by diffraction, thereby leaking into
shadowed regions. Both effects con. tribute to the energy that is found in the shadow
zone.
It was partly this energy leak that made other workers reluctant to accept Gutenberg's
conclusion. In those days seismologists paid little attention to the comparative
amplitudes of earthquake waves. They were primarily interested in travel times, and they
tended to accept any signal, weak or strong, if it appeared in their records at a time
when readings at other seismographic stations led them to expect it.
Moreover, the evidence for the low. velocity layer was by no means clear-cut. The
statistics were assembled from many earthquakes, large and small, shallow and deep. The
data came from seismographs of different designs. In his calculations Gutenberg could make
only approximate corrections for these variations as well as for the local irregularities,
mostly unmapped, in the rock through which different waves traveled.
Underground nuclear explosions finally made
possible a controlled experimental test of Gutenberg's analysis. The time, strength and
location of these events is known so precisely that a single blast provides excellent
data. Furthermore, seismographs today are more numerous, more sensitive and more
standardized than they were in 1926. Studies of several explosions have confirmed the
conclusions Gutenberg extracted so tediously from earthquake records [see illustration on page 7]. Seen in sharper detail, the low-velocity layer extends from
about 60 kilometers to about 250 kilometers. (It is interesting to note that the layer
damps blast waves so effectively that many seismologists think it poses a major difficulty
for the detection of underground nuclear tests.)
Several independent pieces of evidence now support the idea of a low-velocity plastic
layer. One is furnished by surface waves. These are seismic disturbances that follow the
curved surface of the earth (see bottom illustration on page
6) instead of passing through its body, Although the waves travel along the surface,
they "feel" the elastic properties of the underlying material to a depth that
depends on their wavelength; the longer the wave, the deeper it feels [see "Long
Earthquake Waves," by jack Oliver; SCIENTIFIC AMERICAN, March, 1959]. Since in
general elasticity increases with depth, longer waves travel faster than shorter ones, and
waves that start out together are dispersed, or spread out. Detailed analyses of the
dispersion patterns show that elasticity does not increase continuously with depth but
falls off in the region of the low-velocity layer.
Body waves, which pass through the deep interior, provide only a point-by-point
sampling of the outer regions of the earth. Surface waves, on the other hand, contain
information about these regions over their entire path. Recent
studies of surface waves in our laboratory at the California Institute of Technology
and at Columbia University have demonstrated for the first time that the low-velocity
layer is present below the oceans as well as below the continents. Some of the waves
used in the analysis had travelled around the earth as many as seven times. They indicate that the layer is in fact a world-wide phenomenon.
Comparison of oceanic and continental paths shows that the waves are slowed more under the
oceans. Evidently the geological differences between ocean basins and land masses are not
limited to the crust but extend several hundred kilometers into the mantle.
Conclusive proof of the world-wide extent of the low-velocity layer came from the great
Chilean earthquake of May 22, 1960. It was so violent that it set the earth as a whole
into vibration, making it "ring" like a bell. The tone of a bell-that is, the
frequencies at which it vibrates-depends on its elastic properties; a steel and a bronze
bell emit different sounds. From records of the free vibrations following a big earthquake
it is possible, with enormous mathematical labor, to deduce the elastic structure of the
earth. The labor has been performed. It shows that the
low-velocity zone is necessary to account for the observed frequencies.
In an attempt to construct a model of the earth that fits the current seismic data, 1
have been obliged to conclude that the low-velocity zone transmits the horizontal and
vertical vibrations in shear waves at different speeds. A crystalline material in which
the crystal grains were aligned in one direction would behave this way. One mechanism that
could bring about such an alignment is a flow of the material. Others are directional heat
flow and differential stress.
In addition to the purely seismic data, several other phenomena attest to a lowered
rigidity in the material pear the top of the mantle. Variations in atmospheric pressure
cause measurable deflections of the earth's surface. The amount of deflection is much
greater than it would be if the crust and mantle had the same strength. By assuming a weak
layer in the upper mantle the observations can be explained quite well. Moreover, most
earthquakes originate in the first 60 kilometers below the surface, at an average-depth of
25 kilometers. At a depth of more than 60 kilometers the number falls abruptly, indicating
a sudden drop in the strength of the rock.
From 60 kilometers down the frequency of earthquakes decreases steadily, dying away to
zero at about 700 kilometers. This distribution implies that the rock becomes less brittle
all the way from 60 to 700 kilometers and that it does not regain its strength at any
deeper level. The picture agrees with a nomenclature first proposed in 1914 by the U.S.
geologist Joseph Barrell. He spoke of an upper, rigid "lithosphere" (from the
Creek word lithos, meaning stone) and a lower, more plastic "asthenosphere"
(from the Creek word asthenes, meaning weak). Barrell placed the boundary between the two
at a depth of 100 kilometers. Now it appears to be not a sharp boundary but a transition
zone starting at some 60 kilometers.
The concept of strength and weakness in the foregoing discussion applies to the time in
which stresses build up to cause earthquakes. Viewed on this temporal scale the mantle
undergoes a transition from a brittle to a plastic state at about 60 kilometers and
thereafter increases in plasticity. On the much shorter time scale of earthquake-wave
vibrations, however, the material reverts to a stronger, or more elastic, condition at a
depth of more than 250 kilometers. The decrease in velocity at the top of the mantle is
gradual; it is not yet clear whether the base of the low-velocity zone is characterized by
a gradual or an abrupt increase in velocity.
Almost certainly the short-term properties that set apart the low-velocity layer are
determined by the temperature and pressure of the mantle in relation to its melting point
at different depths. In general the elasticity of any material decreases as its
temperature approaches the melting point. But an increase in pressure raises the melting
point and elasticity. Below the surface of the earth both temperature and pressure
increase with depth, and so the two have opposing effects on the proximity to the melting
point as well as on the elastic strength of rock. Presumably at ;~ depth of about 60
kilometers temperature takes the upper hand and the rock begins to approach its melting
point, growing weaker as the depth increases. This trend continues down to some 200
kilometers, where it reverses. Then pressure raises the melting point faster than the
temperature increases and the material becomes more elastic (until the liquid outer core
is reached). A few laboratory experiments on rock under high temperature and pressure seem
to confirm this picture. Extrapolating the rather scanty data indicates a very low
strength at a depth of somewhat more than 100 kilometers.
Hugo Benioff of the California Institute of Technology has discovered a remarkable
indication of discontinuity at the level of the top of the low-velocity zone. In studying
a large number of earthquakes in the Pacific Ocean earthquake belt he was able to connect
certain sequences of earthquakes to single fault structures. One sequence that occurred in
South America between 1906 and 1942 delineates a great fault off the west coast of the
continent. The fault is some 4,500 kilometers long and goes down 600 kilometers-a tenth of
the distance to the center of the earth. The earthquakes related to the fault fall
naturally into three groups: (1) those shallower than 70 kilometers, (2) those from 70 to
250 or 300 kilometers and (3) those from 300 to 600 kilometers (see top illustration on page 6). Analysis of the
earth motions in the quakes showed a marked similarity between the intermediate and deep
groups but no resemblance of these to the shallower group. In particular the motions of
the two deeper groups changed suddenly, and in the same way, in 1921. There was no
corresponding change in the shallower earthquakes. Evidently there is some mechanical
coupling between the lower layers, but these are sharply decoupled from the region above
70 kilometers. Other are-as of the circum-Pacific tectonic belt show similar phenomena.
When the earthquake foci are plotted in three dimensions, those down to 250 kilometers
fall in a plane about 900 kilometers wide, dipping about 33 degrees under the continents
with respect to the surface of the earth. The deep earthquakes, on the other hand, are on
a plane tilted at 60 degrees. Thus, although they are mechanically connected, the
intermediate and deep layers are spatially discontinuous. The dimensions and location of
the intermediate layer correspond closely to those of the low-velocity zone.
An interesting clue to the state of the material in the upper mantle was furnished by
the Soviet volcanologist G. S. Gorshkov in 1957. He found that shear waves from Japanese
earthquakes do not reach the Kamchatka Peninsula when their paths cross the volcanic belt
between Japan and the peninsula. Gorshkov concluded that there must be pockets of liquid
magma at a depth of 55 kilometers that absorb the waves. Apparently in certain regions the
temperature not only approaches the melting point but even exceeds it. Many seismologists
have remarked on the fact that the average wavelength of shear waves is many times longer
than that of compressional waves. The observation could be accounted for by a weak,
perhaps partially molten, layer that absorbs the shorter S waves more than the longer S
waves.
Volcanoes are concentrated in parts of the world where earthquakes are most common, and
the earthquakes actually associated with volcanism mostly originate at depths between 60
and 200 kilometers. This suggests that volcanoes are connected with disturbances in the
region of the low-velocity zone. Therefore the distribution of volcanoes constitutes
direct evidence for the temperature-melting point relation inferred from laboratory
measurements and suggests that the low-velocity layer may he the source of primary
basaltic magma.
Volcanism and the postglacial uplift of the crust constitute the only dynamic, as
opposed to static, geological "experiments" Both indicate fluidity, and some
degree of actual flow, in the material below the crust. Moreover, they are consistent with
the idea of a layer of maximum plasticity in the upper mantle.
Almost all present theories of isostasy and tectonics, including those concerned with
mountain building, faulting and the possible drifting of the continents, focus attention
on the Mohorovicic discontinuity, which divides the crust of the earth from the mantle. If
the picture 1 have tried to outline in this article is correct, the important
discontinuity is farther down, at the ill-defined boundary of the rigid lithosphere and
the weaker asthenosphere. Most of the activity responsible for the broad-scale features of
the earth's surface probably takes place in a low-velocity or plastic layer at the top of
the asthenosphere, extending roughly from 60 to 250 kilometers in depth. In particular the existence of such a plastic layer makes the idea of
continental drift much more plausible than it has seemed heretofore.
