Seismic pumping a hydrothermal fluid

transport mechanism
R. H. SIBSON,

J. MeM.

MOORE

& A. H . R A N K I N

SUMMARY

A consequence of the dilatancy/fluid-diffusion ties of fluid involved are such that significant
mechanism for shallow earthquakes is that hydrothermal mineralisation may result from
considerable volumes of fluid are rapidly each seismically induced fluid pulse, and the
redistributed in the crust following seismic mechanism provides an explanation for the
faulting. This is borne out by the outpourings textures of hydrothermal vein deposits associof warm groundwater which have been ated with ancient faults, which almost invariobserved along fault traces following some ably indicate that mineralisation was
moderate (M5-M7) earthquakes. The quanti- episodic.
HYDROTHERMAL vein deposits, sometimes of economic importance, are often
found in the upper, brittle regions of ancient fault zones. It is a characteristic of
these deposits that their textures usually indicate that mineralisation took place
episodically. To date, little attention has been paid to the transport mechanism
needed to move hydrothermal fluids rapidly and intermittently along fault zones,
so that they attain the requisite state of disequilibrium in the upper crust. Current
ideas on the source mechanism for shallow earthquakes (Frank i965, Nur I972,
Scholz et al. I973) which invoke fluid transport on a large scale in and around
fault zones, provide an answer to this problem.

I. Textures of hydrothermal vein deposits

The formation of hydrothermal vein ores involves the transport of material in
aqueous solutions from a source region to a zone of deposition. Mineral veins then
form by replacement processes, or by crystal growth in rock cavities and open
fractures. M a n y vein systems are associated with faults, and their ore textures
generally indicate that mineral deposition was discontinuous (Bateman i95o ).
Veins in faults commonly consist of cemented breccias resulting from several
periods of movement, in which each episode of shattering was accompanied or
followed by growth of gangue and ore mineral crystals. Vein ores in extension
fractures which are adjacent to faults, but which themselves have suffered no
shear displacement, are texturally distinct, often consisting of laminated aggregates of various minerals with crystal growth zoning and wall rock alteration,
deposited during a series of mineralising episodes. Fig. I illustrates one such
extensional vein in which ten generations of various minerals and several hundreds
of distinct crystal growth zones formed. In the course of development, the veinfilling was separated from its wall rocks and split internally by lateral extension
several times as new episodes of deposition were accommodated. More complex
vein textures may develop from a succession of interspersed shear and extensional
displacements on fractures during mineral deposition.
Jl geol. Soc. Lond. vol. x3x, I975, pp. 653-659, 4 figs. Printed in N. Ireland

654

R. H. Sibson, J. M c M . Moore & A. H. Rankin

Two inferences may be drawn. First, that incremental deposition of minerals

occurred when pulses, rather than a steady flow of hydrothermal fluid passed
through the vein fractures, and second, that the pulses of fluid were associated in
time with increments of shear and extensional displacement on fracture systems.
Any proposed transport mechansim for hydrothermal fluids in and around fault
zones must explain these allied phenomena.

2. The dilatancy/fluid-diffusion model for shallow earthquakes

In recent years, the dilatancy/fluid-diffusion model for energy release in shallow
earthquakes has received increasing attention because of its potential for explaining
a number of observed earthquake precursors (Scholz et al. I973). In its simplest
form the model (Fig. 2) supposes that prior to seismic shear failure along an
existing fault, the region for some considerable distance around the focus of the
subsequent earthquake dilates in response to rising tectonic shear stress (r) by the
opening of extension cracks and fractures normal to the least principal compressive
stress (a~). The development of this fracture porosity causes the fluid pressure (p)
in the dilatant zone to decrease, inducing a slow inwards migration of fluid from
the surrounding crust. At the onset of dilatancy, the drop in fluid pressure causes
a rise in the frictional resistance to shear along the fault (r/ = / , ( a ~ - - p ) , where
is the coefficient of static friction and a, is the normal stress across the fault). As
the migrating fluid fills the cracks, fluid pressure rises again and frictional resistance decreases. Seismic failure eventually occurs when the rising shear stress
equals the frictional resistance. According to an empirically derived relationship,
the precursory dilatant periods for earthquakes of magnitude M5, M6, M 7 and
M8 are 0.3, 2, z3 and 83 years respectively (Whitcomb et al. I973).
The rapid partial relief of shear stress which accompanies earthquake faulting
allows the cracks within the dilatant zone to relax (Scholz & Kranz z974), and
the fluids they contain must be expelled rapidly upwards in the direction of
Fissure Vein Cross Section
H
W

~"

AI

L
R
0
C
K

........
.
:'".'....~~~.,:~:~
....

"" ~ "::~i , '"". ' . '"' . ' ."'"'"

"~
'.',
"~
"-'-?.:; ~':'.~
.'~'~'i'' .;.'.'." ~
"
- ".'.'.'."
.....
~-::I"
~... ~---.
:.~.~'~::.
.:.;.-.:. ~ ~
:-':::"f,~:y)~::
..V,'. " - . . . . . " ~.:.~

'2

~" : ' - : r .""

.....
~,
" " ;.:

:" ".." ."

Jasper
LimoNtic

~ 8
7

X?I

~:::

.~:

" ,--~'~-~:
:,.r" ..~:. . . "....... ~::~::.;,

I (

,~ :; ~';"
".'~.~" :~' ,." . . . . . . : ~ : " k " ~'"~'
~
" ;" "~"
, .
~_
:~. v..~::.~.

..-: :,. ~.:',~:-:::L.~-...-....>

~..
-~ ..,'. "-~:,.
.:
g:f,,.~
~
.:~.~,.:::~?:
"
"

W
A

'~-----~ ~ - . f l l l ~ l l l L - - , ~ - ~ - ~ ' - ~
4,

HX

8~65
7

Q
Crystallk~
ztuar
~

R
0

10

Cryptocrystaline
Silica

Scale
centimetres

Fzo. z. Hydrothermal quartz-jasperoid mineralisationin a fissure vein from Wheal

Crowns, Cornwall. I-V--major episodes of extensional opening; z-xo---separate
generations of mineral deposition.

Seismicpumping

655

easiest pressure relief (Fig. 3). This latter effect should be especially marked with
wrench and normal faults, where a3 is horizontal and the cracks may be expected
to lie in vertical planes. Note, though, that the generally lower differential stresses
associated with normal faulting (Sibson i974) may reduce the extent to which
dilatancy occurs. Upflow from the collapsing dilatant zone must take place through
the fault and adjacent fractures, with flow rates decreasing as the source region of
the earthquake returns to its pre-dilatant state. It should also be noted here that
individual seismic events usually occur as localised shear dislocations Which do
not lead to displacements over the entire surface of an existing geological fault.
Thus, following any one earthquake, both resheared and unsheared areas of the
fault surface may serve as passageways for flow. Irregularities in the fault surface
and intersections with minor fractures will act as channels, concentrating flow
and increasing the effective permeability of the whole fault zone.
Substantial transient outflows of water along spring-lines in the vicinity of faults
are not infrequently observed following moderate shallow earthquakes (Briggs &
Troxell i955, Tchalenko i973). The Matsushiro earthquake swarm, which was
energetically equivalent to a single event of magnitude M6. 3 (Hagiwara & Iwate
1968), resulted from strike-slip movements about a buried wrench fault c. i o km
in length (Kasahara i97o ). It was accompanied by the surface expulsion in one
year of about io 1 litres of warm, Na-Ca-C1 brine, saturated with carbon dioxide
(Tsuneishi & Nakamura i97o ). This outflow has been interpreted as resulting
from the collapse of a focal region with a dilatant strain of about 10-4 (Nur 1974).

B u i l d - u p of
tectonic shear
stress

Onset of
dilatancy

Fluid f i l l s
EQ Collapse
dilatant cracks
of dilatant
zone

TfT_

~T

p.

Inflow

~'

Outflow

Fluid f l o w
around
dilatant
zone

F~G. 2. Synoptic diagram of the seismic pumping process (modified from Scholz
et al. 1973).

656

R. H. Sibson, dr. MeM. Moore ~ A. H. Rankin

3. Hydrothermal mineralisation by seismic pumping

The dilatancy/fluid-diffusion model provides an explanation for the intermittent
flow of hydrothermal fluids in and around fault zones, and suggests that in effect,
seismic faulting acts as a pumping mechanism whereby individual earthquakes are
capable of moving significant quantities of mineralising fluid rapidly from one
crustal environment to another.
If the mechanism is to promote the formation of metalliferous hydrothermal
vein deposits, it is of crucial importance that the fault intersects a suitable source
region such as a pile of volcanic rocks, sediments undergoing lithification or
metamorphism, or a granitic pluton. Potential metallogenetic situations then arise
when the dilatant zone coincides with such a source region. When the zone

Fault

01 ~

x,,~(l,.t

I[" ."" :" : ' : " ." ":i~l

....

--

..

Mineral

/
Dilation

/
I

'Z one

h.
I

~-~

I
A
\

~..

\
X

,-a~--

d~:~-

~ "

.9=,,,-./

/.
Limit of shear dislocationJ
Fluid migration direction
(after fault movement)
~

"%:--::::

Shear displacement vectors

Earthquake focus

Fx o . 3. F l u i d expulsion following t h e collapse o f a d i l a t a n t zone a r o u n d a w r e n c h fault.

F o r simplici W t h e shear dislocation is t a k e n to e x t e n d over t h e cross-sectional a r e a o f a
spherical d i l a t a n t zone.

Seismic pumping

657

collapses after an earthquake, fluids carrying material dissolved from the source
rocks are transported over periods of days to months to different physico-chemical
environments which may bring about mineral deposition. Hydrothermal deposits may develop on the fault itself in both resheared and unsheared regions, and
in connected extension fissures including those created by hydraulic fracturing.
Fluids ejected from the fault plane enter accessible extension fractures across
which there is a lower normal stress.
To demonstrate the viability of the seismic pumping mechansim for the emplacement ofhydrothermal deposits, we consider the potential of a single M6 earthquake
for transporting quartz in solution from a focal depth of xo km to depositional
sites at higher levels. Quartz is chosen because its solubility in water is dependent
largely on temperature and pressure, with chemical factors playing only a minor
role (Holland 1967). An earthquake fault dimension (L) of IO km and a mean
slip of I O-IOO cm are reasonable parameters for an event of this magnitude (Wyss
& Brune 1968 ). Assuming, conservatively, that the dilatant volume (V) is a
sphere (the symmetrical updoming at Matsushiro suggests that this is reasonable
(Hagiwara 1972)) of radius L[2, and that the dilatant strain (AV/V) is IO-s
(Scholz et al. I97~3, Nur 1974) , the volume of fluid released after the earthquake
(AV ~-, lO-5 . L8/2) is about 5 IO9 litres.
The solubility of quartz in pure water along a geothermobaric gradient of
35C and 3oo bars per kilometre is shown in Fig. 4. At a depth of lO km, 5 lO9
litres of water can potentially dissolve ~.~iot0 g of quartz in the dilatant zone.
From the solubility curve it is clear that by the time the expelled fluid cools to
1ooC during ascent, more than 95% of this quartz will be precipitated. The
volume of quartz deposited is equivalent to a sheet vein of approximate dimensions
1 cm 4o m IO km, extending along the length of the earthquake fault, but
channel flow will tend to concentrate this. A fault with a finite displacement of
I km could have produced from i o 8 to to 4 such mineralising fluid pulses.
Depth
2
I

4
I "1

6
i

(kin)
8
10
I

12
I

14
I

Q 71

FIG. 4"

Quartz solubility, Q, (g.SiOs/kg.solution) in

pure water along a gcothcrmobaric gradient of
35C and 3oo bars per kilomctrc (after Holland,
x967).

100

200

300

Temperature

400
(C)

500

658

R. H. Sibson, J. McM. Moore & A. H. Rankin

4. D i s c u s s i o n
Most previous workers have suggested that faults play a passive role in the emplacement of high-level hydrothermal deposits, acting either as permeable conduits
for percolating fluids, or in some eases as impermeable barriers which impound
migrating waters. Here, we suggest that the mechanics of seismic faulting plays
a key role in the intermittent transport of hydrothermal fluids. The calculations
in the previous section show that the potential for hydrothermal ore formation is
very great, provided suitable material is available for solution in the zone of
dilatation, and physico-chemical changes can be effected to bring about the
deposition of minerals during ascent. Because their composition is suited to the
transport of a wide variety of gangue and ore minerals (Roedder 1972), the saline
fluids expelled at Matsushiro are of particular interest in this regard. Interestingly, a study of recent metalliferous sedimentation in the Red Sea (Bignell 1975)
draws attention to the episodic appearance of metal-bearing brines, and to the
association of brine-filled deeps with regions where the median rift is offset along
transform faults. As it is these sections of the transform faults which should be
seismically active, it seems possible that the intermittent faulting accompanying
sea-floor spreading could provide a mechanism for the sporadic discharge of
metal-bearing brines. We would also suggest that the quantities of fluid involved
in the seismic pumping process are such (lO 1 litres ~ 6 I o ~ barrels) that the
mechanism may substantially assist the migration of hydrocarbon fluids in tectonitally active areas.
ACKNOWI.~DO~NTS. We thank Dr N. J. Price, Dr D. J. Shearman, Professor Janet Watson and
Professor G. R. Davis for critical reading of the manuscript.

5-References
BATEMAN,A. M. I95O. Economic mineral deposits, 2rid ed. Wiley, New York. 9x6 pp.
BION~LL, R. D. z975. Timing, distribution and origin of submarine mineralization in the Red Sea.
Trans. Inst. Mining Metallurgy ]384, B I-6.
BI~IGGS,R. C. & TROXELL, H. C. I955. Effect of Arvin-Techapi earthquake on spring and stream
flow. In Earthquakes in Kern County, California, during I952. Calif. Div. Mines Bull. x7x,

8x-98.
F ~ N x , F. C. I965. On dilatancy in relation to seismic sources. Rev. Geophys. 3, 484-503
HAorwAa~, T. x972. Causes of shallow earthquakes occurring in the earth's crust as suggested by
the accompanying land deformation. Phys. Earth planet. Interiors, 6, 25o-5 .
- - . & IWATE, T. x968. Summary of the seismographic observation of Matsushiro swarm earthquakes. Bull. Earthquake Res. Inst. Tokyo Univ. 46, 65 I-6 I.
HOLLAND, H. D. I967. Gangue minerals in hydrothermal deposits. In Barnes, H. L. (ed.)
Geochemistry ofhydrothermal ore deposits. Holt, Rinehart & Winston Inc., New York, 382-436.
K ~ A H ~ , K. x97o. The source region of the Matsushiro swarm earthquakes. Bull. Earthquake
Res. Inst. Tokyo Univ. 48, 58I-6O2.
NuR, A. I972. Dilatancy, pore fluids and premonitory variations oft,#,, travel times. Bull. Seismol.
Soc. Am. 62, x217-22.
~ 1 9 7 4 - Matsushiro, Japan, earthquake swarm: confirmation of the dilatancy-fluid diffusion
model. Geology 2, 217-2 I .
ROEDI)~R, E. 1972. Composition of fluid inclusions. U.S. geol. Survey Prof. Paper 44o-JJ.

Seismic pumping

659

SCHOLZ, C. H. & KRA~Z, R. I974. Notes on dilatancy recovery. J. geophys. Res. 79, 2132-5.
SCHOLZ, C. H., SYKES,L. R. & AQOARWAL,Y. P. x973. Earthquake prediction: a physical basis.
Science x8x, 8o3-Io.
SmSON, R. H. I974. Frictional constraints on thrust, wrench and normal faults. Nature, Lond. 249,
542-4.
TCHAL~NKO, J. S. i973. The Kashmar (Turshiz) x9o3 and Torbat-e Heidariyeh (South) I923
earthquakes in Central Khorassan (Iran). Ann. Geofis. 26, 29-4 o.
TSUN~ISHI,Y. & NAg.~MURA,K. x97o. Faulting associated with the Matsushiro swarm earthquakes.
Bull. Earthquake Res. Inst. Tokyo Univ. 48, 29-5 I.
WHrrCOMB, J. A., GAGMAN',',J. D. & ANDERSON,D. L. x973. Earthquake prediction: variation of
seismic velocities before the San Fernando earthquake. Science xSo, 632-41 .
WYss, M. & BRuN~,J. N. 1968. Seismic moment, stress and source dimensions for earthquakes in
the CaliforniamNevada region. Jr. geophys. Res. 73, 468x-94Received I6 June I975; revised typescript received 23 July I975.
Richard Hugh Sibson, John McMahon Moore & Andrew Hugh Rankin,
Department of Geology, Royal School of Mines, Imperial College,
London SW7 2BP