Content uploaded by Jeremy Tregloan-Reed
Author content
All content in this area was uploaded by Jeremy Tregloan-Reed on Jan 21, 2018
Content may be subject to copyright.
Author's personal copy
Air ionization at rock surfaces and pre-earthquake signals
Friedemann T. Freund
a,b,e,
, Ipek G. Kulahci
b
, Gary Cyr
c
, Julia Ling
d,f
, Matthew Winnick
e,g
,
Jeremy Tregloan-Reed
b,h
, Minoru M. Freund
a
a
NASA Ames Research Center, Code SGE, Moffett Field, CA 94035-1000, USA
b
Carl Sagan Center, SETI Institute, Mountain View, CA 94043, USA
c
San Jose State University Foundation, San Jose, CA 95192-5569, USA
d
NASA Academy 2007, NASA Ames Research Center, Moffett Field, CA 94025-1000, USA
e
Department of Physics, REU Summer 2008, San Jose State University, San Jose, CA 95192-0106, USA
f
Department of Physics, Princeton University, Princeton, NJ 08544, USA
g
Department of Physics and Astronomy, Vassar College, Poughkeepsie, NY 12604-0745, USA
h
Department of Physics, University of Lancaster, Lancaster LA1 4YQ, UK
article info
Article history:
Received 4 January 2009
Received in revised form
10 June 2009
Accepted 10 July 2009
Available online 11 August 2009
Keywords:
Pre-earthquake phenomena
Ionosphere
Air ionization
Corona discharges
Thermal infrared anomalies
Earthquake lights
Animal behavior
abstract
Pre-earthquake signals have been widely reported, including perturbations in the ionosphere. These
precursory signals, though highly diverse, may be caused by just one underlying physical process:
activation of highly mobile electronic charge carriers in rocks that are subjected to ever increasing levels
of stress. The charge carriers are defect electrons associated with O
in a matrix of O
2
. Known as
positive holes or pholes h
d
, they flow out of the stressed rock into the unstressed rock volume, traveling
meters in the laboratory, probably kilometers in the field. At the rock–air interface they cause: (i)
positive surface potential, (ii) field-ionization of air molecules, (iii) corona discharges. The rate of
formation of airborne ions can exceed 10
9
cm
2
s
1
. Massive air ionization prior to major earthquakes
increases the electrical conductivity in the air column and may cause ionospheric perturbations,
earthquake lights, and unusual animal behavior as well as infrared emission.
&2009 Elsevier Ltd. All rights reserved.
1. Introduction
Seismologists often state that earthquakes cannot not be
predicted except within wide statistical margins, typically several
years or decades (Geller, 1997;Kagan, 1997;Keilis-Borok, 2003).
However, non-seismic signals that precede major earthquakes
have been reported from essentially all tectonically active regions
around the world. A partial list of these pre-earthquake signals
includes: ionospheric perturbations: (Hayakawa and Sazhin,1992;
Liperovsky et al., 2000;Pulinets and Boyarchuk, 2004;Shalimov
and Gokhberg, 1998); thermal infrared anomalies: (Saraf et al.,
2008a, b;Tramutoli et al., 2005;Tronin, 2006); earthquake lights:
(Derr, 1973;St-Laurent, 2000); fog, haze and cloud formation:
(Aleksandrov et al., 2001;Guo and Wang, 2008); and unusual
animal behavior: (Tributsch, 1984).
If these diverse and seemingly disjoint signals are truly
precursory, the question arises how their generation may be
linked to the earthquake preparation process.
In this report we present data from laboratory experiments
that can help us gain insight into the generation of several of these
pre-earthquake signals. Specifically we demonstrate that massive
air ionization can take place at the surface of rocks, which are
being stressed at one end. The build-up of stress within the Earth’s
crust prior to major earthquakes may likewise lead to processes at
the Earth’s surface, including massive air ionization, which can be
expected to cause ionospheric perturbations and a host of other
phenomena.
1.1. Stress-activated electronic charge carriers
Laboratory studies have shown that, when deviatoric stresses
are applied to igneous or high-grade metamorphic rocks, electro-
nic charge carriers are activated (Freund et al., 2006). These
charge carriers are (i) electrons and (ii) defect electrons or holes,
the latter also known as positive holes or pholes for short. Both
electrons and pholes derive from pre-existing defects in the
matrix of minerals in igneous and high-grade metamorphic rocks,
ARTICLE IN PRESS
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/jastp
Journal of Atmospheric and Solar-Terrestrial Physics
1364-6826/$ - see front matter &2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jastp.2009.07.013
Corresponding author at: San Jose State University, SETI Institute, Department
of Physics, 515 N Whisman Road, CA 95192-0106, Mountain View, CA 94043,
United States. Tel.: +1650 604 5183; fax: +1 650 604 4680.
E-mail address: friedemann.t.freund@nasa.gov (F.T. Freund).
Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–1834
Author's personal copy
ARTICLE IN PRESS
specifically peroxy defects, O
3
X–OO–YO
3
, where X Y ¼Si
4+
,Al
3+
etc. The peroxy defects in turn derive from small amounts of O
3
X–
OH incorporated into nominally anhydrous minerals when they
crystallize from H
2
O–laden magmas or recrystallize in H
2
O–laden
high temperature metamorphic environments. The incorporation
of ‘‘impurity’’ hydroxyl can be written as hydrolysis of an O
3
X–O–
YO
3
bond:
O
3
X–O–YO
3
+H
2
O3O
3
X–OH HO–YO
3
(1)
During cooling, at temperatures below about 500 1C, hydroxyl
pairs rearrange electronically in such a way as to change the
valence of their oxygens from O
2
to O
, while reducing their
protons from H
+
to H. The H combine to molecular H
2
, while the
O
combine to form the peroxy bond. (Freund, 1985):
O
3
X–OH HO–YO
3
)O
3
X–OO–YO
3
+H
2
(2)
During mechanical deformation, dislocations are mobilized
and/or generated, sweeping through the mineral grains. When
they intersect peroxy defects, the O
–O
bonds break (Freund,
2002). In the process an O
2
from outside the peroxy bond
donates an electron e
0
, which is captured by the broken peroxy
bond. The donor O
2
thereby turns into O
, e.g. physically a defect
electron in the O
2
matrix, h
d
, a mobile electronic charge carrier:
O
3
XOO YO
3
þ½XO
4
4
e
0
donor
3O
3
XOOYO
3
captured e
0
þ½XO
4
3
phole h
100
ð3Þ
Here the donor O
2
is represented by a structural unit [XO
4
]
4
changing to [XO
4
]
3
.
As an electronic state, the h
d
‘‘live’’ at the upper edge of the
valence band, which consists primarily of O 2sp-symmetry energy
levels. The h
d
can propagate along the valence band, presumably
by phonon-assisted electron hopping. Since the valence bands of
all grains in a rock form an energetic continuum, the h
d
can cross
grain boundaries and propagate through sand and soils. Theore-
tically the phase velocity of an h
d
wave is given by the phonon
frequency
n
10
12
s
1
, times the hopping distance l between
oxygen sites, 2.8 10
10
m. Hence the h
d
phase velocity should
be
n
lE280 m s
1
, consistent with the measured speed of propaga-
tion of h
d
waves through different igneous rocks, 3007100 m s
1
(Freund, 2002).
1.2. Rock battery
When stress is applied to a portion of a rock, the number
density of electrons and pholes inside the stressed rock volume
increases. The h
d
charge carriers can flow out of the stressed rock
and into an adjacent unstressed rock, while the electrons, e
0
, stay
behind. The reason for the electrons staying behind is that there
are no energy levels in the unstressed rock, which they could
access. Fig. 1 shows a block of rock that is stressed at one end.
With the outflow of h
d
a potential difference develops between
the stressed and unstressed rock, equivalent to a battery voltage.
The h
d
outflow continues until an equilibrium is reached between
the electric field and the h
d
concentration gradient.
The situation is analogous to an electrochemical battery, which
delivers two types of charge carriers, electrons and cations. The
electrolyte allows cations to flow out. Thus, the electrolyte turns
positive. For electrons to also flow out, a metal contact must be
attached somewhere to the electrolyte. In the set-up depicted in
Fig. 1 the circuit is not closed.
1.3. Air ionization
An additional characteristic feature in Fig. 1 is the build-up of a
surface charge. It forms because, as the h
d
enter the unstressed
rock, they repel each other electrostatically. They build-up a
surface potential, which is a function of (i) the number density of
h
d
, and (ii) the dielectric constant (King and Freund, 1984).
The electric field reaches 40 0,000 V cm
1
on a flat surface of a
dielectric with a dielectric constant
e
¼10 containing 10
18
cm
3
h
d
charge carriers, equivalent to 100 ppm. At edges and corners,
where the radius of curvature is small, the electric field will be
much higher, potentially exceeding the dielectric strength of air.
This raises the question whether it might be possible to field-
ionize air molecules at the rock surface.
2. Experimental
We used gabbro from Shanxi, China, a typical deep crustal,
igneous rock, chemically identically to basalt, with 40 modal%
plagioclase, 30% augitic clinopyroxene surrounded by alteration
rims to amphibole and chlorite, plus 25% opaques, a porosity of
0.3%, and o1% total water, mostly due to hydroxyl-bearing
minerals such as amphiboles.
The experiments were conducted with 30 15 10 cm
3
blocks
with one polished surface and all other surfaces saw-cut. The
blocks were placed inside an aluminum box (50 30 30 cm
3
)
acting as a Faraday cage and fitted with a steel bellow to apply the
load. The pistons were in electrical contact with the rock but
insulated from the Faraday cage and the hydraulic press by
polyethylene sheets with 410
14
O
resistance as depicted in Fig.
2a. Hardened stainless steel ball bearings, 6.3 mm diameter, were
Fig. 1. Stress applied to one end of a rock activates electrons and hole, e
0
and h
d
.
The h
d
flow out of the stressed rock volume into the unstressed volume, creating a
potential difference. The situation is similar to that of an open circuit electro-
chemical battery. The stressed volume is negative and the unstressed volume is
positive. The h
d
charge carriers become trapped at the surface, leading to a positive
surface charge.
Fig. 2. (a) Configuration of the gabbro block inside the Faraday cage. (b) Detail
view of the stainless steel pistons with rowsof stainless steel ball bearings to act as
stress concentrators.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–1834 1825
Author's personal copy
ARTICLE IN PRESS
glued onto the faces of the pistons as depicted in Fig. 2b. The ball
bearings acted as stress concentrators causing the rocks to
massively deform at the contact points. Because of the stress
concentrators, we do not give the stresses in MPa but only report
the load in pounds [lbs]. We loaded the rocks moderately fast,
200–300 lbs/s until failure, using a manually controlled hydraulic
press.
During earlier impact experiments, we observed positive
surface potentials accompanied by corona discharges along the
rock edges (Freund, 2002). Similar positive surface potentials
appeared in response to slow application of stress (Takeuchi et al.,
2006). Values as high as +12 to +17 V were reported at high stress
rates (Enomoto et al., 1993). In the present experiments, to detect
surface potentials and measure airborne ions, we used an Al sheet
(10 20 cm
2
, 1mm thick) as a capacitor plate or ion collector
respectively. The Al sheet was placed above the rock on pieces of
Styrofoam glued to the inside walls of the Faraday cage extending
over the edges as depicted in Fig. 3. The air gap was 5mm. To
collect positive and negative ions we biased the Al sheet at 90
and +90 V respectively, using ten 9 V batteries connected in series.
Two photodiodes were aligned along one edge of the rock as
depicted in Fig. 2a to capture flashes of light resulting from corona
discharges.
Ion currents and surface potentials were recorded with a
Keithley 487 picoammeter and a Keithley 617 electrometer
respectively, using LabView 7.1. The photodiode output was
recorded on a Tektronix TDS 224 oscilloscope at 100 MHz.
Fig. 3a shows the Faraday cage (with the front half of the cover
removed) at the end of experiment #37 after the rock had slightly
tilted. It shows the ion collector plate, which had initially been
5 mm above the rock and parallel to the rock surface. Fig. 3b
shows details of the pistons with the ball bearings after a wedge-
shaped section of the rock had cleaved off. Also seen are the
polyethylene sheets used to electrically insulate the pistons from
the press and one of the leads connecting the pistons to the
picoammeter or electrometer.
3. Results
3.1. Surface potentials
Fig. 4 shows an example of the surface potential using the Al
plate as capacitor. The first sign of a positive potential appeared
when the platen of the hydraulic press started to move upward
(arrow 1). The stress caused by this modest acceleration was
enough to activate some h
d
charge carriers, which led to a positive
surface potential of +300 mV. When the piston of the press made
contact, the potential rose to +700 mV (arrow 2). When the press
began loading the rock, about 200 s into the run (arrow 3), the
Fig. 3. (a/b): two inside views of the Faraday cage after completion of experiment #37, in which an approximately triangular portion the gabbro block split off laterally,
causing the rock to only slightly tilt upon failure. (a) Position of the Al sheet used as ion collector or as capacitor place. (b) The pistons with stainless steel ball bearing stress
concentrators and the polyethylene sheets.
Fig. 4. Surface potential. Double arrow 1: the surface potential started to build-up
as the platen of the press started to rise and caused enough acceleration to activate
some charge carriers. Double arrow 2: first contact with piston of the press. Double
arrow 3: start of loading. The rock did not break at the end of this run. The surface
potential reached +3.4 V, and soon dropped to negative values (0.3V).
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–18341826
Author's personal copy
ARTICLE IN PRESS
surface potential rose rapidly reaching +3V around 10,000 lbs. The
surface potential remained high with further loading but
fluctuated between +2.6 and +3.4 V. Discontinuities occurred in
the loading curve, the first at about 280 s. They mark moments
when the stainless steel ball bearings, e.g. the stress
concentrators, are sinking into the rock, reducing temporarily
the hydraulic pressure. Rock deformation episodes of this type
were observed during all experiments described here.
Close to 300 s into the run, the surface potential around +3 V
abruptly broke down as depicted in Fig. 4, turning negative to
around 0.3 V with intermittent spikes. This general behavior,
first a strong positive surface charge followed by abrupt transition
to slightly negative values, was consistently observed during these
experiments.
3.2. Positive air ions
To record positive ion currents across the air gap the Al plate
was biased at 90 V. Fig. 5a/b shows a typical run with the inset
depicting the electric circuit. Before loading, the background ion
current was in the low pA range. It remained low at low loads.
However, when the load approached 10,000 lbs, after the ball
bearings had already sunk into the rock as indicated by the
discontinuities in the load vs. time curve, a positive air ion current
started to flow. Between 10,000 and 25,000 lbs several load vs.
time discontinuities occurred. The positive air ion current
increased, fluctuating in the 10–25 nA range. The discontinuities
in the load vs. time curve are not correlated to spikes in the ion
current. The area of the collector plate was 200 cm
2
. An ion
current of 20 nA corresponds to an average ion production rate on
the rock surface on the order of 10
9
cm
2
s
1
.
About 2 s before failure, at about 30,000 lbs, a 55 nA spike
occurred, indicating a burst of positive ions from the rock surface.
The rock failed at the 75.5 min time mark. The rock slightly tilted
but did not touch the collector plate. The collector plate continued
to measure an ion current, which increased sharply to 450 nA and
then decreased exponentially over the next 30 s.
Fig. 6 shows another run under negative bias. In this case the
positive ion current started soon after loading, but the overall
level remained moderate at around 10 nA (left inset). In spite of
several episodes with the ball bearings sinking into the rock, the
positive ion current did not increase. It even dropped to near-
background level between 785 and 800 s. At 815 s a sharp,
prominent current spike occurred, reaching 185nA. The rock
failed at 830 s, splitting across its width, causing the rock to drop
from underneath the ion collector plate without touching it. The
ion current increased moderately after failure and then decreased
exponentially over the next 30 s (right inset).
The recurring high levels of positive ions immediately after
fracture and their decay with a characteristic decay time around
30 s suggest that, during fracture, bursts of ions are generated,
filling the Faraday box. The ions slowly drifted to the walls to be
neutralized.
The short current spikes during loading are distinctly different.
Observed often but without clear correlation to the episodes of
rapid rock deformation, these spikes typically last for only a few
seconds as depicted in Fig. 6. Since corona discharges had been
observed during earlier impact experiments (Freund, 2002), we
changed the bias to +90 V to collect electrons and negative ions. In
addition, light emission was monitored with the pair of photo-
diodes situated as depicted in the inset in Fig. 2a.
3.3. Corona discharges
The results obtained under +90 V bias were characteristically
different from those under 90 V bias. Positive ions typically
started to form at relatively low loads as demonstrated in Figs. 5
Fig. 5. (a/b): positive ion current during deformation of gabbro. The discontinuities in the load vs. time curves are caused by a drop in the oil pressure of the hydraulic press
when the stainless steel balls are sinking into the rock. (a): Before failure; (b): whole run.
Fig. 6. Example of a run under 90 V bias marked by several massive deformation
events but low levels of positive air ions. This run produced a particularly large
narrow spike in the positive ion current at 815 s, indicative of a corona discharge.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–1834 182 7
Author's personal copy
ARTICLE IN PRESS
and 6. By contrast negative ions and/or electrons recorded under
+90 V bias appeared only at higher loads, typically
1
2
to
2
3
the load
needed to cause failure.
Fig. 7 shows an example. Up to 20,000 lbs, the current
remained at background levels, in the low pA range. Abruptly, at
the 2434 s mark, the current rose to 100–115 nA, followed by a
continuously high level around 65–70nA, accentuated by
multiple short spikes. The spikes are accompanied by light
pulses, each about 1.5ms long, one of which is shown in the
inset. Light pulses indicate corona discharges.
4. Discussion
The presence of electrically inactive, dormant precursory
peroxy defects in the structure of rock-forming minerals and of
the electronic charge carriers, which they engender when the
rocks are stressed, holds the key to many reported pre-earthquake
signals. Their discovery may be as fundamental for geophysics as
was the discovery of semiconducting properties in amorphous
materials for solid state and applied physics (Ovshinsky and Adler,
1976).
4.1. Surface potential, positive airborne ions and corona discharges
Stressing one end of a block of igneous rock such as gabbro
leads to a series of processes at the unstressed end. First, positive
surface potentials appear uniformly across the rock surface,
increasing rapidly with increasing stress and reaching about
+3 V. Second, massive amounts of positive airborne ions are
collected above the unstressed end of the rock. Third, massive
amounts of electrons and/or negative airborne ions are collected.
The positive surface potentials confirm observations reported
earlier (Freund, 2002) that, when an igneous rock is subjected to
deviatoric stresses, electronic charge carriers are activated. These
charge carriers are pholes h
d
activated as described by Eq. (3) in
Section 1.1. The activation is thought to involve dislocations that
are mobilized in the stressed rock volume. The dislocations
intersect pre-existing peroxy defects in the matrix of minerals and
cause them to break. Electrons e
0
are activated alongside the
pholes h
d
. However, while the pholes can flow out of the stressed
rock volume, the electrons are unable to follow suit. Hence, as
soon as h
d
flow out, the stressed rock charges negatively relative
to the unstressed rock. The unstressed rock becomes positively
charged.
The situation is analogous to that in an electrochemical battery
where cations spread into the electrolyte leaving behind a
negative charge. The electrons, unable to spread into the
electrolyte, can flow out via a metal contact. It is important to
note that, even without closing the circuit, a potential difference
develops, the ‘‘battery voltage’’.
The difference between an electrochemical battery and the
‘‘rock battery’’ as presented here is that the positive charge
carriers are not cations but positive holes, h
d
. Thus, when stressed,
the rock turns into a type of semiconductor battery, which has not
been previously described.
Positive charges on the surface of the rock appear even at low
stress levels as demonstrated in Fig. 4. Under these conditions the
number of h
d
charge carriers available to flow out of the stressed
rock volume may still be small but they suffice to build-up a
surface potential (Freund et al., 2006). As the rock is loaded more,
the surface potential increases, indicating a larger number of
charge carriers activated in the stressed rock volume. At the
moderately fast loading rates used here the surface potentials
reach rapidly values around +3 V.
Fig. 8a/b shows the surface potential and associated electric
fields calculated for a flat surface of a semi-infinite medium with
the dielectric constant
e
¼10 (King and Freund, 1984). Within the
range of charge carrier concentration under consideration the
surface potential is constant but the thickness (d) of the surface
charge layer decreases with increasing h
d
concentration. This
means that the electric field, which builds up directly at the solid–
air interface, increases as ddecreases. At a relatively low charge
carrier concentration of 10
17
m
3
(10 ppm), the calculated value
for Eis 120,00 0 V/cm. At a concentration of 10
18
m
3
, it increases
to 400,000 V/cm. At still higher concentrations or at edges and
corners, Eis expected to soon exceed the dielectric break-down
strength of air, 2–3 10
6
V/cm.
The electric fields due to the accumulation of charge carriers at
and a few tens nanometer below the surface are not the same as
the macroscopic vertical electric fields at the Earth surface that
are normally given in units of V/m (Brown, 1985;Pulinets, 2009).
The electric fields under consideration here are short-range and
act only over short distances across the surface-to-air interface.
Fig. 8c shows a Monte Carlo simulation with 1000 mobile
charges allowed to relax in a dielectric medium. The charge carrier
concentration is very low inside, low on flat surfaces but high
along edges and corners, predicting very high electric fields.
In a system as depicted in Fig. 1, where the battery circuit is not
closed, the charge carriers flowing out of the stressed rock volume
become stagnant in the unstressed rock. They should create a
uniform and constant surface potential independent of the
number of h
d
charge carriers activated (King and Freund, 1984).
However, as Fig. 4 shows, after the surface potential had increased
to +3 V, it started to fluctuate between +2.6 and +3.4 V. Such
fluctuations are consistent with a break-down of the open circuit
approximation.
This suggests that, around +3 V, the electric field reaches high
enough values to extract electrons from neutral gas molecules, e.g.
to field-ionize them, in particular along edges and corners as
depicted in Fig. 9. The most likely candidates for field-ionization
are O
2
and H
2
O, which have relatively low ionization potentials.
The measured ion current as shown in Fig. 5a is on the order of
20 nA, equal to an ion generation rate 10
9
cm
2
s
1
. For each air
molecule that is field-ionized at the rock surface, an electron is
deposited into the rock surface. This must lower the positive
surface potential.
The charge carrier density needed to produce a 1V surface
potential is on the order of 10
9
cm
2
(Takeuchi and Nagahama,
Fig. 7. Current recorded with +90 V bias on the ion collector plate, indicating free
electrons and/or negative ions. Inset: light flash captured by the photodiode during
the large ion current spike confirming a corona discharge.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–18341828
Author's personal copy
ARTICLE IN PRESS
2002). Therefore, if 10
9
cm
2
s
1
air molecules become field-
ionized, they deposit 10
9
cm
2
s
1
electrons into the rock surface.
This is obviously sufficient to cause a significant reduction of the
positive surface potential. At the same time, more h
d
charge
carriers stream from the stressed portion of the rock to the
unstressed portion and, hence, to the surface. These newly
arriving h
d
rebuild the positive charge of the surface almost as
fast as it is reduced. This causes the positive surface potential to
fluctuate wildly as shown in Fig. 4.
Fig. 4 also shows that, at still higher load, the positive surface
potential drops abruptly and turns negative. The emission of light
shown in the inset in Fig. 7 indicates corona discharges. This
reversal of the sign, together with the light blips, suggests the
production of electrons by the corona discharges. These electrons
‘‘rain down’’ onto the rock surface and annihilate the positive
surface charge more effectively than the field-ionization of air
molecules was able to do. In addition, through attachment to
neutral gas molecules, the electrons form negative airborne ions.
Corona discharges are expected to manifest themselves in air
ion current bursts, both under negative and positive bias.
Concurrent electronic excitation and electron–ion recombination
reactions will lead to the emission of visible light.
Ionization of air over the unstressed portion of the rock
explains the break-down of the open circuit approximation of
Fig. 1. As illustrated in Fig. 10, while electrons are delivered to the
surface and recombine with h
d
charge carriers, positive ions will
drift toward the negatively charged pistons, which are in contact
with the stressed rock. This closes the battery circuit.
In summary, the laboratory experiments, designed to measure
surface potentials and airborne ion currents, have provided
Fig. 8. Surface potential (a) and electric field (b) calculated for the surface of a semi-infinite insulator with a dielectric constant of 10 with charge carrier densities of 1 in
100,000 and 1 in 10,000 (King and Freund, 1984). (c): Simulation of 1000 charge carriers inside a dielectric medium,
e
¼10, after reaching an equilibrium distribution due to
their mutual repulsion in the bulk. The charge carrier concentration is projected onto the basal plane.
Fig. 9. Conceptual representation of field-ionization of air molecules at the rock
surface, in particular at edges and corners, where the h
d
densities are highest and
where the electric field will be high enough to extract an electron from a gas
molecule.
Fig. 10. Massive air ionization on the unstressed part of the rock will produce an
electric current running through the air and closing the battery circuit.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–1834 1829
Author's personal copy
ARTICLE IN PRESS
insight into processes at the rock surface when a portion of the
rock is subjected to deviatoric stresses. With increasing stress the
following processes occur sequentially at or above the rock
surface:
(i) trapping of h
d
charge carriers and appearance of positive
surface charges;
(ii) field-ionization of air molecules and generation of positive air
ions; and
(iii) corona discharges with bursts of ion current and flashes of
light.
4.2. Air ionization and pre-earthquake phenomena
We can use these laboratory observations to address phenom-
ena, which have been observed in the field and have been linked
to impending seismic activity.
We start from the simple, but very basic premise that
earthquakes represent the final stage of a long drawn-out process
deep below, in the future hypocentral volume, marked by
increasing levels of deviatoric stresses. How the stresses evolve,
how they increase, decrease or change directions will vary
enormously from case to case. However, close to a catastrophic
rupture, the stresses in the hypocentral volume can be expected to
locally concentrate and to increase fast. Another generally valid
statement is that the hypocentral volume, where the stresses
evolve toward catastrophic rupture, will always be surrounded by
a larger volume of less stressed or unstressed rocks.
Therefore, the experimental set-up used in the experiments
described here can be used to draw parallels to pre-earthquake
situations in the Earth’s crust. This is particularly true for the
electric analog of an open battery circuit with the outflow of
h
d
charge carriers from the most stressed rock volume being the
dominant process.
Under such conditions h
d
charge carriers activated deep below
are expected to spread out of the prospective hypocentral volume
into the surrounding less stressed or unstressed rocks, while the
electrons cannot flow out. Some h
d
will travel upward and reach
the Earth’s surface. The higher the stresses in the hypocentral
volume, the larger the number of h
d
charge carriers flowing out
and the larger the number of h
d
charge carriers reaching the
Earth’s surface. With more h
d
charge carriers arriving, the electric
fields at the surface will increase. Eventually those fields can be
expected to reach values high enough to initiate air ionization at
the Earth’s surface (leading to positive ions) and corona
discharges (leading to free electrons and negative ions alongside
positive ions).
The measured ion currents suggest rates on the order of
10
9
–10
10
ionization events per second and cm
2
. Extrapolated to
km
2
, this is equivalent to an ion current on the order of 10–10 0 A
rising off the Earth surface. Such currents, if they occur in nature,
could constitute a significant part of the global electric circuit,
which connects the solid Earth with the ionosphere (Roble and
Tzur, 1986).
4.3. Perturbations in the ionosphere
Using a ground-based ionosonde to determine the total
electron content (TEC) in the ionospheric foF2 layer, anomalies
were recognized two to three days before most of the MZ6.0
earthquakes in the Taiwan area during 1994–1999 (Liu et al.,
2000). Likewise, using GPS receivers, ionospheric anomalies were
found to occur prior to 16 out of 20 MZ6.0 earthquakes in the
Taiwan region between 1999 and 2002 (Liu et al., 2006). One day
before the M¼6.8 Kythira earthquake in Greece on January 8,
2006, during a quiet period marked by the absence of magnetic
storms, the TEC values above the epicentral region were found to
have increased significantly, up to 50% relative to the surrounding
region and the longtime average (Zakharenkova et al., 2007). In
Japan statistically significant changes in the subionospheric
propagation of low frequency radio waves prior to earthquakes
have been recorded (Maekawa et al., 2006) as well as sporadic
ionospheric E layer anomalies prior to the M¼7.2 Kobe earth-
quake of January 16, 1995 (Ondoh, 2003), and anomalies prior to
the M¼8.3 Tokachi-oki of September 25, 2003 off the coast of
Hokkaido (Hayakawa et al., 2005). Using satellite and ground-
based GPS data anomalous fluctuations in the integrated TEC have
also been described prior to the M¼7.6 Bhuj earthquake in
Gujarat, India (Trigunait et al., 2004) and other events on the
Indian subcontinent (Singh and Singh, 2007). Similar ionospheric
variations were seen prior to the M¼7.8 Colima earthquake of
January 21, 2003 in Mexico (Pulinets et al., 2005).
The ionospheric anomalies extend over several hundred to a
few thousand kilometers. However, no consensus has been
reached as to the cause or causes (Rishbeth, 2007). If a process
exists that links the ground to the ionosphere, powerful enough to
give rise to the reported perturbations, it must extend over large
areas and it will most likely involve changes in thermospheric
chemistry or in the electrodynamic drifts in the lower atmo-
sphere.
Radon emission from the ground has been widely quoted as a
possible cause for changes in the conductivity of the air linked to
pre-earthquake ionospheric anomalies (Ondoh, 2003;Pulinets,
2007, 2009;Yasuoka et al., 2006). The basic concept is that, when
stresses build-up across the ‘‘earthquake preparation zone’’,
radioactive radon is released from the ground. The ‘‘earthquake
preparation zone’’ is assumed to be centered around the future
epicenter with a radius rgiven by the empirical relation
r¼10
0.43 M
km, where M is the magnitude of the earthquake
(Dobrovolsky et al., 1979;Teisseyre, 1997). For magnitude 6, 7 and
8 events, the ‘‘earthquake preparation zones’’ would therefore
cover areas as large as 50 0, 600 and 700 km across.
222
Rn with its half-life of 3.82 days is a progeny of radium,
226
Ra, which in turn derives from
238
U, an element enriched in
granitic rocks.
222
Rn decays by emitting 5.49 MeV alpha particles
to
218
Po with a half-life of 3.11 min, which decays by emitting
6.00 MeV alpha particles to short-lived
214
Pb and other progeny. In
air, each alpha particle of the
222
Rn decay generates 150,000 to
200,000 electron–ion pairs. The entire
222
Rn decay chain in air
therefore produces at most 10
6
ionization events.
While 10
6
ionization events look like a large number, the rate
at which radon is released from the ground, even in areas
dominated by granitic rocks, is very small. On the average, air ion
concentrations due to radon and its progeny range from about
25 cm
3
to 250 cm
3
. Contributions from cosmic rays and
secondary cosmic ray decay products at sea level are equally
low, on the order of 2 ionscm
3
and 15 ions cm
3
respectively
(Hoppel et al., 1986).
Measured over a 3-year period near Pune, India, the median
radioactivity in air has reported to be 9.70 Bq m
3
due to radon
itself and 2.84 Bq m
3
due to its progeny, producing a median air
ion concentration of only 5.5 cm
3
s
1
(Nagarajaa et al., 2003).
Using a network of 20 radon measuring stations along the western
part of the North Anatolian Fault, placed into 0.5–1m deep
trenches or holes, the count rates due to radon decay over a period
of 1-year were found to vary from less than 0.3 to about 6 min
1
(_
Inan et al., 2008). At the same time, a total of 19 earthquakes of
magnitudes 4–5 occurred in the region, some less than
20 km from a radon measuring station. At some stations M44
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–18341830
Author's personal copy
ARTICLE IN PRESS
earthquakes correlate with a small increase in radon release from
the ground but, more generally, the radon release patterns are
found to be broad and not directly correlated to any specific
seismic event.
In California radon transects were taken across creeping,
locked, and freshly ruptured sections of the San Andreas Fault
(King et al., 1993). Along the actively creeping section the radon
release increased 6–11 times over the typically low background
values. Increased radon release was also found over the fault itself,
only tens of meters wide, indicating a very localized effect.
It thus appears that, while radon is definitely coming out of the
ground, its release rate increases only by a factor of about 10 in
seismically active regions, mostly in the immediate neighborhood
of active faults. Hence, the contribution of radon to the air
conductivity is of minor importance, in particular when averaged
over areas as large as those predicted by the ‘‘earthquake
preparation zone’’ concept and necessary to cause the reported
pre-earthquake ionospheric perturbations.
By contrast, much higher rates of air ionization can be
expected to occur when a large number of h
d
charge carriers,
stress-activated at depth, arrive at the Earth surface. If our
laboratory experiments are any indication of the potential
magnitude of the effect, air ionization rates on the order of
10
9
cm
2
s
1
and higher appear possible. Even if a large fraction of
these ions are lost to recombination events, air ion concentrations
orders of magnitudes higher than those achievable by radon decay
are to be expected.
High concentrations of airborne ions have indeed been
reported. Prior to the October 30, 2007 Alum Rock, California,
M5.4 earthquake ultra low frequency (0.01–12Hz) pulsations
were detected with a three axis induction magnetometer located
2 km from the epicenter. The 1–12 s wide pulsations were 10–50
times more intense than 2-year background noise levels, and
occurred 10–30 times more frequently in the 2 weeks prior to the
event than during the previous 22 months. An air conductivity
sensor at the same site saturated for over 13 h during the night
and much of the day before the earthquake. Compared to the
previous year’s average, conductivity patterns at the site the
conductivity was determined not to be caused by moisture
contamination (Bleier et al., 2008, 2009). An increase in the air
ionization was recorded at the same time at another station about
40 km from the epicenter, e.g. within the range of the ‘‘earthquake
preparation zone’’.
An extended network of dedicated air conductivity sensors in
Japan, the PISCO network, has also produced a large amount of
data pointing at episodes of dramatically increased air ionization,
tentatively associated with seismic activity within a radius given
by the ‘‘earthquake preparation zone’’ (Hattori et al., 2008;Wasa
and Wadatsumi, 2003). One such record obtained by a station at
Kanagawa near Yokohama on the Tokyo Bay is given in Fig. 11. The
28 day long record shows a series of positive ion maxima, mostly
by small ions, occasionally accompanied by negative ion maxima,
mostly by large ions. Both positive and negative ions reach
concentrations on the order of 450,000 cm
3
. It is interesting to
note that, while there are sometimes maxima of just positive ions,
the negative ions seem to form only as a corollary to the positive
ions. This is consistent with the sequence of first positive and then
negative ion formation identified in our laboratory experiments.
If airborne ion concentrations at ground level increase
significantly over such large areas, by many orders of magnitude
over the normal background level, the ions can be expected to be
transported upward, mostly due to thermal convection currents
that will result from the release of the latent heat of condensation
of water, when the ions act as condensation nuclei. Thus the
electrical conductivity profile across the entire lower and middle
atmosphere will be modified. This in turn will affect the electric
field distribution between the lower edge of the ionosphere and
the ground (Sorokin et al., 2006).
4.4. Thermal infrared anomalies
Night-time infrared (IR) satellite images have shown areas of
enhanced IR emission from the epicentral regions prior to major
earthquakes (Saraf et al., 2008a, b;Tramutoli et al., 2005;Tronin,
2000). Known as ‘‘thermal anomalies’’, they are also large areas,
often 100–500 km across. They also reportedly appear a few days
before major seismic events. They often fluctuate rapidly in areal
extent and intensity. They reportedly disappear soon after the
main shock and major aftershocks.
The rapidity with which the thermal anomalies appear and
disappear rules out that they are caused by Joule heat rising from
a source below and heating the rocks at the Earth’s surface.
Several alternative processes have been invoked to account for the
apparent temperature increase: rising fluids, due to regional
stresses, leading to the emanation of warm gases (Gornyi et al.,
1988), diffuse CO
2
release from the ground, causing a local
greenhouse effect (Quing et al., 1991;Tronin, 2000), and air
ionization due to radon leading to the release of latent heat during
condensation of water vapor (Pulinets et al., 2006). However, none
of these explanations seem to adequately account for the
characteristic features of the ‘‘thermal anomalies’’ as reported
from satellite data.
A very different process has been proposed based on the activation
of h
d
charge carriers and their diffusion to the surface (Freund et al.,
2007). As outlined above in Eqs. (2) and (3), the dormant precursor of
the h
d
charge carriers consists of peroxy links in which two O
are
tightly bonded together. Breaking the peroxy link costs energy,
estimated to be on the order of 2.4 eV. This energy is supplied by the
mechanical work done in the stressed rock volume and leads to the
activation of h
d
. When the h
d
arrives at the surface, they can
recombine to restitute peroxy bonds. When this happens, energy on
theorderof2.4eVwillbeimpartedtothetwoO
participating in the
recombination process. The two O
will thus be ‘‘born’’ in a
vibrationally highly excited state, equivalent to 30,000 K. The best
way for the two O
to get rid of this large excess vibrational energy is
to radiatively de-excite by emitting photons corresponding to the
transitions between the quantum levels of the vibrational manifold of
the peroxy bond.
The energy for the transition from the 1st excited level to the
0th ground level is known (Ricci et al., 2001). It is equivalent to a
narrow emission band at 930 cm
1
or 10.7
m
m. Accordingly,
transitions between higher excited levels, 2 )1, 3 )2, 4 )3,
etc., will lead to a series of narrow IR emissions on the long
wavelength side of the 10.7
m
m band as indeed observed. Because
these narrow non-thermal IR bands occur within the spectral
window of the broad thermal 30 0 K emission maximum and
current satellite IR sensors cannot spectrally resolve them, they
have been lumped together with the thermal emission. Hence the
name ‘‘thermal anomalies’’.
Rapid fluctuations, both in areal extent and intensity, are a
characteristic feature of pre-earthquake ‘‘thermal anomalies’’
recorded by IR satellites. Similar fluctuations in the IR emission
have been noted during the laboratory study (Freund et al., 2007).
In both situations the fluctuations may be due to waves of h
d
charge carriers arriving at the surface, field-ionizing air molecules
and thereby causing fluctuations of the positive surface potential
as shown in Fig. 4. The surface potential fluctuations in turn lead
to fluctuations in the number of h
d
that might be available for
recombination and, hence, IR emission.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–1834 1831
Author's personal copy
ARTICLE IN PRESS
4.5. Earthquake lights
Fleeting, short-lived luminous phenomena arising from the
ground and apparently related to earthquake activity have been
reported since ancient times (Derr, 1973;Tributsch, 1984). They
are often referred to as earthquake lights, EQLs. Based on reports
from several events in Japan (Musya, 1931) stated: ‘‘the observa-
tions were so abundant and so carefully made that we can no
longer feel much doubt as to the reality of the phenomena’’.
Nonetheless, doubts persisted in the scientific community at least
until the late 1960s when EQLs were photographically documen-
ted during an earthquake swarm near Matsushiro, Japan (Yasui,
1973). EQLs have been reported from Mexico (Araiza-Quijano and
Hern
andez-del-Valle, 1996), Canada (St-Laurent, 2000), and other
seismically active regions of the world.
The most common explanations for EQLs invoke piezoelec-
tricity as the physical process that can produce electric fields
strong enough to cause a dielectric break-down of air and
luminous discharges (Finkelstein et al., 1973;Johnston, 1991;
Ouellet, 1990) or frictional heating of the fault during rupture
(Lockner et al., 1983). However, on the basis of the experiments
presented here a more likely explanation is that electric
discharges at the Earth’s surface are caused by h
d
charge carriers
arriving at the ground–air interface, building up sufficiently
strong electric fields to cause field-ionization of air molecules
and corona discharges. Such ionization events are expected to
occur over areas as large as h
d
charge carriers are able to spread
after they have been stress-activated at depth. Pervasive corona
discharges may cause luminous phenomena and may also be the
cause for increased noise in the radiofrequency range that
reportedly disrupted the telemetric data transfer from a seism-
ometer network in India prior to an earthquake swarm at a
distance of about 150 km (Kolvankar, 2001).
Local outbursts of light from the ground, often called EQLs, may be
due to a condition in the Earth’s crust that is theoretically predicted to
occur when the number density of h
d
in the rocks increases to a point
where the electronic wavefunctions of the h
d
charge carriers begin to
overlap, creating a solid state plasma
(St-Laurent et al., 2006). Such plasmas are expected to be inherently
unstable, leading to a cloud of h
d
charge carriers traveling outward at
speeds around 200 m s
1
(Freund, 2002). When the wave front breaks
through the Earth’s surface, it will ionize the air and produce flashes
of light. Depending on conditions yet to be fully understood, such
process could also lead to ‘‘flames’’ coming out of the ground
(Demetrescu and Petrescu, 1942;Galli, 1910;Mack, 1912)orto
outbursts of light (St-Laurent, 2000).
4.6. Fog, haze and cloud formation
While there is a widespread tendency to associate fog, haze
and cloud formations prior to earthquakes with air ionization due
to radon emission (Araiza-Quijano and Hern
andez-del-Valle,
1996;Pulinets and Dunajeck, 2006;Pulinets et al., 2006), air
ionization at the rock–air interface due to stress-activated h
d
charge carriers appears to provide a more effective mechanism for
the generation of large numbers of airborne ions.
Since most airborne ions will be positive, at least during the
early phase of surface potential build-up, they will be repelled by
the surface and rise into the air. Acting as nuclei for the
condensation of water, they will cause the release of latent heat,
which will lead to thermal updrafts. Ionized and convectionally
unstable columns of air may thus form ‘‘streamers’’, which play a
role in triggering cloud-to-ground lightning strikes (Aleksandrov
et al., 2001).
4.7. Unusual animal behavior
High levels of positive air ions may also have biological
consequences. Although effects of positive and negative airborne
ions on humans and animals have been studied since the 1960s,
the results have not been equivocal due to design faults (Krueger
and Reed, 1976), or lack of proper controls (Hedge and Eleftherkis,
1982). However, a majority of these studies have found that high
levels of positive ions, ranging from 2 10
3
to 1 10
6
ions cm
3
,
have detrimental physiological and behavioral effects, including
respiratory problems, possibly due to the swelling of the trachea,
and increased levels of serotonin (5-hydroxytryptamine, 5-HT)
levels in the blood (Krueger and Kotaka, 1969).
High serotonin levels can result in irritation, headaches,
nervousness, and increased sensitivity to pain (Sulman et al.,
1974). These effects are consistent with reported pre-earthquake
changes in animal behavior. Laboratory data show changes in
mouse circadian rhythms 24h before earthquakes (Yokoi et al.,
2003). Other causes quoted in literature for changes in animal
behavior before earthquakes include low frequency vibrations,
ground tilting, humidity changes, and magnetic field variations
(Kirschvink, 2000).
Tributsch (Tributsch, 1982) suggested that changes in animal
behavior before earthquakes may be caused by charged aerosol
particles due to electrochemical glow. Hoenig (Hoenig, 1979) has
shown that electrons and positive ions with current levels up to
10pA are produced as rocks approach failure under laboratory
settings. Although, in both cases the phenomena were attributed
Fig. 11. Air ionization, broken down into positive and negative ions, small and large, as monitored by the PISCO network in Japan. Data from the Kanagawa Station for a 28
day period in June–July 2008.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–18341832
Author's personal copy
ARTICLE IN PRESS
to piezoelectric effects, it is important to note that piezoelectricity
always produces both positive and negative ions. The selective
production of positive airborne ions as demonstrated in this paper
may amplify the anomalous animal behavior before major earth-
quakes.
Acknowledgments
Supported by NASA grants ‘‘Earth Surface and Interior’’
#NNX08AG81G and ‘‘Exobiology’’ #NNX07AU04G, by the 2007
NASA Ames Academy, and a 2008 NSF–REU grant to the
Department of Physics, SJSU. We thank the staff of the NASA
Ames EEL (Engineering Evaluation Laboratory) Jerry Wang, Lynn
Hofland, and Frank Pichay.
References
Aleksandrov, N.L., Bazelyan, E.M., Carpenter, R.B., Drabkin, M.M., Raizer, Yu P., 2001.
The effect of coronae on leader initiation and development under thunder-
storm conditions and in long air gaps. J. Phys. D: Appl. Phys. 34, 3256–3266.
Araiza-Quijano, M.R., Hern
andez-del-Valle, G., 1996. Some observations of atmo-
spheric luminosity as a possible earthquake precursor. Geofis. Int. 35, 403–408.
Bleier, T., Dunson, C., Maniscalco, M., Bryant, N., Bambery, R., 2008. ULF Pulsations,
Air Conductivity Changes, and Infrared (IR) Radiation Signatures Observed
Prior to the 2008 Alum Rock (California) M5.4 Earthquake, AGU Fall Meeting.
Abstract S53B-1826.
Bleier, T., Dunson, C., Maniscalco, M., Bryant, N., Bambery, R., Freund, F.T..
Investigation of ULF magnetic pulsations, air conductivity changes, and
infrared signatures associated with the October 30, Alum Rock M5.4
earthquake. Nat. Hazards Earth Syst. Sci. in review.
Brown, L.W., 1985. Two experiments to measure the electric field at the Earth’s
surface and the variation of potential difference with height. Phys. Educ. 20,
287–291.
Demetrescu, G., Petrescu, G., 1942. Sur les ph
enom
enes lumineux qui ont
accompagn
e le tremblement de terre de Roumanie du 10 Novembre 1940.
Bull. Sect. Sci. 23, 292–296.
Derr, J.S., 1973. Earthquake lights: a review of observations and present theories.
Bull. Seismol. Soc. Am. 63, 2177–21287.
Dobrovolsky, I.P., Zubkov, S.I., Miachkin, V.I., 1979. Estimation of the size of
earthquake preparation zones. Pure Appl. Geophys. 117, 1025–1044.
Enomoto, Y., Akai, M., Hashimoto, H., Mori, S., Asabe, Y., 1993. Exoelectron
emission: possible relation to seismic geo-electromagnetic activities as a
microscopic aspect in geotribology. Wear 168, 135–142.
Finkelstein, D., Hill, U.S., Powell, J.R., 1973. The piezoelectric theory of earthquake
lightning. J. Geophys. Res. 78, 992–993.
Freund, F., 1985. Conversion of dissolved ‘‘water’’ into molecular hydrogen and
peroxy linkages. J. Non-Cryst. Solids 71, 195–202.
Freund, F., 2002. Charge generation and propagation in rocks. J. Geodyn. 33 (4–5),
545–572.
Freund, F.T., Takeuchi, A., Lau, B.W., 2006. Electric currents streaming out of
stressed igneous rocks—A step towards understanding pre-earthquake low
frequency EM emissions. Phys. Chem. Earth 31 (4–9), 389–396.
Freund, F.T., Takeuchi, A., Lau, B.W.S., Al-Manaseer, A., Fu, C.C., Bryant, N.A.,
Ouzounov, D., 2007. Stimulated thermal IR emission from rocks: assessing a
stress indicator. eEarth 2, 1–10.
Galli, I., 1910. Raccolta e classifzione di fenomeni luminosi osservati nei terremoti.
Bolletino Soc. Sismol. Ital. 14, 221–448.
Geller, R.J., 1997. Earthquake prediction: a critical review. Geophys. J. Int. 131, 425–
450.
Gornyi, V.I., Salman, A.G., Tronin, A.A., Shilin, B.B., 1988. The Earth’s outgoing IR
radiation as an indicator of seismic activity. Proc. Acad. Sci. USSR 301 (1), 67–
69.
Guo, G., Wang, B., 2008. Cloud anomaly before Iran earthquake. Int. J. Remote
Sensing 29 (7), 1921–1928.
Hattori, K., Wadatsumi, K., Furuya, R., Yada, N., Yamamoto, I., Ninagawa, K., Ideta, Y.,
Nishihashi, M., 2008. Variation of radioactive atmospheric ion concentration
associated with large earthquakes, AGU Fall Meeting. Abstract S52A-03.
Hayakawa, M., Sazhin, S.S., 1992. Mid-latitude and plasmaspheric hiss: a review.
Planet. Space Sci. 40, 1325–1338.
Hayakawa, M., Shvets, A.V., Maekawa, S., 2005. Subionospheric LF monitoring of
ionospheric perturbations prior to the Tokachi-oki earthquake and a possible
mechanism of lithosphere–ionosphere coupling. Adv. Polar Upper Atmos. Res.
18, 42–54.
Hedge, A., Eleftherkis, E., 1982. Air ionization: an evaluation of its physiological and
psychological effects. Ann. Occup. Hyg. 25 (4), 409–419.
Hoenig, S.A., 1979. Aerosol anomalies preceding earthquakes. Nature 279 (5709),
169.
Hoppel, W.A., Anderson, R.V., Willet, J.C., 1986. Atmospheric Electricity in the
Planetary Boundary Layer, The Earth’s Electrical Environment. National
Academic Press, p. 149–165.
_
Inan, S., Akg ¨
ul, T., Seyis, C., Saatc-ılar, R., Baykut, S., Ergintav, S., Bas
-, M., 2008.
Geochemical monitoring in the Marmara region (NW Turkey): a search for
precursors of seismic activity. J. Geophys. Res. 113, B03401 03410.01029/
02007JB005206.
Johnston, A.C., 1991. Light from seismic waves. Nature 354, 361.
Kagan, Y.Y., 1997. Are earthquakes predictable?. Geophys. J. Int. 131 (3), 505–525.
Keilis-Borok, V., 2003. Fundamentals of earthquake prediction: four paradigms. In:
Keilis-Borok, V., Soloviev, A.A. (Eds.), Nonlinear Dynamics of the Lithosphere
and Earthquake Prediction. Springer, pp. 1–36.
King, B.V., Freund, F., 1984. Surface charges and subsurface space charge
distribution in magnesium oxide containing dissolved traces of water. Phys.
Rev. B 29, 5814–5824.
King, C.-Y., Zhang, W., King, B.-S.,1993. Radon anomalies on three kinds of faults in
California. Pure Appl. Geophys. 141, 111–124.
Kirschvink, J.L., 2000. Earthquake prediction by animals: evolution and sensory
perception. Bull. Seismol. Soc. Am. 90 (2), 312–323.
Kolvankar, V.G., 2001. Report BARC-2001/E/006: Earthquake sequence of 1991
from Valsad region, Gujarat. Bhabha Atomic Research Centre, Seismology Div.,
Mumbai, India.
Krueger, A., Kotaka, S., 1969. The effects of air ions on brain levels of serotonin in
mice. Int. J. Biometeorol. 13 (1), 25–38.
Krueger, A.P., Reed, E.J., 1976. Biological impact of small air ions. Science 193
(4259), 1209–1213.
Liperovsky, V.A., Pokhotelov, O.A., Liperovskaya, E.V., Parrot, M., Meister, C.-V.,
Alimov, O.A., 2000. Modification of sporadic E-layers caused by seismic
activity. Surv. Geophys. 21 (5–6), 449–486.
Liu, J.Y., Chen, Y.I., Chuo, Y.J., Chen, C.S., 2006. A statistical investigation of
preearthquake ionospheric anomaly. J. Geophys. Res. 111, A05304 05310.01029/
02005ja011333.
Liu, J.Y., Chen, Y.I., Pulinets, S.A., Tsai, Y.B., Chuo, Y.J., 20 00. Seismo-ionospheric
signatures prior to MZ6.0 Taiwan earthquakes. Geophys. Res. Lett. 27 (19),
3113–3116.
Lockner, D.A., Johnston, M.J.S., Byerlee, J.D., 1983. A mechanism for the generation
of earthquake lights. Nature 302, 28–33.
Mack, K., 1912. Das s ¨
uddeutsche Erdbeben vom 16. November 1911, Abschnitt VII:
Lichterscheinungen, W¨
urtembergische Jahrb¨
ucher f ¨
ur Statistik and Land-
eskunde. p. 131.
Maekawa, S., Horie, T., Yamauchi, T., Sawaya, T., Ishikawa, M., Hayakawa, M., Sasaki,
H., 2006. A statistical study on the effect of earthquakes on the ionosphere,
based on the subionospheric LF propagation data in Japan. Ann. Geophys. 24,
2219–2225.
Musya, K., 1931. On the luminous phenomena that attended the Idu earthquake,
November 6th, 1930. Bull. Earthquake Res. Inst. 9, 214–215.
Nagarajaa, K., Prasad, B.S.N., Madhavaa, M.S., Chandrashekaraa, M.S., Paramesha, L.,
Sannappaa, J., Pawarb, S.D., Murugavelb, P., Kamra, A.K., 20 03. Radon and its
short-lived progeny: variations near the ground. Radiat. Meas. 36 (1–6), 413–
417.
Ondoh, T., 2003. Anomalous sporadic-E layers observed before M 7.2 Hyogo-ken
Nanbu earthquake; Terrestrial gas emanation model. Adv. Polar Upper Atmos.
Res. 17, 96–108.
Ouellet, M., 1990. Earthquake light and seismicity. Nature 348, 492.
Ovshinsky, S.R., Adler, D., 1976. Local structure, bonding, and electronic properties
of covalent amorphous semiconductors. Contemp. Phys. 19 (2), 109–126.
Pulinets, S., 2007. Natural radioactivity, earthquakes, and the ionosphere. EOS 88
(20)10.1029/2007EO200001.
Pulinets, S.. Physical mechanism of the vertical electric field generation over active
tectonic faults Adv. Space Res. in press.
Pulinets, S., Boyarchuk, K., 2004. Ionospheric Precursors of Earthquakes. Springer.
Pulinets, S.A., Dunajeck, M.A., 2006. Specific variations of air temperature and
relative humidity around the time of Michoacan earthquake M8.1 September
19, 1985 as a possible indicator of interaction between tectonic plates.
Tectonophysics 431 (1–4), 221–230.
Pulinets, S.A., Leyva Contreras, A., Bisiacchi-Giraldi, G., Ciraolo, L., 2005. Total
electron content variations in the ionosphere before the Colima, Mexico,
earthquake of 21 January 2003. Geofis. Int. 44 (4), 369–377.
Pulinets, S.A., Ouzounov, D., Karelin, A.V., Boyarchuk, K.A., Pokhmelnykh, L.A.,
2006. The physical nature of thermal anomalies observed before strong
earthquakes. Phys. Chem. Earth 31 (4–9), 143–153.
Quing, Z., Xiu-Deng, X., Chang-Gong, D., 1991. Thermal infrared anomaly—-
precursor of impending earthquakes. Chin. Sci. Bull. 36 (4), 319–323.
Ricci, D., Pacchioni, G., Szymanski, M.A., Shluger, A.L., Stoneham, A.M., 2001.
Modeling disorder in amorphous silica with embedded clusters: the peroxy
bridge defect center. Phys. Rev. B 64 (22) 224104-224101–224104-224108.
Rishbeth, H., 2007. Do earthquake precursors really exist?. EOS 88 (29)10.1029/
2007EO290008.
Roble, R.G., Tzur, I., 1986. The Global Atmospheric–Electrical Circuit, The Earth’s
Electrical Environment. National Academic Press, p. 206–231.
Saraf, A.K., Rawa, V., Banerjee, P., Choudhury, S., Panda, S.K., Dasgupta, S., Das, J.D.,
2008a. Satellite detection of earthquake thermal precursors in Iran. Natural
Hazards, doi:10.1007/s11069-11007-19201-11067.
Saraf, A.K., Rawat, V., Banerjee, P., Choudhury, S., Panda, S.K., Dasgupta, S., Das, J.D.,
2008b. Satellite detection of earthquake thermal infrared precursors in Iran.
Nat. Hazards 47, 119–135.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–1834 1833
Author's personal copy
ARTICLE IN PRESS
Shalimov, S., Gokhberg, M., 1998. Lithosphere–ionosphere coupling mechanism
and its application to the earthquake in Iran on June 20, 1990. A review of
ionospheric measurements and basic assumptions. Phys. Earth Planet. Inter.
105 (3–4), 211–218.
Singh, C., Singh, O.P., 2007. Simultaneous ionospheric E- and F-layer perturbations
caused by some major earthquakes in India. Ann. Geophys. 50 (1), 111–122.
Sorokin, V.M., Yaschenko, A.K., Hayakawa, M., 2006. Formation mechanism of the
lower-ionospheric disturbances by the atmosphere electric current over a
seismic region. J. Atmos. Solar Terr. Phys. 68 (11), 1260–1268.
St-Laurent, F., 2000. The Saguenay, Qu
ebec, earthquake lights of November 1988–
January 1989. Seismol. Res. Lett. 71 (2), 160–174.
St-Laurent, F., Derr, J., Freund, F., 2006. Earthquake lights and the stress-activation
of positive hole charge carriers in rocks. Phys. Chem. Earth 31 (4–9), 305–312.
Sulman, F.G., Levy, D., Levy, A., Pfeifer, Y., Superstine, E., Tal, E., 1974. Air-ionometry
of hot, dry dessert winds(Sharav) and treatment with air ions of weather-
sensitive subjects. Int. J. Biometeorol. 18 (4), 313–318.
Takeuchi, A., Lau, B.W., Freund, F.T., 2006. Current and surface potential induced by
stress-activated positive holes in igneous rocks. Phys. Chem. Earth 31 (4–9),
240–247.
Takeuchi, A., Nagahama, H., 2002. Surface charging mechanism and scaling law
related to earthquakes. J. Atmos. Electr. 22 (3), 183–190.
Teisseyre, R., 1997. Generation of electric field in an earthquake preparation zone.
Ann. Geophys. 15, 297–304.
Tramutoli, V., Cuomob, V., Filizzolab, C., Pergolab, N., Pietrapertosa, C., 2005.
Assessing the potential of thermal infrared satellite surveys for monitoring
seismically active areas: the case of Kocaeli (Izmit) earthquake, August 17,
1999. Remote Sensing Environ. 96, 409–426.
Tributsch, H., 1982. When the Snakes Awake. The MIT Press.
Tributsch, H., 1984. When the Snakes Awake: Animals and Earthquake Prediction.
MIT Press.
Trigunait, A., Parrot, M., Pulinets, S., Li, F., 2004. Variations of the ionospheric
electron density during the Bhuj seismic event. Ann. Geophys. 22 (12), 4123–
4131.
Tronin, A.A., 2000. Thermal IR satellite sensor data application for earthquake
research in China. Int. J. Remote Sensing 21 (16), 3169–3177.
Tronin, A.A., 2006. Remote sensing and earthquakes: a review. Phys. Chem. Earth
31 (4–9), 138–142.
Wasa, Y., Wadatsumi, K., 2003. Functional strengthening and employment of
macroscopic anomaly system by e-PISCO ASP. J. Jpn Soc. Inf. Knowl. 13 (2), 41–
47.
Yasui, Y., 1973. A summary of studies on luminous phenomena accompanied with
earthquakes. Mem. Kakioka Magn. Obs. 15 (2), 127–138.
Yasuoka, Y., Igarashi, G., Ishkawa, T., Tokonmai, S., Shinogi, M., 2006. Evidence of
precursor phenomena in the Kobe earthquake obtained from atmospheric
radon concentration. Appl. Geochem. 21 (6), 1064–1072.
Yokoi, S., Ikeya, M., Yagi, T., Nagai, K., 2003. Mouse circadian rhythm before the
Kobe earthquake in 1995. Bioelectromagnetics 24 (4), 289–291.
Zakharenkova, I.E., Shagimuratov, I.I., Krnakowski, A., 2007. Features of the
ionosphere behavior before the Kythira 2006 earthquake. Acta Geophys. 55
(4), 524–534.
F.T. Freund et al. / Journal of Atmospheric and Solar-Terrestrial Physics 71 (2009) 1824–18341834
