WATER ASSISTED DYNAMIC RECRYSTALLIZATION AND

Tectonophysics.
Elsevier
96
( 1983) 125% 151
Science Publishers
WATER
ASSISTED
- Printed
DYNAMIC
IN POLYCRYSTALLINE
JANOS
125
B.V.. Amsterdam
in The Netherlands
RECRYSTALLIZATION
AND
WEAKENING
BISCHOFITE
L. URAI
Instrtuut ooor Aardwetenschappen,
State
lJniversit.p of Utrecht, P.O. Box 80.021, 3508 TA Utrecht (The
Netherlands)
(Received
July 7. 1982; revised version accepted
November
17. 1982)
ABSTRACT
Urai.
J.L..
1983. Water
Tectonophwcs.
Artificially
temperatures
between
assisted
dynamic
prepared
between
specimens
of bischofite
20 and 100°C. strain
0.1 and 28 MPa. Development
experiments.
recrystallization
and weakening
in polycrystalline
bischofite.
96: 125-157.
and results
(MgC1,.6H,O)
rates between
of microstructure
of these were correlated
have been experimentally
10m4 and
lo-”
with strain was studied
with observations
deformed
s- ‘, and confining
at
pressures
by in-situ deformation
made on thin sections
of deformed
samples.
In a first series of experiments
behaviour
was investigated.
the effect of grain size. impurity
Addition
content
flow stress by a factor of 5. This effect was found to be associated
on grain
boundaries.
gram boundaries.
the lattice.
samples
strongly
In a second
was investigated.
power
law creep equation
stresses
below this value.
grain-boundary
migration
rates.
rocks in nature
sensitivity
for values of the differential
mechanisms
were intracrystalline
rotation
fell into two different
of high-angle
due to excess water present
in
of the flow stress of selected
one with a stress exponent
slip. twinning,
and high-angle
grain-boundary
n = 4.5 in the
and grain-boundary
migration.
regimes, one regime being distinguished
of the experimentally
the
of a thin fluid film
stress above 2.0 MPa. and one with n = 1.5 for
by subgrain
The applicability
on the flow
to decrease
due to the movement
plasticity
the strain-rate
was found
with the formation
recrystallization
intracrystalline
Two regimes could be distinguished:
occurred
migration
dynamic
also to enhanced
series of experiments
The main deformation
Recrystallization
enhancing
and possibly
and water content
of about 0.1 wt.% water to dry samples
found
sliding.
The rates of
by extremely
flow law to the behaviour
fast
of bischofite
is discussed.
INTRODUCTION
Bischofite (MgCl, .6H,O) occurs in evaporite sequences. It represents the last
stage in the evaporation
of seawater (Lotze, 1957); lo-100 m thick deposits of
bischofite have been reported from the lower Volga region (Kazantsev et al., 1976),
Gabon (Belmonte et al., 1965). and from the northeastern
Netherlands
(Coelewij et
004O- 195 l/83/$03.00
0 1983 Elsevier Science Publishers
B.V
126
al., 1978). In the Zechstein
(Fulda,
deposits
1931) and it is generally
The easily induced
reported
extrusion
plasticity
pressure
Geller
bischofite
is quite
rare
has long been known.
Dewar
(1894)
to form a wire, and Mtigge (1906) determined
(1925)
of bischofite,
however.
as secondary.
of bischofite
that it could be extruded
(110) twin glide plane.
of Germany,
regarded
studied
and Geller
the temperature
(1924.
dependence
1930) investigated
its
of the
the pressure
dependence
of its incongruent
melting point. Geller’s results showed that bischofite
is indeed the weakest of all salt minerals.
Although
of a limited occurrence.
bischofite layers have been noted for the drilling problems they cause related
high plasticity of this material (Strelets et al., 1968: Mukhin et al.. 1975).
The reason
for investigating
(1) A better understanding
to solving drilling problems
the flow behaviour
Production
EXPERIMENTAL
Specimen
and the University
in Van Eekelen
of Utrecht:
results of these experiments
et al. (1983).
PROCEDURES
preparation
Because of the hygroscopic
done in a room with relative
diameter
date with observations
on the
(Urai et al.. 1980: this paper).
experiments to determine the rheological behaviour of bischofite
as a joint project of the Koninklijke/Shell
Exploration
and
Laboratories
are also published
was twofold:
of its mechanical behaviour could give a contribution
and to the design of radioactive
waste disposal sites
(Herrmann,
1980).
(2) It was possible to correlate the macrorheological
deformation
of bischofite during in-situ experiments
The deformation
were carried out
of bischofite
to the
and approximately
nature of bischofite all specimen preparation
was
humidity
of 15%. Cylindrical
samples. 50 mm in
100 mm long. were prepared
artificially.
Grain
size.
water content and impurity content were independently
varied to obtain samples
with a wide range of combinations
of the different values of these parameters. These
were chosen to cover the variation of these parameters which was expected to occur
in natural bischofite. The following
vary the various parameters:
(1) Grain size:
(a) Fine-grained
compacting
ground
methods
of sample
preparation
were used to
samples (grain size between 0.1 and 1.0 mm) were prepared by
bischofite in an eudometer at 40.0 MPa and 110°C for one day-.
(b) Coarse-grained
samples (grain size of around
10 mm) were prepared
by
casting molten bischofite into an ingot. This preparation
method, however. gav’e
some difficulties.
When
molten bischofite is cooled from above 130°C. between
120” and 117°C it becomes saturated with respect to MgCl, .4H,O. Precipitation
of
this phase occurred very unpredictably
in the form of a few vol.% of needle shaped
crystals.
127
(2) Water content:
Ground
and
bischofite
room
contained
crystal
(grain size less than
temperature
around
water).
until
no
more
weight
loss occurred.
water
(above
the stoichiometrically
0.1 wt.% extra
Samples
1 mm) was dried at 15% relative
with three different
water contents
This
humidity
material
still
determined
were then prepared
as
follows:
(a) Wet samples: the material was allowed to absorb about 0.4 wt.% of atmospheric water before sample preparation
(water content about 0.5%).
(b) Slightly wet samples: by using the dried material (water content about 0.1%).
(c) Dry samples by adding about 1 wt.% of the tetrahydrate
(MgCl, .4H,O)
which absorbed almost all remaining free water during specimen preparation
(water
content < 0.1%) Similar samples were also prepared from material which was kept
at 15% relative
humidity
(3) Impurity
content:
for one year.
(a) Pure samples were prepared
from analytical
grade MgCl Z .6H,O (Baker
grade)
(b) For impure samples, drillcores from the Veendam area (Coelewij et al., 1978)
were used. These contained impurity grains of halite, carnallite and kieserite (in total
less than 2 vol.%) and undoubtedly
(Diarov and Dogalov, 1971).
contained
impurity
atoms
in solid
solution
Thin sections
Thin sections were prepared
and by grinding under a volatile
using a high viscosity cyanoacrylate
resin (loctite)
oil (Shell S4919) Because bischofite recrystallizes at
room temperature
when its surface
avoid introducing
artifacts
Deformation
is scratched
(see Fig. 7) great care was taken to
by this.
apparatus
The specimens
were ground to have plane-parallel
ends,
between end-pieces of a high-strength-low-thermal
conductivity
with thin PTFE sheets to minimize end-effects.
The deformation
apparatus
(Fig.
1B) consisted
jacketed
material
of an internally
heated
in rubber
(cellaron)
pressure
vessel with oil as confining medium, and a piston sealed by O-rings. Axial load was
produced
by a hydraulic
load frame (MTS), operating
either in stroke or load
control modes. Strains were defined with respect to undeformed
sample length. No
correction was made for area changes during deformation,
since the samples were
only deformed to relatively small strains. Temperature
was controlled to within 2” in
the sample.
128
AU
( MPa)
26
24
22
20
16
14
12
Ill
8
LVDl
6
4
Heating
element
Thetma
2
>
,-*
COUple
B
Fig. I. A. Stress-strain
curves for the samples
deformed
s-I_ and a confining
pressure
6O’C. strain rate of 10-j
size, water content
and impurity
wet ones is clearly
shown.
composition
content.
in the first series of experiments.
of 28 MPa, while the samples
The large difference
Solid lines: fine-grained
samples:
of each sample, see Table I. B. Schematic
between
broken
drawing
dry samples
All tests are at
had different
and slightly
lines: coarse-grained
samples.
of the triaxial cell for deforming
grain
wet and
For
2 x 4 inch
samples.
Deformation
experiments
Three series of experiments
(a) First
the influence
were performed:
of specimen
parameters,
grain
size, water
content
and
impurity content, on the flow behaviour was investigated
by testing samples which
had different values of these parameters
under fixed conditions
of deformation:
temperature
= 60°C strain rate = lop5 s- ‘.
(b) In the second series, samples which were impure,
were tested at different strain rates and temperatures.
fine grained
and slightly wet.
(c) In the third series, 0.1 mm thick wafers of bischofite were deformed under the
microscope in an apparatus discribed by Urai et al. (1980). Strain rate was lo-’ s- ‘.
temperature
was varied between 20” and 110°C. Development
of microstructure
was
recorded by time-lapse photography.
129
RHEOLOGICAL
OBSERVATIONS
First series of experiments
The stress-strain
curves obtained
and a list of the experimental
parameters
is described
The accuracy
experiments
from these experiments
are shown
data is given in Table I. The influence
in Fig. 1A
of different
below.
of the stress measurements
were duplicated
was within
and reproducibility
a few tenth of a MPa. Most
was generally
within
5%.
The effect of specimen parameters
The effect of water content. The flow stress of slightly wet and wet samples,
generally going through a slight maximum at 2-3% strain. reached a steady
value above
stress-strain
after
state
4% strain. Dry samples showed markedly different behaviour.
Their
curves were of the work-hardening
type: the flow stress steadily
increased approaching
steady state at 10% strain. (In a few stepping tests conducted
with dry samples, however, constant flow stress was reached at strain rates of 10 --’
C’.)
Also, the absolute value of the flow stress at 8% strain was about five times higher
than in the slightly wet and wet samples. In Fig. 2 the flow stress is plotted versus
water content
occurs
within
for the various
the
first
samples.
It can be seen that the main weakening
0.1% of water
added
to the
dry
material,
additional
lowering of the flow stress is caused by adding another 0.3%.
Another
interesting
phenomenon
was observed
in the experiments
while
with
effect
some
wet
samples. When removing the rubber sleeve, the sample was found to be quite wet on
the outside, though it was assembled dry. Therefore, unlike experiments where some
fluid is put around
the sample before deformation
and is introduced
into the sample
by dilatation pumping, fluids are expelled from the sample in this case. This can be
accounted
for by grain boundaries
containing
brine which arrive at the sample
boundary.
For this brine there will be no transport
mechanism
back into the sample
and the sample boundary
will be gradually
enriched in brine. Another possible
explanation
is that some compaction
takes place during deformation,
by which fluid
is expelled from the sample. This effect may be responsible
for expelling from salt
deposits large amounts of fluids generated during salt metamorphism
(Herrmann.
1980a, b).
The effect of impurities. Impure samples tend to be slightly stronger than pure ones
(see Fig. 2). Although this is a fairly consistent result for all values of water content.
the variation
drawn.
between
duplicated
samples
prevents
firm
conclusions
from
being
130
xxxxx
xxxx
x
xxxx
X
xxxx
xxxxxxxxxxxx
x
131
l
fine grained,
* fine grained.
A coarse
-
Fig. 2. Plot of the differential
0.14.
See text for discussion
0.4
1.0
% excess H20
MgCl2.6H20
the first series of experiments.
pure
grained
0.1
0.01
impure
stress at 8% strain versus excess water content
The main weakening
of the suggested
decrease
occurs
for all samples
for excess water contents
of this effect for water contents
deformed
between
in
0.01 and
below O.Ol’%c.
The effect of grain size. In the comparison
of the fine- and coarse-grained
samples,
only a few coarse-grained
(cast) samples (those with relatively few tetrahydrate
needles) were used for mechanical analysis (see Figs. IA and 2). The results suggest
that initial grain size has no or little effect on the flow stress. Strong support for this
interpretation
comes from the microstructural
observations:
by dynamic recrystallization. the material will adjust its grain size to the imposed deformation
conditions
(temperature,
strain rate. stress) whatever the original grain size is.
132
The effect
of confining
pressure
As has been shown by numerous
the confining
pressure
cracking
cataclastic
increases
cracking
and
their
flow.
flow stress.
is suppressed
This
Above
enhances
a certain
and increasing
flow behaviour.
A number of samples
investigate
workers and for various
from 0.1 MPa to a few hundred
materials.
the ductility
value
the confining
of confining
pressure
and
all the
has little efect on the
have
therefore
been
at atmospheric
the effect of confining
pressure.
Before and after deformation.
of the samples was measured to an accuracy of 0.2 vol.%.
(a) Compacted
samples which were annealed
for several
deformed unconfined
at room
stress--strain
curves are shown
of
suppresses
of the samples
pressure
tested
an increase
MPa generally
days
pressure
to
the v.olume
at 60°C
were
temperature
using a strain rate of 10ei s- ‘_ Their
in Fig. 3. After a maximum at 2% strain. the flow
stress decreased continuously.
These samples showed a volume increase of 5 -6F.
and their translucancy
disappeared
due to the development
of fine cracks.
(b) A sample which was deformed to a strain of 6.5% (in several steps. in creep
mode, under a confining pressure
deformed further at artmospheric
of 28.0 MPa, see second series) was unloaded. and
pressure using the value of stress employed during
the last step. Figure 4 shows the creep curve.
creep the strain
rate decreased
value obtained at a confining
the value employed during
decreased
to a value about
to a value about
After
amount
of primary
state
pressure of 28.0 MPa. The stress was then increased to
the penultimate
confined step. Again. the strain rate
10 times less than in the confined
As was the case for all other confined experiments,
could be detected after the confined steps. whereas
measured
a certain
ten times less than the steady
after the two unconfined
step.
no significant volume increase
an increase of 0.7 v.ol.? vv.as
steps.
In summary, while the material reaches steady state after a few percent strain at a
confining pressure of 28.0 MPa, during an unconfined
experiment there is an initial
hardening
associated with the onset of dilatation
which is followed by progressive
Fig. 3. Tests at different
discussion.
confining
pressures.
The test at 10 MPa confining
pressure
is T20. See text for
133
3& (%)
3.0 MPa
unconfined
time
I
I
200
400
I
600
time (hrs)
Fig. 4.
C‘rsepCUI-WS
of the last (unconfined)
which were arrived
sketch
illustrating
steps of T47. Broken lines show the steady state strainrates
at with the same values of axial pressure
the difference
in creep curves between
and at 28 MPa confming
confined
and unconfined
pressure.
samples.
Inset is a
See text for
discussion.
weakening
of the sample
An experiment
showed
bischofite
that
due to cataclasis
conducted
dilatation
(see inset of Fig. 4).
at a confining
is already
will be independent
pressure
suppressed
of 10.0 MPa (T20, see Fig. 3)
at that
of depth at relatively
pressure.
shallow
Thus
the flow of
levels in the crust.
The second series of experiments
After the first series, the fine-grained,
slightly
wet, impure
samples
were selected
as being representative
for bischofite occurring in nature, and these samples were
used to determine the strain rate sensitivity of the steady state flow stress at different
temperatures.
at a confining
pressure
of 28 MPa. The results
of these experiments
are shown in Fig. 5 and Table II. The procedure used with the stepping tests was
based on the following observation.
After the samples had reached steady state (in
creep mode, after a few percent strain) the load was changed to a lower value. The
sample remained at the same length for some time, then it started deforming and a
new steady state strain rate was arrived at almost immediately
(see Poirier, 1977).
This could be verified at higher strain rates where one could proceed to a few
134
?
135
TABLE
Results
II
of the second
series of experiments
Test
Temp.
Strain rate
Differential
No.
(“C)
(s-l)
stress
strain
(MPa)
(%)
27
60
30
60
31
60
38
41
42
43
45
46
47
48
49
60
60
60
80
40
80
40
80
60
Cumulative
1.01’10~~
5.6
1.00~10~6
3.4
7.26
7.46
1.03. loms
4.5
1.42
1.02. 1om6
2.5
1.78
1.01~10-5
4.0
1.24
1.02.10-h
2.4
1.75
l.03~10-5
20.8
8.03
4.10. 1om6
20.8
10.99
1.10~10-6
18.8
11.26
1.02.10m5
4.9
8.43
4.26. 10m6
4.0
10.10
1.05. l0-h
3.4
10.49
1.03.10-5
5.0
7.22
4.19.10m6
4.0
8.25
9.74.10-7
3.1
8.50
9.62. lo-’
8.3
16.93
4.12.10m6
4.5
2.39
1.02.10m5
5.1
5.00
11.38
1.02. 1o-4
8.8
6.84. 10mh
7.5
5.1
1.71. 1om5
8.9
13.1
6.86. 1om6
4.5
1.69.10-5
6.2
11.9
1.05. lo-’
0.8
12.66
1.95. lo-’
I .4
14.21
4.06. 1O-6
9.8
2.84
2.14. lo-’
4.5
3.59
2.48. IO-’
5.0
4.42
4.51
1.14. IO_’
4.2
5.36
5.65. IO-*
3.2
5.92
1.53.10m*
1.7
5.97
2.29. 10m8
1.7
6.36
912. 10m9
1.0
6.64
1.35.10m’
1.0
3.14
4.64. lo-’
2.2
4.38
4.32. lo-’
0.5
4.96
8.09. lo-’
3.0
6.98
1.38.10-*
0.3
7.24
7.08. 10m6
5.3
10.10
1.40.10-’
2.0
1.46
5.26. 1O-6
6.0
5.11
1.09.10m6
4.0
6.93
136
TABLE
II (continued)
Test
Temp.
Strain rate
Differential
No.
(“C)
(s-l)
stress
strain
(MPa)
(9)
51
52
percent
strain
Cumulative
8.94. lo--’
1.o
7.26
2.13.10~’
1.5
7.66
3.65. IO-”
0.6
7.89
3.55.lo-x
2.0
9.21
8.91. lOmu
1.2
9.47
9.19. lomx
3.0
II.36
1.28.10-R
1.5
Il.88
40
3.26. lo-’
8.0
2.08
40
1.22.l0-R
2.0
2.47
60
4.08. lomn
2.0
3.36
80
5.46.10-’
2.0
4.15
40
1.64.10~’
I.0
4.98
60
3.58. lo-v
I .o
4.89
80
8.22. IO-’
1.0
5.02
100
1.40, lo-’
2.0
7.60
60
1.99~10~*
6.0
2.73
1.26.10-6
6.0
3.97
2.27. lo-’
4.0
4.34
6.62. lo-’
3.0
5.50
1.95.1o-x
2.0
6.14
9.18.10~’
1.5
6.65
4.00.10-9
I .o
6.91
in reasonable
steps at low differential
time, and drawing
endorsement
stress were done where the strain
only a few tenth of percent
from this obsemation.
rate was determined
after
strain.
The value of the steady state flow stress was independent
of the deformation
history, as was shown by returning
to a previous value of strain rate after a few
steps. Also, observations
on relaxation tests (Schmid et al.. 1977: 1980). which can
in terms of a flow law, seem to indicate that the above method will
be interpreted
yield results describing steady state conditions.
Data of each experiment could be fitted by a power law of the form:
i=A(Au)”
(I)
where A and n are constants.
The constants A and n are well constrained
for each sample. although there is a
variation from sample to sample (Table II).
Below values of the flow stress of around 3.0 MPa. at 40-60°C.
and 1.5 MPa at
137
SO”C, the stress
around
dependence
4.4 to about
of the flow law changes
markedly:
dependence
of the flow behaviour
changes
from
in both flow regimes are
1.5. A and n values for each experiment
given in Table III.
The temperature
n
can be described
by the
equation:
exp
P = A,
where H is the apparent activation
The value for H was determined
(1) For all experiments,
energy for creep and R is the gas constant.
in several ways:
the strain-rate
values were extrapolated
to stress values of
10.0 and 3.0 MPa in the high n regime and 2.0 and 0.5 MPa in the low n regime
(using the A and n values of each experiment). These values were than used in a plot
of log P versus l/T. Values of H were 14 + 10 and 16 f 7 kcal/mol,
respectively.
Due to the large scatter between
in the high and low n regimes.
samples
could be made for H values
no distinction
(2) In test no. 51. the steps were done at two stress levels at different
tures.
of H was 8 * 5 kcal/mol.
The value
determine
(3) Geller
pressure
(1924.
\‘alurs
1925) determined
for a number
ble assumptions.
TABLE
Unfortunately,
in which flow regime the measurement
of salts, among
his data
it was not possible
for A and n for each
the temperature
dependence
which bischofite.
can be recalculated
of the extrusion
With a number
to yield
a value
of the activation
low n regime
High n regime
n
45
2.4. lo- ”
5.1
47
2.8. lo- lo
4.2
27
3.6.10m9
4.6
4. I
A
n
0.83.10-s
1.6
30
2.2.10-s
41
5.1.
42
5.6. 1O’9
4.6
49
1o-9
4.1. lo- ‘”
4.1
0.83~10~s
1.8
4.6
3.9, lo-9
2.3
lo- I0
3.1.
l
6.3
43
2.4. 1O-9
4.85
46
0.9. lo-’
2.9
1.3. lo-’
1.1
48
3.0.10~*
3.2
1.23.10-’
1.4
* Gram
of reasona-
experiment
A
content
to
was taken.
111
Test no.
52
tempera-
growth
during
of this sample
preparation
was slightly
was less extensive
less than normal.
than
in other
samples.
possibly
because
water
13X
energy.
These assumptions
are: (a) the rate of extrusion
was the same. (He raised the pressure
The extrusion
uniaxial
pressure
in all Geller’s
up to a value so that extrusion
at a given rate in Geller’s
apparatus
experiments
just began.)
is proportional
flow stress of the same material:
p=Ka
(3)
where K is a constant
and Stenger, 1976).
l/T
(b)
to the
determined
by apparatus
effects and extrusion
rate (see Laue
With these assumptions,
a plot of the logarithm of the extrusion pressure versus
yields a value for H of 20 k 2 kcal/mol
when n = 4.4.
In summary, the value of H is around 15 kcal/mol
with a rather large scatter
(i: 10 kcal/mol)
similar
between
value
while
but
reinterpretation
with less scatter.
of data
from
The present
data
Geller
(1924.
1925) gives a
do not allow
a distinction
H values in the two flow regimes.
MICROSTRUCTURAL
OBSERVATIONS
The Veendam drill cores
From the point of view of applicability
of the present data to the defornlation
bischofite in nature, it is important to describe in some detail the microstructures
of
of
natural
bischofite.
and they contain
The bischofite layers are quite pure (about 98% bischofite)
inclusions
of halite, carnallite
and kieserite (see Fig. 7). The halite and carnallite
form grains of up to 2 mm size. They are generally of a rounded shape.
Figure 6 shows a number
contact
drawings
of the grain boundaries
on polished
and slightly etched surfaces. Generally, some shape preferred orientation
is present.
although cores with equiaxed grains also occur. The grain size is about 10 mm. An
interesting
feature is that a few (about 1%) of the grains are completely idiomorphic.
This was also noted by Miigge ( 1906).
In thin sections
slightly
curved.
the grains
The grains
appear
sometimes
to be undeformed.
contain
growth
and grain
twins.
boundaries
Deformation
are
twins
(Troger, 1971) appear to be very rare. and are generally introduced
by preparation.
Avoiding this is difficult, but because the twins are immediately
apparent
in the
grains as reflecting lamellae, their absence before grinding can be checked for (see
also Mtigge, 1906).
Fluid inclusions
are quite rare inside the grains. Unfortunately
the cores were
leached along the grain boundaries by the drilling fluid. so nothing can be said about
fluids at the grain boundaries.
Thus, the actual water content of natural bischofite
could not be determined.
In halite inclusions.
however. fluid inclusions
(along
growth surfaces) are quite common. Often, groups of grains have only a few degrees
139
Fig. 6. Contact
carnallite
drawing
inclusions.
of the grain boundaries
from a few bischofite
drillcores.
Black dots are halite or
Scale bar is 50 mm. See text for discussion.
difference in orientation
with respect to their neighbours.
They are likely to have
formed by a subgrain rotation process.
Kieserite grains are often idiomorphic
and twinned with grain sizes up to 0.3 mm.
Inside
bischofite
closely
resembling
kieserite
tion
they
are often
the present
walls represent
before
moved
grains,
deformation.
away. Observation
arranged
grain-boundary
the old (diagenetic
By extensive
in walls
configuration
or primary)
recrystallization
of the kieserite
network
which
form
networks
in bischofite.
grain-boundary
the grain
does indicate,
These
configura-
boundaries
however,
have
that the
grain size of the bischofite stayed roughly the same during its history. This has also
been described
for carnallite
(Leng, 1945). When halite grains are present. the
kieserite is often concentrated
inside them.
The undeformed
Compacted
The milled
samples
samples (pure and impure)
bischofite
contains
grains
which
vary in size from
1 mm to a few
microns.
By compaction,
porosity is reduced to zero and the subsequent
grain
equiaxed grains with slightly
growth by annealing
produces optically strain-free,
140
curve grain boundaries,
Annealing
twins were present
and wet ones. Grain
around
frequently
making
an angle of 120 degrees at triple points.
in dry samples.
sizes were around
while they are quite rare in slightly wet
1 mm in slightly
wet and wet materials.
and
0.1 mm in dry samples.
In slightly wet and wet samples,
fluid inclusions.
most of the grain boundaries
As their size was about
5-50 microns.
thick sections to avoid damage.
In the dry samples, the fine-grained
aggregates
contained
they were studied
of tetrahydrate
ara>s of
in 0.5 mm
are transformed
along their edges into hexahydrate,
and the number of fluid inclusions
on grain
boundaries
was much reduced. (See Van Eekelen et al.. 1982. Fig. 1). The samples
had a milky translucent
appearance.
Cast sampies (pure und impure)
The system
MgCI,-H,O
(1938) and Dietzel
to give a saturated
has been
which dissolve at 129°C. During
of a radially
boundaries
uted needles
grown
parallel
extensively
studied
by Grube
and
Brauning
and Serowy (1959). MgCl, .6H,O melts at 117°C incongruently
solution of MgClz. 4H,O with some crystals of M&I,.
4H,O
aggregate
cooling,
to (110). about
of MgCl,
a typical ingot texture is formed. consisting
of elongated
hexahydrate
10 X 5 mm in diameter.
grains.
with manv
grain
with irregulari>- distrib-
.4H,O.
Deformed sampies
Deformation microstruclures
Because in the fine-grained
samples it was impossible
to tell the difference
between old grains and recrystallized
ones, most of the information
on the microstructural
changes during
ples. During preparation
recrystallize
deformation
was obtained from the coarse-grained
samof the thin sections bischofite
could be obsened
to
at room temperature
(Fig. 7). This raised the question
tion. Additional
information
along scratches
of the stability
was obtained
on the surface of the thin section
of the microstructure
from the in-situ
deformation
after deformaexperiments.
The results of this study are described together with the thin sections of deformed
samples, to illustrate and clarify certain aspects which could not be deduced from
thin section
studies
alone.
A marked difference in microstructure
was found between dry and slightly wet
samples. As will be shown later, this could be interpreted
in terms of recrystallization behaviour,
while the deformation
mechanisms
were largely the same for all
water contents.
Deformation mechanisms. Evidence for intracrystalline
slip was found in grains
showing undulose
extinction
(Fig. 8). Subgrains
with wavy boundaries
were frequently developed. The identity of these slip systems is as yet unknown. Twin glide
Fig. 7. Kieserite
indicating
preparatmn.
inclusions
former
in a Bischofite
grain-boundary
positions.
grain
from
The bischofite
The array of fine grains at the bottom
the Veendam
drillcores,
grain is extensively
of the photograph
arranged
twinned
is a preparation
along
during
artifact.
walls
specimen
Scale bar is
1 mm. crossed polarizers.
Fig. 8. Mechanically
(T 29). Many
polarizers.
twinned
of the small
grains
together
recrystallized
with ones showing
grains
contain
growth
undulose
twins.
extinction
in a dry sample
Scale bar is 0.1 mm. Crossed
142
on the (110) system in bischofite
temperatures
the easiest
has been determined
deformation
system
by Mtigge (1906). It is at all
(shear stress is below 0.05 MPa). The
ease of twin glide on this system is also demonstrated
by the presence of kinks
produced by twin glide (Fig. 9). In some thin sections at least three different sets of
twins were observed.
U-stage
work is in progress
to determine
the identity
of these
twin planes.
The presence of grain-boundary
displacements
was determined
by a method
resembling
that of Schmid et al. (1977). After deformation,
on the initially smooth
surface of the sample, due to the softness of the rubber jacket grain boundary offsets
could be seen. The edges of these offsets, however, were relatively rounded, indicating the plastic deformation
in areas adjacent to the grain boundary.
This is caused
by the rapid recrystallization
along shearing grain-boundary
regions, transforming
the grain boundary
into a thin shear zone, as could be seen on thin sections
extending to the edges of the specimen. Because of the above mentioned
problems
only a rough estimate
termined. Using:
of the strain
due to grain-boundary
sliding
could
~,s = cpvd
(4)
where ‘p is a geometrical
Fig. 9a. Sharp boundary
(T40). Recrystallization
grains
be de-
factor, o the average step height at grain-boundary
between
is initiated
(white spots) are artifacts
twinned
and untwinned
at grain boundaries.
due to grinding.
part of a grain:
a “kink”
Scale bar is 0. I mm.. crossed
formed
offsets.
by t% in glide
polarizers.
Very fine
Fig. 9b. In-situ
temperature.
crossed
sequence
illustrating
the formation
Note the shape change
of the initially
of the structure
spherical
shown
in Fig. 9a. Dry sample.
holes in the grains.
room
Scale bar is 0.2 mm.
polarizers.
and d the grain size (Bell and Langdon,
total strain
is believed
1969; Gifkins,
to be due to grain-boundary
1973) about
lo-20%
of the
displacements.
Rec~wtailization.
The development
of new grains by progressive
misorientation
across subgrain boundaries
(Poirier and Nicolas, 1975) was frequently observed in
all samples. The size of these grains is relatively large, from one old grain generally
only a few new grains were formed (see Fig. 13). Subgrain walls were slightly curved
and the subgrains equiaxed, suggesting the presence of more than one slip system.
recrystallization”
(Poirier and Guillope,
This process has been called “rotation
1979). No fluid inclusions
were found on subgrain boundaries
having misorientations below a few degrees.
The migration of high-angle grain boundaries
was the most important
process of
formalion of new grains.
As the in-situ experiments
showed, the migration could take place at two distinct
rates,
about
0.1 mm/min
and about 0.1 mm/hr
while intermediate
velocities were
not observed.
The initiation
of this process could take place either at twin boundaries
144
and
intersections
process
misorientation
boundary
lamellae,
due to high strain
across
a subgrain
can start to migrate
When
growing
frequently
grains,
of twin
is either
elongated
into
grain
regions,
boundary
boundary
(Gottstein
a grain
parallel
and recrystallizing
or at preexisting
in grain
has reached
This
last
or to the fact that
a critical
value
so the
et al., 1976).
containing
twin
lamellae,
to these (Fig. 10). Grain
grains
boundaries.
which were much
the
new
boundaries
larger
than
grains
were
between
twinned
the width
of twin
lamellae showed a characteristic
stepped shape, closely resembling
the microstructure described by Calais et al. (1961) for alpha-uranium.
These steps appear to be
caused by changes in orientation
and interfacial energy along the boundary. Grains
growing
into large old ones in coarse-grained
samples
sometimes
showed
a slight
tendency towards idiomorphism
by having a boundary straight and parallel to (110).
The observation
that most twin lamellae have parallel boundaries
up to the grain
boundaries
boundaries
visible
The
is somewhat
have moved
(lensoid
puzzling. It can be understood
by realizing that most grain
after twinning and the original grain boundaries
are rarely
twins were seen to form in the in-situ
appearance
of new grains
formed
tests).
by migration
recrystallization
stricted to the first stages of deformation
where the coarse-grained
transformed
into fine-grained
material. After this in the dynamically
Fig. 10. Elongated
Crossed
polarizers.
new grains growing
Very fine grains
along twin lamellar
are artifacts
was re-
samples were
recrystallizing
in a larger old grain (T40). Scale bar LS0. I mm.
due to grinding.
145
aggregate
only the migration
of existing
grain boundaries
occurred,
so that in a strict
sense, no new grains were formed.
The relation between microstructure,
High n regime. In general,
microstructure
increasing
as increasing
grain-sizes
grain
At 60°C
boundaries
Mechanically
and
twinned
samples.
wet samples
grains
been
recognized
indicating
were found
inside
the position
of old
new grains,
while
were clear and inclusion-free.
had
or grains
as such using
a strikingly
with
giving the samples an optically strain-free
microstructure
developed during dynamic
have
undulose
different
extinction
microstructure.
were very rare
appearance.
It should be noted that thi:
recrystaliization,
although it would not
microstructural
criteria
which
are generally
i
*;
Fig.
11. Mechanically
grain-boundary
crossed
migration
polarizers.
on
of experiments).
quite common
in the dry samples (see Fig. 8).
sizes around 0.1 mm frequently contained anneal-
in the undeformed
wet
series
of both grain-boundary
migration rates, as was observed in
12), resulted in a bimodal
distribution
of recrystallized
grain boundaries
slightly
had the same influence
(first
{Fig. 11). Arrays of gas and fluid inclusions,
newly formed
The
the water content
temperature.
strongly
deformed
grains were
Recrystallized
grains with grains
ing twins. The presence
in-situ
tests (see Fig.
water content and temperature
twinned
velocity.
old
grain
bimodal
is replaced
grain-size
hy
new
distribution
ones.
Because
is produced.
of the
m
_L
difference
in
T40. scale bar is 0.1 mm,
Fig. 12. In-situ
strongly
sequence
deformed
migration
illustrating
rate of 1 micron/set.
T = 80°C. dry sample.
applied
recrystallization
by two different-grain
old grain new grains are slowly formed
D. Grain
boundary
B, which has the same orientation
Strain rate = IOm5 s
for rocks like the presence
‘. crossed
velocities.
A. In a
at C. B. C. New grain A starts growing
polarizers.
wjith a
growing.
Scale bar is 0.2 mm.
extinction
(see
The microstructure
was formed by the complex interplay of subgrain rotation
grain boundary migration (at the slow and fast rate). In the in-situ experiments.
and
dry
also Means,
samples
showing
undulose
1982).
at 60°C
occasional
grain
common.
Wet
indicating
that
showed
growing
samples
migrating
mainly
rapidly
(Fig.
water
recrystallizing
grain
slow
displayed
an increased
migration.
The dynamically
rapidly
of new grains
as grain A. starts
boundaries,
this
migration
of grain
boundaries
12) while at 100°C rapid growth
behaviour
content
enhances
microstructure
producing
already
at room
the rapid
was thus mainly
the typical
with
an
was very
temperature.
grain
boundary
determined
microstructure
shown
by
in
Fig. 13.
Two aspects of this microstructure
should be mentioned:
(1) Two populations
of grains are present: often along boundaries of a new grain
a group of smaller ones which all have approximately
the same orientation
is
present
(see grains 2 and 3 in Fig. 13). These could be observed
to develop
when an
Fig.
13. The microstructure
different
rates
discussion.
crossed
Grains
grain
resulting
subgrain
having
polarizers.
preparation
old
and
rotation.
from
the complex
(T29:
the same number
The very fine grains,
coarse
belong
interplay
grained.
of grain-boundary
slightly
wet, pure
to the same “orientation
the fine twin lamellae
and
was swept
(a) The transition
mobility
family”.
Scale bar is 1 mm,
the air bubbles
by a rapidly
migrating
grain
boundary.
in the resin are
from slow to fast migration
(e.g., absorbing
of the grain
impurities
boundary
or changes
may drop
Very frequently
leaving
developed
by slow migration
occurs catastrophically.
in grain-boundary
During
fast
structure)
the
so that the reverse catastrophic
and consequently
change
which has just
has a low dislocation
so the driving force for migration is much less in this region.
(c) The mobility of the grain boundary
is also orientation
the
over the smaller
occurs and the migration rate drops to its slow migration value.
(b) The migrating grain boundary
arrives at a part of the grain
recently
at two
See text for
artifacts.
rapid migration stops before the whole grain is consumed,
grains. There may be a number of reasons for this:
migration
migration
sample.)
dependent.
density
If the
rapidly migrating grain boundary arrives at a subgrain boundary across which a few
degrees misorientation
has developed. its mobility may decrease so that the reverse
catastrophic
change in grain-boundary
velocity occurs. In this subgrain the dislocation density may also be much less because of the polygonization
processes.
(2) Generally
a grain boundary
stops migrating
when it arrives at the next
high-angle
grain boundary.
However, new grains are formed by the migration
of
D
C
Ftg. 14. Schematic
drawing
effect of recrystallization
illustrating
development
of the microstructure
shown in Fig. 13. Only
the
is shown.
high-angle grain boundaries.
so grain boundaries
will be reactivated. The point at
which a grain boundary starts to migrate will also depend on the neighbouring
grain,
so it will be possible that after a grain has become elongated by consuming
a few
neighbours,
will consist
through
it is cut in two by a migrating grain boundary. Thus, the microstructure
of “orientation
families”, formed by the propagation
of one orientation
the sample,
boundaries,
while
and modified
this orientation
by subgrain
14). A slight shape preferred
was observed
orientation
in most samples
is destroyed
rotation
perpendicular
after deformation,
flattening
of grains and an influence
boundary
configuration.
by other
and lattice
rotation
migrating
to the compression
showing
grain
processes
the combined
(Fig.
direction
effect of
on the shape of new grains of the existing
grain
~~c~o~~ru~t~~es in rhe low n regime. The microstructure
developed
in the low n
regime is shown in Fig. 15. The main difference compared with samples from the
high n regime is the stronger bimodality
in grain size. resulting
from a more
pronounced
difference in size between the migrated grains and the ones “left over”.
The same microstructure
was seen to develop at higher strain rates in the in-situ tests
at temperatures
above 90°C. The changes in rheology as a function of temperature
and strain rate are thus well reflected in the microstructure.
Gruitt boundaries
In the deformed
samples,
high-angle
grain
boundaries
contained
arrays
of fluid
Fig. 15. Typical
boundag
microstructure
which has stopped
of the samples
halfway
the edges of the big ones are created.
deformed
in the low n regime.
across an old grain. This IS the process
T5 1. crossed
polarizers.
Arrow, indicates
a grain
by which the small grains on
scale bar is 1 mm.
more than the dry ones.
and gas inclusions
(see Fig. 16) wet samples containing
When studying wet or slightly wet samples in-situ, grain boundaries
were optically
clear without
to both adsorb
inclusions
visible fluid inclusions.
and leave behind
connected
During
migration,
fluid inclusions.
to the grain boundary
however,
they could be seen
After deformation
by thin channels
stopped,
fluid
could be observed
to
develop into arrays of isolated fluid inclusions by necking-down
process (Lemmlein
and Kliya. 1960) (Fig. 17). This generally took place within one hour.
Static recystallization
and stability of microstructure
When deformation
is stopped, the samples statically recrystallize
ture consisting
of polygonal,
equiaxed grains, closely resembling
to a microstructhe undeformed
compaction
samples. The extent of this process is strongly dependent
on the
temperature.
When the samples are rapidly cooled to room temperature,
in dry and
slightly wet samples it becomes very slow, so that thin sections of these deformed
samples do give a true picture of the microstructure
during deformation.
In wet samples, however, even at room temperature,
the recrystallized
texture is
overgrown by idiomorphic
grains (Fig. 18). This microstructure
is well known in
ceramics.
it is believed
to develop
only when a small amount
of a second
phase
is
Fig. 16. Fluid and gas inclusions
on grain boundaries.
These were found
slightly wet and wet samples (T34). Scale bar is 0.1 mm.. partly crossed
about
before and after drformatlon
polarizers.
thickness
m
of that secuon
0.5 mm.
Fig. 17. In-situ
argued
sequence
in the present
illustrating
paper,
process
from a continuous
microns,
plane polarized
growth
the arrays
of isolated
fluid inclusions
of fluid and gas inclusions
fluid film which is present
light. Time lapse between
during
in a wet bischofite
at grain boundaries
grain boundary
first and last photograph
sample.
are formed
migration.
As is
by this
Scale bar is 50
is 30 min. T = 50°C.
Fig.
18. Idiomorphic
microstructure
grain
overgrowing
IS also observed
phase on the grain boundaries.
Blach spots are preparation
present
in natural
the microstructure
bischofite.
Note the growth
formed
by dynamic
and is diagnostic
twins. Wet sample.
recrystallization.
for a small amount
scale bar is
This
of a second
I mm. crossed polarizers.
artifacts.
at grain boundaries
(Burke,
1950). This microstructure
is also found
in the
Veendam drill cores (see p. 138) and is strong evidence for natural bischofite being
6‘wet” and having a fluid at the grain boundaries.
This should not negate the
applicability
of the rheological data to natural bischofite. because in the natural
material the thickness of the grain boundary
fluid film will be greater, as the grain
size is much higher.
DISCUSSlOh‘
The experiments
described
above have shown that increasing
the water content
bischofite causes a strong weakening of the material.
As will be argued below, above a certain value of excess water content
of
there is a
fluid film formed on the migrating
grain boundaries
and this strongly enhances
grain-boundary
mobility. This water assisted grain-boundary
migration must be at
least partly responsible for the weakening in slightly wet samples. It can be inferred
from the smooth shape of the stress-strain
curves, that this process is sufficiently out
of phase in different areas of the sample to prevent stress drops due to recrystallization (Glover
and Sellers,
1973).
152
It is by this process
samples
relatively
that, after stepping
rapidly
stored
in the samples,
strain
energy
reach
a new steady
recrystallization
level corresponding
relatively
rapid
processes
alone.
process,
tests can be understood
state.
will continue
Driven
until
will be reached
that the initial
when one considers
by the strain
faster
during
that softening
energy
reaches
As recrystallization
much
hardening
stress. the
the sample
to the new stress value.
this state
It is proposed
to a lower value of differential
than
the
is a
by recovery
unconfined
creep
is caused by recrystalliza-
tion and this is mostly initiated by the migration of high angle grain boundaries.
At
the onset of dilatation,
grain boundaries
open up to form cracks and most of the
sites for grain boundary migration are eliminated this way.
Although
the solubility
microstructural
water:
of water
observations
the absence
indicate
of undulose
in slightly wet and wet samples
This enhanced
the unit
recovery
cell parameters
in the MgCI,.
6H10
lattice
is not
that there is also an intracrystalline
extinction
and the large amount
of subgrain
suggest the ease of climb processes
will also cause weakening.
of wet and
This
rotation
in these samples.
Work is in progress
dry bischofite.
whether or not water is present in the lattice.
Because the slightly wet and wet samples
known.
effect of
to determine
may give an indication
have a fluid
phase
present
during
deformation.
one must also consider the effect of fluid pressure on the flow. stress of
these samples. A number of observations
have to be considered:
(1) In wet samples, fluid is expelled during deformation.
thus the fluid pressure
must be at least equal to the confining pressure in this case.
(2) Tests done at atmospheric
pressure have about the same values of flow stress
as the ones done at a confining
fluid pressure
(3) As will be argued
grain
percent
boundaries
strain)
pressure
of 28.0 MPa. although
in these experiments
can not have been much higher than 0.1 MPa.
below, a continuous
after some recrystallization
so at low strains
strong
difference
already
present
in differential
fluid pressure
stress
fluid film will only be present
has taken
place (that
will not be important.
between
dry
and
slightly
on the
is. after a few
Howev.er. the
vvet samples
is
at low strains.
In summary, as suggested by the difference in strength at low strains. about half
of the difference in flow stress between dry and slightly wet samples is thought to be
due to intracrystalline
effects. while the rest of the difference will be mainly due to
recrystallization.
Fluid pressure is thought to have a minor effect on the flow stress.
As is shown by the variation in A and n values between samples. the relativ-e
importance
of the processes operating parallel-concurrently
in each sample (dislocation motion, recovery, dynamic recrystallization
and grain boundary
sliding). must
differ slightly between each sample (Gifkins, 1970).
Observations
on grain boundaries in thin sections and in-situ tests are interpreted
the following way: during migration
there is a continuous
fluid layer present on
grain boundaries.
Fluid inclusions
encountered
during migration are incorporated
into this layer, while when a grain boundary
is pinned by. e.g., impurity particles.
153
cigar-shaped
grain boundary
fluid inclusions
are left behind.
When
ceases to move, by the process
deformation
of necking
is stopped
and a
down. the film breaks
up
into an array of gas and fluid bubbles (Lemmlein
and Kliya, 1960). This is the
reason why in the thin sections of deformed samples one always only encounters
these bubbles.
On the other hand,
from the fluid inclusions,
in dry samples,
and do not contain
grain
boundaries
fluid themselves
migrate
away
(Fig. 19).
An estimate of the thickness of the fluid film in slightly wet samples was made by
the following method. The total volume of fluid present in an array was measured by
counting bubbles in a grain boundary.
Dividing
area gave the thickness of the fluid layer, about
their total volume by the counted
500-1000 A. For wet samples, this
value is still much higher, up to 1 micron.
In the light of the observation
that grain-boundary
in dry and wet samples. the presence of annealing
been inferred to be due to differences in interfacial
is at least qualitatively
explained.
seems to be completed
in our dry samples,
water content
even further
Because
structure
is strongly
different
twins in dry samples (which has
energy by Aust and Rutter. 1960)
the change
in grain-boundary
it is suggested
that decreasing
properties
the excess
will have less effect on the flow stress than between
0.1
and 0.0 1o/r (see Fig. 2).
The microstructures
deformation
behaviour
Fig. 19. Old grain boundary.
boundary.
This behaviour
in the bischofite
observed
indicated
cores can fully be interpreted
in the experimentally
by an array
of fluid inclusions
is only found in the dry samples.
deformed
is left behind
T40. crossed
in terms of the
material:
polarizers,
subgrain
by a migrating
grain
scale bar is 0.1 mm.
154
rotation
accompanied
of idiomorphic
grains
by the migration
of high-angle
after deformation
has only been observed
This is strong evidence
Another
reason
for natural
for natural
bischofite
bischofite
grain boundaries.
behaving
being
The growth
in the wet samples.
as our wet samples.
“wet”
is that it is present
hetvveen
layers of carnallite and halite (Coelewij et al., 1978) and halite grains in bischofite
contain frequent fluid inclusions. Rock salt generally has water contents in excess of
a half a percent (Herrmann,
1980a, b; Roedder and Bassett. 1981). Because of its
hygroscopic nature, bischofite is unlikely to have lower free-water content.
This process of recrystallization
assisted by a fluid layer on the grain boundary
could
be important
inclusions
of domal
for other
on grain boundaries
salt minerals.
as these are known
(e.g.. Roedder
and Bassett,
to contain
salt is in the order of 0.1% which is of the same order of magnitude
amount necessary for weakening bischofite.
Also, in a recent work, White and White (1981) showed
fluid
1981). The water content
as the
that grain boundaries
in
tectonites are better described using data from ceramics than from metallurgical
studies: in the samples they investigated
there were distorted layers up to a few
hundred
Angstroms
wide at the grain boundaries.
arrays on grain boundaries
are comparable
with
Their observations
of bubble
those on grain boundaries
in
bischofite after deformation.
As the process of fluid inclusions forming from cracks
filled with a thin fluid film is well established for many silicate minerals. it may be
that the processes
described
above for bischofite
are applicable
to other rock-form-
ing minerals, although precipitation
on grain boundaries
can equally well account
for these structures (see also Wilkins and Barkas. 1978). An important
process may
be the transformation
of the fluid from inclusion arrays into a continuous
film. The
presence of arrays of fluid inclusions indicating former grain-boundary
positions in
natural
halite shows that this process does not always occur (like in our dr)
samples). Under suitable conditions. however, this process may take place and cause
weakening. From this point of view, most of data on the deformation
behav,iour of
salt should be taken with caution when trying to apply them to the in-situ behaviour
of salt rocks during
water then is present
Catastrophic
diapirism,
when they were shown to have contained
now (Herrmann,
changes
much more
1980).
in grain-boundary
migration
velocity
due
to impurity
effects have been described for halite (Guillope
and Poirier. 1979) and sodium
nitrate (Tungatt and Humphreys,
1981). These have been interpreted in terms of the
metallurgical
models for impurity controlled grain-boundary
migration (Lticke and
Sttiwe, 1971; Poirier and Guillope,
1979). These models are based on assumptions
on the general nature of grain boundaries
and the process operating during migration. It has been shown that the structure of the grain boundaries
during migration
in bischofite is fundamentally
different from those in metals: there is a fluid layer
several hundreds
of molecules
thick on the grain boundaries.
Grain-boundary
migration occurs by solution due to higher dislocation density on one side (Bosworth.
1981), diffusion through the fluid and precipitation
due to a local oversaturation
on
155
the other side. The catastrophic
due to impurity
resembles
on the growing
that of the migration
of a thermal
gradient
effects
change in grain boundary
gradient
A difference
is the different
Thus, although
surface
of fluid inclusions
(Anthony
and high dislocation
crystal
and Cline.
density
(Kern.
1969). This process
in salt crystals under
1974; Holdoway,
giving an increase
grain orientation
the phenomena
velocity may be explained
1974)
in solubility
the influence
both
thermal
of the mineral.
on both sides of the water layer.
observed
under
the microscope
are the same as in
metals, the underlying
processes are fundamentally
different.
While detailed analysis of the grain-boundary
migration process in bischofite
is in
progress, a few interesting
observations
can be made from the study of tine films
made from the in-situ experiments.
The shape of grain boundaries
during migration
is irregular, somewhere between lobate to serrated. Migration generally continued
until
the other
stopped
side of the grain
halfway
across the grain.
was reached.
However,
in some cases migration
Small parts of the grain boundary
then moved in
the opposite direction. This was interpreted
as being a readjustment
of the shape of
the grain boundary to lower its surface energy. This was also observed to occur in an
in-situ
experiment
Means
(1981).
boundary
done
using
then started
with paradichlorabenzene
high strain
moving
rates (10-I
backwards;
tion is yet given. have also been described
why grain
boundaries
move
in an apparatus,
s-l).
similar
In a few other
processes,
described
for which no explana-
by Means (1982). In an attempt
like this, Means
(1982)
by
cases. the grain
considered
to explain
the influence
of
different rates of straining of the two grains, strain accomodation
problems along
grain boundaries,
and the differences in strain in “new” and “old” parts of the same
grain. Additional
factors determining
grain boundary
movement
can be recovery
processes reducing the dislocation density in a grain in front of a moving boundary,
and the above mentioned surface energy driven movements. Which of these processes
will be dominant is a kinetic problem.
ACKXOWLEDGEMENTS
This work. sponsored
by the Dutch Organisation
for Pure Scientific Research
(ZWO) was done in cooperation
with the Koninklijke-Shell
Research laboratories
in
Rijswijk.
The author
wishes to thank
A. Hulsebos,
H. Groeneweg
and H.A.M.
van
Eekelen for their help with evaluation
of the rheological data, running the experimental program and many helpful discussions. Part of the in-situ experiments
were
done
when the author
was at the Institut
de Physique
du Globe,
Paris.
Prof. J.P.
Poirier. J-C. Mercier and M. Guillope are thanked for many fruitful discussions on
dynamic recrystallization.
The author is greatful to C.J. Spiers, J.P. Poirier, G.S.
Lister. H. Heard and I. van der Molen for discussions
of an early version of the
manuscript.
156
REFERENCES
Anthony,
T.R.
(Editor),
Aust.
and
Cline.
Fourth
K.T. and Rutter.
Trans.
Sydney,
1974. Thermomigratton
on Salt. Northern
J.W.. 1960. Annealing
Metall. Sot. AIME.
Bell. R.L. and
Langdon.
droplet:,
in salt.
In: A.H.
Coogan
twins and coincidence
site boundaries
in htgh-purity
lead.
218: 1023-1028.
T.G..
1969. In: R.C. Gifkins
Y.. Hirtz. P. and Wenger,
In: Salt Basins Around
Society.
(Editor).
Interfaces
Conference.
Butterworth\.
Africa.
R.. 1965. The salt basins of the Gabon
Proc. of a Joint Meeting
and the Congo
of the Inst. of Petroleum
(Braua\tle).
and the Geologtcal
pp. 55-7X.
Bosworth.
W.. 1981. Strain-induced
Burke, J.E.. 1950. Grain
Processes.
M.I.T.
M.M..
preferential
growth
in ceramics.
Press. Cambridge.
Pascal,
In: G. Chaudron
Presses Universitaires
of halite. Tectonophysics.
W.D. (Editor).
78: 5099525,
Kinetics
of High Temperature
P.. 1961. Les facteura
(Editor).
4e Colloque
de la migration
de Metallurgic
des Jomts
Proprietea
de grams
dam
de Joints de Gratnr.
de France.
P.A.J.. Haug. G.M.W.
Netherlands.
dissolution
In: Kingery
Mass.
J. and Lacombe.
I’uranium-alpha.
Coelewij.
of liquid
Ohio Geol. Sot.. 3 13 pp.
pp. 115- 137.
Belmonte.
Calais.
H.E..
Symposium
and Van Kuijk. H.. 1978. Magnesium
Geol. Mijnbouw.
salt exploration
in the northed,tern
57(4): 487-502.
Dewar.
J.. 1894. Note on the viscosity
Diarov.
M. and Dogalov.
of solids. Proc. Chem. Sot..
A.B.. 1971. Bischofite
137 pp.
rocks of the northern
Caspam
louland.
Ir\. .Akad. Sauk
Kaz. SSR. Ser. Geol.. 27(2): 72-76.
Dietzel,
H. and Serowy.
200°C.
F.. 1959. Die Losungsgleichgewichte
Freib. Forschungh..
Fulda.
E.. 1928. Das Kali. Enke, Stuttgart.
Fulda,
E.. 193 I. Erlauterungen
Geller.
A., 1924. Uber
wechselenden
zur Geologischen
das Verhalten
Temperaturen.
Karte van Preussen.
verschiedener
Z. Kristallogr..
Geller,
A.. 1925. Zur Kenntnis
Geller,
A.. 1930. Das Schmelzen
der Salzmetamorphose.
des Systems
hlgCI,-H1O
zutschen
20 und
A. 132: 5-32.
Minerale
fester K&per.
van Salzen bei hohen
Mineral..
Gifkins.
R.C.. 1970. Transitions
Gifkins.
R.C.. 1973. The measurement
Nr. 223 I.
bei hohen
Druchen
und
60: 414-472.
des Fliessdruckes
Fortschr.
blatt Vienenburg.
der Salzlager
Kristallogr.
in creep behaviour.
Z. Kristallogr..
Drucken
Petrogr..
J. Mater.
of grain boundary
62: 395541 I.
und seiner Bedeutung
fur den V’organg
14: 143-166.
Sci. 5: 156-165.
sliding in polycrystalline
specimens.
hlet. Set. J..
7: 15-19.
Glover.
G. and Sellars. C.M.. 1973. Recovery
of alpha-iron.
Gottstein.
Metall. Trans.,
G., ZabardJadi.
tensile
deformation
Alloys.
Nancy.
Grube,
G. and
Carnallit.
Guillope,
Bratming,
H.C..
Fracture
W.. 1938. Uber
Herrmann.
1972. Steady
A.G.,
Substanzen
Herrmann.
A.G.,
die Entwasserung
13 1: 433-459.
van
deformation
recrystallization
on the Strength
Magnesiumchlond
during
and
of Metals
Hertahydrat
Monogr.
und
halite at a pressure
und mineralogische
Geochemische
Fortschr.
Prozesse
radioaktiver
hahte:
an
5557.
Ser., Am. Geophys.
Norddeutschlands.
fur die Endlagerung
during creep of single crystalline
recrystallization
Res., 84810):
1980a. Geochemische
1980b.
of dynamic
Proc. 4th Int. Conf.
state flow in polycrystalline
Geophys.
in Salzdiapiren
Konzequenzen
high temperature
44 (2): 134.
study. J. Geophys.
of Rocks..
during
H., 1976. Investigations
single crystals.
M. and Poirier, J.P., 1979. Dynamic
experimental
Heard,
D. and Mecking.
of copper
Z. Electrochem..
and recrystallization
4: 765-775.
in
Union,
Grundlagen
Mineral..
mat-men
Substanzen
of 2 kilobars.
In: Flou
and
16.
fur die Entlagerung
58(l):
radioakti\er
169-21 I.
Salrablagerungen:
in Salzdiapiren.
Bedeutung
Z. Dtsch.
Geol.
und
Ges..
157
Holdouay.
K.A.
Coogan
1974. Behaviour
(Editor).
Kazantsev.
O.A..
R.. 1969. Crystal
V.A. and Grebennikov,
growth
Bureau.
G.G.
and adsorption.
New York. Vol.8
Laue. K. and Stenger.
conditions
M.O..
temperature.
Berucksichtigung
Lowe. F.. 1957. Steinsalz
K. and Sttiwe.
Metal..
and irradiation.
Soctety.
1976. Discovery
In: A.H.
Vol. 1. p. 303.
of bischofite
in the lower
Sheftal
(Editor).
Growth
of Crystals.
Vol. X
Aluminium-Verlag.
features
Dusseldorf.
of the healing
of a crack
in a crystal
under
lnt. Geol. Rev.. 196012): 125-128.
des Trtimmercarnallits
und Kalisalze.
H.P..
Krugershall
Dtss. Techn.
Borntraeger,
1971. On the theory
zu Teutschental.
Hochschule
Halle mit
Berlin.
Berlin.
of impurity
controlled
grain
houndary
motion.
Acta
19: 1087-99.
W.D..
1981. The concept
Means,
W.D.,
1983. Microstructure
first look. Sander
volume
of a steady
Mukhin.
Geol. Rundsch..
und Nickel.
L.K.. Sobolev.
drilling
78: 179-199.
flow of octachloropropane:
a
in press.
Mineral..
V.M.. Leonidova.
Potrier. J.P.. 1977. Microscopic
Tectonophysics.
in recrystallization
und Deformattonen
Neues Jahrb.
fluids on the deformati~~n
Metall..
state foliation.
and micromotion
0.. 1906. Uber die Kristallform
van Kobalt
Potrier.
In: N.N.
des Carnallitgefiiges.
Means.
Mugge,
N.P.,
1960. Distinctive
Leng. Pet Keng. 1945. Gefugeuntersuchung
Liicke.
Heating
Ohio Geological
pp. 3-23.
H.. 1976. Strangpressen.
and Kliya.
of declining
hesonderer
in Salt during
on Salt. Northern
Int. Geol. Rev., 18: 734-742.
Consultants
Lemmlein.
of Fluid inclusions
Symposium
Yermakov.
Volga region.
Kern.
Fourth
des Bischofiet
Geol. Palaontol.,
A.I.. Federova,
of artificial
und der verwandten
E.P. and Angelopulo,
Bischofite
specimens.
creep models and the interpretation
Chlorure
I: 91 pp.
O.K..
1975. Effect of
Neft. Khoz. 1975(7): 20.-22.
of stress-drop
tests during creep. Acta
25: 913-917.
J.P. and Guillope.
M., 1979. Deformation
induced
recrystallization
of minerals.
Bull. Mineral..
102: 67-74.
Poirier.
J.P. and Nicolas.
tion of subgrains,
Roedder.
E. and Bassett.
11s significance
Schmid.
S.M..
Boland,
J.N.
storage
and
recrystallization
to mantle
R.L.. 1981. Problems
in nuclear-waste
Tectonophysics.
Schmtd.
A., 1975. Deformation-induced
with special reference
peridotites.
in determination
siting. Geology,
Paterson,
MS.,
due to progressive
misorienta-
J. Geol., 83: 7077720.
of water content
of rock-salt
samples
and
9: 525-530.
1977. Superplastic
flow
in fine-grained
limestone.
43: 2577291.
S.M.. Paterson.
tion in Carrera
M.S. and Boland.
marble.
Streiets. G.A.. Filatov.
Tectonophysics,
B.S.. Luban.
J.N., 1980. High temperature
flow and dynamic
recrystallira-
65: 245-280.
V.Z. and Eremeev. Yu.A..
1968. Plastic flow of magnesium
salts tn oil
wells. Neft. Khoz.. 46: 18-22.
Trciger.
W.E.. 1971. Optische
Bestimmung
der Gesteinsbildende
Minerale.
Teil 2. Schweitzerbart.
Stutt-
gart.
Tungatt.
P.D.
sodium
Urai.
and
nitrate-
Humphreys.
an anatogue
J.L.. Humphreys.
recrystallization
Van
Eekelen.
1981. An in-situ
F.J. and Burrows,
of rhombohedral
H.A.,
Mechanical
F.J..
for calcite. Tectonophysics,
Hulsebos.
Behaviour
SE.,
camphor.
A. and
Urai.
of Salt. Pennsylvania
investigatton
1980. In-situ
J. Mater.
J.L.,
hehaviour
of
studies
of the deformation
and dynamic
Sci., 15: 1231-1240.
1983. Creep
State University,
White. J.C. and White. S.H.. 1981, On the structure
of the deformation
78: 661-675.
of bischofite.
University
of grain boundaries
Proc.
1st Conf.
on the
Park. Pa.. in press.
in tectonites.
Tectonophysics.
78
613-628.
Wilkins.
R.W.T.
tectonites.
and
Contrib.
Barkas.
Mineral.
J.P..
1978. Fluid
Petrogr..
inclusions.
65: 2933299.
deformation
and recrystallizati(~n
in granite