Tectonophysics. 96 ( 1983) 125% 151 Elsevier Science Publishers B.V.. Amsterdam - Printed in The Netherlands

125

WATER ASSISTED DYNAMIC RECRYSTALLIZATION AND WEAKENING

IN POLYCRYSTALLINE BISCHOFITE

JANOS 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 assisted dynamic recrystallization and weakening in polycrystalline bischofite.

Tectonophwcs. 96: 125-157.

Artificially prepared specimens of bischofite (MgC1,.6H,O) have been experimentally deformed at

temperatures between 20 and 100°C. strain rates between 10m4 and lo-” s- ‘, and confining pressures

between 0.1 and 28 MPa. Development of microstructure with strain was studied by in-situ deformation

experiments. and results of these were correlated with observations made on thin sections of deformed

samples.

In a first series of experiments the effect of grain size. impurity content and water content on the flow

behaviour was investigated. Addition of about 0.1 wt.% water to dry samples was found to decrease the

flow stress by a factor of 5. This effect was found to be associated with the formation of a thin fluid film

on grain boundaries. strongly enhancing dynamic recrystallization due to the movement of high-angle

gram boundaries. and possibly also to enhanced intracrystalline plasticity due to excess water present in

the lattice. In a second series of experiments the strain-rate sensitivity of the flow stress of selected

samples was investigated. Two regimes could be distinguished: one with a stress exponent n = 4.5 in the

power law creep equation for values of the differential stress above 2.0 MPa. and one with n = 1.5 for

stresses below this value.

The main deformation mechanisms were intracrystalline slip. twinning, and grain-boundary sliding.

Recrystallization occurred by subgrain rotation and high-angle grain-boundary migration. The rates of

grain-boundary migration fell into two different regimes, one regime being distinguished by extremely fast

migration rates. The applicability of the experimentally found flow law to the behaviour of bischofite

rocks in nature 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 deposits of Germany, however. bischofite is quite rare

(Fulda, 1931) and it is generally regarded as secondary.

The easily induced plasticity of bischofite has long been known. Dewar (1894)

reported that it could be extruded to form a wire, and Mtigge (1906) determined its

(110) twin glide plane. Geller (1925) studied the temperature dependence of the

extrusion pressure of bischofite, and Geller (1924. 1930) investigated 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 to the

high plasticity of this material (Strelets et al., 1968: Mukhin et al.. 1975).

The reason for investigating the flow behaviour of bischofite was twofold:

(1) A better understanding of its mechanical behaviour could give a contribution

to solving drilling problems and to the design of radioactive waste disposal sites

(Herrmann, 1980).

(2) It was possible to correlate the macrorheological date with observations on the

deformation of bischofite during in-situ experiments (Urai et al.. 1980: this paper).

The deformation experiments to determine the rheological behaviour of bischofite

were carried out as a joint project of the Koninklijke/Shell Exploration and

Production Laboratories and the University of Utrecht: results of these experiments

are also published in Van Eekelen et al. (1983).

EXPERIMENTAL PROCEDURES

Specimen preparation

Because of the hygroscopic nature of bischofite all specimen preparation was

done in a room with relative humidity of 15%. Cylindrical samples. 50 mm in

diameter and approximately 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 methods of sample preparation were used to

vary the various parameters:

(1) Grain size:

(a) Fine-grained samples (grain size between 0.1 and 1.0 mm) were prepared by

compacting ground 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 bischofite (grain size less than 1 mm) was dried at 15% relative humidity

and room temperature until no more weight loss occurred. This material still

contained around 0.1 wt.% extra water (above the stoichiometrically determined

crystal water). Samples with three different water contents were then prepared as

follows:

(a) Wet samples: the material was allowed to absorb about 0.4 wt.% of atmo-

spheric 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 for one year.

(3) Impurity content:

(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 contained impurity atoms in solid solution

(Diarov and Dogalov, 1971).

Thin sections

Thin sections were prepared using a high viscosity cyanoacrylate resin (loctite)

and by grinding under a volatile oil (Shell S4919) Because bischofite recrystallizes at

room temperature when its surface is scratched (see Fig. 7) great care was taken to

avoid introducing artifacts by this.

Deformation apparatus

The specimens were ground to have plane-parallel ends, jacketed in rubber

between end-pieces of a high-strength-low-thermal conductivity material (cellaron)

with thin PTFE sheets to minimize end-effects.

The deformation apparatus (Fig. 1B) consisted of an internally heated 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

6

4

2

LVDl

Heating element > Thetma ,-* COUple

B

Fig. I. A. Stress-strain curves for the samples deformed in the first series of experiments. All tests are at

6O’C. strain rate of 10-j s-I_ and a confining pressure of 28 MPa, while the samples had different grain

size, water content and impurity content. The large difference between dry samples and slightly wet and

wet ones is clearly shown. Solid lines: fine-grained samples: broken lines: coarse-grained samples. For

composition of each sample, see Table I. B. Schematic drawing of the triaxial cell for deforming 2 x 4 inch

samples.

Deformation experiments

Three series of experiments were performed:

(a) First the influence 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, fine grained and slightly wet.

were tested at different strain rates and temperatures.

(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 from these experiments are shown in Fig. 1A

and a list of the experimental data is given in Table I. The influence of different

parameters is described below.

The accuracy of the stress measurements was within a few tenth of a MPa. Most

experiments were duplicated and reproducibility was generally within 5%.

The effect of specimen parameters

The effect of water content. The flow stress of slightly wet and wet samples, after

generally going through a slight maximum at 2-3% strain. reached a steady state

value above 4% strain. Dry samples showed markedly different behaviour. Their

stress-strain 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 for the various samples. It can be seen that the main weakening effect

occurs within the first 0.1% of water added to the dry material, while some

additional lowering of the flow stress is caused by adding another 0.3%.

Another interesting phenomenon was observed in the experiments with 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 between duplicated samples prevents firm conclusions from being

drawn.

130

xxxxx

xxxx x x

xxxx X

xxxx

xxxxxxxxxxxx

131

l fine grained, impure

* fine grained. pure

A coarse grained

0.01 0.1 0.4 1.0

- MgCl2.6H20 % excess H20

Fig. 2. Plot of the differential stress at 8% strain versus excess water content for all samples deformed in

the first series of experiments. The main weakening occurs for excess water contents between 0.01 and

0.14. See text for discussion of the suggested decrease of this effect for water contents 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 recrystalli-

zation. 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 workers and for various materials. an increase of

the confining pressure from 0.1 MPa to a few hundred MPa generally suppresses

cracking and cataclastic flow. This enhances the ductility of the samples and

increases their flow stress. Above a certain value of confining pressure all the

cracking is suppressed and increasing the confining pressure has little efect on the

flow behaviour.

A number of samples have therefore been tested at atmospheric pressure to

investigate the effect of confining pressure. Before and after deformation. the v.olume

of the samples was measured to an accuracy of 0.2 vol.%.

(a) Compacted samples which were annealed for several days at 60°C were

deformed unconfined at room temperature using a strain rate of 10ei s- ‘_ Their

stress--strain curves are shown 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 of 28.0 MPa, see second series) was unloaded. and

deformed further at artmospheric pressure using the value of stress employed during

the last step. Figure 4 shows the creep curve. After a certain amount of primary

creep the strain rate decreased to a value about ten times less than the steady state

value obtained at a confining pressure of 28.0 MPa. The stress was then increased to

the value employed during the penultimate confined step. Again. the strain rate

decreased to a value about 10 times less than in the confined step.

As was the case for all other confined experiments, no significant volume increase

could be detected after the confined steps. whereas an increase of 0.7 v.ol.? vv.as

measured after the two unconfined 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 confining pressures. The test at 10 MPa confining pressure is T20. See text for

discussion.

133

3-

& (%)

unconfined

3.0 MPa

time

I I I 200 400 600

time (hrs)

Fig. 4. C‘rsep CUI-WS of the last (unconfined) steps of T47. Broken lines show the steady state strainrates

which were arrived at with the same values of axial pressure and at 28 MPa confming pressure. Inset is a

sketch illustrating the difference in creep curves between confined and unconfined samples. See text for

discussion.

weakening of the sample due to cataclasis (see inset of Fig. 4).

An experiment conducted at a confining pressure of 10.0 MPa (T20, see Fig. 3)

showed that dilatation is already suppressed at that pressure. Thus the flow of

bischofite will be independent of depth at relatively shallow 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 II

Results of the second series of experiments

Test

No.

Temp.

(“C)

Strain rate

(s-l)

Differential

stress

(MPa)

Cumulative

strain

(%)

27 60 1.01’10~~ 5.6 7.26

1.00~10~6 3.4 7.46

30 60 1.03. loms 4.5 1.42

1.02. 1om6 2.5 1.78

31 60 1.01~10-5 4.0 1.24

1.02.10-h 2.4 1.75

38 60 l.03~10-5 20.8 8.03

4.10. 1om6 20.8 10.99

1.10~10-6 18.8 11.26

41 60 1.02.10m5 4.9 8.43

4.26. 10m6 4.0 10.10

1.05. l0-h 3.4 10.49

42 60 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

43 80 4.12.10m6 4.5 2.39

1.02.10m5 5.1 5.00

1.02. 1o-4 8.8 11.38

45 40 6.84. 10mh 7.5 5.1

1.71. 1om5 8.9 13.1

46 80 6.86. 1om6 4.5 4.51

1.69.10-5 6.2 11.9

1.05. lo-’ 0.8 12.66

1.95. lo-’ I .4 14.21

47 40 4.06. 1O-6 9.8 2.84

2.14. lo-’ 4.5 3.59

2.48. IO-’ 5.0 4.42

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

48 80 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

49 60 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)

51 40

40

60

80

40

60

80

100

52 60

Test

No.

Temp.

(“C)

Strain rate

(s-l)

8.94. lo--’

2.13.10~’

3.65. IO-”

3.55.lo-x

8.91. lOmu

9.19. lomx

1.28.10-R

3.26. lo-’

1.22.l0-R

4.08. lomn

5.46.10-’

1.64.10~’

3.58. lo-v

8.22. IO-’

1.40, lo-’

1.99~10~*

1.26.10-6

2.27. lo-’

6.62. lo-’

1.95.1o-x

9.18.10~’

4.00.10-9

Differential Cumulative

stress strain

(MPa) (9)

1 .o 7.26

1.5 7.66

0.6 7.89

2.0 9.21

1.2 9.47

3.0 II.36

1.5 Il.88

8.0 2.08

2.0 2.47

2.0 3.36

2.0 4.15

I.0 4.98

I .o 4.89

1.0 5.02

2.0 7.60

6.0 2.73

6.0 3.97

4.0 4.34

3.0 5.50

2.0 6.14

1.5 6.65

I .o 6.91

percent strain in reasonable time, and drawing endorsement from this obsemation.

steps at low differential stress were done where the strain rate was determined after

only a few tenth of percent 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

be interpreted in terms of a flow law, seem to indicate that the above method will

yield results describing steady state conditions.

Data of each experiment could be fitted by a power law of the form:

i=A(Au)”

where A and n are constants.

(I)

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

SO”C, the stress dependence of the flow law changes markedly:

around 4.4 to about 1.5. A and n values for each experiment in both

given in Table III.

137

n changes from

flow regimes are

The temperature dependence of the flow behaviour can be described by the

equation:

P = A, exp

where H is the apparent activation energy for creep and R is the gas constant.

The value for H was determined in several ways:

(1) For all experiments, 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 samples no distinction could be made for H values

in the high and low n regimes.

(2) In test no. 51. the steps were done at two stress levels at different tempera-

tures. The value of H was 8 * 5 kcal/mol. Unfortunately, it was not possible to

determine in which flow regime the measurement was taken.

(3) Geller (1924. 1925) determined the temperature dependence of the extrusion

pressure for a number of salts, among which bischofite. With a number of reasona-

ble assumptions. his data can be recalculated to yield a value of the activation

TABLE 111

\‘alurs for A and n for each experiment

Test no. High n regime

A n

low n regime

A n

45 2.4. lo- ”

47 2.8. lo- lo

27 3.6.10m9

30 2.2.10-s

41 5.1. lo- I0 42 5.6. 1O’9

49 3.1. 1o-9 52 l 4.1. lo- ‘” 43 2.4. 1O-9

46 0.9. lo-’

48 3.0.10~*

5.1

4.2 0.83.10-s 1.6

4.6

4. I 6.3

4.6

4.1 0.83~10~s 1.8

4.6 3.9, lo-9 2.3

4.85

2.9 1.3. lo-’ 1.1

3.2 1.23.10-’ 1.4

* Gram growth during preparation was less extensive than in other samples. possibly because water

content of this sample was slightly less than normal.

13X

energy. These assumptions are: (a) the rate of extrusion in all Geller’s experiments

was the same. (He raised the pressure up to a value so that extrusion just began.) (b)

The extrusion pressure at a given rate in Geller’s apparatus is proportional to the

uniaxial flow stress of the same material:

p=Ka (3)

where K is a constant determined by apparatus effects and extrusion rate (see Laue

and Stenger, 1976).

With these assumptions, a plot of the logarithm of the extrusion pressure versus

l/T 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) while reinterpretation of data from Geller (1924. 1925) gives a

similar value but with less scatter. The present data do not allow a distinction

between 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 of

bischofite in nature, it is important to describe in some detail the microstructures of

natural bischofite.

The bischofite layers are quite pure (about 98% bischofite) and they contain

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 the grains appear to be undeformed. and grain boundaries are

slightly curved. The grains sometimes contain growth twins. Deformation 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 drawing of the grain boundaries from a few bischofite drillcores. Black dots are halite or

carnallite inclusions. 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 grains, they are often arranged in walls which form networks

closely resembling the present grain-boundary configuration in bischofite. These

kieserite walls represent the old (diagenetic or primary) grain-boundary configura-

tion before deformation. By extensive recrystallization the grain boundaries have

moved away. Observation of the kieserite network does indicate, however, 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 samples

Compacted samples (pure and impure)

The milled 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

growth by annealing produces optically strain-free, equiaxed grains with slightly

140

curve grain boundaries, frequently making an angle of 120 degrees at triple points.

Annealing twins were present in dry samples. while they are quite rare in slightly wet

and wet ones. Grain sizes were around 1 mm in slightly wet and wet materials. and

around 0.1 mm in dry samples.

In slightly wet and wet samples, most of the grain boundaries contained ara>s of

fluid inclusions. As their size was about 5-50 microns. they were studied in 0.5 mm

thick sections to avoid damage.

In the dry samples, the fine-grained aggregates of tetrahydrate 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 has been extensively studied by Grube and Brauning

(1938) and Dietzel and Serowy (1959). MgCl, .6H,O melts at 117°C incongruently

to give a saturated solution of MgClz. 4H,O with some crystals of M&I,. 4H,O

which dissolve at 129°C. During cooling, a typical ingot texture is formed. consisting

of a radially grown aggregate of elongated hexahydrate grains. with manv grain

boundaries parallel to (110). about 10 X 5 mm in diameter. with irregulari>- distrib-

uted needles of MgCl, .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 micro-

structural changes during deformation was obtained from the coarse-grained sam-

ples. During preparation of the thin sections bischofite could be obsened to

recrystallize at room temperature along scratches on the surface of the thin section

(Fig. 7). This raised the question of the stability of the microstructure after deforma-

tion. Additional information was obtained from the in-situ deformation experiments.

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 recrystalliza-

tion 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 fre-

quently developed. The identity of these slip systems is as yet unknown. Twin glide

Fig. 7. Kieserite inclusions in a Bischofite grain from the Veendam drillcores, arranged along walls

indicating former grain-boundary positions. The bischofite grain is extensively twinned during specimen

preparatmn. The array of fine grains at the bottom of the photograph is a preparation artifact. Scale bar is

1 mm. crossed polarizers.

Fig. 8. Mechanically twinned grains together with ones showing undulose extinction in a dry sample

(T 29). Many of the small recrystallized grains contain growth twins. Scale bar is 0.1 mm. Crossed

polarizers.

142

on the (110) system in bischofite has been determined by Mtigge (1906). It is at all

temperatures the easiest deformation system (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, indicat-

ing 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 of the strain due to grain-boundary sliding could be de-

termined. Using:

~,s = cpvd (4)

where ‘p is a geometrical factor, o the average step height at grain-boundary offsets.

Fig. 9a. Sharp boundary between twinned and untwinned part of a grain: a “kink” formed by t% in glide

(T40). Recrystallization is initiated at grain boundaries. Scale bar is 0. I mm.. crossed polarizers. Very fine

grains (white spots) are artifacts due to grinding.

Fig. 9b. In-situ sequence illustrating the formation of the structure shown in Fig. 9a. Dry sample. room

temperature. Note the shape change of the initially spherical holes in the grains. Scale bar is 0.2 mm.

crossed polarizers.

and d the grain size (Bell and Langdon, 1969; Gifkins, 1973) about lo-20% of the

total strain is believed to be due to grain-boundary 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.

This process has been called “rotation recrystallization” (Poirier and Guillope,

1979). No fluid inclusions were found on subgrain boundaries having misorienta-

tions 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 of twin lamellae, or at preexisting grain boundaries. This last

process is either due to high strain in grain boundary regions, or to the fact that

misorientation across a subgrain boundary has reached a critical value so the

boundary can start to migrate (Gottstein et al., 1976).

When growing into a grain containing twin lamellae, the new grains were

frequently elongated parallel to these (Fig. 10). Grain boundaries between twinned

grains, and recrystallizing grains which were much larger than the width of twin

lamellae showed a characteristic stepped shape, closely resembling the microstruc-

ture 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 is somewhat puzzling. It can be understood by realizing that most grain

boundaries have moved after twinning and the original grain boundaries are rarely

visible (lensoid twins were seen to form in the in-situ tests).

The appearance of new grains formed by migration recrystallization was re-

stricted to the first stages of deformation where the coarse-grained samples were

transformed into fine-grained material. After this in the dynamically recrystallizing

Fig. 10. Elongated new grains growing along twin lamellar in a larger old grain (T40). Scale bar LS 0. I mm.

Crossed polarizers. Very fine grains are artifacts 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, water content and temperature

High n regime. In general, increasing the water content had the same influence on

microstructure as increasing temperature. At 60°C (first series of experiments).

strongly deformed grains were quite common in the dry samples (see Fig. 8).

Recrystallized grains with grains sizes around 0.1 mm frequently contained anneal-

ing twins. The presence of both grain-boundary migration rates, as was observed in

in-situ tests (see Fig. 12), resulted in a bimodal distribution of recrystallized

grain-sizes {Fig. 11). Arrays of gas and fluid inclusions, indicating the position of old

grain boundaries in the undeformed samples. were found inside new grains, while

newly formed grain boundaries were clear and inclusion-free.

The slightly wet and wet samples had a strikingly different microstructure.

Mechanically twinned grains or grains with undulose extinction were very rare

giving the samples an optically strain-free appearance. It should be noted that thi:

microstructure developed during dynamic recrystaliization, although it would not

have been recognized as such using microstructural criteria which are generally

i m

*; _L

Fig. 11. Mechanically twinned old grain is replaced hy new ones. Because of the difference in

grain-boundary migration velocity. bimodal grain-size distribution is produced. T40. scale bar is 0.1 mm,

crossed polarizers.

Fig. 12. In-situ sequence illustrating recrystallization by two different-grain boundary velocities. A. In a

strongly deformed old grain new grains are slowly formed at C. B. C. New grain A starts growing wjith a

migration rate of 1 micron/set. D. Grain B, which has the same orientation as grain A. starts growing.

T = 80°C. dry sample. Strain rate = IOm5 s ‘. crossed polarizers. Scale bar is 0.2 mm.

applied for rocks like the presence of new grains showing undulose extinction (see

also Means, 1982).

The microstructure was formed by the complex interplay of subgrain rotation and

grain boundary migration (at the slow and fast rate). In the in-situ experiments. dry

samples at 60°C showed mainly slow migration of grain boundaries with an

occasional grain growing rapidly (Fig. 12) while at 100°C rapid growth was very

common. Wet samples displayed this behaviour already at room temperature.

indicating that an increased water content enhances the rapid grain boundary

migration.

The dynamically recrystallizing microstructure was thus mainly determined by

rapidly migrating grain boundaries, producing the typical microstructure shown 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 resulting from the complex interplay of grain-boundary migration at two

different rates and subgrain rotation. (T29: coarse grained. slightly wet, pure sample.) See text for

discussion. Grains having the same number belong to the same “orientation family”. Scale bar is 1 mm,

crossed polarizers. The very fine grains, the fine twin lamellae and the air bubbles in the resin are

preparation artifacts.

old grain was swept by a rapidly migrating grain boundary. Very frequently the

rapid migration stops before the whole grain is consumed, leaving over the smaller

grains. There may be a number of reasons for this:

(a) The transition from slow to fast migration occurs catastrophically. During fast

migration (e.g., absorbing impurities or changes in grain-boundary structure) the

mobility of the grain boundary may drop so that the reverse catastrophic change

occurs and the migration rate drops to its slow migration value.

(b) The migrating grain boundary arrives at a part of the grain which has just

recently developed by slow migration and consequently has a low dislocation density

so the driving force for migration is much less in this region.

(c) The mobility of the grain boundary is also orientation dependent. 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 disloca-

tion 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

C D

Ftg. 14. Schematic drawing illustrating development of the microstructure shown in Fig. 13. Only the

effect of recrystallization 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, it is cut in two by a migrating grain boundary. Thus, the microstructure

will consist of “orientation families”, formed by the propagation of one orientation

through the sample, while this orientation is destroyed by other migrating grain

boundaries, and modified by subgrain rotation and lattice rotation processes (Fig.

14). A slight shape preferred orientation perpendicular to the compression direction

was observed in most samples after deformation, showing the combined effect of

flattening of grains and an influence on the shape of new grains of the existing grain

boundary configuration.

~~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 microstructure of the samples deformed in the low n regime. Arrow, indicates a grain

boundag which has stopped halfway across an old grain. This IS the process by which the small grains on

the edges of the big ones are created. T5 1. crossed polarizers. scale bar is 1 mm.

and gas inclusions (see Fig. 16) wet samples containing more than the dry ones.

When studying wet or slightly wet samples in-situ, grain boundaries were optically

clear without visible fluid inclusions. During migration, however, they could be seen

to both adsorb and leave behind fluid inclusions. After deformation stopped, fluid

inclusions connected to the grain boundary by thin channels 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 to a microstruc-

ture consisting of polygonal, equiaxed grains, closely resembling the 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 before and after drformatlon m

slightly wet and wet samples (T34). Scale bar is 0.1 mm.. partly crossed polarizers. thickness of that secuon

about 0.5 mm.

Fig. 17. In-situ sequence illustrating growth of isolated fluid inclusions in a wet bischofite sample. As is

argued in the present paper, the arrays of fluid and gas inclusions at grain boundaries are formed by this

process from a continuous fluid film which is present during grain boundary migration. Scale bar is 50

microns, plane polarized light. Time lapse between first and last photograph is 30 min. T = 50°C.

Fig. 18. Idiomorphic grain overgrowing the microstructure formed by dynamic recrystallization. This

microstructure IS also observed in natural bischofite. and is diagnostic for a small amount of a second

phase on the grain boundaries. Note the growth twins. Wet sample. scale bar is I mm. crossed polarizers. Blach spots are preparation artifacts.

present 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 of

bischofite causes a strong weakening of the material.

As will be argued below, above a certain value of excess water content 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 recrystalliza-

tion (Glover and Sellers, 1973).

152

It is by this process that, after stepping to a lower value of differential stress. the

samples relatively rapidly reach a new steady state. Driven by the strain energy

stored in the samples, recrystallization will continue until the sample reaches the

strain energy level corresponding to the new stress value. As recrystallization is a

relatively rapid process, this state will be reached much faster than by recovery

processes alone. It is proposed that the initial hardening during unconfined creep

tests can be understood when one considers that softening 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 of water in the MgCI,. 6H10 lattice is not known.

microstructural observations indicate that there is also an intracrystalline effect of

water: the absence of undulose extinction and the large amount of subgrain rotation

in slightly wet and wet samples suggest the ease of climb processes in these samples.

This enhanced recovery will also cause weakening. Work is in progress to determine

the unit cell parameters of wet and dry bischofite. This may give an indication

whether or not water is present in the lattice.

Because the slightly wet and wet samples 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 pressure of 28.0 MPa. although in these experiments

fluid pressure can not have been much higher than 0.1 MPa.

(3) As will be argued below, a continuous fluid film will only be present on the

grain boundaries after some recrystallization has taken place (that is. after a few

percent strain) so at low strains fluid pressure will not be important. Howev.er. the

strong difference in differential stress between dry and slightly vvet samples is

already present 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 (disloca-

tion 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 fluid inclusions are left behind. When deformation is stopped and a

grain boundary ceases to move, by the process of necking 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, in dry samples, grain boundaries migrate away

from the fluid inclusions, and do not contain fluid themselves (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 their total volume by the counted

area gave the thickness of the fluid layer, about 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 structure is strongly different

in dry and wet samples. the presence of annealing twins in dry samples (which has

been inferred to be due to differences in interfacial energy by Aust and Rutter. 1960)

is at least qualitatively explained. Because the change in grain-boundary properties

seems to be completed in our dry samples, it is suggested that decreasing the excess

water content even further will have less effect on the flow stress than between 0.1

and 0.0 1 o/r (see Fig. 2).

The microstructures in the bischofite cores can fully be interpreted in terms of the

deformation behaviour observed in the experimentally deformed material: subgrain

Fig. 19. Old grain boundary. indicated by an array of fluid inclusions is left behind by a migrating grain

boundary. This behaviour is only found in the dry samples. T40. crossed polarizers, scale bar is 0.1 mm.

154

rotation accompanied by the migration of high-angle grain boundaries. The growth

of idiomorphic grains after deformation has only been observed in the wet samples.

This is strong evidence for natural bischofite behaving as our wet samples.

Another reason for natural bischofite being “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 for other salt minerals. as these are known to contain fluid

inclusions on grain boundaries (e.g.. Roedder and Bassett, 1981). The water content

of domal salt is in the order of 0.1% which is of the same order of magnitude as the

amount necessary for weakening bischofite.

Also, in a recent work, White and White (1981) showed 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. Their observations of bubble

arrays on grain boundaries are comparable with 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 diapirism, when they were shown to have contained much more

water then is present now (Herrmann, 1980).

Catastrophic changes 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 migra-

tion. 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 change in grain boundary velocity may be explained

due to impurity effects on the growing crystal surface (Kern. 1969). This process

resembles that of the migration of fluid inclusions in salt crystals under the influence

of a thermal gradient (Anthony and Cline. 1974; Holdoway, 1974) both thermal

gradient and high dislocation density giving an increase in solubility of the mineral.

A difference is the different grain orientation on both sides of the water layer.

Thus, although the phenomena 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 side of the grain was reached. However, in some cases migration

stopped halfway across the grain. 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 done with paradichlorabenzene in an apparatus, described by

Means (1981). using high strain rates (10-I s-l). In a few other cases. the grain

boundary then started moving backwards; similar processes, for which no explana-

tion is yet given. have also been described by Means (1982). In an attempt to explain

why grain boundaries move like this, Means (1982) considered 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 experi-

mental 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. and Cline. H.E.. 1974. Thermomigratton of liquid droplet:, in salt. In: A.H. Coogan

(Editor), Fourth Symposium on Salt. Northern Ohio Geol. Sot.. 3 13 pp.

Aust. K.T. and Rutter. J.W.. 1960. Annealing twins and coincidence site boundaries in htgh-purity lead.

Trans. Metall. Sot. AIME. 218: 1023-1028.

Bell. R.L. and Langdon. T.G.. 1969. In: R.C. Gifkins (Editor). Interfaces Conference. Butterworth\.

Sydney, pp. 115- 137.

Belmonte. Y.. Hirtz. P. and Wenger, R.. 1965. The salt basins of the Gabon and the Congo (Braua\tle).

In: Salt Basins Around Africa. Proc. of a Joint Meeting of the Inst. of Petroleum and the Geologtcal

Society. pp. 55-7X.

Bosworth. W.. 1981. Strain-induced preferential dissolution of halite. Tectonophysics. 78: 5099525,

Burke, J.E.. 1950. Grain growth in ceramics. In: Kingery W.D. (Editor). Kinetics of High Temperature

Processes. M.I.T. Press. Cambridge. Mass.

Calais. M.M.. Pascal, J. and Lacombe. P.. 1961. Les facteura de la migration des Jomts de grams dam

I’uranium-alpha. In: G. Chaudron (Editor). 4e Colloque de Metallurgic Proprietea de Joints de Gratnr.

Presses Universitaires de France.

Coelewij. P.A.J.. Haug. G.M.W. and Van Kuijk. H.. 1978. Magnesium salt exploration in the northed,tern

Netherlands. Geol. Mijnbouw. 57(4): 487-502.

Dewar. J.. 1894. Note on the viscosity of solids. Proc. Chem. Sot.. 137 pp.

Diarov. M. and Dogalov. A.B.. 1971. Bischofite rocks of the northern Caspam louland. Ir\. .Akad. Sauk

Kaz. SSR. Ser. Geol.. 27(2): 72-76.

Dietzel, H. and Serowy. F.. 1959. Die Losungsgleichgewichte des Systems hlgCI,-H1O zutschen 20 und

200°C. Freib. Forschungh.. A. 132: 5-32.

Fulda. E.. 1928. Das Kali. Enke, Stuttgart.

Fulda, E.. 193 I. Erlauterungen zur Geologischen Karte van Preussen. blatt Vienenburg. Nr. 223 I.

Geller. A., 1924. Uber das Verhalten verschiedener Minerale der Salzlager bei hohen Druchen und

wechselenden Temperaturen. Z. Kristallogr.. 60: 414-472.

Geller, A.. 1925. Zur Kenntnis des Fliessdruckes fester K&per. Z. Kristallogr.. 62: 395541 I.

Geller, A.. 1930. Das Schmelzen van Salzen bei hohen Drucken und seiner Bedeutung fur den V’organg

der Salzmetamorphose. Fortschr. Mineral.. Kristallogr. Petrogr.. 14: 143-166.

Gifkins. R.C.. 1970. Transitions in creep behaviour. J. Mater. Sci. 5: 156-165.

Gifkins. R.C.. 1973. The measurement of grain boundary sliding in polycrystalline specimens. hlet. Set. J..

7: 15-19.

Glover. G. and Sellars. C.M.. 1973. Recovery and recrystallization during high temperature deformation

of alpha-iron. Metall. Trans., 4: 765-775.

Gottstein. G., ZabardJadi. D. and Mecking. H., 1976. Investigations of dynamic recrystallization during

tensile deformation of copper single crystals. Proc. 4th Int. Conf. on the Strength of Metals and

Alloys. Nancy.

Grube, G. and Bratming, W.. 1938. Uber die Entwasserung van Magnesiumchlond Hertahydrat und

Carnallit. Z. Electrochem.. 44 (2): 134.

Guillope, M. and Poirier, J.P., 1979. Dynamic recrystallization during creep of single crystalline hahte: an

experimental study. J. Geophys. Res., 84810): 5557.

Heard, H.C.. 1972. Steady state flow in polycrystalline halite at a pressure of 2 kilobars. In: Flou and

Fracture of Rocks.. Geophys. Monogr. Ser., Am. Geophys. Union, 16.

Herrmann. A.G., 1980a. Geochemische und mineralogische Grundlagen fur die Entlagerung radioakti\er

Substanzen in Salzdiapiren Norddeutschlands. Fortschr. Mineral.. 58(l): 169-21 I.

Herrmann. A.G., 1980b. Geochemische Prozesse in mat-men Salrablagerungen: Bedeutung und

Konzequenzen fur die Endlagerung radioaktiver Substanzen in Salzdiapiren. Z. Dtsch. Geol. Ges..

13 1: 433-459.

157

Holdouay. K.A. 1974. Behaviour of Fluid inclusions in Salt during Heating and irradiation. In: A.H.

Coogan (Editor). Fourth Symposium on Salt. Northern Ohio Geological Soctety. Vol. 1. p. 303.

Kazantsev. O.A.. Yermakov. V.A. and Grebennikov, N.P., 1976. Discovery of bischofite in the lower

Volga region. Int. Geol. Rev., 18: 734-742.

Kern. R.. 1969. Crystal growth and adsorption. In: N.N. Sheftal (Editor). Growth of Crystals. Vol. X

Consultants Bureau. New York. Vol.8 pp. 3-23.

Laue. K. and Stenger. H.. 1976. Strangpressen. Aluminium-Verlag. Dusseldorf.

Lemmlein. G.G. and Kliya. M.O.. 1960. Distinctive features of the healing of a crack in a crystal under

conditions of declining temperature. lnt. Geol. Rev.. 196012): 125-128.

Leng. Pet Keng. 1945. Gefugeuntersuchung des Trtimmercarnallits Krugershall zu Teutschental. Halle mit

hesonderer Berucksichtigung des Carnallitgefiiges. Dtss. Techn. Hochschule Berlin.

Lowe. F.. 1957. Steinsalz und Kalisalze. Borntraeger, Berlin.

Liicke. K. and Sttiwe. H.P.. 1971. On the theory of impurity controlled grain houndary motion. Acta

Metal.. 19: 1087-99.

Means. W.D.. 1981. The concept of a steady state foliation. Tectonophysics. 78: 179-199.

Means, W.D., 1983. Microstructure and micromotion in recrystallization flow of octachloropropane: a

first look. Sander volume Geol. Rundsch.. in press.

Mugge, 0.. 1906. Uber die Kristallform und Deformattonen des Bischofiet und der verwandten Chlorure

van Kobalt und Nickel. Neues Jahrb. Mineral.. Geol. Palaontol., I: 91 pp.

Mukhin. L.K.. Sobolev. V.M.. Leonidova. A.I.. Federova, E.P. and Angelopulo, O.K.. 1975. Effect of

drilling fluids on the deformati~~n of artificial Bischofite specimens. Neft. Khoz. 1975(7): 20.-22.

Potrier. J.P.. 1977. Microscopic creep models and the interpretation of stress-drop tests during creep. Acta

Metall.. 25: 913-917.

Potrier. J.P. and Guillope. M., 1979. Deformation induced recrystallization of minerals. Bull. Mineral..

102: 67-74.

Poirier. J.P. and Nicolas. A., 1975. Deformation-induced recrystallization due to progressive misorienta-

tion of subgrains, with special reference to mantle peridotites. J. Geol., 83: 7077720.

Roedder. E. and Bassett. R.L.. 1981. Problems in determination of water content of rock-salt samples and

11s significance in nuclear-waste storage siting. Geology, 9: 525-530.

Schmid. S.M.. Boland, J.N. and Paterson, MS., 1977. Superplastic flow in fine-grained limestone.

Tectonophysics. 43: 2577291.

Schmtd. S.M.. Paterson. M.S. and Boland. J.N., 1980. High temperature flow and dynamic recrystallira-

tion in Carrera marble. Tectonophysics, 65: 245-280.

Streiets. G.A.. Filatov. B.S.. Luban. 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. and Humphreys. F.J.. 1981. An in-situ investigatton of the deformation hehaviour of

sodium nitrate- an anatogue for calcite. Tectonophysics, 78: 661-675.

Urai. J.L.. Humphreys. F.J. and Burrows, SE., 1980. In-situ studies of the deformation and dynamic

recrystallization of rhombohedral camphor. J. Mater. Sci., 15: 1231-1240.

Van Eekelen. H.A., Hulsebos. A. and Urai. J.L., 1983. Creep of bischofite. Proc. 1st Conf. on the

Mechanical Behaviour of Salt. Pennsylvania State University, University Park. Pa.. in press.

White. J.C. and White. S.H.. 1981, On the structure of grain boundaries in tectonites. Tectonophysics. 78

613-628.

Wilkins. R.W.T. and Barkas. J.P.. 1978. Fluid inclusions. deformation and recrystallizati(~n in granite

tectonites. Contrib. Mineral. Petrogr.. 65: 2933299.