You are on page 1of 12

Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

View Article Online / Journal Homepage / Table of Contents for this issue

Indirect Measurements of the Effect of Stress on the


Velocity of Dislocations

BY D. HULLAND F. NOBLE
Dept. of Metallurgy, The University, Liverpool, 3
Received 15th June, 1964
Three indirect methods of determining the variation of dislocation velocity with stress are de-
scribed and their limitations discussed. Experiments on single crystals of silicon iron have shown
that the three methods give comparable results. The variation of the parameter rn, in the empirical
relation II = (./TO)" between velocity and stress, has been measured as a function of strain using a
stress relaxation method and the results considered in relation to.observations on the macroscopic
deformation processes in single crystals. Additional experimental observations using the relaxation
method are presented for polycrystalline silicon iron, pure iron and tantalum, and for tantalum
single crystals.

Gilman 1 found that the rapid variation of dislocation velocity with applied stress
in lithium fluoride could be described by
v = vo exp (-A/z), (1)
where vo is the elastic shear wave velocity of the material and z the applied stress.
Stein and Low 2 also found that their velocity data for dislocations in iron+3& <;
silicon could be expressed in this way, though an extrapolation to 112 = 0 did not
give the shear wave velocity for this material. Over the lower range of dislocation
velocities the relation
D = (.Cl.to)", (2)
where 70 is the resolved shear'stress required to produce a dislocation velocity of
1 cm sec-1 and m is a constant characteristic of the temperature and material, has
been found to represent adequately the lithium fluoride and silicon iron results.
Schadler 3 has confirmed this relationship for tungsten,
Both relations (1) and (2) are purely empirical and, in particular, refer to results
at a particular temperature. Johnston and GiIman 4 found that the temperature
dependence could be represented by
v = (z/zo)mexp (-E/kT), (3)
where E is the activation energy associated with dislocation movement. This relation
was found to be applicable to silicon iron only over a small range of stress and
temperature.
In attempts 6 to study dislocation mobility by indirect experiments, the velocity-
59

stress relationship in eqn. (2) has been assumed to apply and the value of the exponent
m obtained in such experiments compared with the value obtained by direct observa-
tion. The basis of the indirect experiments is the relation
E = bpv (4)
where 4 is the plastic strain rate and p is the density of moving dislocations. Assuming
p is constant at constant strain and substituting from eqn. (2),
4 = Kz", (5)
25 1
View Article Online

252 EFFECT O F STRESS O N D I S L O C A T I O N VELOCITY

where K = bp/z" = constant. Differentiation of eqn. (5) gives


6'In k/i? In z = rn = 8In 618 In 0, (6)
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

where c is the applied tensile stress. Writing


B = 80 exp (- H(r)/kT), (7)
an activation volume v* can be associated with the deformation process such that 7
D* = kT(8 In yaz*)*, (8)
so from eqn. (6),
v* = mkT/a. (9)
Thus, by determining the change of strain rate with stress, m can be obtained
using eqn. (6) and the activation volume using eqn. (9). The result will be meaningful
only if K i n eqn. (5) is truly constant throughout the test, i.e., there is no change in
the effective dislocation density. In addition, machine transients associated with
strain rate change tests, used for evaluating rn and o*, may lead to erroneous results.
In an attempt to determine the magnitude of the above limitations we have used three
types of indirect test and compared the values of m obtained. The three tests were :
(1) measurement of the strain rate dependence of the flow stress,s (eqn. (6));
(2) measurement of the change in strain rate produced by an increment of load to a
creeping specimen, (eqn. (6)) ; (3) measurement of the relaxation curves produced
by stopping the tensile test of a plastically deforming specimen in an elastically
hard machine.8 Assuming a constant density of moving dislocations, rn can be
determined from the relaxation curve as follows. The elastic strain rate of the
specimen and machine at time t after the start of relaxation is given by
1 do
&,= --
E dt'
where 0 is the stress at time t and Eis the elastic constant of the specimen and machine.
In the absence of an imposed strain rate the plastic strain rate kP is equal to the elastic
strain rate :

so from eqn. (5),


1 do -
---- - Ka".
E dt
Integrating gives
In Q = [l/(l -m)J In (I +At)+constant (12)
m is obtained directly from a plot of In o against In t provided A t 9 1.
Although these tests are basically the same experiment performed in different
ways it is important to note the differences of detail. For clarity some typical results
are shown in fig. 1-3. These tests were made on single crystals of iron +3+ % silicon
at 293°K. The orientation of the crystals were identical, as shown in inset diagram,
and it was possible to perform tests 1 and 3 described above on the same specimen.
In the first test, large changes of strain rate are required to produce small changes in
the flow stress so that errors in measurement may be large and machine transients
often produce complicating effects. In addition, because of the large strain rate
change required to produce measurable flow stress changes, a change in the moving
dislocation density is most likely in this test. The latter effect can be minimized by
using the second type of test ; here small, but very accurately known, load increments
View Article Online

D . H U L L A N D F . NOBLE 253
can be applied to the specimen in a dead loading machine. Changes in strain rate
are large (fig. 2) even for small load increments. Machine transients are not necessarily
eliminated in this test but can be completely eliminated in the third type of test
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

(fig. 3). The value of m is obtained from the full relaxation curve, and good agree-
ment is obtained with eqn. (12), as demonstrated in fig. 4. Since it was assumed in
the derivation of eqn. (12) that the dislocation density remained constant this type of
experiment provides a direct check on this postulate.

elongation, %
FIG.1.-Change in flow stress associated with change in strain rate for a single crystal of silicon
iron at 293°K.

In table 1 the values of rn for iron +3$ % silicon single crystals (orientation as
in fig. 1) determined using the three tests are listed. As shown later, m is independent
of strain during the Luders strain but changes during work hardening and depends
on strain ; the work-hardening values refer, therefore, to a given strain ( 0.03). N

The results are in fairly good agreement and give confidence that each method gives a
true value of pn for the specimen. It seems most unlikely that a substantial change
in the mobile dislocation density, as postulated by Guard,s can occur to the same

TABLECOM COMPARISON OF VALUES OF m DETERMINED BY THREE INDIRECT METHODS ON


IRON+ 33 % SILICON SJNGLE CRYSTALS (FOR ORIENTATION see fig. 1).
Liiders work-hardening

-
Test region region, strain 4 - 0 3
(1) strain rate change 54 75
(2) load increments 53 - 84
(3) stress relaxation 58 N 77
View Article Online

254 EFFECT OF STRESS O N D I S L O C A T I O N VELOCITY

(b)
W o r k Hardening Region
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

Luders . Region

0 50 100 I50 200 250


time (sec)
FIG. 2.-Change in strain rate associated with change in applied load, (a) Liiders-strain region,
(b) work-hardening region, for a single crystal of silicon iron at 293°K.

0
strain time
FIG.3.-Stress relaxation curve for a single crystal of silicon iron at 293°K
View Article Online

D . H U L L A N D F . NOBLE 255
extent in the three types of test to give similar values of m. If one accepts that the
values of rn measured are due solely to the movement of dislocations then these values
may be used as a basis for assessing the relevancy of direct observations of dislocation
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

mobility to bulk deformation processes.

log time (sec)


FIG.4.-Log plot of stress relaxation curves for single crystal of silicon iron at 293°K.

Determinations of m in the work-hardening region are complicated due to the


back stress acting on the moving dislocations as a result of work hardening. When
such a stress exists it is incorrect to use the applied stress in the indirect evaluation
of m. For purposes of comparison with " direct " values of rn a back stress factor
must be subtracted from the applied stress to give the stress experienced by a disloca-
tion moving into undeformed material as in the etch-pit experiments. Stein and
Johnston6 calculated values of m from strain rate change tests using the applied
stress and they extrapolated work-hardening values of m to zero strain as a means of
eliminating the back stress factor. However, the present results appear to cast
doubt on the validity of applying such a technique at low strains.

SOME EXPERIMENTAL OBSERVATIONS


In the context of the present paper the principal reason for undertaking a study
of the relation between dislocation velocity and stress has been to understand the
yield behaviour of single crystals of silicon iron. The previous section has shown
that the stress relaxation test, which is easy to do experimentally and to analyze,
can be used with confidence to determine rn and it seemed appropriate to explore the
applicability of the approach to polycrystalline silicon iron and other b.c.c. metals.
Therefore, we describe some of the features of yielding in single crystals and the
variation of m in a variety of metals.
Single crystals of silicoii iron, with a rectangular cross-section, were cut from large
grain-size polycrystalline sheet. Prior to testing they were electropolished and
annealed at 800°C. The general features of yielding were similar at all crystal
View Article Online

256 EFFECT OF STRESS ON DISLOCATION VELOCITY

orientations and testing temperatures and it was possible to follow the formation
and growth of slip lines during deformation on the polished surfaces of the crystals
using an optical microscope attached to the tensile machine. Plastic deformation is
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

initiated usually at both specimen grips in the form of discrete, coarse, slip lines.
Yielding occurs by the spread of the deformation in four ways : (i) the slip lines grow
in length across the specimen; (ii) further slip occurs within the slipped volume
producing larger steps on the surface; (iii) the slip lines broaden, often into wide
bands; (iv) new slip lines are nucleated further along the specimen and between
existing slip lines.

E
i
w

-
4

0.1 1.0 2.0


log time (sec)
FIG.6.-Log plot of stress relaxation curves for single crystal of silicon iron tested at 293°K
showing variation of slope with strain, see-caption fig. 8.

In this way a stable slip band configuration, as shown in fig. 5, is established and
this propagates down the gauge length as a Luders band. No discontinuity in this
process is observed during the yield drop (if one is present). Macroscopic work
hardening occurs when the Luders region completely covers the specimen and there
are no more deformation-free regons. Since an important contribution to the
Luders strain is the movement of slip lines into undeformed parts of the crystal
some direct correlation is expected between directly and indirectly determined values
of m.
In fig. 6 a typical set of results from stress relaxation tests at different strains has
been plotted. rn is independent of strain during the Luders strain and then increases
sharply at the onset of work hardening. There is a slow increase of m during work
hardening. The variation of rn with strain was found to be almost independent of
orientation. Two specific orientations were studied in more detail. These were
chosen such that, at 293"K, slip was restricted to (a) two { 1 lo} (1 11) systems, i.e., the
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47. View Article Online

FIG. 5.. -Distribution of slip bands at the Luders front in a single crystal of silicon iron ( x 100).
[Toface page 256.
View Article Online

D . H U L L AND F. NOBLE 257


orientation shown in fig. 1 and the one used for the comparison experiments in the
first section of this paper, and (b) to a single ( 112) (1 11) system, i.e., the tensile axis
was 9" from [llO] along the [110]-[lll] boundary. The summary of the results in
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

table 2 demonstrates that there is only a small, and probably insignificant, difference
between m in these two extreme cases. This agrees with Erickson's etch-pit observa-
tions 9 that the stress dependence of the velocity of dislocations moving in (1 12)
planes is the same as for dislocations moving in (110) planes, as measured by Stein
and Low.2 Table 2 also gives values of m determined by stress relaxation at 77°K
and shows that in this material there is only a small variation of m with temperature
cf. Stein and Low 2, which is only significant in the Luders strain region.

TABLE2.-vARIATION OF m WITH STRAIN FOR IRONf3$% SILICON SINGLE CRYSTALS


DETERMINED BY THE RELAXATION METHOD

(a) Crystal oriented for (110) slip tested at 293°K; Luders strain = 0-038
strain true tensile stress m
kg mm-2
0.005 323 64
0-014 32.3 64
0.027 31.9 61
0.052 33.1 76
0.083 35.5 79
0.133 38.0 80
0.197 40.5 81
0.236 42.0 85
(b) Crystal oriented for (112) slip tested at 293°K; luders strain = 0.004
strain true tensile stress m
kg mm-2
0.0002 31-4 59
0.0017 31-4 62
0.016 326 68
0.041 346 71
0.088 37-1 82
0.260 46.0 86
(c) Crystals oriented for (1 10) slip tested at 77°K ; Luders strain = 0-054
strain true tensile stress m
kg mm-a
0405 70 76
0.025 69.5 73
0.044 69.5 74
0.085 76.2 81
0.121 80.5 82
0.176 87.0 82

An attempt was made to obtain m using an alternative direct method to the


restrictive etch pit technique. Single crystal specimens were strained in a dead
loading machine in the Liiders strain region. A series of photographs was taken of
the specimen surface as the specimen was pulse loaded for different times. In this
way the position and movement of individual slip lines was measured. By pulse
loading at different applied loads it was possible to obtain a wide range of slip line
velocities (assuming that the velocity is merely the distance moved by a slip line during
a fixed time of pulse loading). A typical set of results is given in fig. 7. These results
represent the arithmetic mean of the velocity of a large number of individually
View Article Online

258 EFFECT OF STRESS O N DISLOCATION VELOCITY

measured slip lines in a given specimen. It was found that the slip lines did not move
at a uniform velocity and for a given time of pulse the velocity of the individual slip
lines varied over a wide range. In spite of the limited reproducibility of results nz,
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

determined by this method, was usually between 40 and 50, and a line corresponding
to rn = 40 is marked in fig. 7.

Foo 500 4.00

log velocity
FIG.7.-Variation of the velocity of the tips ofslip bands as a function of stress for a single crystal
of silicon iron tested at 293°K.

For comparison and information, the results of stress relaxation tests on poly-
crystalline silicon iron, pure iron (0.004 % C), and tantalum, and single crystals of
tantalum are listed in table 3. The polycrystalline silicon iron gave comparable,
though slightly higher, values of m to those in single crystals. The polycrystalline
iron and tantalum showed exceptionally marked increases of m at the onset of work

TABLE3.--(U) VARIATION OF n2 WlTH STRAIN FOR POLYCRYSTALLINE SPECIMENS


material grain diamcter Liiders region work-hardening
mm region
Ta 1 . 5 10-2
~ 26 67-78
Fe 2.4 x 10-2 21 71-102
Fe4-34 % Si 1.8 x 10-2 89 106-112

(b) VARIATION
OF m WITH STRAIN FOR SINGLE CRYSTALS OF TANTALUM
temp. OK LGde,s region work-hardening
region
293 11 13-23
77 87 87
View Article Online

D . H U L L A N D F. NOBLE 259
hardening and fig. 8 demonstrates this particularly well. In the tantalum single
crystals, m was appreciably lower in the Luders-strain region than in the poly-
crystalline material, although this might be due to compositional changes. The
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

orientation of the crystal was such that the Luders strain was short. The crystals
deformed in a similar way to the silicon-iron single crystals. In contrast to the silicon
iron there was a marked dependence of rn on temperature.

.4 1-0 2.0

log time
FIG.8 . 4 tress relaxation curves of polycrystallinepure iron at 293°K showing pronounced changein
slope at the end of Luders strain, (log u is replaced by A u in this example, see reference 8).

DISCUSSION

All the results show a definite discontinuity in the indirectly determined values of
m obtained in the Luders and work-hardening regions. In silicon-iron single crystals
this discontinuity is slight but reproducible. The values of rn obtained in the Luders
region are constant and lower than values obtained in the work-hardening region.
In the work-hardening region, m becomes greater with increasing strain when the
applied stress is used in calculating rn and hence extrapolation in the work-hardening
region as discussed previously may be justified. However, in crystals which show a
Luders region, extrapolation to zero strain can only be performed by ignoring the
constant values obtained in this region. The justification for such a procedure is
far from clear.
Good agreement with the results of direct observation experiments is expected
only when exactly the same process is being studied in each case. Since the direct
method involves the propagation of a slip line into undeformed material this process
must be responsiblefor the whole of the observed strain rate in the indirect experiments
if agreement is to be achieved. Undoubtedly, this process is contributing to the
Luders-strain rate in the deformation of a single crystal but the contribution to the
strain rate by the propagation of a limited number of slip lines is likely to be small in
View Article Online

260 E F F E C T O F STRESS ON D i S L O C A T l O N V E L O C l T Y

proportion to the strain rate contribution due to continuing deformation in the existing
slip lines. The process being studied in the indirect test in the Liiders region is
therefore the mobility of dislocations in an already deformed region with a small
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

contribution from the dislocations at the tips of the slip lines moving into undeformed
material. In the steady state-region of Liiders propagation a constant value of m
should be obtained since there will be a fixed contribution to the plastic strain rate by
the various mechanisms. When work hardening begins, the small contribution
from the slip-line tips vanishes and a discontinuity in values of m occurs at this point.
Because the contribution from the slip-line tips is small similar values of m should be
obtained in both Liiders and work-hardening regions because essentially the same
process is being studied in each case. It is not clear why values of m should be
higher even at the yield point for dislocations moving within slip bands though the
reason is probably due to the velocity distribution which exists within the bands and
the complex stress field through which the dislocation moves.
Evaluation of the values of m obtained in the work-hardening region assuming the
effective stress on the dislocations was constant at the value at the end of the Luders
region did not give a constant m but the values obtained were more nearly constant
than if the applied stress were used.
The values of m obtained in the tests on polycrystalline silicon iron follow much
the same pattern as those for the single crystals but the values in general are slightly
higher. As with single crystals, a slight but reproducible discontinuity in values of m
between Luders and work-hardening regions is observed. Constancy of m in the
Luders region is found as would be expected. The results may be explained by
analogy with the single crystal results, namely, that in the Luders region by far the
greater contribution to the strain rate is due to the movement of dislocations in already
yielded grains behind the Luders front. The reason for the slightly lower values
obtained in the Luders region is the small but measurable contribution of the dis-
location movement in the grains in the process of YieldingattheLiidersfront. Such an
explanation implies a fairly wide Luders-strain profile and a diffuse band front in
silicon iron. At low enough strains, therefore, the band may not be fully developed
and the contribution from the yielding grains in the Liiders front will be correspond-
ingly enhanced. The lower values of m which were obtained close to the yield point
may be the result of such a situation.
Unfortunately, no directly observed values are available for pure iron or tantalum
so the results of relaxation experiments can only be compared with the corresponding
tests on silicon iron. For the polycrystalline case the change in values of m obtained
in Luders and work-hardening regions is very large-about three times. If we
accept the above reasoning on Luders-band profiles, then these results suggest that
in polycrystalline pure iron and tantalum the Liiders profile is extremely sharp and
that the yielding grains in the front are the only ones contributing to the deformation
rate.
As with silicon iron, the values of rn obtained in the Liiders region for polycrystal-
line tantalum are somewhat higher than for the single crystal results. This dis-
crepancy is presumably due to the effect of really measuring a weighted average of m
in the polycrystalline case as discussed by Li.10
The activation volume associated with any particular value of m can be calculated
from eqn. (9). The results show a considerable discontinuous change in u* between
the Luders and work-hardening regions, particularly in polycrystalline pure iron and
tantalum. These results contradict those of Conrad 7 who found that in pure iron
v* was independent of the deformation process occurring. However, in silicon iron
the indirectly determined values of m in the Liiders region are closer to those determined
View Article Online

D . HULL AND F . NOBLE 261


by direct observation than the indirectly determined values in the work-hardening
region. Thus, the results are not inconsistent with Conrad's conclusion that yield
propagation is controlled by the overcoming of the Peierls barriers as opposed to an
Published on 01 January 1964. Downloaded by University of Michigan Library on 25/10/2014 04:31:47.

unpinning or cross-slip mechanism.

We are grateful to many colleagues for helpful discussion. One of us (F. N.)
acknowledgesthe generous financial assistance provided by a Leverhulme Fellowship.

1Ghan, in
Fracture, Wiley, 1959.
2 Stein and Low, J. Appl. Physics, 1960,31, 362.
3 Schadler, Acta. Met., to be
published.
4 Johnston and Gilman, J. Appl. Physics, 1959,30, 129.
5 Guard, Acta Met., 1961, 9, 163.
6 Johnston and Stein, Acta Met., 1963, 11, 317.
7 Conrad, J. Iron Steel Inst., 1961,198, 364.
* Noble and Hull, Acta Met., to be published.
9 Erickson, J. Appl. Physics, 1962,33,2499.
10 Li, U.S. Steel Report No. 1129, 1963.

You might also like