Home Search Collections Journals About Contact us My IOPscience

The development of highly magnetostrictive rare earth-iron alloys

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1994 J. Phys. D: Appl. Phys. 27 1

(http://iopscience.iop.org/0022-3727/27/1/001)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 128.103.149.52
This content was downloaded on 26/07/2015 at 03:40

Please note that terms and conditions apply.

J. Phys. D: Appl. Phys. 27 (1994) 1-11. Printed in the UK

REVIEW ARTICLE

I The development of highly

I magnetostrictive rare earth-iron alloys
D C Jiles
Ames Laboratoryt, Iowa State University, Ames, IA 50011, USA
Received 23 November 1992, in final form 23 August 1993

Abstract. This paper reviews recent developments in highly magnetostrictive

terbium-dysprosium-iron alloys. This includes discussion of domain Structure,
the magnetization process, microstructure, sample preparation, magnetostriction,
effectsof differentchemical compositions on magnetic properties of the alloys,
and possible applications of these materials. It has been found over recent years
that there has been a Steady improvement in material properties, both in terms of
increased strain amplitude and reduction in variability from sample to sample. The
material is now finding applications in magnetostrictive transducers, actuators and
adaptive vibration control systems.

1. Introduction and of DyFez is +2.1 x IO6 J m-3, in comparison KI of

Although the rare earth metals terbium and dysprosium
have very large magnetostrictions, as shown in the work
of Rhyne and Legvold [I] and Clark et ai [2], they also
have large anisotropies and low transition temperatures.
Therefore they are unsuitable as practical materials
in magnetostrictive devices and for applications that
generally need to operate at ambient temperature under
the action of relatively small fields of a few kiloampkres
per metre. The ordering temperatures can he raised
by alloying these elements with iron to form the Laves
phase intermetallic compounds TbFez and DyFez. These
alloys have high magnetostrictions at room temperature,
but also have high anisotropy, which necessitates high
field strengths, typically greater than H = 2 x lo6 A m-'
(25 kOe) to saturate the magnetization as shown by Clark
[3]. Therefore these compounds also remain unsuitable
as practical materials in devices and applications.
The anisotropy constants of TbFez and DyFez are,
however, opposite in sign and therefore the use of an
alloy consisting of Tb, Dy and Fe formed the next step
in the search for a practical material for magnetostrictive
applications. The pseudo-binary alloy (Tb0,3Dy,,,)Fe~
was first suggested as a likely candidate material in
the work of Clark [3]. At 298 K (25°C) the first
anisotropy constant Kl of TbFez is -7.6 x lo6 J m-3
t Ames Laboratory is operated for the US Department of Energy
by Iowa State University under contract W-740.5-ENC-82. This
work was supported by the Office of Basic Energy Sciences.
Division of Materials Sciences.
Tbo.3Dyo,FeZ is only -0.06 x lo6 J m-3, which leads to
higher permeability and lower coercivity. Therefore, at
room temperature, the properties of this pseudo-binary
alloy, particularly the rate of change of magnetostriction
with field dh/dH, are much better suited for applications
than those of the binary compound.
Early work on this material by Abbundi and Clark
[4] showed magnetostrictions along the [ I l l ] axis as
large as 2000 x at 240 K under a field of
1.2 MA m-I (15 kOe), as shown in figure 1. At
room temperature of 298 K (25 "C) the magnetostriction
under the same field strength was 1640 x From
the shape of the curves, this was assumed to be the
saturation magnetostriction along this axis, a result
that has been confirmed by subsequent observations.
Anisotropy measurements on the alloy by Clark et a1
I51 were able to show that the anisotropy constants
were very low (these were later found to be typically
KI = -0.6 x IO5 J KZ = -2.0 x lo5 J m-3
at room temperature, although these values are strongly
dependent on chemical composition) while the material
maintained its high hulk magnetostriction. However,
it was also shown that at low temperatures (23 K <
T < 283 K) the value of Kt is positive, leading to
easy axis alignment along the (100) axes and hence
very low magnetostriction. At about 283 K (lO"C),
KI becomes zero, and above this temperature K I
becomes negative, so that the (111) axes become the
directions of easy alignment. This leads to much higher
magnetostriction at temperatures above 10"Cand below
the Curie temperature of typically 653 K (380°C).

0022-3727/94/010001+11$07.508 1994 IOP Publishing Ltd

D C Jiles
0 400 800 1200
H(M.O!"l
Figure 1. Variation of magnetostriction of Tb0.27Dy0.73Fe2
along the 11111 direction with applied field at various
temperatures (after Abbundi and Clark 141).
2. Domain structure and the magnetization
process
Despite these improvements in anisotropy of !he alloy
it still required relatively high fields to saturate it, and
so in no sense could it be described as a soft magnetic
material. Therefore it was clear that some understanding
of the magnetization mechanism would be important in
improving the performance of the material in device-
related applications. The need for such understanding
was noted for example by Clark et al [6] in their domain
studies on single crystal Tb0,27Dy0.73Fe2, via synchrotron
x-ray topography. Using a (110) oriented disc, they
were able to show that the slow approach to saturation
resulted from formation of an immobile domain structure
consisting of 109" domain walls between domains
oriented along easy (111) axes inclined to the surface
of the disc. The component of magnetization in the
plane of the specimen is the same for each of the
domains, and so under the action of an in-plane field
the structure was immobile for the field strengths used
in their study, H = 0.239 x lo6 A m-' ( E = 0.3 Tesla).
Their observations indicated that quite rapid changes
in magnetization occurred with field strengths of up
to H = 0.1 1 x IO6 A m-' (1.4 kOe), but, beyond
that, the rise to saturation, which occurred at about
H = 0.32 x lo6 A m-I (4 kOe), was slow.
Further x-ray topography studies of the domain
structure and defects of single-crystal Tb0,27Dy0.nFe1,95
were carried out by Lord et a1 [7]. This study
was concerned with a specimen containing a high
density of defects in the form of oxide inclusions and
microcracks. The main conclusion was that this material
contained a high density of (1 11) Widmanstztten platelet
precipitates, which were thought to have the composition
We3 where R is a rare earth. These were the main
2
defects found in the material, and were considered to be
the primary cause of high coercivity.
Domain observations can also be made by optical
techniques. This usually involves use of polarized
light as in the magneto-optic Kerr effect. However,
work by Parpia et al [SI has shown that in Tb-Dy-
Fe, the magnetic domains can even be observed directly
in unpolarized light because of the surface distortions
arising from large magnetostriction.
The role of microstructure is of course crucial
in defining the domain configuration in this material,
and consequently in determining the bulk magnetic
properties such as permeability and magnetostriction. It
was observed by Abell and Lord [9] that, although some
understanding of the magnetization mechanism has been
obtained from domain observations, interpretation is
nevertheless complicated by the presence of inclusions,
defects and other metallurgical phases in the material.
It is these effects among others that have impeded
technological development of this material because of
resulting variations in magnetic properties from sample
to sample. The presence of additional phases arises as a
result of deviations from the stoichiometriccomposition.
Subsequent domain observations using optical dif-
ferential interference contrast microscopy were reported
by Lord et al [lo]. Here again, as in the work of
Parpia et al [SI, distortion of the single-crystal surface
resulting from magnetostrictive strains was used to im-
age the domains. As in the work of Clark et al, the
domain configurations could be interpreted in terms of
low energy 71" and 109" domain walls with components
of magnetization normal to the surface. Observations
by these authors indicated that little change in the do-
main configuration occurred for field strengths of up to
H = 32 kA m-' (400 Oe) along the (100) direction
in the plane of the (011) disc specimens, however, at
fields of H = 64 kA m-' (800 Oe) significant changes
in domain structure were observed. When the field was
applied along the { I l l ) axis in the plane of the (112)
disc specimen the domain configuration was found to
change at lower field strengths, which appeared to be
about a factor of two lower. In particular, although little
change was observed in fields of up to H = 16 kA m-I
(200 Oe) changes occurred at fields above this strength
and the specimen appeared to be completely saturated at
H = 64 kA m-' (800 Oe). A stereographic projection
of the various crystallographic axes onto the (112) plane
is shown in figure 2.
Recent domain observations in Tb0.3Dyo.7Fez
twinned single crystals by Teter et a1 [ 1 I ] used both
differential interference contrast and x-ray topography.
It was shown that good correlation exists between the
interpretation of domain observations and the measured
bulk magnetization and magnetostriction when the field
and strain are along the [112] axis. However, no expla-
nation of the effects of applied stress and field along this
direction was given. Neither were there results on the
application of field and measurement of magnetostric-
tion along the [ I 1 11 direction. This highlights the need
for the development of quantitative theoretical models to

Figure 2. Stereographic projection of the (112)plane,
showing the various principal crystallographic directions
(after Lord et a/ [IO]).
c110, <Ill> <loo> <Ill>410,
Figure 3. Torque magnetometer results for a specimen of
(after Sat0 et a/ [12]).
Tb0.27Dy0.7SFe2
account for the observed properties, which can be used
to predict results in situations for which no experimental
data yet exist.
The crystallographic orientation of the magnetization
vectors within the domains is dependent on temperature.
Results of Sato et al [12] have shown that below 23 K
the magnetic moments are oriented along the (1 10) axis.
Between 23 K and 285 K they are oriented along the
(100) axes, and above 285 K they are oriented along the
(111) axes. The torque magnetometer measurements of
Sato et al are shown in figure 3.
3. Microstructure
Although the effects of microstructure on extrinsic
(or structure-sensitive) magnetic properties are difficult
to describe theoretically, they are very important
in determining the bulk magnetic properties such
as permeability and hence the dependence of both
magnetization and magnetostriction on field. This
remains true even though the microstructure has
little effect on the intrinsic magnetic properties such
as saturation magnetization Ms,the magnetostriction
Highly magnetostrictive rare earth-iron alloys
coefficients AIm and A l l l , or the magnetocrystalline
anisotropy constants Kt and K2.
The main influence of microstructure on magnetic
properties arises through its effects on domain wall
motion. Here the term microstructure is used to mean
the distribution of different phases within the material
and their morphologies. Anisotropy and texture are
therefore separate from microstructure. The distribution
of the different phases can have different effects on the
pinning of domain wall motion. On the other hand, the
distribution of these phases will have little effect on the
magnetocrystalline anisotropy of the primary phase and
so will not greatly influence rotation of the magnetization
direction within domains in the primary phase.
It has been established by Lord et a1 [7] that
the presence of plate-like particles, also known as
Wiedmansatten zones, adversely affects the magnetic
properties of this material. The Wiedmanstatten platelets
are composed of an iron-rich phase of composition
We3. In order to avoid the presence of these
particles, recent material has been produced, which
has a nominal chemical composition (Wex), with iron
slightly lower than the stoichiometric value of x =
2. Suppression of the Wiedmanstatten zones in this
way has led to improved properties. Although it is
felt that the stoichiometric composition is perhaps the
ideal composition, this is difficult to achieve in practice
because of loss of the rare earth components during
synthesis.
A number of microstructural studies of Tb-Dy-Fe
alloys have been conducted. The main concerns have
been to establish an optimum chemical composition of
the alloys to maximize permeability and rate of change
of magnetostriction with field dh/dH. Abell and Lord
[9] investigated the effects of heat treatment on the
density and distribution of precipitates, and the relation
of these to the magnetization process. This study was
similar to the earlier study by Joyce et al [13], which
showed that magnetization and magnetostriction of these
alloys were strongly affected by iron content in the range
1.7 ix c 2.3. The dependence of magnetostriction on
iron content is shown in figure 4. Joyce et al were also
concerned with the coefficient d33. the rate of change
of magnetostriction with field, which also exhibited a
similar double peak with iron content close to 2.
In the work of Abell and Lord [9], alloys were
investigated that had constant rare earth ratio Tb:Dy,
but differing iron compositions in the range 1.4 < x <
2.5. The as-cast microstructures of these alloys were
radically affected by the iron content. For compositions
with x < 2.0, the second phase was predominantly a
rare-earth-rich phase, which was present as both intra-
and intergranular precipitates. For x > 2.0 there
were acicular particles of the iron-rich phase, which
was thought to be We,. In the range 1.9 c x <
2.1 WiedmansVdtten-type (plate-like) precipitates were
found to be present at all compositions except x = 2,
the stoichiometric composition. Subsequent changes
in microstructure of these materials were found for
compositions in the range 1.7 < x < 2.3.
3
D C Jiles
\
600
I
1.7 i.6 1'9 2.0 2.1
x (Iron Conlonl)
Figure 4. Variation of maximum ObSeNed magnetostrictive
with iron content x (after Joyce et
strain in Tbo.27Dy0.73FeX
a/ 1131).
In the study of Tb,-,Dy,Fe2 alloys for y = 0, 0.25,
0.5, 0.6, 0.65, 0.73, 0.8 and 1.0, it was found that there
were additional phases present for all values of y except
y = 0.73. This is somewhat surprising, since at first
sight one would expect that the iron:rare earth ratio
alone controlled the presence of additional phases, and
not the terbium to dysprosium ratio. The conclusion
from this investigation was that, when varying the
Tb:Dy ratio, the optimum microstructure appeared to
occur at a chemical composition Tb0.27D0.73FQ. It was
also found that the We2 phase constitutes the matrix
material in all of these cases, and that, for x < 2.0,
secondary recrystallization occurs, involving a small
amount of remaining rare-earth-rich liquid on the grain
boundaries, following primary solidification of We2. A
tertiary recrystallization process occurs for x > 2.0 after
solidification of We3 and W e phases.
In conclusion, Ahell and Lord deduced that, at
least for arc-casting of small ingots, the stoichiometric
composition Tbo.27D0.73Fe2 seemed to he the optimum
composition because this gave a single-phase material.
This was different from conclusions reached by others
using float zoning or Bridgman techniques, in which it
has been found that the presence of unwanted additional
phases, in particular We3 platelets, was minimized
by using starting material with x % 1.95. However,
this latter result may simply reflect the loss of rare
earth material between the starting composition and the
final composition of the specimen. This is a well-
known problem in the synthesis of this material and is
demonstrated, for example, in the results of figure 5.
Subsequent microstructural characterization of Tbo.2,
Do.7sFe, (1.95 < x < 2.05) was reported by Westwood
er a1 [ 141, using scanning electron microscopy (SEM).
The backscattered electron micrographs with x = 1.95,
x = 2.0 and x = 2.05 showed significant differences.
Once again the x = 2.0 microstructure gave no evidence
of second-phase material in the as-cast ingots, while
x z 2.0 showed the presence of the iron-rich We3
4
-
2
0-
72
- . ~ TARGET
VALUE
L.
7i
x la9
Figure 5. Variation of the rare earth-to-iron ratio
along the length of a specimen of nominal composition
Tbo.nrDyo.nFel.s (after McMasters et a/ [17]).
phase, often as Wiedmanstatten precipitates within the
grains, together with large acicular dendrites of We3.
The x < 2.0 microstructure revealed the presence of the
rare-earth-rich phase, which accumulated on the grain
boundaries, with some rare earth oxide attached to any
microcracks present within the grains.
Jenner er a1 [ 151 have also conducted microstructure
studies on this material, in which they have imaged
the domain structures and tried to correlate these with
the microstructure. The study was made with different
Tb:Dy ratios, the specimens having been prepared by
contained float-zone growth.
4. Sample preparation
The phase diagrams of the Tb-Fe and Dy-Fe systems are
shown in figure 6. The methods of sample preparation
of this alloy fall broadly into three categories: (i)
float zone solidification; (ii) Bridgman growth; and
(iii) Czochralski growth. Several different methods of
sample preparation have been tried. Simple as-cast
material prepared by argon arc melting, as in the work
of Ahell and Lord [9], can be refined for example by
the horizontal zone method employed by McMasters
[ 161, or the free-standing float-zone solidification method
as used later by McMasters et a1 [17]. Alternatively,
single crystals can be grown by the Czochralski method
[IS] or the Bridgman method [19,20]. The best results
have been obtained with the free-standing float-zone
technique, although some problems still remain, the most
notable being the reproducibility of magnetic properties
from sample to sample.
It has proved very difficult to grow (1 11) direction
oriented single crystals of the material in any but the
smallest volumes. This is because on solidification,
the dendrites of the material grow naturally along the
(1 12) directions with no side branching out of the ( 1 11)
sheet planes [19], as shown in figure 7. This leads
to oriented platelets, as shown in figure 8. It would
clearly be advantageous to be able to grow (1 11) oriented
single crystals because the magnetostriction is greatest
along the (111) directions. Under present capabilities it

m n
Alomic pensnl Dy~pmsium
Figure 6. (a) Phase diagram of the TbFe system. (b)
Phase diagram of the Dy-Fe system.
<112>
L A
Figure 7. Diagram of dendritic growth front along the (1 12)
direction in Terfenol (after Verhoeven et a/ 1191).
is only possible to grow large specimens (for example
40 mm long x 5 mm diameter rods) with the (112)
crystallographic axis along the growth direction.
Another problem that arises is twinning of the
crystals. The presence of twins is very difficult to
eradicate because of the low energy needed to produce
them. This in itself also leads to a reduction in properties
such as saturation magnetostriction because it is no
longer possible to have the optimum [ill] axis aligned
along a particular direction throughout the whole solid.
More recent methods of production have sought to
find less expensive methods of producing high-quality
Highly magnetostrictive rare earthiron alloys
Figure 8. Diagram of dendritic platelets formed in
T&,27Dy0.73Fe2(after Clark et a/ [28]).
material. Verhoeven et al [21] have successfully used
directional solidification, in which the molten material
was poured from the bottom of a crucible into a
cylindrical quartz mould. The quartz mould was pre-
heated but contained a water-cooled copper pedestal at
the bottom. The molten alloy then began solidifying
directionally upwards. Comparisons were made between
the directionally solidified specimens and those produced
by the float-zone technique. The directionally solidified
materials were in general of lower quality (typically
20% lower strain amplitude) than the float-zone-melted
materials.
5. Thermomagnetic post-process annealing
A post-process thermomagnetic anneal has been used
recently to improve further the magnetic properties
by causing preferential alignment of magnetic domains
normal to the unique axis of the specimen. Although the
properties of the directionally solidified material were in
general not quite so good as the float-zoned material,
the process did produce fairly high-performance material
and it seems to be a method that can usefully be used to
optimize the performance-to-cost ratio.
It has been found by Verhoeven et al [22] that a
final anneal of rod-shaped ( 1 12) oriented specimens in
a magnetic field of H = 0.72 x lo6 A m-' (9 kOe)
aligned along the (111} direction normal to the rod axis,
at a temperature of 950°C for 1 h, led to significant
improvements in the magnetostrictive properties of the
bulk material. In subsequent measurements at mom
temperature, the magnetostriction at H = 0.2 x
IO6 A m-' (2.5 kOe) was more than doubled from
800 x to 1990 x as a result of this treatment.
The interpretation of this interesting result is that the
magnetic field preferentially aligns magnetic domains
along the (111) direction normal to the unique axis
of the specimen. When this alignment of domains is
achieved under elevated temperatures it results in a relief
of strains. After subsequent cooling followed by removal
of the field any other alignment of the domains will cause
an increase in the internal stress in the material as a
result of magnetostrictive strains along other directions.
5
D C Jibs
Therefore, preferential alignment of the domains along
this perpendicular direction remains even after the field
is removed. However, it is likely that the fractional
occupancy of domains in this direction is somewhat
reduced once the field is removed because of the high
magnetostatic energy, which arises from alignment of
magnetic moments along the short dimension of a
specimen rather than along the rod axis.
6.Magnetostriction and t h e magnetization
process
The bulk magnetostriction A is caused by non-180"
domain processes occurring during the magnetization
process. It is therefore advantageous to arrange for
the magnetization vectors within the domains to align
perpendicular to the intended field direction in the
demagnetized state. At present material is grown as
twinned crystals with a (1 12) crystallographic direction
oriented preferentially along the unique axis of the
cylinder. Since the magnetostriction is largest along the
(1 11) axes it is widely assumed that the properties of
the material would be improved if the material could be
grown with a ( I 11) crystallographic direction oriented
preferentially along the unique axis of the cylinder.
Measurements of the magnetization and magne-
tostriction of this and related alloys up to 1980 have
been summarized in the review article by Clark [23].
At that time the value of A L I I was known to be about
1640 x however, the value of htw, which is at
least an order of magnitude smaller, was unknown and
difficulties were encountered in its measurement in the
presence of such a large strain along the (1 1 I) axes.
In order to observe the largest bulk magnetostriction
in a given specimen, it is necessary to minimize the
amount of 180" domain wall motion or domain rotation,
since these processes make no contribution to the change
in length of the sample. This can be achieved by
aligning the domain moments perpendicular to the field
direction using a compressive uniaxial stress. Clark
and Savage [23] were first to report measurements of
the bulk magnetostriction h(= dl/l) under conditions
of applied stress of up to 24 MPa. This resulted in
a greatly enhanced magnetostriction of h = 1450 x
at 12.5 MPa and under a magnetic field of
H = 100 kA m-I (1.25 kOe). The increased bulk
magnetostriction under applied compressive load has
opened up a range of possibilities for utilization of these
alloys in practical applications such as transducers and
positioning devices.
Subsequent work by Clark et a1 [25]reported values
of magnetostriction of up to h = 1800 x for
compressive stresses of greater than 20 MPa in fields
of H = 160- kA m-I (2 kOe) in oriented twinned
'single' crystal Tbo.z7D0.,3Fe2. It was also found in
this work that the magnetostriction A under compressive
stress deviated significantly from the ideal or theoretical
dependence on M Z , which is expected when all domains
have been aligned perpendicular to the field axis in zero
6
i -
li Diu 5W x 10.81 ~ MPa
0
33
I
105
14
240 0 240
H W.m.'l
Figure 9. Variation of bulk magnetostrictive strain 1 with
magnetic field H under variousapplied compressive stress
levels U (afterJiles et a/ [30]).
field by an applied stress as shown by Jiles [ 2 6 ] . This
deviation from quadratic dependence on M is caused by
the magnetocrystalline anisotropy.
The highly magnetostrictive behaviour of these
materials is retained at higher temperatures, as discussed
by Clark and Crowder [27] who measured the properties
at temperatures up to 375°C.In their paper it was
shown that the anisotropy in magnetostriction is very
large, A l l , a 50Alw in the related compound DyFez.
and therefore, to ensure high magnetostriction, it is
essential to operate the material at temperatures at which
the (111) axes are the easy directions, and at which
the anisotropy constants Kl and Kz are low (to ensure
high permeability). The theoretical limit of the bulk
saturation magnetostriction is $ I I I = 2460 x [281,
which can be obtained if all domains are initially aligned
perpendicular to the field axis in zero field (by the
effect of the applied compressive stress); if the [ I l l ]
axis is aligned along the direction of application of the
magnetic field and measurement of strain, and the final
orientation of moments is in a single domain along the
[ 11 I] direction.
Recent improvements in the material have led to an
enhancement of the magnetostrictive properties. Results
on these have been reported by Clark et al [29].
These high-performance materials exhibit an interesting
discontinuous variation of magnetostriction with field
H. The properties of several specimens of this latest
material were investigated thoroughly by Jiles et al
[ZO]using simultaneous measurement of magnetic field,
induction, magnetostriction and Barkhausen effect under
on-line computer control [31]. In one specimen, these
measurements gave a magnetostriction amplitude of
(2200 ?c 20) x at a field of H = 240 kA m-I
(3 kOe) under a compressive uniaxial stress of 14 MPa.
This result, which is shown in figure 9, was significantly
higher than had been observed under similar conditions
on previous material, and represents a limiting value for
magnetostriction measured along the (1 12) direction.
The change in magnetization when the magnetization
changes from the perpendicular [ l l i ] direction to the
parallel [ 1121 direction is, to a first approximation, given
by the equation given by Jiles and Thoelke (Journal of
Magnetism and Magnetic Materials, to be published).

- signal intensity with applied field H at various applied
80 0 80
H W.m"l
Figure 10. Variation of Barkhausen signal intensity with
applied field H at various applied compressive stress
levels U (after Jiles e t a / [30]).
3 and so there does not appear to be an irreversible change
+ 3hl IIC(COS e,, COS ef,j+l- COS ei,, COS ei,j+l)
,=I
x cos +j cos @j+l
where &,, are the angles of initial orientation, and 0f.j
are the angles of final orientation.
Assuming values of A l l 1 = 1640 x and AIM) =
50 x the largest change in magnetostriction along
the [I 121 direction occurred for rotation of domains from
the [fill or [ I l i ] directions into the [ I l l ] direction.
This gave rise to an induced magnetostriction of Ah =
2190 p strain. When the moments were deflected further
in the [ I 121 direction by the action of a stronger field,
the magnetostriction actually declined to 2063 p strain.
Further investigation of this specimen by Jiles, Os-
tenson and Owen [32] indicated that the Barkhausen
spectrum, figure IO, had two peaks in count rate, which
seemed to indicate two principal magnetization mech-
anisms. Magnetoacoustic emission studies, figure 1 1,
revealed additional fine structure within the two count
rate peaks, which so far has not been satisfactorily ac-
counted for, but may be a result of twinning of the crystal
alignment in the specimen, which is known to occur in
all of these specimens. An interesting feature was the
behaviour of the magnetostriction of the unstressed ma-
terial as the field was swept from the demagnetized state
at H = 0 to H = 240 kA m-' (3 kOe) and then cycled
at this amplitude.
In the unstressed material an unexplained effect has
been observed in which the magnetostriction begins at
A = 0 in the demagnetized state (H = 0, B = 0)
but after magnetizing it never returns to A = 0 despite
Highly magnetostrictive rare earth-iron alloys
I I I
160 80 80 160
kAm.7
Figure 11. Variation of magnetoacoustic emission
compressive stress levels U (afterJiles et a/ [32]).
both H and B returning to zero as the field is cycled.
This has been observed in all specimens to date. It
appears to be due to an irreversible change in the
domain configuration, so that the original demagnetized
domain configuration is never regained in the cyclic
state. However, in the stressed material the situation is
quite different. In this case, the h = 0 state is regained
in the domain configuration. These effects have been
described by Sablik et a1 [33].
Recent Barkhausen effect measurements have been
analysed using pulse-height spectrum analysis as in the
work of Jiles and Hariharan [34]. This showed the
distribution of Barkhausen emission pulse amplitudes,
and provided additional information on the magnetic
properties of materials. In particular it has shown that,
as the compressive stress was applied to the material,
there was a marked increase of Barkhausen activity at
2.5 n Wb s-'. This was the result of increasing the
domain occupancy of easy directions perpendicular to
the fieldstress axis.
7. Applications
The material is beginning to find applications in
sonar transducers, anti-vibration mountings, positioning
devices, dynamic load detection and magnetometry.
These applications have been discussed in a recent
review [35]. Applications-oriented measurements of the
materials properties have been made by Pitman eta1 [36l
and by Pitman [37]. Pitman, for example, has shown
that the strain amplitude under bias fields in excess of
50 kA m-l (625 Oe) exhibits a peak under an applied
compressive load of 12 ma, as shown in figure 12.
For applications such as low-frequency acoustic
transducers, in which energy conversion is the primary
concern, the property of principal interest is the
7
D C Jiles
wain (10.61
,000
......._..I.0iaEBOW.T
.. ....... 0ia3 70 w m
Applied SVBSI IMP4
Figure 12. Dependence of magnetostrictive strain on
applied stress at various levels of magnetic field bias (after
Pitman 1351).
9iaSFiEld1X4.W1J
Figure 13. Dependence of the coupling coefficient on
bias field for ( a ) TbonDyo.73Fet.9;
and (b) Tbo,3Dyo,,Fel,s
(after Jenner et a/ [15]).
magnetomechanical coupling factor k33. This is defined
by the equation
where po is the magnetic constant (4z x IO-’ H m-l),
fir is the relative permeability, E is the elastic modulus,
A is the magnetostriction and H is the magnetic field.
k:, is the fraction of the magnetic energy which can
be converted to elastic energy per cycle in an ideal
(namely lossless) situation. The value of k33 is optimized
by designing materials with a high magnetostriction-
to-anisotropy ratio. Furthermore, k33 varies with bias
field strength in a given sample, as shown by Jenner
et a1 [15], figure 13. This means that, in order to
maximize the energy conversion efficiency, transducers
of the material should be operated undere a bias field
of typically 50 kA m-l (625 Oe). Values in the range
k33 = 0.4-0.75 have been reported [38].
Another parameter of importance for applications,
such as magnetostrictive magnetometers, in which the
sensitivity of strain to applied magnetic field is a primary
concern, is the rate of change of magnetostriction with
8
Applied Field (k”)
Figure 14. Variation of the strain derivative &(= dA/dH)
as a function of applied field H at various compressive
stress levels for Tbo.sDyo.7Fer.s(after Jenner et a/ [40]).
field, dA/dH. This is often denoted d , or sometimes d3?,
the value of dA/dH along the unique axis of a cylinder
of the material,
4 3 = ($).
Specimens can be operated close to the maximum
value (dA/dH),, by application of a bias field, supplied
for example by a permanent magnet, so that the largest
magnetostrictive change in length can be obtained
for a small AC field amplitude. In other forms of
detection, in which the change in magnetization is
used to detect a strain, for example in magnetostrictive
vibration detectors, the parameter of primary interest is
the generation coefficient g [39]:
where M is the magnetization, U is the stress, E is the
elastic modulus and H is the magnetic field. In these
applications, the larger the value of g, the more sensitive
the device.
Performance of the Tb-Dy-Fe alloy for DC actuator
devices has been studied under quasi DC conditions
by Jenner er al [40], in which the strain derivative
d3,(dA/dH) and the coupling coefficient k33 were
investigated as a function of both magnetic field and
applied stress, figures 14 and 15. The optimum operating
conditions were found to be dependent on the specific
application, as may be expected. The field strengths
at which peaks occurred in (dA/dH) were shown to
vary with stress, and in particular these field strengths
increased with applied compressive stress, with the
largest of the maxima occurring under a stress of 5 MPa
and a field of 15 kA m-!.
In most cases the primary consideration for DC
applications, such as positioning devices, is the
maximum strain A, and the maximum strain derivative
(dA/dH),,. Ideally, these should be as large as
possible and occur at the lowest field strengths possible.
Schulze et a1 [41] have investigated these parameters
as a function of stress up to 20 MPa and magnetic
field strengths of 18 kA m-’. The largest value

0.2 I
0 40 80
I
120
Applied Field ( W m )
Figure 15. Variation of the coupling coefficient with
applied field H at various compressive stress levels for
Tbo.sDyo.,Fer.s(afler Jenner et a/ [40]).
of (dh/dH), reported to date has been 0.312 x
A-' m by Verhoeven et al 1221.
The composition of the alloys, in particular the iron
content, strongly affects d33 and can result in changes as
large as a factor of two, as reported by Joyce ef al [ 131.
This means that the field range over which the non-180"
domain processes (magnetostrictive processes) occur is
strongly dependent on the iron content. Optimum
performance is obtained when this range is a minimum,
since this will lead to the largest values of d33. In
comparison the magnetostriction is less sensitive to the
iron content, although it was found to vary by 20% for
iron contents in the range 1.85 < x < 2.10, where the
iron content x is defined by Tb0.27Dy0.73Fex.A further
study of magnetomechanical coupling k33 in these alloys
has been reported by Abell et d [42], in which the
values of k33 were calculated from the measurement of
resonance and anti-resonance in the electrical impedance
of a coil wrapped around the specimen.
The resulting plots of impedance versus frequency at
various bias field strengths are shown in figure 16. The
peaks represent resonance, h, and the valleys represent
antiresonance, fa. The value of kg3 was then calculated
from the equation
The calculated values of k33 as a function of bias field
are shown in figure 17.
Savage et al [43], have investigated macroscopic
strain as a function of magnetic field and applied stress
in this alloy. From this it was possible to calculate the
external work done by the material. It was found that the
magnetomechanical coupling coefficient k is inadequate
for descriptions of the material performance in high-
strain applications because k is derived on the basis of
small strains and the approximations involved do not
cany over to high strains.
For AC applications such as adaptive vibration
control, additional factors such as eddy current effects
Highly magnetostrictive rare earthiron alloys
Bias lield W m " )
0 40 60 80 100 $20 140 160 180
Frequency (kHz)
Figure 16. Variation of the electrical impedance
with frequency for a coil wrapped on a specimen of
Tb0.27Dyo.73Fe2. The curves show both resonance
(maxima) and anti-resonance (minima) at different
frequencies as the bias field was changed. The coupling
coefficient can be calculated from the location of the
resonance frequency f and the anti-resonance frequency
fa, as described in the text (after Abell et a/ [42]).
100 200 300
Bmrfleld (ram-')
Figure 17. Variation of the coupling coefficienth,as
calculated from the impedance curves, with bias field for
Tb0.27Dy0.73Fe2 (after Abell et a/ [42]).
and skin depth, need to be taken into account. The
performance of the material under these conditions
has been investigated by Greenough and Schulze 1441,
who have shown the frequency-dependence of both
relative permeability and magnetostrictive strain 1451.
It is clear from this work that, under zero bias field,
the dependence of complex permeability on frequency,
based on calculation of the classical eddy current effect,
does not lead to the observed result, as shown in
figure 18, and that therefore some corrections need to
be made to take into account the domain configuration.
However, under a bias field of 15 kA m-I, the
agreement between the observed complex permeability
and the expected value based on classical eddy current
calculations is much improved. It is considered that this
9
D C Jiles
1
3 s
-____.
....A -*-A
....I
D
5 both cases, this leads to deterioration of the magnetic
Lap I
Figure 18. Variation of the magnitude and imaginary
components of the complex relative permeability with
frequency for Tbo,aDyo,7Fe2 (after Greenough and Schulze
Wl).
is probably due to the slowness of non-180" domain
processes in responding to higher frequency excitations
because of the magnetostrictive work done in these
processes.
The movement of the surface of a rod of Tt-Dy-Fe
under the action of an AC field was subsequently found
to be non-uniform in the work of Schulze etal [46]. This
work indicated that the characterization of the strain of
the material using a surface strain gauge is not truly
representative of the bulk response of the material under
AC conditions.
8. Conclusions
This paper has outlined some of the recent developments
in the highly magnetostrictive terbium-dysprosium-iron
alloy, and provides a brief update on work conducted
since the review of Clark [23]. The main conclusions
from this survey of recent work are that several obstacles
still need to be overcome before the material will see
widespread engineering applications. These are
(i) to reduce the variability in properties arising from
identical processing procedures,
(ii) to understand better the relationship between
structure and magnetic properties of these materials,
(iii) to extend the viable range of frequencies by
reducing conductivity, and
(iv) to enhance the magnetostriction amplitude
AA and in particular the differential magnetostriction
dA/dH.
It is clear from earlier work that the anisotropy
of the material is largely controlled by the terbium-to-
dysprosium ratio. The anisotropy is minimized at the
composition Tb0.27Dy0,73Fex for a given iron content.
This leads to specimens with high relative permeability.
On the other hand, as shown by Galloway et a1 1471,
in order to obtain a high dh/dH some anisotropy is
beneficial. The Tb :Dy ratio 0.3 :0.7 exhibits the largest
values of d33 and k33 for a given iron content.
The rare earth-to-iron ratio is critical in controlling
the amounts of second-phase material present in a
10
sample. The optimum composition is without doubt
the stoichiometric composition Mez. However, this is
difficult to achieve because of the more rapid loss of the
rare earth elements compared with iron during synthesis.
This means that the final composition of the material
is significantly different from the initial composition.
Too much rare earth gives rise to a rare-earth-rich
phase, which accumulates preferentially on the grain
boundaries. Too much iron leads to Wiedmanst2tten
precipitates of the RFe, phase within the grains. In
properties. The work of Joyce et a1 1131 has shown
that both the strain amplitude (at a given field strength)
and the derivative d33 are maximized at iron contents
close to the stoichiometric, although a curious double
peak was reported in both cases. Subsequent work by
Greenough et a1 [48] also showed that, among a group
of five specimens with x = 1.87 and eight specimens
with x = 1.95, the value of d33 was, with one exception,
always higher in the x = 1.95 group. This serves to
confirm the conclusion.
In order to further increase the strain amplitude along
the unique axis of a (112) oriented cylinder of the
material, techniques have been developed to enhance
the occupancy in the demagnetized state of the (1 11)
axes perpendicular to the cylinder axis. This has been
achieved by annealing the specimens at typically 950°C
in a magnetic field oriented along either the [ 11i] or
the [ i i l ] axes. This increased occupancy of the [ I l i ]
and/or [Til] axes leads to a greater change in strain when
the domain magnetizations are rotated into the [ 1111 and
later [I 121 axes under the action of an applied field.
The result of these processing improvements means
that (112) oriented specimens can be produced, which
exhibit strain amplitudes of up to 2200 x with
maximum d33 in the vicinity of 300 nA m-' and with
coupling coefficients of k33 = 0.7.
Acknowledgment
This work was supported by the US Department of
Energy, Office of Basic Energy Sciences, under wntract
W-7405-ENG-82. The work was performed at Ames
Laboratory.
References
[I] Rhyne 1 J and Legvold S 1965 Phys. Rev. A 138 507
[2] Clark A E, DeSavage B and Bozorth R 1965 Phys. Rev. A
138 216
131. Clark A E 1974 AIP Conf: P m . 18 (New York:
.
American Institute of Physics) p IO15
141 Abbundi R and Clark A E 1977 Trans. Maan. 13 1519
i$ Clark A E, Abbundi R and Gillmor W R 1578 IEEE
Trans. Magn. 14 542
[61 Clark G F, Tanner B K and Savage H T 1982 Phil. Mag.
B 46 331
[71 Lord D G, Savage H T and Rosemeier R G 1982 J. Magn.
Magn Mater. 29 137
[SI Parpia D Y. Tanner B K and Lord D G 1983 Nature 303
684

[91 Abell J S and Lord D G 1986 J. Less-Common Mer. 126
107
[IO] Lord D G, Ellion V, Clark A E, Savage H T, Teter J P
and McMasters 0 D 1988 IEEE Trans. Magn. 24 1716
[ I l l Teter J P, Mahoney K. Al-Jiboory M, Lord D G and
McMasters 0 D 1990 Proc. 35th Conf. on Magnerism
and Magnetic Mareriais, San Diego, CA, Ocrober 1990
1121 Sat0 K, lsikawa Y, Mori K, Clark A E and Callen E 1986
J. Magn. Magn. Mater. 54 875
[I31 Joyce V, Abell J S, Greenough R D and Pitman K C 1986
J. Magn. Magn. Mater. 54-7 877
[I41 Westwood P, Abell J S and Pitman K C 1988 IEEE Trans.
Mngn. 24 1873
[I51 Jenner A G I, Lord D 0 and Faunce C A 1988 IEEE
Trans. Magn. 24 1865
[I61 Savage H T, Abbundi R, Clark A E and McMasters 0 D
1979 J. AppI Phys. 50 1674
[171 McMasters 0 D,Verhoeven I D and Gibson E D 1986
J. Maen M a m Mafer. 54-7 849
[IS] Czochrlski 19f7 2.Phys, Chem. 92 219
[I91 Verhoeven J D,Gibson E D and McMasters 0 D 1987
Metall. Trans. A 18 223
[20] Teter J P, Clark A E and McMasters 0 D 1987 J. Appl.
Phys, 61 3787
.1211_Verhoeven J D.Gibson E D. McMasters 0 D and
Ostenson J E 1990 Merali. Trans. A 21 2249
127.1 Verhoeven J D, Ostenson J E, Gibson E D and
McMasters 0 D 1989 J. Appl. Phys. 66 112
[23] Clark A E 1980 Ferromagnetic Materials vol 1,
ed E P Wohlfarth (Amsterdam: North-Holland) p 531
[24] Clark A E and Savage H T 1983 J. Magn. Magn. Mater.
31 849
[25] Clark A E, Savage H T and Spano M L 1984 IEEE Trans.
Magn. 20 1443
[26] Jiles D C 1991 Introduction 10 Magnetism and Magnetic
Materinls (London: Chapman and Hall) pp 103-4
[27] Clark A E and Crowder D N 1985 IEEE Trans. Magn 21
1945
[ZS] Clark A E, Verhoeven J D, McMasters 0 D and
Gibson E D 1986 IEEE Trans. Magn. 22 973
[29] Clark A E,Teter J P and McMasters 0 D 1988 J. Appl.
Phys. 63 3910
Highly magnetostrictive rare earthiron alloys
1301 Jiles D C, Ostenson J E and Owen C V 1988 J. Appl.
Phys. 64 5417
[31] Jiles D C 1988 J. Appl. Phys. 63 3946
1321. Jiles D C, Ostenson I E and Owen C V 1989 J. Phvsiaue
. . .
49-68 1939
[33] Sablik M J, Jiles D C and Barghout L 1990 J. Appl. Phys.
67 5019
[34] Jiles D C and Hariharan S 1990 J. Appl. Phys. 67 5013
[35] Jiles D C 1990 Development and characterization of
the highly magnetostrictive alloy %.3Dyo.7Fel.9 for
use in sensors and actuators New Materials and Their
Applicarions ed D Holland (Bristol: Institute of Physics)
[36] Pitman K C. Branwood A and Piercy A R 1988 Proc.
2nd Int. Conf: on Giant Mangetostrictive Alloys and
Amorphous Alloys for Actua& and Sensors,-Marbella,
Spaii, October 1988
Pitman K C 1989 Admiralty Research Establishment,
Report ARE TM (UMT) 89403 (Holton Heath: ARE)
Atzmony U,Dariel M P, Bauminger E R, Lebenbaum D,
Nowik I and Ofer S 1973 Phys, Rev. B 7 4220
Jiles D C 1993 Introduction to rhe Electronic Properties of
Matenhls (London: Chapman and HaU) ch 15
Jenner A G, Greenough R D, Wilkinson A J and
Parvinmehr A 1990 IEEE Trans. Magn. 26 2589
Schulze M P. Greenoueh R D and Gallowav N 1992 IEEE
Trans. Main. 28 31<9
AbeU I S. Butler D. Greenoueh R D. Jovce V and
Pitman'K C 1986 J. Magn.-Magn.'M&r. 62 6
Savage H T,Clark A E and Spano M L 1984 IEEE Trans.
Magn. 20 1449
Greenough R D and Schulze M P 1990 IEEE Trans.
Magn. 26 2586
Greenough R D,Schulze M P, Jenner A G I and
Wilkinson A J 1991 IEEE Trans. Magn. 27 5346
Schulze M P,Chung R, Greenough I D, Greenough R D,
Jiles D C and Weber R 1992 IEEE Trans Magn. 28
3162
WI Galloway N, Schulze M P, Greenough R D and Jiles D C
1993 Appl. Phys. Lett. 63 1
[481 Greenough R D, Schulze M P , Jenner A G I and
Wilkinson A I 1991 IEEE Trans. Magn. 27 5346
11