Ballistic magnetoresistance of electrodeposited nanocontacts in thin film and micrometer wire gaps

Article abstract of DOI:10.1016/j.jmmm.2003.12.290

In this paper, we review the recent advances and progress in ballistic magnetoresistance (BMR) in magnetic nanocontacts electrodeposited in thin films and micrometer gaps. We report the influence of magnetostriction in the measurements under different configurations and substrates, as well as the contribution of the magnetic material forming the contacts. To avoid the magnetostriction effect, we have fabricated magnetic nanocontacts in Cu wires and Cu films. Similar BMR results can be observed in these systems. Our results show that the BMR effect should depend on the microproperties of the nanocontacts and should not be related with the macroproperties of the electrodes. The magnetostriction results, measured by an atomic force microscopy system with a built-in electromagnet, clearly show that there is no direct relationship between the displacement (caused by the magnetostriction effect) and the value of BMR. In fact, we present large magnetoresistance values for permalloy, coinciding with displacements in the latter's structure less than 1 nm, which is the smallest clearly observable shift allowed by our atomic force microscope. Repetitions of hundreds of R(H) curves are presented for different materials with different coercive fields. The interpretation of the results is based on the formation of an interfacial transparent layer (non-stoichiometric oxide, sulfur, etc.) at the nanocontact where the theory can explain large magnetoresistance values.

Full text of DOI:10.1016/j.jmmm.2003.12.290

ARTICLE IN PRESS
Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
Ballistic magnetoresistance of electrodeposited nanocontacts in
thin film and micrometer wire gaps
N. Garcia*, H. Cheng, H. Wang, N.D. Nikolic, C.A. Guerrero,
A.C. Papageorgopoulos
Laboratorio de Fisica de Sistemas Pequenos y Nanotechnologia CSIC, Serrano 144, E-28006 Madrid, Spain
Abstract
In this paper, we review the recent advances and progress in ballistic magnetoresistance (BMR) in magnetic
nanocontacts electrodeposited in thin films and micrometer gaps. We report the influence of magnetostriction in the
measurements under different configurations and substrates, as well as the contribution of the magnetic material
forming the contacts. To avoid the magnetostriction effect, we have fabricated magnetic nanocontacts in Cu wires and
Cu films. Similar BMR results can be observed in these systems. Our results showthat the BMR effect should depend
on the microproperties of the nanocontacts and should not be related with the macroproperties of the electrodes. The
magnetostriction results, measured by an atomic force microscopy system with a built-in electromagnet, clearly show
that there is no direct relationship between the displacement (caused by the magnetostriction effect) and the value of
BMR. In fact, we present large magnetoresistance values for permalloy, coinciding with displacements in the latter’s
structure less than 1 nm, which is the smallest clearly observable shift allowed by our atomic force microscope.
Repetitions of hundreds of RðHÞ curves are presented for different materials with different coercive fields. The
interpretation of the results is based on the formation of an interfacial transparent layer (non-stoichiometric oxide,
sulfur, etc.) at the nanocontact where the theory can explain large magnetoresistance values.
r 2004 Elsevier B.V. All rights reserved.
PACS: 68.37.Ps; 73.40.Àc; 75.60.Àd; 75.70.ÀI
Keywords: Electrochemical deposition; Ferromagnetic nanocontacts; Magnetoresistance; Magnetostriction
1. Introduction
that was called ballistic magnetoresistance (BMR) [3,4].
Recently, other experiments have shown that this
phenomenon can also be extended to magnetite [5] and
to CrO2. These studies have, furthermore, reported a
universal behavior [6], with the results explained using
the theory of Cabrera and Falicov [3,7].
Magnetoresistive effects are currently studied for
applications in newtechnologies after the discovery of
giant magnetoresistance [1,2], which has made a strong
impact on magnetic reading heads at present in the
market. At the limit of further integration (in terabits
per square inch) as well as in the nanometer range, much
stronger effects are needed. In experiments performed in
1999, we showed that it is possible to have magnetor-
esistive effects up to 300% in Ni, Co and Fe
nanocontacts of atomic size by the ballistic transport
In addition, experiments were performed for Ni
nanocontacts deposited in small gaps of Ni wires
forming a ‘‘T’’ geometry. These have shown magnetor-
esistive effects of 700% ([8], this paper shows normal
(negative) BMR low R for large fields). Recently, MR
values up to 3000%, 4000% and, in fact, effects reaching
practically to infinity were observed (We should mention
here that very large variation of resistance with magnetic
field up to practically infinity have been obtained with
Ni wires but this needs further understanding. However,
*Corresponding author. Tel.: +34-91-561-88-06; fax: +34-
91-563-15-60.
E-mail address: nicolas.garcia@fsp.csic.es (N. Garcia).
0304-8853/$ - see front matter r 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmmm.2003.12.290

ARTICLE IN PRESS
N. Garcia et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
1723
it is obtained inverse MR (see J.M. De Teresa, Science
286 (1999) 507) (large resistances at large fields). The
possibilities of magnetostriction effects have been
discussed in this contest, see Ref. [9]), although these
values of many thousands may be an illusion. One
important point, however, needs to be addressed
regarding the corresponding experiments of Ref. [8]
versus Ref. [9]. The former has shown negative BMR
while the latter inverse or positive MR, that is to say a
large resistance (R) at a large applied field. Moreover, as
the nanocontacts are grown on macroscopic Ni wires, it
is thinkable to assume that the magnetoelastic deforma-
tions may alter the nanocontact, and hence contribute to
the change of R by the applied field, and this needs to be
investigated. We have performed experiments measuring
MR and magnetostriction with an atomic force micro-
scope (AFM), and clearly observed that the existing
magnetostrictive changes cannot be assigned directly to
the cause of MR. Strictly speaking, however (although
hard evidence is lacking here), magnetoelastic effects
cannot be totally discarded when magnetic wires are
used as the support. In short, we have observed that
magnetostriction does not change resistance [10] in 90%
of the samples measured. The case where the magnetos-
triction may not be the cause of the generally observed
BMR was also explained in early experiments with Ni
nanocontacts deposited on Cu wires, which exhibited
magnetostriction deformations 100 times smaller than
Ni and from the very large values in permalloy [10] wires
that have 30 times smaller deformations than Ni. Lately,
during the last several months, we have been working on
Fe, Co and permalloy contacts electrodeposited in Cu
wires, but more importantly in Cu films as well. In the
latter case the magnetostrictive effect is unobservable
measured by our newAFM technique using a Dimen-
sion 3100 scanning probe microscope (SPM) with an
adapted magnetic field source [10], while at the same
time the BMR can reach values up to 400%. When we
use pure Cu wires or Cu plates as substrates, however,
with contacts of deposited permalloy we observe up to
300% MR, which is very reproducible in hundreds of
RðHÞ curves. In this case, the alleged deformations are
evidently belowthe minimum observable shift (1 nm) of
our AFM, with estimations for the deformation no
larger than 0.2 nm. These experiments in thin films may
be relevant from the technological point of viewbecause
they may permit the mass production and integration of
devices.
of Cu wire is 0.5 mm. We, furthermore, arrange them in
two different configurations (the ‘‘T’’ configuration and
the configuration with two Cu wire rings), and get a
nanocontact between two separate parts using electro-
chemical deposition. The details can be seen in our
previous publications. For the contacts grown on films,
the process for sample preparation is one of the
traditional methods used to get printed circuit boards.
It essentially entails drawing on the virgin board (20 mm
of copper deposited in an epoxy resin) with a waterproof
pen, and subsequently immersing the board in a solution
of chloridric acid and sodic perborate. This solution
basically attacks the exposed parts and leaves the
painted areas unaffected. After all the copper is removed
from the exposed parts, we take out the sample and
wash it in water. The process is completed by taking out
the paint with alcohol. We use insulating glue to cover
most of the surface of the Cu films, leaving only a small
area near the contact region for deposition. A schematic
diagram and the microscope picture are shown in
Fig. 1a. In the microscope picture, the yellow–red part
indicates the Cu film and the black part the substrate.
Before chemical deposition, we cleaned the sample by
ultrasonic immersion in alcohol and pure water. We also
controlled the thickness of Cu films by mechanical and
chemical methods, as shown in Fig. 1b, where the
bronze part indicates the Cu film and the thickness is
about 20 mm. We can then growFe, Co and permalloy
contact on Cu films with different thickness by etching
the Cu. The gap width where the magnetic material is
grown can, furthermore, be varied between 100 and
2 mm by depositing more Cu in a starting gap of 100 mm.
The magnetic nanocontact is, subsequently, grown
and its size is controlled by the contact resistance
(RðOÞ ¼ 12900=10a2; where a is the size of the contact in
nm. For R ¼ 1 O, aE30 nm). This estimation is done
assuming that each contact atom contributes with a
conducting channel assuming ballistic conduction for
the spin. In other words, the distance that the spin
travels without reversing has to be larger than a: Taking
into account that the spin reversing distances are of the
order of 30–40 nm, it is reasonable to assume ballistic
conductance for the spin in contact with resistances
larger than 1 O. We have attempted to visualize the
nanocontact by scanning electron microscope (SEM)
techniques, despite the difficulty in localizing the exact
position of the contact. Despite the latter difficulties,
however, we show in Fig. 1c an SEM micrograph of the
contact area in the deposit, even though this is taking
place through a particle of E1 mm size. The nanojunc-
tion cannot be identified here since it may be covered by
the particle. Similarly, the possibility of many nanojunc-
tions formed at the same time also cannot be excluded in
this scenario. This will be discussed below.
2. Ballistic magnetoresistance
In the present study, we have studied Fe, Co and
permalloy nanocontacts on Cu wires and Cu films. For
Cu wires, the preparation method is similar to that of
our previous work on Ni and NiFe wires. The diameter
Iron was deposited from 225 g/l FeSO4 ꢀ 7H2O in
120 g/l Na2SO4 ꢀ 10H2O, with a pH of 3, at a voltage

ARTICLE IN PRESS
N. Garcia et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
1724
current–field configurations. The resistance measuring
the current was, moreover, in the range of 0.1–1 mA,
and the maximum applied magnetic field was 4000 Oe.
To get rid of the magnetostriction effect from the
macroelectrodes (such Fe, Ni or NiFe wires etc.), all
these experiments were carried out on pure Cu wires and
Cu films. In the following graphs, we show the current–
field relation in each case individually. All experiments
are performed at room temperature (RT). We observed
the sample with an optical microscope (Olympus BX 50)
and an SEM (Philips FEG XL-30).
The work on the Cu macrowires is similar to our
previous work on Ni wires. The BMR curves shown in
Figs. 2a and b are observed in a continuous measure-
ment on the same sample where we can see a 400%
magnetoresistance response in a 100 Oe field. What is of
greater significance is that all the loops during measure-
ments are very similar, with the only difference between
them being a small change in the BMR value. In
addition, the interval between the initial loop and the
seventh loop is about several hours. This indicates that
the BMR effect is stable for at least that amount of time.
For the Cu wires the magnetostriction is unobservable
with our AFM, since the magnetoelastic change in the
length is less than 1 nm [10]. For deposited Fe, the
volume of the latter is in the micrometer range. In Fig.
2d, we show the case presented in Fig. 2b for small fields.
Here, the resistance steps in the lowfield that
correspond to the sudden depinning of spins can be
seen clearly. We consider that these steps may be related
with the interaction between the spin and local domain
wall or the spin reorientation in the nanocontact under
the magnetic field. We also measured the BMR in a new
configuration; see the inset of Fig. 2c. In this configura-
tion, despite the fact that there is no macro ‘‘tip’’ (as in
‘‘T’’ configuration), there is also a large BMR effect.
That means the ‘‘tip’’ has no contribution to the latter
BMR effect. The most important thing should be the
existence of a nanocontact between the two separated
electrodes that do not necessarily need to be magnetic.
Fig. 1. Cu film with ‘‘T’’ configuration. (a) A schematic
diagram of Cu film sample and the microscopy picture near
contact area. (b) The cross-section microscopy pictures of the
Cu films. (c) SEM micrograph of the contact area in film. A
nanojunction was formed by a magnetic microcluster during the
electrodeposition process.
3. Experiments with Cu films as substrates
of À4 V. Cobalt was deposited from 0.25 M CoSO4 in
0.50 M Na2SO4, where pH=3.5, at a voltage of À3 V.
Finally, permalloy was electrodeposited from: (a)
NiSO4 ꢀ 6H2O (200 g/l), (b) NiCl2 ꢀ 6H2O (5 g/l), (c)
H3BO3 (25 g/l), (d) FeSO4 ꢀ 7H2O (8 g/l), and (e) s-
benzoic acid sulfimide (3 g/l). For permalloy deposition
the pH was 3.5, while in all the above cases the anode
was a pure platinum wire. The electrodeposition time
was, furthermore, the shortest time period required to
establish contact between two copper parts. Electro-
deposition of Fe and Co was performed utilizing a
Hewlett Packard E 3612 DC power supply, while the
magnetoresistance measurements were done in different
We have extended our experiment from macrowires to
films. The experimental data shown in Figs. 3a–c
demonstrate for the first time BMR in films. It is clear
that the large and symmetric BMR curves can also be
observed in films. These experimental results provide
strong evidence that, as we have demonstrated before,
both magnetostriction and the movement of the macro-
wires could not produce the BMR effect. In a thin film
system, where the Cu film is totally fixed tightly by the
substrate and where we do not see any magnetostriction
(deformations are smaller than 0.5 nm, i.e. less than the
sensitivity of our AFM [10]). Here, the magnetoelastic

ARTICLE IN PRESS
N. Garcia et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
1725
60
35
H
The initio loop
The 7th loop
Cu
wires
50
40
30
20
30
25
20
15
10
-600
-400
-200
0
200
400
600
-600 -400
-200
0
200
400
600
(a)
(b)
H(Oe)
H(Oe)
6.0
5.5
5.0
4.5
60
B
50
40
30
20
D
4.0
3.5
3.0
2.5
C
A
2.0
-200 -160 -120 -80 -40
0
40 80 120 160 200
-1000
-500
0
500
1000
(c)
(d)
H(Oe)
H(Oe)
Fig. 2. (a) and (b) BMR curves showing values up to about 400% in Fe nanocontact in Cu–Cu wires system. Magnetic field is
perpendicular to the plane of the sample. A schematic picture of relationship between the sample and the applied magnetic field is
shown in the up-left corner of (a). (a) Shows the initial loop and (b) shows the seventh loop in a continuous measurement on the same
sample. (c) BMR curves measured in the two Cu wire rings system. The maximum BMR value is about 150%. A schematic picture is
shown in the up-left corner of (c). To show the big BMR effect more clearly in (d), we show the BMR curve in (b) in small magnetic
field (200 Oe). The loop sequence follows as: A–B–C–D. One can clearly see the spin rotation process in detail. All experiments were
carried out in situ, in the electrolytic solution.
deformations are simply reduced to the response of the
magnetic material deposited to conform to the contact.
The amount of this material ranges between 50 to 10 mm
of deposit to close the gap in the Cu film. The
deformation estimations for Ni and permalloy range
between 0.6 to 0.2 nm for Ni and 0.06 to 0.02 nm for the
permalloy, and cannot be observed with the AFM
measuring magnetostriction. It is highly unlikely,
furthermore, that these deformations could change the
resistance of a 5 O contact that has a section size of
25 nm to 15 O with a contact size of 15 nm.
It can be seen clearly that the biggest BMR value
presented here, under a field less than 100 Oe, is nearly
200%, which exceeds any other MR effects in the thin
film system at lowfield values. Inset 1 in Fig. 3b
evidently shows the sensitive resistance response under
the magnetic field, where the resistance dramatically
changes from 3 to 8 O within 200 Oe. Another important
fact is that we measure the sample in air (not in the Fe or
Co solution). This is good for mass fabrication, since the
effect of the solution on the BMR during measurements
can be removed. Moreover, we have measured the
sample in different current–field configurations, as
shown in Figs. 3a and b. BMR curves have a good
magnetic field response in both configurations. In
addition, we show one of the initial experiments on Co
in Fig. 3c. From this figure, one can see that the Co
nanocontact can also give a good BMR response. We
have also observed BMR with Ni deposited in the same
Cu films.
The effect is observed in 60% out of over hundred of
samples that have been prepared, and is observed for
different geometries in both the Cu wires and the films.
The samples, furthermore, lasted during measurements
from several hours to one day, and their resistance
changed with time in such a way that the BMR can be
larger or smaller as time goes by. The values we
observed cannot be easily explained by domain wall
scattering even if the total spin is conserved because the
density of states at the Fermi level for spin up and down
cannot provide such big BMR values. The explanation
has to be found in the formation of some dead layer or
compound that may contain O or other elements
introduced during the electrodeposition process or in

ARTICLE IN PRESS
N. Garcia et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
1726
8
7
6
5
4
3
7
6
5
4
3
2
Fe
Fe
H
H
8
7
6
5
4
3
Cu Films
-300 -150
0
150 300
-1000 -500
0
500 1000
-1000 -500
0
500 1000
(a)
H(Oe)
(b)
H(Oe)
Co
7
6
5
4
H
-1000 -500
0
500 1000
(c)
H(Oe)
Fig. 3. (a) and (b) BMR curves showing values up to nearly 300% in Fe nanocontact in Cu–Cu thin films system. A schematic picture
of relationship between sample and applied magnetic field is shown in the up-right corner of (a) and (b). The inset one of (b) is the curve
in lowfield (400 Oe). (c) BMR curves measured in Co nanocontact in Cu–Cu thin films system. A schematic picture is shown in the up-
right corner of (c). All these experiments were carried out in air.
the oxidation of the contact over time that modified the
density of states or polarized the oxygen [8,9,12]. We
observe that the contact resistance changes with time as
well as its BMR so that the effect may not be due to pure
Fe, Co or Ni effects, and some compound may modify
the transport through the contact. Another thing we
found is that with time the field reversing spin also
changed, while we observed in some cases that a 10 Oe
field has a strong MR effect (10% per Oe that is very
important for application in devices).
moved easily by the applied field, and Fig. 4b is the case
of a permalloy nanocontact that shows a very compact
structure connected by a 1 mm size particle. In the upper
right corner of Figs 4a and b are the corresponding SEM
scans of the aforementioned contacts. It is worth noting
that in the latter case no motion whatsoever is detected
from the contact when measured by the AFM under an
applied magnetic filed. We measured the sample several
times, each measurement corresponding to one hundred
loops, with the time interval for each measurement being
half an hour. For purposes of clarity, we only show the
first 10 loops of each measurement. As is shown, these
nanocontacts exhibit a very stable behavior. The loops
(the BMR value and the shape of curves) measured in
different times are reproducible, and we have success-
fully measured many samples in this way. Some samples
can maintain reproducible values for several days,
although stability during months or several weeks has
not been obtained yet. Nevertheless, the stability and
reproducibility of the curves is striking. Furthermore,
although the curves correspond to contacts with
completely different morphologies, they still have the
same reproducibility. Another important point is that
the change in R takes place in a coercive field that for Fe
4. Reproducibilityand stabilityof the RðHÞ loops
For applications and basic physics, there is another
important issue regarding the BMR effect: the stability
and reproducibility of BMR-RðHÞ loops. We have
performed some experiments and discussed them before.
At this point, however, we would like to present the
latest results in our experiments. We have found that the
nanocontacts deposited on Cu films are very stable and
can keep a constant BMR value for several hours at
least. Fig. 4a is the case of an Fe nanocontact that, as
shown in the figure, has a dendritic structure that may be

ARTICLE IN PRESS
N. Garcia et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
1727
The metal wires used in the magnetostriction measure-
ments were Ni, permalloy Fe21Ni79, commercial Cu as
well as pure Cu. Whereas for most cases the deposited
material was the same as that of the wire; there was one
important exception: permalloy deposited on pure Cu
wire in a configuration where the Cu wires face each
other tip to tip. We will elaborate on this application
further in this report. In essence, measurements involve
placing the sample between the poles of the electro-
magnet inserted under the scanner so that scanning
occurs in the same direction as the expected magnetoe-
lastic change. When the field is applied the wire expands
or contracts (depending on the material and field
direction relative to the length of the wire), and the
topography undergoes a near instant shift. This easily
observable shift can be measured directly from the scan
to at least 1 nm. Detailed results are described elsewhere
[10], while an illustration of the technique is shown in
the figure described below.
Iron nanocontact on Cu film
Uptriangle: 1-10 loops
Square: 100-110 loops
Circle: 200-210 loops
12
11
10
9
8
7
6
5
-1000
-500
0
500
1000
(a)
H(Oe)
Permalloy nanocontact on Cu film
Square: 1-10 loops
18
16
14
12
10
Uptriangle: 100-110 loops
The reliability of our measurement method was first
tested using free magnetic wires (as well as non-magnetic
ones for reference) under an applied magnetic field with
excellent agreement with other methods recorded in the
literature [13,14]. We then went on to measure the
magnetostriction of Ni and permalloy wires in the ‘‘T’’
configuration used in most of the BMR measurements
(many of which were actually measured under the
AFM). Fig 5a depicts a schematic diagram of the wire
configuration, including the tip position and clamping
used for all ‘‘T’’-configured contacts and particularly for
the Ni wire, described below. Fig 5b shows an optical
microscope image (using the optical microscope built
into the SPM) of the ‘‘T’’ contact area of the Ni wires in
the 600 mm magnification range and includes the scan
direction and actual tip position during measurement
(scans were performed around 100–200 mm from the
contact itself). The contact area (of deposited Ni) was
measured at 30 mm, while the resistance was monitored
over the contact throughout the measurements. The
value of this resistance, in fact, represents the effective
conductive portion of the contact. From previous
studies we have found it to correspond to a conducting
contact of 30 nm [9], in obvious contrast with the total
contact geometry. Most of the total contact area is
actually comprised of non-conductive oxides.
100
-300
-200
-100
0
300
200
(b)
H(Oe)
Fig. 4. (a) and (b) showthe BMR curves of iron and permalloy
contacts on Cu film, respectively. In the case of iron, up-triangle
is 1–10 loops, square is 100–110 loops, circle is 200–210 loops.
In the case of permalloy, we measured the sample two times,
each time one hundred loops. We showhere the first ten of each
time: square is 1–10 loops and up-triangle is 100–110 loops. The
inserts at the top right of both curves depict SEM scans of the
contacts in mention with the scale shown on the corresponding
photos.
is around 250 Oe while for permalloy this value is
around 25 Oe, which is understandable due to the fact
that this latter material is softer. Thus, we have obtained
very repetitive loops showing 100 changes in R at 25 Oe.
Fig 5c shows the scanned surface near the contact
when no field is applied, while Fig 5d illustrates the shifts
occurring during repetitive field applications of 80 Oe
and also the axis directions relative to the motion of the
wire, which evidently shifts in the negative direction. It
is, furthermore, significant that the topographic pattern
returns to its original position when the magnetic field is
turned off. For an applied field of 250 Oe the shift is
about half the value of that observed with a single wire
(À16 nm/mm) when the field is parallel to the wire’s
length. The actual measured shift was around 100 nm for
5. Magnetostriction
Magnetostriction measurements were conducted with
a Dimension 3100 SPM and a Nanoscope IV Controller.
The magnetic field was applied to the sample plane with
a laboratory-designed and -constructed electromagnet
with a 12 mm gap. Field strengths for most measure-
ments were measured with a DTM-133 Digital Tesla-
meter at the sample plane, while the DC power supply
current could vary the resulting field to within o2 Oe.

ARTICLE IN PRESS
N. Garcia et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
1728
sample were of the same order as permalloy wires under
respective parallel and transversal fields.
In contrast, Cu wires of 99.999% purity exhibited no
observable magnetostriction. This is important because
the next application was conducted on pure Cu wires
with deposited permalloy forming the contact. The
importance of this application, as mentioned earlier in
this report, is that this particular sample exhibited a
100% BMR response. The application is promptly
illustrated in Fig. 6. As can be seen at the top portion
of Fig. 6a, in this configuration the wire tips were
aligned so that the two separate wires ran in the same
direction, while the permalloy contact crossed a 20 mm
gap between the Cu wires, as shown in the optical
microscope picture of the contact area in Fig. 6a. When
placed under the AFM and with a near instantaneously
applied field (up to 500 Oe in this case), no movement
was observed. The resulting scan (450 nm range with
multiple field applications) is shown in Fig. 6b in a high
contrast color scale. We have mentioned that the AFM
cannot reliably showmagnetoelastic changes in sample
length less than 1 nm. When, however, there is a even
shift that is too small to be measured the scan is
disrupted along at least one scan line. We can see these
shift lines clearly in the case of the Ni contact under an
applied magnetic field, shown in Fig. 5d. This sharp line
is absent in the scan of Fig. 6b, and the image itself does
not differ at all from one where no field is applied. We
have similar images when applying fields to non-
magnetic samples such as pure Cu and Pt wires [10]. It
is not surprising that no shift was observed here, given
that pure Cu has no magnetoelastic shift and that the
permalloy thickness at the contact was 20 mm. This
should exhibit, according to our measurements, 0.1 nm
displacements much lower than that we could measure.
Fig. 5. (a) Schematic diagram of the ‘‘T’’-contact wire
configuration with glued and free portions of the wires shown.
(b) An AFM video camera image (600 mm range), of a ‘‘T’’-
configured Ni contact, with indicated scan position and
direction of scan during field applications. (c) and (d) Two
1 mm scans of the ‘‘T’’-contact configuration of the Ni wire.
Scan (c) shows a portion of the Ni surface without an applied
magnetic field, and (d) depicts the same surface topography
with the magnetic field (H ¼ 80 Oe), applied and removed five
times during the scan (scan direction is from top to bottom).
Axis orientation of the wire and the direction of the
magnetoelastic shift are shown in white.
the 6 mm wire used in this case. Although one would
expect that a total contraction of 100 nm would at least
change the contact resistance, if not break the contact,
the monitored resistance remained stable during multi-
ple consecutive field applications. In other words,
magnetoelastic deformation does not alter the resistance
across the contact and more importantly does not
automatically imply magnetoresistance.
For permalloy Ni79Fe21 ‘‘T’’-configured contacts the
fractional change in the length of the wire whose tip
makes the contact was found to be the same value as for
the free wire (Dl=l ¼ À2 nm/mm). The permalloy con-
tact also remained intact with a stable resistance during
deformation, which is not surprising given the results
from Ni, and the total movement of 20 nm exhibited by
the 10 mm permalloy wire. With commercial Cu single
wires, on the other hand, motion was observed when a
field was applied. Our commercial Cu wire was analyzed
by EDAX where the presence of 3% Co, Ni and Fe
impurities was detected. It is evident from this analysis
that even minute amounts of impurities cause appreci-
able changes in the values of magnetoelastic deformation
in a paramagnetic material such as Cu. In particular, the
changes in length measured in the commercial Cu
Fig. 6. Magnetostriction measurements on pure Cu wires with
permalloy Ni79Fe21 deposited at the gap (20 mm). (a) The top
portion shows a schematic of the wire configuration and below
is an optical microscope picture in the 500 mm range of the
contact itself where scanning took place. The wire lengths are
from top to bottom on the picture. (b) Shows a 450 nm scan of
the permalloy contact under repetitive applications of
a
magnetic field (500 Oe) during scanning. This field causes no
effect on the image, indicating no magnetostriction.

ARTICLE IN PRESS
N. Garcia et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1722–1729
1729
Although not measurable in itself, however, if there was Acknowledgements
a shift, its presence would have been detected as a sharp
line across the scan.
This work has been supported by the Spanish
DGICyT.
We should mention that the important point is not to
obtain monstrously large BMR effects. What is needed
is to have reproducibility and large changes at small
fields. The work presented here does obtain reproduci-
bility up to 60% with BMR ranging between 10% and
400%, with large BMR at low fields and for films that
may allowmass fabrication and miniaturization. How-
ever, the duration of our samples is still not enough for
application at present stage. One point of our future
work is to explore the reason of this instability and find
the optimal parameters for BMR. Also the BMR effect
shown in films is in clear agreement with much earlier
whiskers experiment [10]. One thing that should be
stressed is that, in the Cu films, we never observe BMR
values of thousands percent as we do with Ni and
permalloy wires where magnetostatic deformations exist
[8,9]. But neither can we say that the deformations give
automatically large variation of resistance, the thou-
sands percent of resistance variations only happen in 2%
of the samples we have measured.
References
[1] M.N. Baibich, J.M. Broto, A. Fert, F. Nguyen Van Dau,
F. Petroff, P. Etienne, G. Creuzet, J. Chazelas, Phys. Rev.
Lett. 61 (1988) 2472.
[2] G. Binasch, P. Gru.nberg, F. Saurenbach, W. Zinn, Phys.
Rev. B 39 (1989) 4828.
!
[3] N. Garcıa, M. Munoz, Y.-W. Zhao, Phys. Rev. Lett. 82
(1999) 2923;
G. Tatara, Y.-W. Zhao, M. Munoz, N. Garcıa, Phys. Rev.
Lett. 83 (1999) 2030.
!
!
[4] N. Garcıa, M. Munoz, Y.-W. Zhao, Appl. Phys. Lett. 76
(2000) 2586.
[5] J.J. Verluijs, A.M. Bari, J.M.D. Coey, Phys. Rev. Lett. 87
(2001) 026601.
!
[6] S.H. Chung, M. Munoz, N. Garcıa, W.F. Egelhoff, R.D.
Gomez, Phys. Rev. Lett. 89 (2002) 287203.
[7] G.G. Cabrera, L.M. Falicov, Phys. Stat Sol (B) 61 (1974)
539.
Also, at this point, we should report that we have
measured hundreds of samples [4,8,11] on contacts
deposited in evaporated thin films (say up to 100 nm
thickness) and never see BMR effects except in one case.
In addition to contacts (with resistances smaller than
500 O), there are unexpected large BMR values with
pure metals [3,7]. We believe that the large MR in
smaller resistance contacts is provided by the formation
of a very thin non-stoichiometric oxide, sulfate and/or
other compound during the electrodeposition process
[8,9], where recent full calculations support this idea [12].
This may be the reason why we cannot obtain any MR
when the contacts are made by evaporation of thin films.
!
[8] N. Garcıa, M. Munoz, G.G. Qian, H. Rohrer, I.G.
Saveliev, Y.-W. Zhao, Appl. Phys. Lett. 79 (2001) 4550.
[9] H.D. Chopra, S.Z. Hua, Phys. Rev. B 66 (2002) 0204403;
!
Hai Wang, Hao Cheng and N. Garcıa arXiv:cond-mat/
0207516v1.
[10] A.C. Papageorgopoulos, H. Wang, C.A. Guerrero, N.
Garcia, J. Magn. Magn. Mater. 268 (2004) 198.
[11] N. Garcia, G.G. Qian, I.G. Saveliev, Appl. Phys. Lett. 80
(2002) 1785.
[12] N. Papanikolaou, arXiv:cond-mat/0210551v1.
[13] A.H. Morrish, The Physical Properties of Magnetism,
Wiley, NewYork, NY, 1965.
[14] R.M. Bozorth, Ferromagnetism, D. Van Nostrand Co.
Inc., NewYork, NY, 1951.