Magnetostatic Interaction in Nanowires
MASSA( CHUSETTS INST-fMTE
0 F TECHNOLOGY
by
Saima Afroz Siddiqui
EP 2 5 2014
I
B.S., Electrical and Electronic Engineering
Bangladesh University of Engineering and Technology (2011)
LI
Submitted to the Department of Electrical Engineering and Computer
Science
in partial fulfillment of the requirements for the degree of
Master of Science in Electrical Engineering and Computer Science
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
September 2014
@ Massachusetts Institute of Technology 2014. All rights reserved.
Signature redacted
A uth or ..............................................................
Department of Electrical Engineering and Computer Science
August 29, 2014
Signature redacted
Certified by..
...........
Marc A. Baldo
Professor of Electrical Engineering
Thesis Supervisor
Signature redacted
......
/ LeleOA. Kolodziejski
Chairman, Department Committee on Graduate Students
Accepted by..............
BRARIES
2
Magnetostatic Interaction in Nanowires
by
Saima Afroz Siddiqui
Submitted to the Department of Electrical Engineering and Computer Science
on August 29, 2014, in partial fulfillment of the
requirements for the degree of
Master of Science in Electrical Engineering and Computer Science
Abstract
Nonvolatile memory and logic devices rely on the manipulation of domain walls in
magnetic nanowires, and scaling of these devices requires an understanding of domain
wall behavior as a function of the wire width. Due to the increased importance of
edge roughness and microstructure in narrow lines, domain wall pinning increases
dramatically as the wire dimensions decrease and stochastic behavior is expected
depending on the distribution of pinning sites. This work reports on the field driven
domain wall statistics in sub-100 nm wide nanowires made from Co films of 8 nm
thickness made by an electron beam lithography and etching process that minimizes
edge roughness.
Thesis Supervisor: Marc Baldo
Title: Professor of Electrical Engineering
3
4
Acknowledgments
First and foremost I'd like to thank my research advisor, Marc, for his interest and
encouragement. I'd have been unable to complete this work without his help. I'd
also like to thank my co-advisor Caroline for her ideas and suggestions for the experiments. Jean Anne Currivan has been very helpful with her feedback and saved me a
lot of time by providing necessary and much-appreciated assistance with experimental design. Sumit Dutta has helped me with the simulation. Scott Speakman has
provided his thoughtful insight for characterizing thin films with X-ray diffraction.
I'd like to acknowledge the rest of The Soft Semiconductor Group that I've worked
alongside of: Nicholas Thompson, Phil Reusswig, Matthias Bahlke, Brian Modtland,
Tony Wu, Dan Congreve, Markus Einzinger and Paul Azunre for assistance and helpful discussions along the way.
I can't thank Ahmad enough for being a wonderful partner and also for his comprehensive attention to detail of my work. Outside of those that assisted with theory
and experimental work, I'd like to thank my mother and father for their support and
for helping me realize that graduate school in engineering was right for me.
5
6
Contents
1 Introduction
15
2
Thin Film Deposition
21
2.1
In plane Anisotropy Film ..............................
23
2.2
Perpendicular Anisotropy Film .....
3
.....................
Thin Film Deposition
29
3.1
Lithography Processes
3.2
Line Edge Roughness ................................
38
3.3
M agnetic Properties
41
..........................
. . . . . . . . . . . . . . . . . . . . . . . . . . .
4 DW Motion
5
23
29
45
4.1
High Anisotropy Film . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
4.2
Magnetic Field induced Stochastic Domain wall Motion ........
46
4.3
Modeling of Domain Wall Stochastic Nature ..............
53
Conclusion
59
7
8
List of Figures
1-1
Magnetoresistance resulting from GMR effect [1].
1-2
Cartoon structure of (a) spin valve and (b) magnetic tunnel junstion [2]. 16
1-3
FIB image of a magnetic nanowire network containing one NOT gate,
. . . . . . . . . . .
16
one AND gate, two fan-out junctions, and one cross-over junction [3].
18
1-4
STT/DW logic device [4].
. . . . . . . . . . . . . . . . . . . . . . . .
19
1-5
Cartoon of vertical racetrack memory device [5]. . . . . . . . . . . . .
19
2-1
In-plane hysteresis loop of CoFeB thin film . . . . . . . . . . . . . . .
23
2-2
Grazing Incidence Angle Measurement for MgO deposited on Si substrate with native oxide. . . . . . . . . . . . . . . . . . . . . . . . . .
2-3
Grazing Incidence Angle Measurement for MgO deposited on Si substrate with native oxide with long integration time.
. . . . . . . . . .
2-4
Out-of-plane hysteresis loop of Ta(5)/CoFeB(1)/MgO(1.2)/Ta(2).
2-5
(a) In-plane and (b) out-of-plane hysteresis loop of different thickness
. .
Co60 Fe2 0 B2 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3-1
25
25
26
27
(a) Single layer and (b) bilayer lift-off patterning process and resulting
line edge roughness in the nanowire. The film used in the patterning
process is sputter deposited Ta (5 nm)
in-plaqne Anisotropy film. [6]
/
NiFe (10 nm)
/
Au (5 nm)
. . . . . . . . . . . . . . . . . . . . . .
30
3-2
SEM image of a patterned arc with single layer PMMA lift-off process.
30
3-3
SEM image of a 9 nm pitch nested-L structures patterned in 10 nm
thick HSQ [7]. ..
.. ....
. .. .. . . . . . . . . . . . . . . . . . ..
9
31
3-4
(a) In-plane Anisotropy and (b) perpendicular anisotropy film for patterning.
3-5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
Bilayer resist processing steps - (a) spin PMMA and HSQ on top of the
film stack, (b) E-beam to expose the HSQ and then development of
it, (c) Oxygen plasma etch to remove PMMA all over the film except
under the HSQ (d) Ion-milling to pattern the film using the resist stack
as etch mask, (e) Removal of the resist on top of the patterns using
hot NMP and sonication.
3-6
. . . . . . . . . . . . . . . . . . . . . . . .
33
: SEM image of the nanowires (a) after oxygen RIE, (b) after ionmilling, and (c) after resist removal for 50 nm, 100 nm and 150 nm
wide wires.
3-7
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
VSM of the as deposited film and the films after oxygen RIE for both
(a) in-plane and (b) perpendicular anisotropy films. . . . . . . . . . .
3-8
33
34
Cross-sectional SEM images of bilayer resist stack after 1 minute 02
reactive ion etch (RIE, Step 4), for HSQ widths a) 468 nm, b) 261 nm,
c) 95 nm, and d) 61 nm. This RIE time is not long enough for the
wider top two HSQ masks. . . . . . . . . . . . . . . . . . . . . . . . .
3-9
35
Helium-ion microscope images of CoFeB nanowires of average width
(a) 64.9 nm, b) 52 nm, c) 39 nm, and d) 27 nm patterned with bilayer
resist processes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36
3-10 Top-down SEM image of the different width and pitch nanowire pat-
terned with 125 keV. . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
3-11 (a) Top-down SEM Images of 50 nm wide CoFeB nanowire, (b) The
two edges of the nanowire in (a) and (c) Gaussian distribution of the
edge variation along the length of the nanowire in (a). . . . . . . . . .
39
3-12 3a and the RMS value of line edge roughness measured for the nanowires
patterned with (a) 30 keV and (b) 125 keV accelerating voltages.
. .
40
3-13 Atomic force microscopy and the magnetic force microscopy images of
(a) spin-ice system, (b) nanorings and (c) oval-shaped structure. . .
42
3-14 AMF and MFM of 50 nm wide wires with different pithes. . . . . . .
43
10
4-1
(a) High anisotropy Co in-plane anisotropy film stack and (b) the hysteresis loop of the deposited film along the in-plane direction (red) and
perpendicular to the surface direction (black). . . . . . . . . . . . . .
46
4-2 MFM image of 200 nm wide and 8 pm diameter circular rings after
applying 220 Oe along the y-direction with scanning direction (a) from
top to bottom, and (b) from bottom to top and (c) after applying 250
Oe along the y-direction with scanning direction from bottom to top.
4-3 SEM image of the concentric rings . . . . . . . . . . .. . . . . . . . .
47
48
4-4 AFM Image of 60 nm wide fifteen concentric rings, and MFM image
of (b) the domain walls in the rings after magnetized the rings along
the x-direction; zoom in view of (c) the tail to tail domain wall and
(d) the head to head domain wall. . . . . . . . . . . . . . . . . . . . .
49
4-5 MFM image of the 30 nm wide eleven concentric rings after applying
(a) 220 Oe, (b) 250 Oe, (c) 300 Oe, (d) 350 Oe and (e) 400 Oe along
the y-direction. The field of view in each figure is 12x 12 pm 2.
. . . ..
50
4-6 MFM image of the 50 nm wide twelve concentric rings after applying
(a) 250 Oe, (b) 350 Oe, (c) 450 Oe, and (d) 500 Oe along the ydirection. The field of view in each figure is 12 x 12 pm2 . . . . . . . .
51
4-7 MFM image of the 30 nm wide fifteen concentric rings after applying
(a) 220 Oe, (b) 250 Oe, (c) 300 Oe, (d) 350 Oe and (e) 400 Oe along
the y-direction. The field of view in each figure is 12x 12 pm2 .
. . . ..
4-8 Stray field distribution from various width nanowire at distances.
.
52
53
4-9 Histogram plot of the distances travelled by tail to tail DWs in 60 nm
wide concentric rings after applying (a) 220 Oe, (b) 250 Oe, (c) 300
Oe, (d) 350 Oe and (e) 400 Oe along the y-direction.
. . . . . . . . .
54
4-10 The distribution of DW position in the nanowires and their distribution follows Poisson's model for (a) non-interacting and (b) interacting
nanowires. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55
4-11 The mean distance of the DW travels with the applied field in different
width nanowires with different spacing. . . . . . . . . . . . . . . . . .
11
56
12
List of Tables
2.1
The list of peaks for MgO thin film.. . . . . . . . . . . . . . . . . . .
13
24
14
Chapter 1
Introduction
Spintronics has emerged as a new technology where the electron spin carries information instead of electron charge. This offers opportunities for a new generation of
devices combining standard microelectronics with spin-dependent effects which arises
from the interaction between the spin of the carrier and the magnetic properties of
the material.
Traditional approaches to using spins are based on the alignment of
a spin with an applied magnetic field or any reference magnetization orientation in
the magnetic film. Device operation needs some alignment of the spin with electric
current. This new degree of freedom allows many advantages over the conventional
devices such as
nonvolatility, increased data processing speed, decreased electron
power consumption and increased integration densities [2].
The discovery in 1988 of the giant magnetoresistive effect (GMR) is considered
the beginning of the new spin-based electronics. [1, 8]. GMR is observed in artificial
thin film materials composed of alternate ferromagnetic and nonmagnetic layers. The
resistance of the material is lowest when the magnetic moment in the ferromagnetic
layers are aligned in the same direction and highest when they are aligned in opposite
directions.
Figure
1-1 shows the magnetoresistance of a [(Fe 30 A)/(Cr 9 A)]40
superlattice of 4.2 K. The current is along [110] and the field is in the layer plane
along the current direction (curve a), in the layer plane perpendicular to the current
(curve b), or perpendicular to the layer plane (curve c).
15
..........
R/R(H=O)
I
(Fe 30A/Cr9A),
0.0
c
.
0.6
-40
-20
0
20
40
Magnetic field (NG)
Figure 1-1: Magnetoresistance resulting from GMR effect [1].
Andi F*rr=
ARR -20%%-0%
aturation field
10-300*
10-M 00
i
i
6
(b)
(a)
Figure 1-2: Cartoon structure of (a) spin valve and (b) magnetic tunnel junstion [2].
A spin valve is a GMR based device, which has a thin nonmagnetic metal sandwiched between two ferromagnetic layers. One of the magnetic layers is pinned so
that the magnetization in the layer is relatively insensitive to moderate magnetic
fields [9]. Figure 1-2(a) shows a cartoon structure of spin valve.
A magnetic tunnel junction (MTJ) is similar to spin-valve only a very thin insulating layer replaces the nonmagnetic metal [10, 11]. The tunneling resistance is modulated in the same way as spin-valve. Figure 1-2(b) shows a cartoon structure of MTJ.
MTJ forms the building block of magnetic random access memory (MRAM), which is
a nonvolatile, high-density and high-speed technology. MRAM is manufactured using
a complementary metal-oxide semiconductor (CMOS) compatible process and therefore indirectly affect microelectronic logic by allowing large amounts of high-speed,
16
high-density, nonvolatile memory to be embedded with the semiconductor processor.
The emerging field of magnetic logic tries to redesign the principle of operation of
microelectronic logic at the lowest level to make use of ferromagnetism.
A number of MTJ-based magnetic logic scheme have been proposed. In one scheme
[12, 13, 14], information enters into the logic gate by the currents in multiple bit
lines. The free layer of the MTJ will rotate into the direction of the net magnetic
field from the combined currents, effectively acting as a nonlinear summing element.
This in turn changes the resistance of the junction, which can be used to control the
current in subsequent bit lines. This scheme has a number of advantages, including
that the devices are based on existing MTJ technology and that the logic function can
be programmed by changing the magnetization direction of the reference layer in the
MTJ. Even more appealing is reconfigurability of this scheme, because the functiondefining magnetic hard layer can be reversed in nanoseconds, allowing hardware to
adaptively track the optimized architecture for the computation [15]. This scheme
has also some disadvantages such as data change require high current densities which
in terms requires high magnetoresistance ratios and large transistors. A variation on
this theme exists, in which an MTJ is used to bias a conventional electronic logic
gate [16]. In this case, the MTJ is only used to define the logic function; the actual
computation is performed entirely in classical electronics.
As an alternative to MTJ logic scheme is the domain-wall logic, which uses no transistors and exhibits very little heating caused by data switching. A domain wall
(DW) is a mobile interface between regions of oppositely aligned magnetization. It
can be propagated through complex structures of nanowires with the application of
magnetic field or current. As in high anisotropy magnetic nanowire, the magnetization preferred to align only a particular direction or opposite to it, these two possible
directions form the basis of binary information scheme with a DW as a transition
edge of the changing signal. Allwood et. al. have implemented logical NOT, logical
AND, signal fan-out and signal cross-over elements with simple geometric design of
magnetic wires using DW motion in them [3]. Figure 1-3 shows the nanowire network
integrating of all four logic elements.
17
Figure 1-3: FIB image of a magnetic nanowire network containing one NOT gate,
one AND gate, two fan-out junctions, and one cross-over junction [3].
Another DW logic device [4] operates by motion of a domain wall in a ferromagnetic
wire. Figure 1-4 shows the cartoon of the logic device. The DW motion is caused by
a spin transfer torque (STT) effect of a current along the wire. So the device is also
called STT/DW device [17]. As the domain wall moves, the magnetization below the
MTJ stack switches, resulting in high or low resistance of the stack. Then, a current
from the "clock" terminal to the output through the MTJ is either high or low. The
output current is used to drive the inputs of next stages.
On the other hand, DW memory has the prospect of on the fly truly 3D device
[5]. One such 3D device is the racetrack memory (RM) in which magnetic domains
are used to store information in tall columns of magnetic materials arranged perpendicularly on the surface of a silicon wafer. The vertical RM is shown in Fig. 1-5.
For proper and repeatable operation of the DW logic and memory devices, the precise depinning and the motion of the DW is the most important and fundamental
issues [18, 19, 20, 21]. The exact position of the DWs must be determined for these
applications. But due to the intrinsic and extrinsic pinning sites, the DW depinning
shows the stochastic behavior and this is the major challenge to be overcome to apply
the scheme of DW motion to the future memory devices [5, 22]. One attempt to con18
........ ......
........
..
- ......
ON State: Output Current High
OFF State: Output Current Low
Figure 1-4: STT/DW logic device [4].
J-u1.
..
A
Mk
A
Figure 1-5: Cartoon of vertical racetrack memory device [5].
19
trol the DW motion is to create artificial trapping sites within magnetic nanowires
[23, 24, 25]. Experimental studies on predicting DW depinning reported by direct
and indirect probing like magnetic X-ray microscopy, macroscopic hysteresis loops
and magnetoresistance measurements reported so far have focused on these artificial
trapping sites [26, 27, 28, 29, 30]. But the DW need to be moved reliably between
engineered pinning sites, without ever getting stuck on the inevitable structural defects which include crystalline defects, grain boundaries, edge roughness. Although
the roughness plays a positive role by introducing feedback mechanism between the
DW velocity and the out-of-plane magnetization component [31], it introduces the
stochastic nature of the DW motion in the nanowires. In continuous magnetic films,
the DW travels in small discontinuous steps a well-known phenomenon [321, caused
by the pinning and subsequent release of the DW from defects in the film. Similarly,
DW motion is troubled by the defects in the nanowires a basic building block for the
DW logic and magnetic devices.
Recently, experimental studies have been done on a few hundred nanometer wide magnetic wires and the stochasticity of DWs in these nanowires is shown to be suppressed
at very low fields where the DW is no more under the precessional regime [33, 34].
The scaling of the above devices that rely on the manipulation of DWs in magnetic
nanowires requires an understanding of domain wall behavior as a function of the wire
width. Due to the increased importance of edge roughness and microstructure in narrow lines, DW pinning increases dramatically as the wire dimensions decrease and
stochastic behavior is expected depending on the distribution of pinning sites. Thus,
statistical observation of DW depinning and propagation in sub-100 nm nanowires
without engineered trapping sites is of great technological importance.
This study is directly related to the direct observation of stochastic behavior of DW
depinning in sub-100 nm Co nanowires spaced by only few tens of nanometer. Here
we show the DW stochastic pinning in two different types of nanowire rings one type
of nanowire rings is interacting with each other and the other type is non-interacting.
We also show that the stochasticity of the DW pinning can be expressed by Poisson's
distribution, which matches very well for the interacting DW case.
20
Chapter 2
Thin Film Deposition
Most ferromagnetic thin films display a uniaxial anisotropy in plane irrespective of
their deposition method. This anisotropy may or may not be obvious from measurement of bulk hysteresis loops. Then the magnetized domain regions within the film
can be seen from domain patterns using the MOKE or magnetic force microscopy
(MFM). The domain images show a tendency for domain walls to lie along preferred
directions in the plane of the film. It is well known that the application of a magnetic
field during the deposition has the effect of ensuring that the easy axis of magnetization of the whole film is in the one direction. The magnetic anisotropy may originate
from the crystal structure of the material, the stress and strain in the film, shape of
a magnetic structure or interfaces. Only the magnetocrystalline anisotropy and the
interface anisotropy will be discussed in this chapter since these are the reasons for
introducing anisotropy in the magnetic films studied in this thesis.
Magnetic materials tend to display a directional dependence of their properties,
and this is a consequence of the magnetic anisotropy. The magnetic anisotropy describes the preference of the magnetization to lie in a particular direction. The magnetic anisotropy in crystalline materials has clearly been demonstrated where certain
crystallographic directions are easy directions of magnetization, whilst others are hard
directions. This has been ascertained by the measurement of magnetization curves.
In the absence of any applied external magnetic field the magnetization prefers to
21
lie along an easy axis, since this minimizes the magnetic energy of the system. The
easy and hard directions can be distinguished by the magnetic field needed to achieve
magnetic saturation. A prime example is a single crystal of Fe, where the <100>
directions are easy axes of magnetization, whilst the <111> directions are said to
be hard axes of magnetization [35].
This form of magnetic anisotropy is referred
to as the crystal anisotropy, or magnetocrystalline anisotropy, which is intrinsic to
the material. The crystal anisotropy originates from the spin-orbit interaction where
the electron spin is coupled to the electron orbit. When a magnetic field is applied
to rotate the electron spins, it also attempts to reorient the electron orbit, which is
strongly coupled to the crystal lattice. The magnetic field or energy needed to rotate
the electron spins (magnetization) is known as the crystal anisotropy energy, which
is the energy needed to overcome the spin-orbit interaction. The expression for the
crystal anisotropy energy density for a cubic crystal can be expressed by the following
equation [35].
Ea = Ko + K1(a2a2 + a2 a2+2 a a) + K2 (a2a2a2) +
where KO, K 1 , K
2
...
(2.1)
are the anisotropy constants, which are material dependent and
a's are the directional cosines of the angle between the magnetization and crystal
axes. Co is an example of magnetic film with magnetocrystaline anisotropy whose
easy axis is along the c axis of the crystal.
Another type of anisotropy involves the shape of the film, which is called the
surface anisotropy. In very thin films (<5nm) the surface or the film/substrate interface can give rise to a significant surface anisotropy [36, 37]; this can strongly
influence the magnetic properties when the surface to volume ratio is comparable
with or larger than that of the film thickness. The surface anisotropy is due to the
abrupt change in the structural and chemical environment at the surface and can
cause the magnetization to point out of the plane of the film. The perpendicular to
the plane anisotropy in thin CoFeB originates from the interface anisotropy [38]. The
22
0.6
0.40.2-
0-0.2-0.4-
-
-0.6
-0.81
-0.5
0
0.5
Field (00)
1
x4W
Figure 2-1: In-plane hysteresis loop of CoFeB thin film
saturation magnetization of the film is 0.5 memu.
2.1
In plane Anisotropy Film
The in-plane magnetic anisotropy (IMA) film is comparatively easier to deposit with
sputter deposition. The IMA CoFeB thin film is deposited using the UHV sputtering
system at base pressure 5x10-8 Torr. The deposition rate of the CoFeB is 0.67
nm/sec at the pressure of 2 mTorr with 100 W DC power. 5 nm Ta film is used as
the seed layer and 3 nm of Au is used as the capping layer for the CoFeB film. The
in-plane hysteresis loop of the IMA CoFeB film is shown in Fig. 2-1.
2.2
Perpendicular Anisotropy Film
In recent years a great effort has been devoted to the study of magnetic films and
multilayers in which there is a change of the preferential orientation of the magnetization from the commonly observed in-plane direction to that perpendicular to the film
23
plane. This phenomenon, usually referred to as perpendicular magnetic anisotropy
(PMA), originates from the competition between the magnetostatic energy and the
out-of-plane anisotropy energy. Using this effect, it is possible to tailor and synthesize structures in which the magnetization can be normal or parallel to the film, by
varying the thickness of the magnetic film and choosing an appropriate combination
of materials. This is very important for technological applications such as solid-state
electronic devices and magnetic recording media.
PMA has been recently discovered in ultrathin CoFeB (t < 1.5 nm) films in
Table 2.1: The list of peaks for MgO thin film.
No. h k I d[Anstrom] 2Theta[deg] I[%]
1
1 1 1
2.431
36.947
10
2
2 0 0
2.106
42.909
100
3
2 2 0
1.489
62.306
52
Ta/CoFeB/MgO structures.
logical interest.
This observation is of both fundamental and techno-
The deposition technique for MgO is different from that of other
material described so far since it is an oxide. MgO is deposited using RF sputtering.
If DC sputtering were used it would require 1012 V to sputter insulators. So a 13.56
MHz AC source is used to avoid charge build-up on the substrate.
The deposited MgO film is characterized using grazing incidence angle measurement
with Rigaku X-ray spectroscopy system. The quick measurement showed a very weak
peak at an angle of 42.9* (peak for (200) plane) (Figure 2-2). So a long measurement
was done to get a better signal to noise ratio and identify the crystal orientation of
the film. The result is shown in Figure 2-3. Three desired peaks of MgO were found
in the measurement, which indicates the MgO film is an amorphous one. The peaks
are listed in Table 2.1. Films were deposited with two different powers at pressure 2
mTorr. Both the deposited film showed the similar amorphous MgO characteristics.
The lower power was chosen for further deposition of the film since high power may
cause crack and other damage to the MgO target.
The next step is to deposit Ta/CoFeB/MgO film stack to check the perpendicular anisotropy.
Figure 2-4 shows the VSM measurement of the deposited Ta(5)/
24
-
- - -.....................
Counts
Middle
2000-
1000-
. . . . . ...
10
. . .
. . . ..
.
0
40
30
Postion (20) (Copper (Cu))
20
60
50
Figure 2-2: Grazing Incidence Angle Measurement for MgO deposited on Si substrate
with native oxide.
Counts
800
200WB0min-30nm
250W-74mn-40nm
400-
410
s0
G0
70)
Poskion r*2Thetaj (Copper (Cu))
Figure 2-3: Grazing Incidence Angle Measurement for MgO deposited on Si substrate
with native oxide with long integration time.
25
x1
8r
6
-4
E
0
N
-.-2
-6
Field (Oe)
X10
Figure 2-4: Out-of-plane hysteresis loop of Ta(5)/CoFeB(1)/MgO(1.2)/Ta(2).
CooFe 2oB 2 o(1)/ MgO(1.2)/ Ta(2) film. The results show that the magnetic layer is
completely dead, a well-known phenomenon [39]. Also since it is hard to get a uniformly thick film without the rotation of the sample and controlling the thickness to
the order of nanometer is hard, a different sputtering system is used to deposit the
PMA film with the facility of rotating sample holder.
The thickness of the MgO layer was fixed to 1.8 nm for depositing PMA films in
this system. CoFeB is varied from 0.6 - 0.9 nm leaving all the other film thicknesses
constant. The base pressure of the system is 1 x 10-
7
Torr. All the metallic thin films
were deposited at 1 mTorr pressure. Only MgO was deposited at 2 mTorr pressure
with 100 W of RF power. The sample stage was rotated during the deposition of
all the films.
The in-plane and the perpendicular to plane VSM measurement of
the hysteresis loop of the deposited film is shown in Figure 2-5. The results show
that only 0.7 nm thick CoFeB has perpendicular anisotropy. As the thickness of the
CoFeB film increases from 0.7 nm to 0.9 nm, the perpendicular anisotropy of the film
also decreases and the film anisotropy becomes in-plane when the thickness of the
film reaches 0.9 nm. The magnetization, M, of the film also increases for the thicker
magnetic films. M, for 0.6 nm film is very small compared to the other cases. It may
26
I.Oxi
4.
-
6. Ox10.a
.0x10
I
I
I.0xi0
0.0
0.0
f.
0x10.
-2.Ox10.' --
I.0x10
Oxi 00
4
-4.
0x10
--
-
-4
-A-4.
4
-
I
2.Ox10
-10000
Ta(5yCFB(O.8yMgO(1.8YTa(5)
Ta(5YCFB(0.7)/MgO(1.8yTe(5)
Ta(5YCFB(O.8YMgO(1.8yTa(5)
x1. Ta(5yCFB(O.9YMgO(1.8)Ta(5)
10000
6"0
0
-5w0
Ta(5)/CFB(O.6yMgO(1.8)/Ta(5)
+
4
-1
i.0ox1
I 2e-+-
--
-10000
0
-5000
Ta(5yCFB(O.7yMgO(1.8yTa(5)
Ta(5)/CFB(O.8yMgO(1.8YTa(5)
Ta(5YCFB(0.9yMgO(1.8YT8(5)
a000
10600
Field (0.)
Fild (0e)
(a)
(b)
Figure 2-5: (a) In-plane and (b) out-of-plane hysteresis loop of different thickness
CooFe 2oB 2 0.
happen that the thickness is not enough to create continuous film.
27
28
Chapter 3
Thin Film Deposition
3.1
Lithography Processes
Usually the patterning of magnetic nanowires either includes the lift-off process or the
focused ion-beam [40, 41] or lift-off processes. The edge roughness of the nanowires
patterned with focused ion-beam is the worst one.
The lift-off patterns also do not have very good edge roughness.
Figure 3-1(a)
shows that during lift-off process there is a lot of sidewall deposition on the resist and
when the resist is lifted off, it leaves hills and valleys of materials at the two edges of
the nanowire (Figure 3-1(b)). The RMS edge roughness of the nanowire is 40.8 nm
[6]. Transport in these nanowires will surely be less than ideal due to the scattering
at the defects. People have also tried the bilayer resist lift-off process which include
two different positive resists with two different etch rate in the developer solvent
[42]. Figure 3-1(c) shows that due to the undercut in the bottom resist, the liftoff
process improves the line edge roughness. The RMS line edge roughness measured
for the patterned nanowire is 12.3 nm, which is way better than that from single
layer resist patterning process (Figure 3-1(d)). Still the thickness of the nanowire is
not uniform over the nanowire width due to the deposition of the material into the
undercut, which may cause a significant reduction in the anisotropy of the films at
the edges. The film used for the nanowires shown in Figure 3.1 is sputter deposited
29
Mona= --
(a) Single layer resist:
0
(c) Si-layer resist:
0
APMMA
PMM
PMMMA
(b)
(d)
Figure 3-1: (a) Single layer and (b) bilayer lift-off patterning process and resulting
line edge roughness in the nanowire. The film used in the patterning process is sputter
deposited Ta (5 nm)
/
NiFe (10 nm)
/
Au (5 nm) in-plaqne Anisotropy film. [6]
Figure 3-2: SEM image of a patterned arc with single layer PMMA lift-off process.
Ta(5)/ NisoFe 2 o(10)/ Au(5). The SEM image of an arc patterned with PMMA resist
lift-off process is shown in Figure 3-2.
To get a fine edge roughness and still have a uniform thickness of the nanowire, a
negative resist need to be used. Considering the scalability of the hydrogen silsesquioxane (HSQ), it is chosen for the present study [7, 43]. It is an excellent negative-tone
resist for high-resolution electron beam lithography down to 4.5 nm half-pitch. Figure
3-3 shows the SEM image of a 9 nm pitch nested-L structures patterned in 10 nm
thick HSQ using Raith 150 at 10 keV accelerating voltage. However, after exposure
and development, the etch resistance of crosslinked HSQ increases, [44] and its sub-
30
Figure 3-3: SEM image of a 9 nm pitch nested-L structures patterned in 10 nm thick
HSQ [7].
(b) 0.8 nm ComFe2B2
(PMA)
(a) 10 nm ComFe2B2
(IMA)
Figure 3-4: (a) In-plane Anisotropy and (b) perpendicular anisotropy film for patterning.
sequent removal requires a hydrofluoric acid dip or a CF4 reactive ion etch (RIE);
both can damage an underlying metallic thin film. This motivated our development
of a bilayer resist process which combines the high etch resistance of HSQ with an
underlying PMMA layer that allows removal of the HSQ features using a solvent such
as n-methyl-2-pyrrolidone (NMP) or acetone, after etching the metal film.
Figure 3-4 shows the two different films used for the patterning process in the
study. The film stack with Ta(5)/ CooFe 2oB 20(10)/ Au(3) is used as the IMA film
and the other stack with Ta(5)/ Co6 oFe2 oB 2 o(0.7)/ MgO(1.8)/Ta(5) is used as the
PMA film.
Figure 3-5 shows the process flow for patterning the magnetic structures. All pat-
31
terning was done on silicon substrates with a native oxide. In step 1, 2 % PMMA in
anisole was spun at 4 krpm for 60 s and baked at 180*C for 90 s to produce a film
thickness of 30 nm. Then 2 % HSQ in methyl isobutyl ketone was spun at 4 krpm for
60 s and baked at 110*C for 60 s to produce a film thickness of 30 nm. In step 2, the
HSQ was exposed. For line widths > 50 nm, a Raith 150 electron beam lithography
tool was used at 30 keV electron energy and 400-800 C/cm2 dose. The samples were
developed using 2.4 %tetramethylammonium hydroxide in water (CD-26) developer
for 2-4 min. For < 50 nm line width, an Elionix F-125 electron beam lithography
%
tool was used at 125 keV with dose 32 mC/cm 2 and developed using 4 % NaCl/1
NaOH in water for 20 s since this salty developer has the highest contrast value among
the results reported by Yang and Berggren [45]. The electron beam exposure of the
HSQ caused scission in the PMMA beneath the HSQ, which increased its solubility,
so HSQ development had to be carried out using nonsolvents for PMMA. In step 3,
an 02 RIE was performed at base pressure 1x 10- Torr, using 10 sccm oxygen and
90 W, for 60-180 s depending on the reference voltage. The etch time was chosen to
produce an undercut of the PMMA under the HSQ. In step 4, ion beam etching was
used to transfer the pattern into the CoFeB using Ar ions at base pressure 1 x 10-7
Torr, with Ar flow of 1.5 sccm, 10 mA beam current, and a 2 cm beam diameter. In
step 5, the PMMA/HSQ mask was removed by placing the sample in NMP at 135*C
or Microposit Remover 1165 at 80*C for 90 min, sonicating for 60 min, and continuing the process three times. This dissolved the PMMA and removed the HSQ. The
extended removal time is required due to redeposition of the HSQ on the sidewalls of
the structures during ion milling, which reduced the surface area of PMMA exposed
to the solvent.
Figure 3-6 shows scanning electron microscope (SEM) images of three different
width nanowires during different steps of the process. The nominal widths of the
three different nanowires were 50 nm, 100 nm and 150 nm. Figure 3-6(a) shows a
cross section of the double-resist stack after step 3. The HSQ was 30 nm thick, with
50-70 nm thick PMMA underneath. We ensured the HSQ thickness was thicker than
32
a. Spin on
Bilayer Resists
b. Expose and
develop HSQ
C.
02 RIE
d. Ion mill to
desired shape
e. Remove resist
with solvent
Figure 3-5: Bilayer resist processing steps - (a) spin PMMA and HSQ on top of the
film stack, (b) E-beam to expose the HSQ and then development of it, (c) Oxygen
plasma etch to remove PMMA all over the film except under the HSQ (d) Ion-milling
to pattern the film using the resist stack as etch mask, (e) Removal of the resist on
top of the patterns using hot NMP and sonication.
(a)
(b)
(c)
5o nm
lo nm
150 nm
Figure 3-6: : SEM image of the nanowires (a) after oxygen RIE, (b) after ion-milling,
and (c) after resist removal for 50 nm, 100 nm and 150 nm wide wires.
33
As deposited
(a)
0.1,
02,
-
0
0.2
0.1
-
I,
After 02 RIE
0.5
S0.4
0.3
0.2
-0.1
S
-0.
-0.1
-0.2
-3
44.
-
(b)
15W
-100
0
100
E0
-100 0 100 X
200
I0
Field, H (0e)
Field, H (0e)
1500
1000
1000.
500.
S
0
0
-SW
-1000
.1000.
V-1500
-100
-50
0
50
Field, H (Oc)
100
i0
-50
0
50
100
Field, H (0e)
Figure 3-7: VSM of the as deposited film and the films after oxygen RIE for both (a)
in-plane and (b) perpendicular anisotropy films.
the 18 nm thick film, since the ion beam etch rate was found to be similar for HSQ
and the metallic film. The 02 RIE process time of 120 s was chosen to produce an
undercut in the PMMA so that the high-resolution HSQ mask defined the eventual
feature size in the metal. Figure 3-6(b) shows the sample after step 4. The mask
shows a tapered cross section after ion beam etching. Figure 3-6(c) shows a top-down
view of the CoFeB wires after step 5 with widths 56 nm, 109 nm and 155 nm matching
almost the nominal width of the wires.
Figure 3-7 demonstrates the hysteresis loop measured by VSM for both IMA and
PMA films after deposition and after processing step 3.
This is the last step of
processing while we still have the continuous film before ion-milling. The saturation
magnetization of 10 nm CoFeB IMA film of 55 mm 2 size decreases from 0.45 memu
to 0.4 memu which indicates that 120 s 02 RIE caused a slight reduction in the
saturation magnetization, M, from 1880 emu/cm 3 to 1560 emu/cm 3 due to oxidation
of the regions of the film surface exposed to the plasma. But this is still the same
order of magnitude of M, as in the literature [46] and the parts of the film under the
34
-
-..
.........
I-- ...
....
....
..........
Figure 3-8: Cross-sectional SEM images of bilayer resist stack after 1 minute 02
reactive ion etch (RIE, Step 4), for HSQ widths a) 468 nm, b) 261 nm, c) 95 nm, and
d) 61 nm. This RIE time is not long enough for the wider top two HSQ masks.
HSQ/PMMA are protected from topdown oxidation. The M, of the PMA 0.8 nm
CoFeB film also reduces from 540 emu/ cm 3 to 520 emu/cm 3 due to oxidation which
is also in the same order of the reported CoFeB PMA films [47].
Figure 3-8 shows the effects of the linewidth on the undercut of the PMMA.
The bilayer resist in this example was made from higher concentration solutions of
4 % PMMA in anisole and 4 % HSQ in methyl isobutyl ketone keeping all other
parameters the same as above to improve visibility of the stack during SEM. The
HSQ was exposed at 10 keV and developed to form lines with wHSQ- (a) 468 nm, (b)
261 nm, (c) 95 nm, and (d) 61 nm ( 5 %) followed by step 4 02 RIE for 60 s. The
HSQ is slightly tapered in (a) and (b), which could be due to backscattered electron
exposure. The SEM images show that while this RIE time produced an undercut of
35
Figure 3-9: Helium-ion microscope images of CoFeB nanowires of average width (a)
64.9 nm, b) 52 nm, c) 39 nm, and d) 27 nm patterned with bilayer resist processes.
PMMA for wHSQ <100 nm in (c) and (d), it was insufficient to produce an undercut
in the wider lines of (a) and (b), i.e., the 02 RIE time must be matched to the desired
wire width. The linewidth dependence of the PMMA undercut may originate from
differences in the electron beam exposure of the PMMA affecting its response to the
oxygen RIE.
Images from a helium-ion microscope (HIM) of metal wires made using the bilayer
resist etching process are shown in Fig. 3-9. Isolated CoFeB wires of average w. (a)
64.9 nm, (b) 52 nm, (c) 39.3 nm, and (d) 27 nm were obtained with completeremoval
of the bilayer mask after ion beam etching. These widths matched the HSQ mask
widths.
Figure 3-10 shows the SEM images of closely spaced arrays of patterned lines.
.
The lines were exposed with a 125 keV electron beam and area dose 6.4 mC/cm2
This dose is lower than that used for isolated lines (32 mC/cm 2) due to proximity
effects of the nearby lines. The proximity effect was similar to that of a single layer
of HSQ, since only the HSQ is developed and not the PMMA. The wire width in the
table shows the nominal width of the wires and the pitch is also the nominal pitch
36
After
Resist
remov
al
30
After
02 RE
After
Resistremov
at
50
After
02 RI
After
Resist
remov
al
Figure 3-10: Top-down SEM image of the different width and pitch nanowire patterned with 125 keV.
37
of the wires in the design files. Although the 20 nm lines with 40 nm pitch was resolved in the e-beam writing, after doing the ion-milling and the resist removal of the
resists, the wires axe merged together. The same thing happened for the 30 nm wires
with 60 nm (2w) pitch. But the 50 nm wide nanowires was resolves with 100 nm
pitch although the wire width became ~60 nm and the pitch remains almost around
the nominal one. All the other closely spaces arrays were resolved nicely after doing
ion-milling and removal of resist. It is ensured that the sample is oriented during
ion-milling so that the beam comes along the length of the nanowires. But this is
done only with bare eyes.
3.2
Line Edge Roughness
Figure 3-11 describes the calculation of 3- of line edge roughness (LER) of the
nanowires patterned with the bilayer resist lithography processes. At first a SEM Image of the nanowire is taken and then its edge deviation in nanometer along the wire
is calculated using SUMMIT Litho Analysis Software [48]. Then the edge deviation
is fitted using the Gaussian function and the 3a- is calculated from the distribution.
Also the root-mean square (RMS) of the LER, rLER is calculated from the edge deviation found analyzing the data using the following equation.
RMS=
n
(3.1)
where, xi is the edge deviations of the wires at different points along the length of
the wire and n is the no of data points.
For each nominal w, five SEM images (pixel size 0.89 nm) were taken of different
wires patterned on the same substrate. Figure 3-12 shows the 3a and RMS LER r
versus wire width w. We find rRMs e 2 nm independent of wire width. These wires
were up to 1 mm long, and the rRMS was the same at several different locations
measured along the length. The LER of the patterned HSQ layer was 0.6-0.7 nm, so
38
..
.....................
(a)
8
(b)
-Top Edg
4
0 2
0
-2
Bom
Edge
-~4
W-
A
200 40060 O
oo
1000
x (nm)
10
(c)
WEa
$6
A
0
3a
0 -2
0-4
~1 %
0.05
01
0.15
0.2
0.25
Figure 3-11: (a) Top-down SEM Images of 50 nm wide CoFeB nanowire, (b) The two
edges of the nanowire in (a) and (c) Gaussian distribution of the edge variation along
the length of the nanowire in (a).
39
......
.......
....-
" _ ........
...
8
12
10-n3a
RM
10
30
RMS
E
S
4-
4
-
_j
j
LU
-2
2
2
0
50
100
150
200
Line Width (nm)
250
90
(a)
20
30
40
Line Width (nm)
50
60
(b)
Figure 3-12: 3a- and the RMS value of line edge roughness measured for the nanowires
patterned with (a) 30 keV and (b) 125 keV accelerating voltages.
there was an increase of 1.3-1.4 nm in LER when the pattern was transferred from
the HSQ to the metal by ion beam etching. Figure 3.12(a) shows the LER for the
wires patterned with Raith 150 at accelerating voltage of 30 eV and the HSQ was
developed with CD-26 solvent.
The wire widths cannot be reduced below 50 nm for the parameters used. The
electron beam voltage was increased from 30 keV to 125 keV for wires < 50 nm to
reduce the electron beam diameter and salty developer was used for HSQ. The LER
of the wires are in the same range for both of these accelerating voltages. Although
the wires patterned with 125 keV accelerating voltage have less variation in the wire
widths comparing the 50 nm width wire patterned with the 30 keV one. The RMS
LER,
rRMs
is 2.7 nm for the 20 nm wide nanowires with 60 nm pitch and
rRMS
is
1.97 nm for 30 nm wide wires with 90 nm pitch, which is similar to those of isolated
lines of same width. We conclude that the proximity of multiple lines does not affect
the LER of small-pitched CoFeB lines patterned using this method.
40
3.3
Magnetic Properties
The magnetic properties of the nanowires and other small patterns can be identified with MFM. It is important to identify the magnetic properties of the patterned
structure since the processing steps can damage the properties and create magnetically dead layer. Experimental studies on characterizing the magnetic properties of
the patterned structures involve either macroscopic hysteresis loops (VSM) or magnetoresistance measurements [49, 50]. For the VSM measurement, the structures need
to closely spaces and a large area (~,l mm2 ) need to be patterned for signal detection [51]. Since we never patterned very big area, the VSM is not a good option for
the patterned structures in our case. On the other hand, the electrical measurement
requires extra patterning of contacts. To avoid the extra step, MFM was chosen for
the magnetic characterization, which also allow us to directly image the domains in
the nanowires. The MFM measurement requires a magnetic probe which can interact
with the magnetic structures while scanning and this interaction can be measures in
terms of force which generates a two dimensional mapping of the each point in the
scanning area.
Figure 3-13 shows the atomic force microscopy (AFM) and the MFM of three
different structures patterned with the bilayer resist lithography process. The scan
height of the MFM measurement was 25-30 nm and the probes used was the low
magnetic moment ones. Figure 3-13(a) (right) shows the MFM image of the spin-ice
system and the bright and dark contrast from the end of the small magnetic bars
is indicative of an in-plane magnetic dipole corresponding to each bar as expected
from its shape anisotropy; the stray field from the north and south poles is imaged
as bright and dark contrast. The next AFM image shows the nano-rings of 500 nm
outer diameters and 260 nm of inner diameters (Figure 3-13(b) left). The lack of any
bright and dark contrast in the MFM images of the nanorings shows that the domain
in the rings is the vortex one, which reduces the mamgnetostatic energy of the system
(Figure 3-13(b) middle). The rightmost MFM image in Figure 3-13(b) is the zoom
41
(a)
(b)
(c)
Figure 3-13: Atomic force microscopy and the magnetic force microscopy images of
(a) spin-ice system, (b) nanorings and (c) oval-shaped structure.
42
(a)
(b)
Figure 3-14: AMF and MFM of 50 nm wide wires with different pithes.
in view of only one nanoring. When the width of the magnetic wire becomes very
large, the domain structures in the wire changes and depending on the magnetostatic
energy and the exchange interaction energy, a complex domain wall may form within
the wire. Figure 3-13(c) (right) shows such a complex domain wall structure, which is
called a vortex wall. There are two vortex walls in the parabolic magnetic structure.
The CoFeB is a very soft magnetic material and the domains in the nanowires
can easily be moved with the stray magnetic field from the MFM probes. Figure
3-14 shows the AFM (left) and MFM (right) images of the 50 nm wide wire arrays
with different pitches. Before measuring with MFM, the wires were saturated along
their lengths. The white contrast from the bottom end of all the wires confirms the
magnetic behavior. A very long black contrast is visible in a few nanowires which
indicates domain walls in the wires. But since the domain walls cannot be so long in
the wires with such small width, it is the dragging of the DWs along the length of
the wires due to magnetostatic interaction of the probe with the DW that causes the
contrast.
43
44
Chapter 4
DW Motion
4.1
High Anisotropy Film
In chapter 3, the MFM image of the nanowires made with soft magnetic material
CoFeB proves that the magnetic probes are affecting the domain walls. Since we
focus on the MFM as our characterizing tool, clearly the CoFeB nanowires cannot
be an ideal choice. So a comparatively hard material Co is chosen for studying the
DW motion in the nanowires. Figure 4-1(a) shows the film stack that will be used
for any further patterning. It comprises of a 5 nm Ta as a seed layer, 8 nm Co as
the magnetic hard layer and 3 nm of Au as the capping layer. The in-plane and
perpendicular to the plane hysteresis loops of the deposited film is shown in Figure
4-1(b). The saturation magnetization of the film is 700 emu/cm3 which is almost half
of that found in literature [52] and the coercive field is 27 Oe. The perpendicular to
plane hysteresis loop did not saturate, so it is only the minor hysteresis loop.
45
......
.....
..........
..- ...........
800-
(b)
600
-
(a)
400200-
-
0
-200
400
-
-600
-8001
1
-0.5
0
Feld (Oe)
0.5
1
xl0
Figure 4-1: (a) High anisotropy Co in-plane anisotropy film stack and (b) the hysteresis loop of the deposited film along the in-plane direction (red) and perpendicular
to the surface direction (black).
4.2
Magnetic Field induced Stochastic Domain wall
Motion
8 ym diameter nanorings are patterned with the deposited Co film using the same
process described in chapter 2. Figure 4-2 shows the MFM images of the nanoring
after applying two different magnetic fields along the y-direction.
The elongated
dark contrast is still visible in Figure 4-2(a) and (b). If we consider the slow scanning
direction of the rings, then the initial position of the DW can be identified. When
scanning along the negative y-direction (Figure 4-2(a)), the wall starts from the
point when the dark contrast is started and the DW gets pinned at the point where
the dark contrast stops continuing. In the next scan, along the y-direction (Fig 42(b)), the dark contrast starts from the same point where it stopped and continued
until it is pinned by the edge defects. In Fig 4-2(c), the two DWs get stuck to some
places after applying 250 Oe, so that the stray field from the probes is not enough to
depin them. Clearly, the edge defects of the nanorings are acting as pinning sites for
the DWs although the edge roughnesses of the rings are very small.
46
..
Scanning
direction
t
(a)
(b)
(c)
Figure 4-2: MFM image of 200 nm wide and 8 pum diameter circular rings after
applying 220 Oe along the y-direction with scanning direction (a) from top to bottom,
and (b) from bottom to top and (c) after applying 250 Oe along the y-direction with
scanning direction from bottom to top.
To check the DW pinning in the nanowires patterned with the described process,
multiple concentric rings with very small spacing have been chosen. The rings are
very good starting point for studying DW motion with magnetic field since it is easy
to introduce domain walls in these structure [50]. The SEM image of the patterned
concentric rings has been shown in Fig. 4-3. The width of the wires is 30 nm and the
spacing between two concentric rings is 60 nm.
Three different width and inter-wire distant nanowire rings have been considered
-
in this study
(i) 30 nm wide with 70 nm inter-wire distances,
(ii) 50 nm wide with 50 nm inter-wire distances and
(iii)60 nm wide with 40 nm inter-wire distances.
Based on the interaction of the DWs in the neighboring nanowires, the above
three different widths nanowires can be divided in to two types(i) Non-interacting nanowire rings
(ii) Interacting nanowire rings
47
.. .......
......
..
..
.. ..
..
..
..
..
..
Figure 4-3: SEM image of the concentric rings.
The first two are non-interacting nanowires - the DW in one nanowire rings
does not interact with its neighboring DWs. The last one is the interacting type of
nanowires as the distance between them is very small. So the DWs in these nanowire
rings are found to move together. The division of the nanowires will become clear at
the end of this chapter.
At first a high magnetic field of 3 kOe is applied along the x-direction of the sample
to create two DWs in each ring to form onion states [53, 54]. Figure 4-4 shows the
AFM and the MFM images of the 60 nm wide nanowires. The onion states in these
wires are shown in Fig. 4-4(b) wide concentric rings. There are fifteen concentric
rings. The bright dots in the MFM image represent head-to-head (HH) DWs and
the dark dots are of tail-to-tail (TT) DWs. In this study, only the TT DWs are used
for locating the pinning sites in the nanowire rings. Figure 4-4(c) and (d) show the
zoomed in MFM image of the TT DWs and HH DWs. These DWs are supposed to be
aligned along the x-direction as a strong field is applied at that direction. But Fig.
4-4(c) and (d) shows that, when the nanowire rings relaxes after applying a high field,
both the HH and the TT DWs are not created along a straight line rather they relaxes
in a scattered way. This implies that the edge roughnesses in the nanowire rings are
acting as trapping sites for the DWs. So, after removing the aligning field, the DWs
48
Figure 4-4: AFM Image of 60 nm wide fifteen concentric rings, and MFM image of
(b) the domain walls in the rings after magnetized the rings along the x-direction;
zoom in view of (c) the tail to tail domain wall and (d) the head to head domain wall.
are created in each nanowire ring at places where it is energetically favorable.
After creating the DWs - all TT walls in the left side and all HH walls in right
side of the rings, the DWs then translated along the rings by applying magnetic field
along the y-direction. When the applied field is high enough to depin the DWs from
their pinning sites, the DWs start moving and eventually they became trapped in the
pinning sites which 'requires higher energy to overcome.
Once a DW trapped in a
pinning site it requires higher magnetic field to depin from that sites. Every time the
DWs are trapped in pinning sites, the nanowires rings are reset to their initial onion
states before applying higher magnetic field. Their position is confirmed by MFM
and then they are translated with higher magnetic field along y-direction. The same
experiment is done by not resetting the DWs before their further translation along
the nanowire rings and only applying higher magnetic field along y-direction. These
two processes of translating DWs do not show much variation.
49
.
...
.....
.
..........
- - ----
Figure 4-5: MFM image of the 30 nm wide eleven concentric rings after applying (a)
220 Oe, (b) 250 Oe, (c) 300 Oe, (d) 350 Oe and (e) 400 Oe along the y-direction. The
field of view in each figure is 12 x 12 pm2
The nanowire rings experience the magnetic stray field from the MFM probes
during scanning. But as there is no elongation of the TT DWs (black dots denote the
attractive forces between the DWs and the MFM probes) in the rings, the stray field
from the MFM probes is not disturbing the domains in the rings. So we can safely
discard the effect of MFM probes on the DW motion and consider the imaging as a
noninvasive process.
The translation of DWs in the nanowire rings require very high magnetic fields
compared to the coercive field of the continuous Co film. It took over 200 Oe to move
the DWs from their initial position. This high value of propagation field reflects the
influence of edge roughness in the nanowire rings. Figure 4-5 shows the MFM images
of 30 nm wide nanowire rings after applying different magnetic filed along y-direction.
Both the HH and TT DWs start moving in the directions of the nanowire rings to
allow the expansion of the domain aligned with the applied magnetic field.
50
The
...........
.
............
Figure 4-6: MFM image of the 50 nm wide twelve concentric rings after applying (a)
250 Oe, (b) 350 Oe, (c) 450 Oe, and (d) 500 Oe along the y-direction. The field of
view in each figure is 12x12 ym2
general trend is that with higher applied field the DWs move further.
But some
of the DWs remain at the same position with higher applied field. The TT DWs
marked by black circle did not move even after applying 350 Oe. But it moved from
the pinning sites after applying magnetic field of 400 De.
So the DW needs some
magnetic field in between 350 Oe and 400 Oe to overcome these pinning sites. For
simplicity, we assume the DW requires 400 Oe magnetic fields to hop out from the
pinning sites.
In other word, the pinning site requires energy of 400 Oe magnetic
fields to be overcome. Thus the pinning strength of different sites of the nanowire
rings can be measured by analyzing the MFM images of the nanowire rings after
applying different magnetic fields. From the MFM images, it is clear that the higher
energy pinning sites are located at further distances from one another compared to
the lower energy pinning sites.
Figure 4-6 shows the MFM images of the 50 nm wide nanowire rings after applying
51
...
.............
I.
Figure 4-7: MFM image of the 30 nm wide fifteen concentric rings after applying (a)
220 Oe, (b) 250 Oe, (c) 300 Oe, (d) 350 Oe and (e) 400 Oe along the y-direction. The
field of view in each figure is 12 x 12 pm2
four different magnetic filed to move the DWs. The DW motion in this case is similar
to that in case for 30 nm wide nanowires. The similar experiment is done for 60 nm
wide nanowire rings and it is found that in this case, the DWs in the neighboring
nanowires couple together while moving (Figure 4-7). This clearly shows that DWs
are coupled with magnetostatic interaction with each other.
To check the stray field of a DW to its neighbors,' uanowires with 30 nm, 50 nm
and 60 nm widths are simulated with the Object Oriented MicroMagnetic Framework
(OOMMF). The magnetic stray fields from the DWs at some distances are shown in
Fig. 4-8 and it is found to be less than 70 Oe for 30 nm wide nanowires at distances
of 60 nm and 130 Oe for 50 nm wide nanowires at distances of 50 nm.
Clearly, the magnetic stray field is less than the strength of the pinning sites for
these two different types of nanowire rings and they are the non-interacting type of
52
j flf
-800
30nm wide
wide
-50nm
60nm wide
w 600
0
Il400
200
00
0
20
40
60
80
S- Ywireedge (nm)
Figure 4-8: Stray field distribution from various width nanowire at distances.
nanowires. So, we can assume that DWs in these two types of concentric rings are
behaving by their own and the pinning sites only come from the lithography processes
only. The stray magnetic field from a DW in the other type of nanowire rings (60
nm width with 40 nm inter-wire distances) is 230 Oe to its neighbors at 40 nm distances, which is found to be equivalent to the minimum magnetic field to depin the
DWs from the pinning sites. So in this type of nanowire rings, the DWs need more
energy to overcome once they are trapped at particular pinning sites. In this type of
nanowires, the stray field from one DW provides remote pinning for the DW next to it.
4.3
Modeling of Domain Wall Stochastic Nature
Modeling of DW stochastic behavior requires the measurement of their travelling
distances with different applied field. Figure 4-9 shows the distances travelled by the
DWs after applying magnetic fields in histograms. These histograms show position
of the pinning sites of certain energy.
We calculated the probability of fining a pinning site of a particular energy at
some places in the nanowires. Figure 4-10 (a) shows the pinning sites distribution
53
~-___________
(a)
(b )4
(C)
=3
=3
.8
0
O6
1
o
2
3
5
6
7
4
diatwic beeled by Ur dornir. wall (pm)
(d)7
(e)
1
2
3
4
5
6
7
dlawic traeled by UT domatrs wds (pm)
CoITL0
0
1st3a
5
F
4
=3
0
o
1
2
3
4
5
e
7
dstane triveled by TT domidra wval (pm)
o0
1
2
3
4
6
7
5
distaic traeled by UT domairs wat (pm)
Figure 4-9: Histogram plot of the distances travelled by tail to tail DWs in 60 nm
wide concentric rings after applying (a) 220 Oe, (b) 250 Oe, (c) 300 Oe, (d) 350 Oe
and (e) 400 Oe along the y-direction.
in 30 nm wide nanowire rings. The markers are found analyzing the actual DWs
positions in the MFM images of the nanowire rings.
The data are normalized at
each magnetic field applied. The distances between a particular energy pinning sites
along a distances in the nanowire rings increases with their strength. So most of the
pinning sites resulting from lithography processes in a small nanowire do not require
high energy to depin the DWs from their traps.
The dotted line in Fig. 4-10(a) shows the Poisson fit of the probability. The color
correspondent to the applied field described in the legend. The mean distances of
the Poisson's fit are taken from the experimental data at each magnetic field. The
Poisson fit is plotted from the following equation
P = AX e-A
X!
(4.1)
where, P is the probability of a DW being pinned, A is the mean distances of the
DWs travelled along the nanowire rings and x is the distances travelled by DWs. The
Poisson's distribution does not fit very well for this non-interacting type of nanowires.
54
(a)
0 2200Oe
0.8'4~
.i
50
~.
+ 300 Oe
13 400 Oe
X 410 Oe
0.6
tg
%.
%
f. 0 %%
-
2
2
6
'0
8.4
Separation between pinning sites (pm)
%
%.
1,
(b)
0.8
4 IV0+
+C 4
---
..
--- 200e
%%P
$0.6
\
0
x
\
%
x
0.2
--
--- 4100e
.a.4+
SX\
2
2200Oe
6
4
2
Separation between pinning sites (pm)
8
Figure 4-10: The distribution of DW position in the nanowires and their distribution
follows Poisson's model for (a) non-interacting and (b) interacting nanowires.
55
5
4-
32;
1w=30nm,d=60nm
900
250
A w=50nm,d=50nm
w=60nm,d=40nm
300
350
400
450
Magnetic Field (0e)
Figure 4-11: The mean distance of the DW travels with the applied field in different
width nanowires with different spacing.
Figure 4-10(b) shows the same data for interacting type of nanowires (60 nm wide
with 40 nm inter-wire distances). Poisson's distribution can calculate the probability
of the DWs trapped at some place in the nanowire quite well. So, the pinning sites
in the nanowire rings follow Poisson's distribution at a particular applied field and
the mean of the distribution is a function of applied field for a particular width of
nanowire rings at a specified temperature
A = f(H, w, T)
(4.2)
where, H is the applied field, w is the width of the nanowire rings and T is the
temperature. All the experiments in this letter were done at room temperature. So
the distribution of the pinning sites in nanowire rings with temperature is beyond the
scope of this letter.
Figure 4-11 shows the mean distances of the pinning sites for three different widths
of nanowire rings with applied magnetic field. The error bar shows the variation in
56
the mean distances from multiple measurements of the same nanowire rings. The
stochastic motion of the DWs reveals in the nanowire rings as they end up at different
positions after carrying out the same experiment multiple times. From Fig. 4-11,
the mean distances of the pinning sites increases with their increase in strengths of
pinning. The applied magnetic field to depin the DWs from a particular site is the
measure of strength for that pinning site. So, the nanowire used to make a DW device
need to be less than a micron to get rid of the probability for higher energy pinning
sites. Otherwise, the probability of higher energy pinning sites in the wire will be
high and it would require higher magnetic field or current to depin the DWs from
those sites, which would require higher energy to operate the DW devices.
57
58
Chapter 5
Conclusion
The DW devices and memories seek the deterministic relation between DW motions
with some known quantity (either magnetic field or current). Although silicon technology should be capable of processing at much lower power, one should not dismiss
the power of DW devices on the basis of its stochastic limitations. The current study
of DW stochastic motion have identified a length scale below which the motion should
be more or less uniform providing that the nanowires in that scale have only moderate
width deviation.
The stochastic behavior in the DW motion may come from thermal effect, the
edge variation in the width of the nanowires or any other interaction from the environment. In the current study, it was not possible to separate the contribution
of temperature from the edge variation to the stochastic behavior. To identify the
contribution of temperature to the shochasticity, one needs the facility to scan the
DW position at lower/higher temperature. Insitu Magnetooptic Kerr effect (MOKE)
microscopy can be an option.
Besides, the DW velocity with magnetic field and current can be measured in
these sub-100 nm nanowires. The use of these nanowires in making actual devices
may prove the concept of low energy requirement for DW devices.
59
60
Bibliography
[1] Mario Norberto Baibich, JM Broto, Albert Fert, F Nguyen Van Dau, F Petroff,
P Etienne, G Creuzet, A Friederich, and J Chazelas. Giant magnetoresistance of
(001) fe/(001) cr magnetic superlattices. Physical Review Letters, 61(21):2472,
1988.
[2] S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. von Molnar,
M. L. Roukes, A. Y. Chtchelkanova, and D. M. Treger. Spintronics: A Spin-Based
Electronics Vision for the Future. Science, 294(5546):1488-1495, November 2001.
[3] D A Allwood, G Xiong, C C Faulkner, D Atkinson, D Petit, and R P Cowburn. Magnetic domain-wall logic. Science (New York, N. Y.), 309(5741):168892, September 2005.
[4] J.A Currivan, Youngman Jang, M.D. Mascaro, M.A Baldo, and C.A Ross. Low
energy magnetic domain wall logic in short, narrow, ferromagnetic wires. Magnetics Letters, IEEE, 3:3000104-3000104, 2012.
[5] Stuart S P Parkin, Masamitsu Hayashi, and Luc Thomas. Magnetic domain-wall
racetrack memory. Science (New York, N. Y.), 320(5873):190-4, April 2008.
[6] J. Zhang, J. A. Currivan, L. Tryputen, M. A. Baldo, and C. A. Ross. Applied
Physics Letter, 2014 (Submitted).
[7] Joel K. W. Yang, Bryan Cord, Huigao Duan, Karl K. Berggren, Joseph Klingfus,
Sung-Wook Nam, Ki-Bum Kim, and Michael J. Rooks. Understanding of hydrogen silsesquioxane electron resist for sub-5-nm-half-pitch lithography. Journal of
Vacuum Science & Technology B: Microelectronics and Nanometer Structures,
27(6):2622, 2009.
[8] J Barnas, A Fuss, RE Camley, P Griinberg, and W Zinn. Novel magnetoresistance effect in layered magnetic structures: Theory and experiment. Physical
Review B, 42(13):8110, 1990.
[9] B. Dieny, V. S. Speriosu, S. Metin, S. S. P. Parkin, B. a. Gurney, P. Baumgart,
and D. R. Wilhoit. Magnetotransport properties of magnetically soft spin-valve
structures (invited). Journal of Applied Physics, 69(8):4774, 1991.
61
[10] Jagadeesh Subbaiah Moodera, Lisa R Kinder, Terrilyn M Wong, and R Meservey.
Large magnetoresistance at room temperature in ferromagnetic thin film tunnel
junctions. Physical Review Letters, 74(16):3273, 1995.
[11] Terunobu Miyazaki and Nobuki Tezuka. Spin polarized tunneling in ferromagnet/insulator/ferromagnet junctions. J. Magn. Magn. Mater., 151(3):403 - 410,
1995. Spin Polarized Electron Transport.
[12] G. Reiss, H. Brckl, a. Htten, H. Koop, D. Meyners, a. Thomas, S. Kmmerer,
J. Schmalhorst, and M. Brzeska. New materials and applications for magnetic
tunnel junctions. Physica Status Solidi (a), 201(8):1628-1634, June 2004.
[13] A Ney, C Pampuch, R Koch, and K H Ploog. Programmable computing with a
single magnetoresistive element. Nature, 425(6957):485-7, October 2003.
[14] R. Richter, H. Boeve, L. Bdr, J. Bangert, G. Rupp, G. Reiss, and J. Wecker.
Field programmable spin-logic realized with tunnelling-magnetoresistance devices. Solid-State Electronics, 46(5):639-643, May 2002.
[15] G. a. Prinz. Magnetoelectronics. Science, 282(5394):1660-1663, November 1998.
[16] Siegfried Methfessel and Daniel C Mattis. Magnetic semiconductors. Handbuch
der Physik, 18:389-562, 1968.
[17] D.E. Nikonov and IA Young. Overview of beyond-cmos devices and a uniform
methodology for their benchmarking. Proceedings of the IEEE, 101(12):24982533, Dec 2013.
[18] M. Klsui, C. Vaz, J. Rothman, J. Bland, W. Wernsdorfer, G. Faini, and E. Cambril. Domain Wall Pinning in Narrow Ferromagnetic Ring Structures Probed by
Magnetoresistance Measurements. PhysicalReview Letters, 90(9):097202, March
2003.
[19] A. J. Zambano and W. P. Pratt. Detecting domain-wall trapping and motion at
a constriction in narrow ferromagnetic wires using perpendicular-current giant
magnetoresistance. Applied Physics Letters, 85(9):1562, 2004.
[20] M. Klsui, C. Vaz, J. Bland, W. Wernsdorfer, G. Faini, E. Cambril, L. Heyderman, F. Nolting, and U. Rudiger. Controlled and Reproducible Domain Wall
Displacement by Current Pulses Injected into Ferromagnetic Ring Structures.
Physical Review Letters, 94(10):106601, March 2005.
[21] Masamitsu Hayashi, Luc Thomas, Rai Moriya, Charles Rettner, and Stuart S P
Parkin. Current-controlled magnetic domain-wall nanowire shift register. Science
(New York, N. Y.), 320(5873):209-11, April 2008.
[22] Guido Meier, Markus Bolte, Rene Eiselt, Benjamin Krdger, Dong-Hyun Kim,
and Peter Fischer. Direct Imaging of Stochastic Domain-Wall Motion Driven by
Nanosecond Current Pulses. Physical Review Letters, 98(18):187202, May 2007.
62
[23] Masamitsu Hayashi, Luc Thomas, Charles Rettner, Rai Moriya, Xin Jiang, and
Stuart Parkin. Dependence of Current and Field Driven Depinning of Domain
Walls on Their Structure and Chirality in Permalloy Nanowires. Physical Review
Letters, 97(20):207205, November 2006.
[24] Peter Lendecke, Rene Eiselt, Guido Meier, and Ulrich Merkt. Temperature dependence of domain-wall depinning fields in constricted Permalloy nanowires.
Journal of Applied Physics, 103(7):073909, 2008.
[25] Aakash Pushp, Timothy Phung, Charles Rettner, Brian P. Hughes, See-Hun
Yang, Luc Thomas, and Stuart S. P. Parkin. Domain wall trajectory determined
by its fractional topological edge defects. Nature Physics, 9:505-511, 2013.
[26] Mi-young Im, Lars Bocklage, Peter Fischer, and Guido Meier. Direct Observation
of Stochastic Domain-Wall Depinning in Magnetic Nanowires. 147204(April):14, 2009.
[27] Colm C Faulkner, Michael D Cooke, Dan A Allwood, Doroth6e Petit, Del Atkinson, and Russell P Cowburn. Artificial domain wall nanotraps in Ni 81 Fe 19
wires Artificial domain wall nanotraps in Ni 81 Fe 19 wires. 6717(2004):20-23,
2004.
[28] M. Klhui, H. Ehrke, U. Rdiger, T. Kasama, R. E. Dunin-Borkowski, D. Backes,
L. J. Heyderman, C. A. F. Vaz, J. A. C. Bland, G. Faini, E. Cambril, and
W. Wernsdorfer. Direct observation of domain-wall pinning at nanoscale constrictions. Appl. Phys. Lett., 87(10):-, 2005.
[29] W. Kleemann, J. Rhensius, 0. Petracic, J. Ferr6, J. Jamet, and H. Bernas. Modes
of Periodic Domain Wall Motion in Ultrathin Ferromagnetic Layers. Physical
Review Letters, 99(9):097203, August 2007.
[30] Y. Yokoyama, Y. Suzuki, S. Yuasa, K. Ando, K. Shigeto, T. Shinjo, P. Gogol,
J. Miltat, a. Thiaville, T. Ono, and T. Kawagoe. Kerr microscopy observations of
magnetization process in microfabricated ferromagnetic wires. Journalof Applied
Physics, 87(9):5618, 2000.
[31] Yoshinobu Nakatani, Andre Thiaville, and Jacques Miltat. Faster magnetic walls
in rough wires. Nature materials, 2(8):521-3, August 2003.
[32] Dong-Hyun Kim, Sug-Bong Choe, and Sung-Chul Shin. Direct Observation of
Barkhausen Avalanche in Co Thin Films. PhysicalReview Letters, 90(8):087203,
February 2003.
[33] Xin Jiang, Luc Thomas, Rai Moriya, Masamitsu Hayashi, Bastiaan Bergman,
Charles Rettner, and Stuart S P Parkin. Enhanced stochasticity of domain wall
motion in magnetic racetracks due to dynamic pinning. Nature communications,
1(3):25, January 2010.
63
[34] Manuel Mufioz and Jose L Prieto. Suppression of the intrinsic stochastic pinning
of domain walls in magnetic nanostripes. Nature communications, 2(May):562,
January 2011.
[35] B. D. Cullity and C. D. Graham. Introduction to magnetic materials. Second
Edition. IEEE Press., 2009.
[36] Gradmann. 1986.
[37] Mattheis. 1999.
[38] S Ikeda, K Miura, H Yamamoto, K Mizunuma, H D Gan, M Endo, S Kanai,
J Hayakawa, F Matsukura, and H Ohno. A perpendicular-anisotropy CoFeBMgO magnetic tunnel junction. Nature materials, 9(9):721-4, September 2010.
[39] K. Oguz, P. Jivrajka, M. Venkatesan, G. Feng, and J. M. D. Coey. Magnetic
dead layers in sputtered Co[sub 40]Fe[sub 40]B[sub 20] films. Journal of Applied
Physics, 103(7):07B526, 2008.
[40] J. Lohau, A. Moser, C. T. Rettner, M. E. Best, and B. D. Terris. Writing
and reading perpendicular magnetic recording media patterned by a focused ion
beam. Applied Physics Letters, 78(7):990, 2001.
[41] Stephan Breitkreutz, Irina Eichwald, Josef Kiermaier, Gaspard Hiblot, Gyorgy
Csaba, Wolfgang Porod, Doris Schmitt-Landsiedel, and Markus Becherer. Controlled domain wall pinning in nanowires with perpendicular magnetic anisotropy
by localized fringing fields. Journal of Applied Physics, 115(17):17D506, May
2014.
[42] Y. D. Park, K. B. Jung, M. Overberg, D. Temple, S. J. Pearton, and P. H. Holloway. Comparative study of Ni nanowires patterned by electron-beam lithography and fabricated by lift-off and dry etching techniques. Journal of Vacuum
Science & Technology B: Microelectronics and Nanometer Structures, 18(1):16,
2000.
[43] Sung-Wook Nam, Michael J. Rooks, Joel K. W. Yang, Karl K. Berggren, HyunMi Kim, Min-Hyun Lee, Ki-Bum Kim, Jae Hwan Sim, and Do Yeung Yoon.
Contrast enhancement behavior of hydrogen silsesquioxane in a salty developer.
Journal of Vacuum Science & Technology B: Microelectronics and Nanometer
Structures, 27(6):2635, 2009.
[44] Marcus Rommel and Jurgen Weis. Hydrogen silsesquioxane bilayer resistsCombining high resolution electron beam lithography and gentle resist removal. Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Struc-
tures, 31(6):06F102, 2013.
[45] Joel K. W. Yang and Karl K. Berggren. Using high-contrast salty development of
hydrogen silsesquioxane for sub-10-nm half-pitch lithography. Journalof Vacuum
64
Science & Technology B: Microelectronicsand Nanometer Structures, 25(6):2025,
2007.
[46] Dexin Wang, Cathy Nordman, Zhenghong Qian, James M. Daughton, and John
Myers. Magnetostriction effect of amorphous CoFeB thin films and application
in spin-dependent tunnel junctions. Journal of Applied Physics, 97(10):10C906,
2005.
[47] Michihiko Yamanouchi, Albrecht Jander, Pallavi Dhagat, Shoji Ikeda, Fumihiro
Matsukura, and Hideo Ohno. Domain Structure in CoFeB Thin Films With Perpendicular Magnetic Anisotropy. IEEE Magnetics Letters, 2:3000304-3000304,
December 2011.
[48] http://www.lithometrix.com.
[49] P. Schio, F. Vidal, Y. Zheng, J. Milano, E. Fonda, D. Demaille, B. Vodungbo,
J. Varalda, a. J. a. de Oliveira, and V. H. Etgens. Magnetic response of cobalt
nanowires with diameter below 5 nm. Physical Review B, 82(9):094436, September 2010.
[50] T. Hayward, T. Moore, D. Tse, J. Bland, F. Castano, and C. Ross. Exquisitely
balanced thermal sensitivity of the stochastic switching process in macroscopic
ferromagnetic ring elements. Physical Review B, 72(18):184430, November 2005.
[51] Elizabeth Friend. Interaction of Stress and Magnetic Properties in Patterned
Copper-Nickel-Copper Thin Films. PhD thesis.
[52] Eric E. Fullerton, J. S. Jiang, M. Grimsditch, C. H. Sowers, and S. D. Bader.
Exchange-spring behavior in epitaxial hard/soft magnetic bilayers. Phys. Rev.
B, 58:12193-12200, Nov 1998.
[53] J. Rothman, M. Klhui, L. Lopez-Diaz, C. Vaz, a. Bleloch, J. Bland, Z. Cui, and
R. Speaks. Observation of a Bi-Domain State and Nucleation Free Switching in
Mesoscopic Ring Magnets. Physical Review Letters, 86(6):1098-1101, February
2001.
[54] S. Li, D. Peyrade, M. Natali, a. Lebib, Y. Chen, U. Ebels, L. Buda, and
K. Ounadjela. Flux Closure Structures in Cobalt Rings. Physical Review Letters,
86(6):1102-1105, February 2001.
65