mgo mtj biosensors for immunomagnetic lateral-flow detection
TRANSCRIPT
1
MgO MTJ biosensors for immunomagnetic lateral-flow
detection
Ricardo Jorge Penelas Janeiro
Dissertação para obtenção do Grau de Mestre em
Engenharia Física Tecnológica
Júri
Presidente: Prof. Dr. João Carlos Carvalho de Sá Seixas
Orientador: Prof. Dr. Susana Isabel Pinheiro Cardoso de Freitas
Vogal: Prof. Dr. Paulo Jorge Peixeiro de Freitas
Outubro de 2010
ii
iii
Acknowledgement
Com chegada ao fim da tese de mestrado, etapa final do curso, impõe-se o dever e felicidade
de agradecer a todos os que permitiram empreender toda esta jornada académica. Dever porque sem
todos eles quem sabe o que poderia ter acontecido pelo caminho ardiloso que é a vida de um jovem
universitário; Dever porque de todos eles o meu ser se formou, por causa da interacção com cada um
se moldou. Alegria porque poder agradecer significa não só o concluir de um ciclo, como a existência
de a quem agradecer, sinal de um caminho não percorrido isoladamente.
Começo por agradecer à instituição que permitiu a realização deste trabalho, o INESC-MN,
que me proporcionou o meu primeiro contacto com a Ciência: a verdadeira ciência crivada de
dificuldades, imprevistos e cujos resultados nem sempre premeiam o esforço dispendido, mas que é
recompensadora quando é atingido um objectivo. Agradeço ao Professor Paulo Freitas e à
Professora Susana Freitas pela oportunidade de trabalhar no INESC-MN e pela orientação científica
e experimental proporcionada. Um agradecimento também ao Dr. Ricardo Ferreira pela paciência e
disponibilidade com encarou todas as infindáveis questões. Obrigado também ao resto da equipa
científica do INES-MN, dos seus engenheiros aos alunos pela disponibilidade e ajudas prestadas
sempre que solicitados. Um especial obrigado aos alunos de mestrado de curso de física pelos
excepcionais momentos de companheirismo (Cláudio, Raquel, Zita).
Prosseguindo na enumeração daqueles que marcaram a etapa que agora termina é
imperioso referir o conjunto de amigos com quem partilhei diariamente a experiência académica no
IST (made me a man). Desde o primeiro grupo com que travei conhecimento / já conhecia – Mike,
Ana, Romão – às mais recentes crianças que incautamente entraram neste curso – Patos e Amigos
Lmta – passando pelo mais experiente e já Mestre José Gustavo Rebelo, e pelo caríssimo D. Diogo
Capelo, até aquela que pertence a uma classe só dela - Raquel (obrigado coisa :) ) o meu muito
obrigado por tudo sendo que muitas vezes o proporcionarem uma gargalhada foi o suficiente e ideal.
Agradeço a toda a minha família por todo o apoio e confiança depositada em mim e nas
minhas capacidades (pobres ingénuos :p). Um muito obrigado aos meus pais, elementos
fundamentais e sempre presentes em qualquer momento.
Não me escudando na minha confessa e evidente inaptidão para a “escrita livre e pessoal”
desde já as minhas desculpas às inúmeras pessoas que aqui ficaram por referir.
Obrigado Mãe, Obrigado Pai, Obrigado Raquel =)
Ricardo Janeiro
Dezembro de 2010
iv
v
Resumo
Nos tempos que correm, é evidente o potencial das nanotecnologias, as quais são capazes
de fornecer os meios necessários para a detecção e rastreamento de patogéneos na forma de
sensores de tamanho micrométrico muito sensíveis.
O objectivo deste trabalho é a fabricação de sensores magnetoresistivos, com resposta linear
capazes de executarem testes de reconhecimento através da detecção de nanopartículas
funcionalizadas, magneticamente polarizadas e previamente ligadas a biomoléculas. Os sensores
magnetoresistivos usados foram Junções de Efeito de Tunel, as quais apresentam por norma um
valor elevado de magnetoresistência. Neste trabalho junções individuais foram ligadas em série,
formando vectores de 360 elementos individuais. Tal foi feito com o propósito de ganhar na
detectividade dos dispositivos.
Várias amostras foram processadas e magnetoresistências da ordem de 70% foram
conseguidas, assim como sensibilidades de 1.24%/Oe na região linear.
Palavras chave: Sensores magnetoresistivos de efeito de túnel, junções de efeito de túnel
de MgO, ruído, fluxo lateral, séries de junções de efeito de túnel.
vi
vii
Abstract
In the modern times is clear the potential of the nanotechnologies, which are capable of
providing the necessary means for the detection and screening of pathogens in the form of very
sensitive micron size sensors.
The aim of this work is the fabrication of magnetoresistive (MR) sensors with linear response,
capable of performing tests of recognition by the detection of functionalized nanoparticles magnetically
polarized which had been previously linked to biomolecules. The magnetoresistive sensors used were
Magnetic Tunnel Junctions which present a high value of magnetoresistance. In addition, in this work
individual Magnetic Tunnel Junctions were connected in series, forming arrays of three hundred and
sixty individual elements. Such was done with the purpose of gain in the devices’ detectivity.
Several samples were patterned and magnetoresistances of the order of 70% were achieved,
as well as sensitivities of 1.24%/Oe in the linear range.
Keywords: Tunneling magnetoresistance (TMR) sensors, Mgo MTJ, noise, lateral
flow, MTJ series.
viii
ix
Contents
Acknowledgement ................................................................................................................................... iii
Resumo ....................................................................................................................................................v
Abstract................................................................................................................................................... vii
Figure List ................................................................................................................................................ xi
Tables List ............................................................................................................................................. xiii
1. Introduction ...................................................................................................................................... 1
1.1. Thesis’s framework – motivation ............................................................................................. 1
2. Theoretical background – Magnetism ............................................................................................. 3
2.1. Magnetic field and Magnetization ............................................................................................ 3
2.2. Magnetic Materials Classification ............................................................................................ 4
2.3. Micromagnetism for ferromagnetic material ............................................................................ 7
2.3.1. Exchange energy ................................................................................................................. 7
2.3.2. Magnetocrystalline Anisotropy – Anisotropic Energy .......................................................... 7
2.3.3. Shape anisotropy – Demagnetizing Energy ........................................................................ 8
2.3.4. Zeeman Energy ................................................................................................................. 10
2.3.5. Interlayer coupling forces .................................................................................................. 10
2.4. Magnetoresistance ................................................................................................................ 11
2.4.1. Anisotropic Magnetoresistance (AMR) .............................................................................. 11
2.4.3. Tunneling Magnetoresistance Effect ................................................................................. 15
3. MgO MTJ biosensors for immunomagnetic lateral-flow detection – System and sensor design .. 23
3.1. Reader system design ........................................................................................................... 23
3.2. Sensor design ........................................................................................................................ 25
3.2.1. Noise and detectivity implications .................................................................................. 26
3.2.2. Application: theoretical preview ..................................................................................... 28
4. Micro fabrication – Systems, techniques and processes protocols............................................... 29
4.1. Sputtering Deposition Systems: Magnetron .......................................................................... 29
4.1.1. Nordiko 2000 ................................................................................................................. 31
4.1.2. Nordiko 7000 ................................................................................................................. 32
4.1.3. Ultra High Vacuum-UHVII – Oxide Sputtering System.................................................. 33
4.2. Ion Beam Systems ................................................................................................................ 34
4.3. Pattern Transfer - Techniques and Processes ...................................................................... 36
x
4.3.1. Pattern Transfer Techniques ......................................................................................... 36
Etching ........................................................................................................................................... 36
Lift-off ............................................................................................................................................. 36
4.3.2. Direct Write Laser Optical Lithography (DWL) .............................................................. 37
4.4. Annealing setup ..................................................................................................................... 40
5. Characterization methods – Measurement techniques ................................................................. 43
5.1. Vibrating Sample Magnetometer (VSM) ................................................................................ 43
5.2. Manual Transport Measurement Setup ................................................................................. 43
5.3. Profilometer ........................................................................................................................... 45
5.4. Ellipsometer ........................................................................................................................... 46
6. Microfabrication – Process steps................................................................................................... 47
6.1. Stack deposition .................................................................................................................... 47
6.2. Bottom contact definition ....................................................................................................... 48
6.3. MTJ pillar definition ................................................................................................................ 50
6.4. Pillar lateral isolation .............................................................................................................. 51
6.5. Top contact definition – Metallization .................................................................................... 51
6.6. Sample insulation .................................................................................................................. 52
6.7. Sample dicing ........................................................................................................................ 52
6.8. Annealing ............................................................................................................................... 52
6.9. Encapsulation ........................................................................................................................ 52
7. Results ........................................................................................................................................... 53
8. Conclusions and Future work ........................................................................................................ 63
Bibliography ........................................................................................................................................... 65
A. Appendix ........................................................................................................................................ 67
B. Appendix ........................................................................................................................................ 69
Run Sheet for MTJs Array fabrication: Influenza_ 2010 ................................................................... 69
xi
Figure List
Figure 1.1: Lateral flow recognition. ........................................................................................................ 2
Figure 2.1: Diamagnetic material ............................................................................................................ 4
Figure 2.2: Paramagnetic material .......................................................................................................... 4
Figure 2.3: Ferromagnetic material ......................................................................................................... 5
Figure 2.4: Antiferromagnetic material .................................................................................................... 6
Figure 2.5: Ferrimagnetic material .......................................................................................................... 6
Figure 2.6: Magnetization and demagnetizing field ................................................................................. 8
Figure 2.7: a) single domain; b) 2 domains; c) 4 domains. ..................................................................... 9
Figure 2.8: Domain Wall .......................................................................................................................... 9
Figure 2.9: Néel coupling....................................................................................................................... 10
Figure 2.10: Oscillatory character of RKKY coupling ............................................................................ 11
Figure 2.11: Transfer curve of a GMR multilayer. ................................................................................. 12
Figure 2.12: Spin valve structure ........................................................................................................... 13
Figure 2.13: Scheme of electron scattering in a GMR device. .............................................................. 14
Figure 2.14: Magnetic tunnel jucntion structure. ................................................................................... 15
Figure 2.15: Couplings in the MTJ stack. .............................................................................................. 15
Figure 2.16: Tunnelling phenomenon .................................................................................................... 16
Figure 2.17: Scheme of a MTJ composed by a pinned layer and a free layer with parallel anisotropies
............................................................................................................................................................... 18
Figure 2.18: MTJ transport behaviour to the parallel anisotropy. .......................................................... 19
Figure 2.19: Scheme of a MTJ composed by a pinned layer and a free layer with crossed anisotropies
............................................................................................................................................................... 20
Figure 2.20: MTJ transport behaviour to the crossed anisotropy: linear response. .............................. 21
Figure 3.1: Scheme- Magnetoresistive lateral flow detection – wheel concept. ................................... 23
Figure 3.2: Wheel device prototype ....................................................................................................... 24
Figure 3.3: Mask used in this work ........................................................................................................ 25
Figure 3.4: MTJ series: array of MTJs; each individual element has a 30x2µm2 area. ........................ 26
Figure 3.5: 3D scheme of the MTJs array’s. ......................................................................................... 26
Figure 3.6: Expected output: particles magnetized vertically. ............................................................... 28
Figure 4.1: Scheme of a magnetron system. ........................................................................................ 29
Figure 4.2: Relation between the wear of a target and the magnetic field lines. .................................. 30
Figure 4.3: Nordiko 2000 deposition camber schematic. ...................................................................... 31
Figure 4.4: Nordiko 7000 - disposition of the all process modules, dealer and load-lock. .................... 32
Figure 4.5: Schematic side view of UHVII. ............................................................................................ 33
Figure 4.6: Picture of UHVII................................................................................................................... 34
Figure 4.7: Schematic view of a IBD system: Z configuration. .............................................................. 35
Figure 4.8: Etching process. .................................................................................................................. 36
Figure 4.9: Lift-off process ..................................................................................................................... 36
Figure 4.10: Picture of the SVG tracks. ................................................................................................. 38
xii
Figure 4.11: Optical lithography – photo-resist mask creation. ............................................................. 39
Figure 4.12: Figure of the DWL system. ................................................................................................ 39
Figure 4.13: Picture of the used annealing setup. ................................................................................. 40
Figure 4.14: Temperature cycle in the annealing process. ................................................................... 41
Figure 5.1: VSM system. ....................................................................................................................... 43
Figure 5.2: Electrical scheme for measurements with 2 (on the left) and four (the right one) probes. . 44
Figure 5.3: INESC-MN profilometer system .......................................................................................... 45
Figure 5.4: ............................................................................................. 45
Figure 5.5: Picture of the ellipsometer. .................................................................................................. 46
Figure 6.1: Standard MTJ stack used ................................................................................................... 47
Figure 6.2: Resulting photoresist after the 1st lithography revelation. ................................................... 48
Figure 6.3: Bottom contact definition: resulting structure after the ion milling. ...................................... 48
Figure 6.4: Picture of bottom contact. ................................................................................................... 49
Figure 6.5: Left - Wet bench’s picture: ultrasound machine and hot bath; Right - microstrip solution. . 49
Figure 6.6: Pillar junction definition. Left - After the 2nd
lithography; Right – After the 2nd
etch ............ 50
Figure 6.7: Picture of the pillar junction. ................................................................................................ 50
Figure 6.8: MTJ pillar and bottom contact insulation ............................................................................. 51
Figure 6.9: Top contact definition ......................................................................................................... 51
Figure 6.10: Picture of the top electrode. .............................................................................................. 52
Figure 6.11: Sample patterned and mounted on a chip-carrier............................................................. 52
Figure 7.1: Close view of one MTJ series with the 4 contacts identified as a, b, c and d. .................... 53
Figure 7.2: ZarMTJ1 transfer curve ....................................................................................................... 53
Figure 7.3: ZarMTJ2 transfer curve. ...................................................................................................... 54
Figure 7.4: ZarMTJ3 transfer curve ....................................................................................................... 54
Figure 7.5: ZarMTJ5 transfer curve. ...................................................................................................... 54
Figure 7.6: ZarMTJ1 and ZarMTJ3; measurement for 360 elements. .................................................. 55
Figure 7.7: The RxA of the sensors in ZarMTJ1 and ZarMTJ2 samples. ............................................. 55
Figure 7.8: ZarMTJ1 - TMR vs number of junctions. ............................................................................. 56
Figure 7.9: Transfer curve between contacts of two independent series. ............................................. 57
Figure 7.10: N2#1 transfer curve. .......................................................................................................... 58
Figure 7.11: N2#2 transfer curve. .......................................................................................................... 58
Figure 7.12: N2#4 transfer curve. .......................................................................................................... 58
Figure 7.13: N2#1 - TMR vs number of junctions ................................................................................. 59
Figure 7.14: TMR vs Vbias ...................................................................................................................... 60
Figure 7.15: TMR of a MTJ decreasing with the applied voltage .......................................................... 61
Figure 7.16: Graph - Response of one sensor under magneto nanoparticles excitation. ..................... 62
Figure .A.1: Mask used .......................................................................................................................... 67
Figure A.2: Top: Series detail ................................................................................................................ 68
Figure A.3: 3D scheme of MTJ’s connected in serious ......................................................................... 68
xiii
Tables List
Table 4.1: Deposition setpoints of Nordiko 2000. ................................................................................. 31
Table 4.2: Standard conditions operation in the used modules ............................................................ 33
Table 4.3: Deposition conditions of UHV2. ............................................................................................ 34
Table 4.4: Ion milling used set values for assist gun of both IBD systems. .......................................... 35
Table 7.1: Variations in the stack of samples with Si substrates. ......................................................... 53
Table 7.2: Variations in the stack of samples with glass substrates. .................................................... 58
xiv
1
1. Introduction
The last century saw the advent of non-classical science, beyond the limits of the Newtian
physics and Maxwell’s electromagnetism. These scientific progresses were instrumental in the
development of several technological areas, having allowed, among others, major advances in
electronics, whose importance has assumed a dominant role nearly in all the areas of human
intervention. In this context of technological progression a pre-requisite for the success of any system:
it’s contained area and volume. From this requirement born microelectronics. driving the need for a
continued miniaturization; Currently, it is no longer just the need of space that dictate the importance
of micro / nanotechnologies, which has extended from their traditional applications’ area in computer
systems and automated control systems, to the most recent biosciences. A family of small-scale
devices that is particularly interesting is the magnetoresistive family, whose applications include
sensors for read heads in hard disk drives, memories, biosensors and diagnostic techniques using
brain mapping.
In fact, in biosciences context, which are currently showing themselves as a scientific area
with major scientific and social impact, it’s clear the potential of the magnetoresistive (MR) sensors, as
a way of providing the necessary means for the detection and screening of health threatening
pathogens. Was in order an application under the biorecognition and detection that this work was
developed, using the advantages of a particular type of magnetoresistive sensors: the magnetic tunnel
junctions (MTJ’s) sensors – high spatial resolution provided by their small size, and higher sensitivities
and output values comparing with the others MR sensors.
1.1. Thesis’s framework – motivation
Done in the INESC-MN facilities, with special emphasis in the INESC clean-room, the work
done in this thesis is integrated in an international project which has, as ultimate goal, to develop a
new diagnostic tool for influenza virus detection based on an immunochromatographic assay with
functionalized magnetic nanoparticles as markers of the virus and ultra sensitive field transducers for
their detection and quantification.
Actually, the present methods for human influenza virus based on cell cultures have a
response time frame of some days, too long to help physicians to make clinically relevant decisions
during the first days of the disease, when the treatment will be more effective. There are already some
immnunochromatography based tests, which yield a response in less than 30 minutes, but they suffer
from low sensitivity (only 50 to 70% of true influenza cases are detected).
2
These rapid immunochromatographic tests are fast assays based on the immunological
capture of a ligand attached to an active microparticle when it flows over a solid support where an anti-
ligand has been immobilized.
The chosen approach for this project takes advantage of the exploitation of magnetic
properties of magnetic nanoparticles as active particle, which will be detected and quantified by ultra-
sensitive magnetic sensors. The first developments in field sensors lead to inductive readers, in which
an electrical response is induced in some coils placed around the field source: if a magnetic flux
variation through the coils exists, an electromotive force is created, like stated in Faraday’s law. The
coils method has a low sensitivity so in the project the coils are replaced by magnetoresistance based
sensors (magnetoresistive sensors).
Resuming, the final goal of the project is the creation of an integrated platform including a
system that enables the processing of samples allegedly virus contaminated and the system of
detection and quantification.
The working principle is like follows: first the testing sample containing the virus is put in
contact with magnetic nanoparticles functionalized with a specific antibody to the particular virus
strain. The fluid is then put in contact with a porous membrane strip; the fluid sample passes through
the membrane until the virus attached to the magnetic markers are recognized by membrane
immobilized antibodies; finally the presence of a particular virus strain is detected through the sensing
of the magnetic fringe field created by the bound magnetic nanoparticles, using a magnetic sensor. Is
worthy to note that this concept can be used to detect other virus than influenza one.
Figure 1.1: Lateral flow recognition. a) the fluid sample is put into contact with magnetic particles; b)virus is attached to nanoparticles antivirus; c) fluid sample migrates through the membrane strip.
3
2. Theoretical background – Magnetism [1],[2],[3],[12],[14],[17]
The magnetic field and magnetic properties of materials play a major role in the behaviour of
devices that were fabricated and studied during this thesis. Therefore this first section will introduce
and remember some basic concepts and definitions.
2.1. Magnetic field and Magnetization
A magnetic field is a field of force that arises from two sources: electric currents (generically
electric charge moving) and materials with an intrinsic nonzero magnetic moment (magnetic
materials). In fact, the magnetism of the matter is due to the movement of electrons from atoms, which
are responsible by the microscopic magnetization currents postulated by Ampère before knowing the
structure of matter, so we can see the common ground in these two sources of field: it results from
electric charge movement.
Because of the nonexistence of magnetic charges every magnetic field source has a dipolar
form (the first term appearing in the dipolar treatment of a distribution of magnetic sources is the
dipolar), and so, magnetic materials, being a magnetic field source, have a magnetic moment
associated just like current loops. Through the dipolar moment is possible to define the magnetization
of a material:
m is the magnetic moment of the material and V its volume.
The two types of sources contributing to the magnetic field can be related by the following
expression:
, µ0 free space permeability
with H taking into consideration the free currents influence, while B considers all the currents ( free
and magnetization currents; so B takes into account the magnetization of the medium).
The influence of the H field over the magnetization of magnetic materials is the information
contained in magnetic susceptibility:
4
2.2. Magnetic Materials Classification
Diamagnetism
The atoms of diamegnetic materials do not present a permanent magnetic moment in absence
of applied magnetic field, since they don’t have unpaired electrons on their completely filled orbital
shells. However, when a magnetic field is applied a small intensity response occurs: these materials
have a negative small susceptibility, so they tend to align the magnetic moment of their atoms against
the applied field.
The diamagnetic behaviour is present in every material, but due to its small magnitude is often
negligible or covered by stronger magnetic effects.
Paramagnetism
In the absence of a magnetic field, the atoms of a paramagnetic material have a permanent
magnetic moment due to the unpaired electrons on their partially filled shell. Since these atoms, as
magnetic dipoles, poorly interact with one another they get a random orientation due to thermal
agitation. In the presence of an external magnetic field the moments will gradually align along with it
as the intensity of the field increases. So paramagnetic materials have a positive, but small
susceptibility.
So, for both diamagnetic and paragmetic materials, when a magnetic field is applied a
magnetization proportional to that field is produced:
and so the magnetic field is:
Figure 2.1: Diamagnetic material: Left - In absent of applied field; Right - Under applied field.
Figure 2.2: Paramagnetic material in field absence (left) and under its influence (right).
5
Diamagnetic materials:
- ;
- is thermally independent;
- has a very small absolute value;
Paramagnetic materials:
- ;
- is generally small (however greater than the absolute values of diamagnetic materials): for
low temperatures when the thermal agitation decreases significantly paramagnetic materials
exhibit higher values.
Ferromagnetism
Ferromagnetic materials exhibit a large permanent magnetization even when a magnetic field
is not present. Like in paramagnetism, the atoms of ferromagnetic materials have unpaired electrons.
The main difference between these two types of magnetism lies on the stronger interaction existing
between ferromagnetic atoms – the exchange interaction. If a magnetic field rises, the individual
magnetic moments tend to align with the field, so the susceptibility is positive and typically large.
Above a certain temperature, the Curie temperature, ferromagnetic materials behave like
paramagnetic ones.
Figure 2.3: Ferromagnetic material – moments align without an applied field.
6
Antiferromagnetism
In antiferromagnetic materials the exchange interaction between their atoms results in
individual magnetic moments ordered in such way that moments of consecutive atoms in the lattice
are aligned anti-parallel. Since in these two sub-lattices the moments have an equal magnetic
magnitude, the net magnetic moment in the absence o field is zero. This kind of materials has a small
and positive susceptibility.
Antiferromagnetic materials behave like paramagnetic ones above the Néel
temperature (it is the analogous of the Curie temperature in ferromagnetism); each
ferromagnetic also has a characteristic blocking temperature above which the strong
interlayer exchange interaction vanishes and an antiferromagnetic layer loses its “ability” to
pin a ferromagnetic layer.
Ferrimagnetism
The magnetic moments exhibit alternate orientation like in ferromagnetism. However because
there is more than one type of magnetic ion in the material, not all magnetic moments have the same
absolute value. This means that they have a non null net magnetization.
Figure 2.4: Antiferromagnetic material – antiparallel alignment of the moments; no net magnetization.
Figure 2.5: Ferrimagnetic material
7
2.3. Micromagnetism for ferromagnetic material
Ferromagnetic and antiferromagnetic materials are key elements in the behaviour of the
devices focused by thesis. So to better understand their properties it is a good approach to make
some considerations about the interactions and the associated energies that occur in or with these
materials.
2.3.1. Exchange energy
The spontaneous ordering of ferromagnetic atomic moments is mainly due to the exchange
interaction of their atoms, which is independent of the total magnetic moment of the sample.
According to the theoretical atomic dipoles model for ferromagnets, the main difference
between ferromagnets and paramagnets is that the each permanent dipole of the ferromagnets
interact strongly with their nearest neighbours dipoles. The interaction energy between two sequenced
atoms is described by the Heisenberg model:
is the coupling constant between the spins of the neighbour atoms, and is usually considered
constant through a material and Si and Sj are the spins of each neighbour.
The total exchange energy of a material implies sum over all pairs of nearest neighbours;
generalizing to the continue approximation the exchange energy of entire crystalline lattice is
a is the lattice constant.
If the magnetization varies too rapidly in a short range the exchange energy will be very high,
so in order to keep it in a minimum value the magnetic dipoles have a preference to remain aligned
with each other.
2.3.2. Magnetocrystalline Anisotropy – Anisotropic Energy
Crystalline materials are magnetically anisotropic because there is a preferential direction for
orientation of the dipoles along the main crystallographic axis of the structure. This means that exist a
sort of internal field which forces the net magnetization to align with certain axis of the crystalline
structure, defining the so called easy axis or easy direction.
The magnetocrystalline anisotropy energy, Ek, is the work needed to rotate the sample
magnetization to a certain direction out of the easy axis direction, and is calculated as follows:
8
where K is an energy density, and α is the angle between the magnetization and the easy axis.
Is clear to see that this energy will have a minimum for α=0, which means that the
minimization of this energy implies the magnetization to prefer align with the easy axis. This
contributes to the “memory effect” of ferromagnets (after the magnetic field get off the material will
tend to align their dipole with the easy axis).
2.3.3. Shape anisotropy – Demagnetizing Energy
The demagnetizing energy corresponds to the interaction between the magnetization of the
material and the demagnetizing field, Hd. His expression follows below:
In order to understand the appearance of the demagnetizing field a good strategy could be to
establish the analogy with the electric field. A polarized material creates a distribution of electric
charges at the surfaces whose normal has the same direction of the polarization vector P. That
charges are responsible for the rise of an electric field/electrostatic potential (Poisson’s equation). With
the magnetization a similar phenomenon happens: on the surfaces, at the material boundaries (those
that are perpendicular to some magnetization component) the magnetization is no longer continuous,
and “magnetic charges” appear. These charges are the sources of the demagnetizing field that will
oppose to the normal magnetization of the material; this means that this energy tries to become the
magnetization parallel to “charged” surfaces.
Minimizing the demagnetizing energy corresponds to rotate the magnetic atomic dipoles of the
sample, so that the number of charges created on surfaces is minimal. That makes the material
become divided into different magnetic domains oriented in reverse directions. This way, the magnetic
charges formed by one domain will cancel the charges of neighbour domains, reducing Hd and Ed.
Figure 2.6: Magnetization and demagnetizing field: the field is greater the closer are the “magnetic charges”.
9
If the demagnetizing energy was the only one ruling the behaviour of ferromagnetic materials
they would divide themselves in as many domains as necessary to completely eliminate the
demagnetizing field resulting in a zero total magnetization.
However, indeed, the exchange energy avoids this to happen, because it has the inverse
effect: its minimization cause the atomic moments of neighbours to align with each others. So, it is the
balance of these two energies is responsible by the creation of magnetic domains separated by
domain walls, which are regions where the orientation of the magnetic dipoles that belong to them is
not constant, rotating from the orientation in the first domain to the orientation of the second (Figure
2.8: Domain WallFigure 2.8). The thickness of these domain walls is also influenced by the anisotropy
energy: this energy forces the moments to align by the easy axis, so it would “prefer” no walls at all,
but the exchange energy would prefer a wall so big as the material so that neighbour dipoles would be
as parallel as possible; so anisotropy energy will try to create a thin wall and exchange energy will try
to enlarge it.
It is important to note that wasn’t written that there are magnetic monopoles. Isolated magnetic
charges doesn’t exist ( ), so that charges referred above always appears in pairs constituting a
dipole. In fact if one tries to split the north pole from the south pole of a magnet will end with two
smaller magnets, each with one north and one south pole.
Figure 2.7: a) single domain; b) 2 domains; c) 4 domains.
Figure 2.8: Domain Wall
10
2.3.4. Zeeman Energy
There is one last energy term that describes the magnetic properties of a ferromagnet: the
Zeeman energy. The Zeeman energy is related to the interaction between the magnetization of a
material and the external applied field (Ha) and is given by:
This energy represents the amount of work necessary to rotate the magnetization by a certain
angle with respect to the applied field, and would be a minimal value when the magnetization became
align along the magnetic field direction.
2.3.5. Interlayer coupling forces
2.3.5.1. Néel Coupling
It is right to say that the Néel coupling is present in every ferromagnetic interlayer
discontinuity, because it is associated with roughness of the ferromagnetic interfaces (it is reasonable
to say that exist always some level of roughness in an interface). The magnetostatic interactions
between the “magnetic charges” at the ferromagnetic interface cause a ferromagnetic coupling.
The Neel energy is given by:
2.3.5.2. RKKY Coupling
Ruderman-Kittel-Kasuya-Yodsida coupling or indirect oscillatory exchange interaction is an
exchange interaction between two ferromagnetic layers separated by a non magnetic spacer. The
wave functions of the atoms of the two ferromagnetic layers interact with each other in a way mediated
by the spacer, whose thickness is responsible for the oscillatory character of the interaction between a
ferromagnetic and an anti-ferromagnetic coupling.
Figure 2.9: Néel coupling.
11
2.4. Magnetoresistance
Magnetoresistance is the property of a material to change the value of its electrical resistance
when the value of the applied magnetic field changes. The magnitude of this effect can be expressed
numerically as follows and is presented as a percentage:
2.4.1. Anisotropic Magnetoresistance (AMR)
The AMR effect consists in the change of electrical resistivity with the orientation of
magnetization in respect with the direction of the electrical current in the material. Thus, the resistivity
of a magnetic material depends on the angle between the magnetization and the current direction and
can be described by the following equation:
For most materials the resistance is maximized when the current is parallel to the
magnetization and minimized when it is perpendicular.
To measure the amplitude of this effect (to know how much the resistivity varies) is necessary
measure the maximum and the minimum, so it is necessary to saturate the magnetization in parallel
and perpendicular directions. With these two values it’s possible to calculate the anisotropic
magnetoresistivity ratio:
Typically AMR is somewhere between 2 a 6%.
Figure 2.10: Oscillatory character of RKKY coupling with the spacer thickness. The strength of the ferromagnetic and anti-ferromagnetic coupling decreases with the spacer thickness.
12
2.4.2. Giant Magnetoresistance (GMR)
Giant magnetoresistance is an effect observed in thin film structures composed of alternating
ferromagnetic and non magnetic layers. The observed effect is a significant change in the electrical
resistance of such structures depending on whether the magnetization of adjacent ferromagnetic
layers are parallel or anti-parallel aligned. The overall resistance is low for parallel alignment and high
for anti-parallel alignment. Therefore the intensity of the GMR effect is given by:
Multilayer GMR
In this structure two or more ferromagnetic layers are separated by a non-ferromagnetic
spacer (FM/AFM/FM structure - e.g. Fe/Cr/Fe – the first GMR sample). To a certain thickness of the
spacer, the RKKY coupling between the adjacent ferromagnetic layers becomes anti-ferromagnetic,
making it, in the absence of a magnetic field, energetically preferable for the magnetizations of
adjacent ferromagnetic layers to align anti-parallel. When an external filed is applied and reaches a
certain value which makes energetically favourable to break the inter layer coupling the ferromagnets
will align parallel to each other and parallel with the field. This rotation and alignment of the
ferromagnetic layers are followed by a drop in the electrical resistance of the structure. In the
multilayer configuration with 10 or more stacks MRs of about 10% were achieved.
Figure 2.11: Transfer curve of a GMR multilayer. The red arrows represent the magnetization of each layer to different applied fields.
13
Spin Valve
In spin valve GMR two ferromagnetic layers are separated by a thin non-magnetic metallic
layer spacer, and an antiferromagnetic material is contiguous to one of the ferromagnetic layers (the
pinned layer).
In the spin valve configuration the electrical current flows in plane with the layers: Current In
Plane - CIP geometry.
The antiferromagnetic material is an exchange bias, which sets an exchange anisotropy at the
interface with the ferromagnet by the exchange interlayer coupling. This way the anti-ferromagnetic
material sets and pines the magnetization of the ferromagnetic material providing a fixed reference.
The other ferromagnetic layer is free to rotate with the applied field and align along it. With this
configuration the spin valve resistance has a maximum value when the magnetization of the free layer
and the pinned layer are anti-parallel and a minimum value when they are parallel.
It is the spin-dependent conductance in the ferromagnetic layers that supports the GMR effect.
This spin dependence can be explained by the so called “two channel model”. This model states that
the spin-flip scattering process that takes place in ferromagnets have a much larger time-scale than all
other processes contributing to electrical resistance. Thus it is assumed that the electrons conserve
their spin, and that there are two distinct families of electrons, which are distinguished by their state of
spin and can be considered independently. In GMR devices most of the current will flow through the
non-magnetic material, which is usually the layer with the lowest resistivity. In a non-magnetic material
the population of conduction electrons, the ones with energy close to Fermi level, is constituted by an
equal number of electrons with spin up and with spin down. Since the scattering of electrons is mostly
dominated by collisions between electrons of the same spin is possible to understand that in nom-
magnetic materials there is no spin orientation granted with a higher scattering rate. This is so while
the electrons are confined to the non-magnetic material, but when they approach the interface
ferromagnet/non-ferromagnet that is no longer true because the density of states at the Fermi level is
asymmetric with respected to the spin. This means that according to the ferromagnet magnetization,
near the Fermi level the number of spins aligned parallel with the magnetization direction will be higher
Figure 2.12: Spin valve structure; the current flows in the layer’s plane.
14
than the number of spins with the anti-parallel alignment. As result, and considering, as already
mentioned, that the scattering is dominated by collisions between electrons with the same spin, when
the electrons enter the ferromagnet the scattering probability will increase for one of the spin directions
and decreased for the other, depending on the magnetization direction. When the magnetization of the
two layers are anti-parallel the two spins orientations are scatter at the same rate (scattering of one of
the spin orientations is increased in one of the ferromagnet and the other spin orientation scattering
increase in the other ferromagnet; the same applies to the complementary spin orientation in each
ferromagnets, which the scattering rate decreases). If the magnetizations of both ferromagnetic layers
are parallel, the GMR device structure will no longer be symmetric with respect to spin orientation.
Now the two ferromagnetic layers will have a larger population of one of the spin types (the same type
for both ferromagnets) in the vicinity of the Fermi level depending again on the orientation of the
magnetization. The scattering for this particular spin type will be strongly decreased and its
contribution for the current flowing through the device will be very high. To the opposed spin direction
the scattering will be increased and its contribution for the current very small.
The typical MR signal of a spin valve reaches the 9%.
To a better understand this model the relation between the band structure of a three layer
ferromagnetic / anti-ferromagnetic / ferromagnetic device and electron scattering it can be seen the
figure bellow, which have at the right side the equivalent resistor model (Figure 2.13).
Figure 2.13: Scheme of electron scattering in a GMR device.
15
2.4.3. Tunneling Magnetoresistance Effect
Magnetic Tunnel Junction (MTJ)
A magnetic tunnel junction is a structure constituted by two ferromagnetic layers separated by
an insulator layer thin enough to allow the electrons tunneling. In the MTJ’s structure the current flows
perpendicularly to the layer’s plane (current perpendicular to plane geometry – CPP geometry).
Figure 2.14: Magnetic tunnel junction structure; the current flows perpendicularly to the planes.
Like in the spin valve configuration one of the ferromagnetic layers is pinned, providing a fixed
reference while the other is free to move under an external field application.
In the MTJs used in this work the pinned layer role is performed by a synthetic anti-
ferromagnetic (SAF) structure, which is composed by two ferromagnetic layers separated by non-
magnetic one. With this SAF structure is created an alternative way to have a pinned layer to the
simple anti-ferromagnet / ferromagnet coupling. Hence, the anti-ferromagnetic layer is a exchange
bias material and pines the Ferromagnet Layer 1 which pins the Ferromagnet Layer 2 by RKKY
coupling, which competes with the ferromagnetic Neel coupling, several orders of magnitude inferior to
RKKY coupling ( Figure 2.15). This way a stronger pining is achieved than using just the inter layer
exchange coupling.
Figure 2.15: Couplings in the MTJ stack.
16
Being strictly forbidden in classical physics, the tunnelling phenomenon rises from the double
corpuscular-wave character of the electrons. The wave-like behaviour raises a non-null probability of
the electrons cross a trough an energetically forbidden barrier (the oxide insulator in the MTJ
structure).
When no voltage is applied (V=0) between the two ferromagnets (the free and the pinned
layer), the Fermi level of the two electrodes is the same and no current flows. If a voltage is applied
between the electrodes the energy levels of one electrode will be shifted eV relatively to the energy
levels of the other electrode, and a current start to flow through the barrier. This means that electrons
in the filled states near the Fermi level at the ferromagnet with the higher energy levels (FM1) tunnel
trough the insulator barrier to the empty states of the lower energy level ferromagnet (FM2)
Tunnelling Magnetoresistance
The first hypothesis in the model used to explain the tunneling magnetoresistance is that the
spin is conserved during the tunneling.
Assuming that the spin is conserved a spin up and a spin down current can be discriminated,
because spin up electrons coming from one ferromagnet only will be accepted by the unfilled spin up
states of the other ferromagnet and the same to spin down electrons.
The second hypothesis is that contribution to the tunneling current is directly proportional to
the density of states in FM1 times the density of states in FM2 at the Fermi level.
This phenomenon is well described considering the conductance, G, through the barrier.
In ferromagnets materials there is an imbalance in the density of states of spin up (U) and spin
down (D) direction that creates the magnetic moment in these materials.
When the two electrodes have a parallel magnetization, and supposing, for instance, that in
FM1 the majority of electrons with energy near the Fermi level have spin up they will tunnel to the
unfilled spin up states of the FM2 in which are also in majority (since the electrodes have the same
magnetization); regarding the spin down electrons will travel from a minority band to a minority band
(the band, which have a lesser density of states). So the conductance for the parallel configuration can
be written as:
where Ui and Di are the spin Up and spin Down densities of state in the ferromagnet i (i=1,2).
Figure 2.16: Tunnelling phenomenon
17
In the case of an anti-parallel magnetization of the ferromagnets the electrons are tunneling
from a minority band for a majority one, and vice versa. Thus, the conductance between electrodes
with parallel magnetization is higher than in the anti-parallel configuration.
Then the conductance in anti-parallel configuration is:
The difference in density of states at the Fermi level is expressed by:
P is the tunneling spin polarization.
The conductance change is:
Knowing that the resistance is the inverse of the conductance (R = 1 /G) we get to the
resistance change relatively to its lowest value (this is the TMR definition):
TMR is usually presented as a percentage which implies to multiply the previous expression
by 100. It is possible see that TMR higher for higher polarized electrodes, so they should highly
polarized materials.
Since the resistance is the inverse of the conductance from is possible to state that
a MTJ has a low resistance state when the ferromagnets have their magnetization parallel to each
other, and a high resistance state when they are in the anti-parallel configuration.
18
2.4.3.1. MTJ output response – linear and square signals
The resistance variation between the maximum and minimum values in a MTJ can fallow two
different rules: the MTJ can have a square of a linear response to the magnetic field. Linear responses
are used in sensors type MTJ, and the square behaviour is useful in data storage industry in order to
fabricate memories.
To obtain the desired response there are two distinct “easy axis configurations” to the MTJ: a
parallel anisotropy and a crossed one.
Parallel Anisotropy
The pinned and the free layer have both the same easy axis orientation:
The energy terms of the free layer for this configuration are:
Zeeman term:
Crystalline anisotropy term:
Demagnetizing field of the free layer term:
Demagnetizing field of pinned layer term:
Neel energy term:
The total energy of the free layer is:
In the previous equation is the magnetization of the free layer,
is the demagnetizing
field of the free layer and is the demagnetizing field of the pinned layer, which is given by
where N the demagnetizing factor and depends on the geometrical shape; finally we have
.
Figure 2.17: Scheme of a MTJ composed by a pinned layer and a free layer with parallel anisotropies.
19
In order to this energy to have a minimal value it is required that:
and
So the derivatives are:
For
there are 3 possible solutions:
1-
2-
3-
Substituting the first solution in the second derivative:
As the resistance variation of an MTJ follows the law the solutions of the
free layer’s energy minimization can show the transport behavior of such device. Two different cases
are obtained (Figure 2.18).
Figure 2.18: MTJ transport behaviour to the parallel anisotropy. Left: square response; Right: linear response.
Concluding, in parallel anisotropy is possible to obtain hysteretic transfer curves if ,
or linear curves if
. Therefore, if is desired a linear response from sensors that use these
parallel orientation for the easy axis of the free and fixed layer a strong demagnetizing field is required,
which is achieved by ensuring that the sensors present a large aspect ratio.
20
Crossed Anisotropy
The easy axis of the free layer has a direction perpendicular to the orientation of the easy axis
of the free layer:
Figure 2.19: Scheme of a MTJ composed by a pinned layer and a free layer with crossed anisotropies.
The energy terms of the free layer for this configuration are the same as previous with
exception to the anisotropy energy term, which is now , and so the total energy of the free
layer is:
Doing the same analysis as before the solutions that minimize the energy are:
1-
2-
Substituting the first solution in the second derivative:
So in the case where the pinned and free layer have crossed anisotropies the sensor
response will be always linear.
21
Figure 2.20: MTJ transport behaviour to the crossed anisotropy: linear response.
22
23
3. MgO MTJ biosensors for immunomagnetic lateral-flow
detection – System and sensor design [17],[19]
3.1. Reader system design
The sensitive part of the system is a magnetoresistive sensor which will detect the fringe field
produced by the magnetic particles. In order to the sensor identify the field created by the band with
the particles it is necessary that the band passes close to the sensor. This sweeping movement can
be done using two different methods:
1- Is used a static MR sensor on chip and the test strip mounted on a plastic wheel is brought in
contact with the sensor surface as it passes through it.
2- The second configuration uses also a static MR sensor mounted in a tape head and the
immobilized particles are pressed and passes through the read head (like a magnetic tape
passing over a read head).
Figure 3.1: Scheme- Magnetoresistive lateral flow detection – wheel concept; Picture- Close view of
the real wheel device.
In the case of the wheel device it is an functional prototype developed in INESC-MN and is
currently at INA (Instituto de Nanociencia de Aragón – Nanocience Institute of Aragon).
24
The so-called magnetic particles used in this project don’t have a magnetic moment per si;
they are paramagnetic particles so their magnetic moment in the absence of an external magnetic field
is null, meaning that they do not create any field. In order to magnetize the nanoparticles the wheel
device has a set of electromagnetic coils and a permanent magnet: the coils magnetize the particles
horizontally while the magnet magnetize them vertically.
Electromagnetic
coils
Wheel
MR sensor
Figure 3.2: Wheel device prototype
25
3.2. Sensor design
For this project MTJ devices are used as sensors (not as memories) it is required a linear
response and a low noise level. These two factors have conditioned the MTJ samples design (Figure
3.3).
In the mask used the are 21 sensors series, each with 360 individual elements (24 columns
with 15 elements each), placed in a central position of the die sample; 2 contacts are at the sides so
the stripe can be sweep freely and without any constrains over MTJ’s (such as wires bonding the
contacts to the chip-carrier). The mask also presents test structures: the two single element MTJs on
the right lower corner allow the comparison between their RxA and TMR values and RxA and TMR of
the series.
In order get a linear response the free layer of the sensors have a large aspect ratio: each
individual element is a rectangular pillar with 2 by 30 µm2. Concerning the noise level the MTJs were
arranged in arrays, constituting therefore series of MTJs so a lower detectivity could be achieved
Figure 3.3: Mask used in this work: sensors at the center and contacts at the sides (Appendix A for greater detail).
26
Figure 3.5: 3D scheme of the MTJs array’s: 4 individual MTJs linked in serious.
3.2.1. Noise and detectivity implications
Noise is a physical phenomenon present in every electrical circuit. Being generated by every
components present in circuits, either passive or active, it manifests itself by voltage fluctuations
across components and fluctuations in the current that flows through them.
The MTJs sensors signals are characterized by a transfer curve of their resistance as function
of the field. To get the value of their resistance at some applied field a current is forced through the
MTJ and the voltage at its terminals is measured. Consequently, in the specific case of MTJs, the
voltage fluctuations have the consequence of adding an uncertainty to the output value and so to the
resistance value. Thereby, the minimal variation in the magnetic field which can be detected depends
directly on the noise level.
The detectivity, D ( , is defined as
and is the minimum magnetic field magnitude that can be detected. It is directly proportional to the
output noise of the sensor at a certain band width, SV ( , and ΔV/ΔH is the slope of the transfer
curve at linear region (is the sensor sensitivity). The TMR of MTJs decreases with the applied bias
voltage, so the voltages applied should be small in order to operate with the maximum TMR ratio (at
the maximum TMR ratio of one sample we get the minimum D value).
Figure 3.4: MTJ series: array of MTJs; each individual element has a 30x2µm2 area.
=
27
The square of the spectral noise voltage density of in each individual junction is:
In this last equation the first term represents the white nose and the second term the 1/f noise,
where is the electron charge, I the baising current, V the voltage between the sensor electrodes of
an individual junction, T the temperature, kB the Boltzman constant, α the Hooge like parameter, A the
junction’s area and f the is the frequency. The white noise incorporates the thermal noise, which
results from the random thermal motion of electrons, and the shot noise, which results from the current
flowing trough discontinuities in the circuit, and is given by the Hooge model like expression.
If we consider a device with N junctions in series, each junction with a resistance r, with a
driven current I equal for all elements we have Voltage = N R I. Therefore the square of spectral noise
voltage density of such a series is given by:
For the same configuration (an array of N individual junctions) the sensitivity is:
So the square of the detectivity is:
and in the limit V<<kBT, the detectivity is then:
This last equation shows that for an array of N tunnel junctions the value of D decreases with
, meaning that the minimum field which is possible to detect is decreasing with this factor. This is
so because the square of the noise spectral density increases with N while the sensitivity (not the
square of the sensitivity) increases with N.
28
500 1000 1500 2000 2500 3000displacement m
0.0005
0.0000
0.0005
0.0010
V V
Functionalized stripe
Nanoparticle
3.2.2. Application: theoretical preview
In order to predict the outcome of a measurement of the magnetic particles over an array of
magnetic tunnel junction was assumed that the particles can be approximated to a magnetic dipole.
Also it was assumed that the particles are perfectly spherical and that the dipole center is at the
geometrical center of the particle. The magnetic field created by a magnetic dipole at the position r
from the dipole center is as follows:
As already described the magnetization of the particles can be either vertical or horizontal
depending on the use of the electromagnetic coils or the permanent magnet existing in the reader
system prototype.
In this thesis the theoretical prediction was made only for particles magnetized perpendicularly
to the sensor plane (vertical magnetization – along the zz direction) since this is the configuration
being currently tested with the reader device at INA.
As the magnetoresistive sensors used (which are indeed the focus of this work) only sense the
transverse in-plane component of the field created by the particle, only this component is relevant,
which was arbitrarily identified like the xx component.
Considering a spherical particle of radius r and magnetic moment
the xx
component of the magnetic field created is given by:
For the calculus of the expected behaviour of an array of MTJs was considered that the
magnetic particles had a susceptibility of 4.81, a 125 nm radius, and that they are placed at the
margins of the functionalized stripe. It was also assumed that the magnetic field at particles position
was a 5000Oe vertically oriented field (the correct technical data was not available). It was considered
a sensor series with 148 µm wide, and a pulling movement of a 740 µm wide stripe over the sensors.
The xx coordinate of the initial point was 1184 µm distant from the sensor. Due to the limitations of the
available computer and the highly time/processor consuming program only some points were
calculated.
Figure 3.6: Expected output: particles magnetized vertically (z coordinate =1000Å).
(µm)
V (V)
29
4. Micro fabrication – Systems, techniques and
processes protocols [13],[15],[16]
4.1. Sputtering Deposition Systems: Magnetron
Sputtering is a physical deposition process, whose foundations are based on the momentum
transfer between ions in a plasma a given target surface, against which they are accelerated. One way
of being able of deposit films by sputtering is using a magnetron system.
Figure 4.1: Scheme of a magnetron system.
The sputtering happens in a vacuum chamber, which contains the magnetron with the target
of the material to deposit and the substrate where it is intended to deposit. In the chamber is
introduced a gas (in a controlled way), which is subject to an electric field. This field is created by
applying a negative voltage to the target, which accelerates ions, always present in the gas despite
their overall neutrality. Depending on your energy, these ions are able to ionize neutral atoms by
colliding with the. These just-ionized atoms, are now charged particles, and so they starts to feel the
electric field originated by the target loads, thus they accelerates towards it, and participate also in the
ionization of new atoms, in what is a cascade process of successive ionizations. In the mean time the
electrons released in each ionization event also participate in this process, while having the
energy/momentum needed. These electrons would be repelled by the target; however thanks to the
presence of permanent magnets positioned behind the target, the electrons are confined to
trajectories in the vicinity of the target, along the field lines of the magnetic field. Thus, the process
described above, through the collisions of the considered species (electrons and positive ions), will
possibly results in the ignition of the plasma near the target. The magnetic confinement allows this to
happen with a lower concentration of atoms and therefore the pressure lower; however the use of
permanent magnets has the disadvantage of causing an irregular wear on target, revealing more
pronounced in the area where the ionization rate is higher, which corresponds to the region where the
magnetic field lines are parallel to the target surface (Figure 4.2)
30
Figure 4.2: Relation between the wear of a target and the magnetic field lines.
The ions that reach the target have energies/momentums in a wide range, depending on the
history of its route (when the ionization happened and the collisions they suffered), so the energy
available to be transferred to the target also belongs to a quite large range. Consequently the result of
a collision with the target is variable and directly dependent on the energy of ions. To lower energies
the impact of ions results in a simple perturbation of the structure of the target, resulting in an atomic
rearrangement that generates heat, which dissipates in the water cooling system existing on
magnetrons. In an intermediate energy range, the impact can cause the ionization of target atoms,
causing the ejection of electrons, captured by the magnetic field and therefore involved in the process
of ionization of plasma. Finally, for energies equal or greater than the bidding energy of target’s atoms,
the collision results in the emission of several of these atoms. Once they are neutral particles, the
target atoms travel freely over the magnetron chamber, and they are deposited on the substrate
forming a film that was intended to deposit.
This type of systems presents a major handicap concerning atomic species responsible by the
target sputtering: ions are charged particles, so that if the target is not conductor, the charges will
accumulate at the target, so after a while the field generated by them will shield the target and even
repel the other ion, stopping the sputtering process; this problem does not arise in case the target is a
conductor, because the ion charge is neutralized by electrons from the source of power. The solution
to this accumulation is doing the target biasing with an RF source instead of applying a DC voltage:
on the negative phase of the signal ions are attracted to the target, when it happens sputtering and
positive phase those that have accumulated on the target are neutralized.
The substrate, usually grounded, may also be applied an RF voltage. In this case, the
substrate becomes a target with which ions impact, thus making it possible the etching of the
substrate (sputter etch), which is useful both in the definition of structures at micro as in control
roughness of the deposited films.
31
4.1.1. Nordiko 2000
The Nordiko 2000 is an automated deposition sputtering system constituted by a deposition
chamber and a load-lock, which allows samples to be loaded and unloaded without breaking the
vacuum inside the main chamber (typical system pressure: 7x10-8
Torr).
The deposition chamber has six magnetrons positioned on top of the substrate table, with 3
inch diameter each. Between the substrate and the magnetrons is the shutter, a rotating disk with a
slot that can be placed beneath the targets. The substrate table has 12 slots for substrate holders,
spread by four quadrants of 3 slots each. These, three are water-cooled and electrically grounded,
and one is heated and electrically isolated from the rest of the table so that can be connected to
ground or to a power supply allowing surface treatment by sputter etch. One slot in one of the water-
cooled quadrants has an array of magnets, providing a 30 Oe magnetic field for easy axis definition
when magnetic materials are deposited.
Target # Material Magnetron parameters Ar Flow (sccm) Pressure (mTorr) Dep. rate (Å/s)
1 Mn60%Pt40% DC:40mA 9 5.0 0.64
2 Ru DC:45mA 8 5.0 0.44
3 Ta DC:45mA 10 4.5 0.52
4 Co40%Fe40%B20% DC:45mA 9 5.0 0.44
5 MgO RF:130W 9 18.0 0.142
6 CoFe RF:35W 8 5.0 0.36
Table 4.1: Deposition setpoints of Nordiko 2000.
Figure 4.3: Nordiko 2000 deposition camber schematic.
32
4.1.2. Nordiko 7000
Nordiko 7000, like Nordiko 2000, is also a fully automated commercial magnetron. Being able
of handle 6 inch industrials wafers, consist in a central dealer camber, connected to four process
modules, and a load-lock (Figure 4.4).
Figure 4.4: Nordiko 7000 - disposition of the all process modules, dealer and load-lock.
In the load-lock, pumped by a turbo pump, is achieved a “primary” vacuum, with a base
pressure about 5.0x10-6
Torr. The dealer and each process chamber are pumped with cryogenic
pumps, reaching base pressures of the order of 5x10-9
Torr. A robot arm, placed inside the dealer,
carries the wafers from the load-lock to each process module (Figure 4.4).
As can be seen in the previous figure, each module has a different purpose.
Module 1 – Used to perform flash annealing processes (wasn’t used in the present work).
Module 2 – Perform a soft sputter etch, usually used to remove small a thickness of natural oxidized
material, before the deposition of TiW(N) and Al; a argon (Ar) plasma and two RF power sources are
used, one applied directly to the target and primarily responsible by speed up the ions towards the
target (RF1 source), and another source exclusively responsible for the plasma maintenance by
promoting his ionization.
Module 3 – Carries out a reactive sputtering deposition of TiW(N). During a reactive sputtering at least
one gas is introduced in the process chamber with a purpose beyond the maintenance of the plasma.
In this case, the deposition is made from a TiW target with a mixed Ar/N2 plasma, wich provides the N2
required to react with the target material and become incorporated in the deposited film. This material
was used has passivation layer because:
- Is a hard material, protecting the sample against physical damaged;
- It is dense, offering a good chemical protection against corrosion caused by substances used
in sample fabrication (like the revelator used after the lithography);
- Is a dark metal, so it’s used as anti-reflective coating for the direct opto-lithography, minimizing
reflected light.
33
Module 4 – Used to the deposition of Al98.5Si1.0Cu0.5. It’s used an Ar plasma powered by a DC power
supply.
The plasma is maintained with a DC power supply, and the standard deposition conditions can
be seen in the table above.
Operation RF1 RF2 DC Pressure Gas Flow
Module2 Soft Sputter
Etch
70W 40W - 3.0mTorr 50 sccm Ar
Module3 TiW(N) - - 0.50kW 3.0mTorr 50 sccm Ar+10 sccm N2
Module4 Al - - 2.0kW 3.0mTorr 50 sccm Ar
Table 4.2: Standard conditions operation in the used modules
4.1.3. Ultra High Vacuum-UHVII – Oxide Sputtering System
Another sputtering deposition machine, UHVII is used to deposit Al2O3. This system, a 6 inch
magnetron, has a single chamber, so there is no load-lock, and so it isn’t possible to load or unload
samples without vent the process chamber; this has a consequence materialized in the time need to
reach the optimal 3x10-7
base pressure for deposition: roughly 12h hours. An Ar plasma powered by a
RF source sputters a ceramic Al2O3 target placed above a 6 inch static and water-cooled substrate
table. Obviously 6 inch wafers can be used in UHVII, as well as smaller area samples; trough this 6
inch usable area the uniformity of the deposition varies about 10% between the center and the edge of
the table.
Figure 4.5: Schematic side view of UHVII.
34
Figure 4.6: Picture of UHVII.
4.2. Ion Beam Systems
A Ion Beam Deposition (IBD) systems can be used for deposition of thin films and for ion
beam milling, which consists in a non selective dry etch process. In IBD systems, a highly energetic
ion beam is used to remove material from a target that will deposit on a substrate a thin film in a
deposition process, or used to remove material from the sample itself in the etching process. The ion
beam is created by an ion source, called ion gun, within which is created a plasma by an RF power
supply. The ions are extracted from the gun by a set of three charged grids, which pull out the ions
from the plasma and accelerate them towards the vacuum chamber as a uniform collimated beam that
will impact a target.
There are two ion guns in the IBD system: the deposition gun and the assist gun. The
deposition gun points to the targets made of the material to be deposited. In the deposition process
the ions that hit the target are less energetic than in the sputter process, implying a lower deposition
rate.
In the assist gun, the beam points directly to the substrate and can be used to ion milling
process. Besides assist and deposition guns, there are two neutralizer guns inside the chamber, which
emit electrons to neutralize the accumulated ions on the target and on the insulating substrates .The
assist and deposition guns, the substrate and the target are disposed in a Z configuration.
Material RF Pressure Gas Flow Deposition Rate (Å/s)
Al2O3 200 W 3 mTorr 45 sccm Ar ~0.1905 at the center ~0.1714 at the edge
Table 4.3: Deposition conditions of UHV2.
35
Figure 4.7: Schematic view of a IBD system: Z configuration. The substrate table has a permanent magnet array mounted around it, producing a 40 Oe
magnetic field that defines the easy axis during the deposition. This table can be rotated depending on
whether a deposition or an etch is being made. The substrate holder also rotates during deposition
and milling processes in order to achieve a better uniformity throughout the sample. The target holder
has a hexagonal prism shape with one different target in each face, and rotates according to the
material that will be deposited, letting a specific target exposed to the ion beam, while the others
targets remain protect from contamination by a shutter.
There is also another shutter protecting the sample until assist and deposition guns have the
ion beams stable, accordingly to the set parameters. The chamber vacuum is obtained with a turbo-
molecular pump and a cryogenic pump, achieving a working base pressure of 10-8
Torr.
There are two IBD systems in INESC, the Nordiko 3600 and the Nordiko 3000, both installed in a
class 100 clean-room. They are both very similar to each other, where the main difference is that
N3600 is a much larger system being capable to handle 8 inch substrates, instead of the 6 inch
provided by N3000. During this work both facilities were only used as ion milling tools.
System operation RF V+ V- Gas Flow I
N 3000 Ion milling 58W 500V 200V 8 sccm Ar 30 mA
N 3600 Ion milling 160W 735V 350V 10 sccm Ar 105 mA
Table 4.4: Ion milling used set values for assist gun of both IBD systems.
36
4.3. Pattern Transfer - Techniques and Processes
For the fabrication of devices, and considering they are allowed with sizes ranging from
microns, it’s fundamental the ability to selectively remove or deposit material from a substrate. In this
work were two the techniques used to do so, namely lift-off and etching techniques.
4.3.1. Pattern Transfer Techniques
Etching
This process starts with a substrate in which was deposited the material to be patterned. That
material is selectively covered with a mask, so, this way, fractions of the sample become protected.
Then the not protected material is removed, and at the end the mask is itself removed, leaving the
substrate with the desired shape.
1) 2)
3) 4)
Lift-off
As well as in etching, also in lift-off process the ultimate goal is to obtain a material with a
certain pattern. Again, it still is used a mask but it is laid on a substrate without the material intended to
shape; that material is deposited over the substrate with the mask already on it, and then the mask is
removed taking with it all the material covering it, resulting in a substrate with a patterning material and
free material areas where previously was laid the mask.
1) 2)
3) 4)
Figure 4.8: Etching process.
Figure 4.9: Lift-off process
37
4.3.2. Direct Write Laser Optical Lithography (DWL)
In order of being able perform any of the two techniques described above a mask must be
transferred to the substrate. There are two types of masks that can be used, hard masks and soft
masks, and several methods to create and apply them. In this work all the masks were created using
direct write laser optical lithography process which includes mask design, vapour prime, photo-resist
coating, lithography exposure and photo-resist development. The masks were designed in a CAD
software, converted in a set of binary files and transferred to the lithography system hard drive.
4.3.2.1. Vapour prime system
This is a system placed in the clean-room that helps to ensure that the optical lithography is
successfully done. Basically the vapour prime system is an oven, pumped to remove any unwanted
material and heated to prevent any residual water molecules of remaining in the sample surface.
When the oven is completely pumped down a gas is introduced in his chamber: hexamethyldisilizane
(HDMS), is inserted to promote a better adhesion of the photo-resist onto the sample; after 5 minutes
in this hexamethyldisilizane atmosphere approximately one monolayer of HDMS will be coated at the
sample surface. The steps that compose the process which the sample undergoes during the vapour
prime are described in detail below (program 0).
1. Wafer dehydratation and purge oxygen from the chamber. Vacuum, 10 Torr, 2 minutes.
N2 inlet, 760 Torr, 3 minutes x 3
times.
2. Priming: Vacuum, 1 Torr, 3 minutes.
HDMS, 6 Torr, 5 minutes.
3. Purge prime exhaust: Vacuum, 4 Torr, 1 minute.
N2 inlet, 500 Torr, 2 minutes.
Vacuum, 4 Torr, 2 minutes.
4. Return to atmosphere pressure: N2 inlet, 3 minutes.
38
4.3.2.2. Silicon Valley Group Track – Coating and Developing Track
After the vapour prime, the samples are coated with the photo-resist, which is done in the
Silicon Valley Group Track. Being a fully automated system the track performs two main tasks:
deposition of 1.5 µm thick positive photo-resist (PFR 7790G 27cP) and its development after it had
been exposed. This photoresist is made of a resin and a photo-reactive compound dissolved in a
solvent that is also dissolution inhibitor. Because of the photo-reactive compound the photo-resist is
light sensitive, becoming unstable under exposure to certain light wavelength. The thickness of the
photo-resist is defined by the angular velocity in the spinning module in which the samples are coated
with photo-resist: for 40 seconds the spinning module makes 2800 revolutions per minute.
Figure 4.10: Picture of the SVG tracks.
Conditions used for the coating and developing processes:
Coating – Recipe 6/2
- Dispense of photoresist at 800 rpm for 5 seconds.
- Spin at 2.5 krpm for 40 seconds, then at 1.6 krpm for 5 seconds.
- Clean photoresist form the wafer border at 1 krpm for 2 seconds.
- Spin at 1.5 krpm for 15 seconds.
- Bake at 100ºC for 60 seconds.
Developing – Recipe 6/2
- Bake at 110ºC for 60 seconds.
- Cooling for 30 seconds.
- Water spray rinse at 500 rpm for 1 second.
- Dispense of the developer (Ethyl lactase – Pth70eg) at 500 rpm for 5 seconds.
- Development for 60 seconds with the wafer stopped.
- Rinse with DI water at 1 krpm for 20 seconds.
- Drying with wafer rotating at 3.5 krpm for 30 seconds.
39
4.3.2.3. Optical Lithography Exposure
The optical lithography exposure is performed by a direct-writ laser system DWL 2.0
Lasararay, which makes use of a 442 nm Helium-Cadmium laser of 120 mW. The laser sweeps the
sample, turning on and off concordantly with a predefined mask designed by the user with proper CAD
software (AutoCAD). The write lens is focused on the sample surface by an air pressure auto-focus
system, and exposure energy can be adjusted accordingly with reflectivity of the substrate material.
Since the used photo-resist is a positive one, it became unstable in the areas where the laser was on.
Following the sample exposure a photo-resist development is necessary: in this step the unstable
parts of the photo-resist vanish in contact with the developer(Ethyl lactate-commercial name Pth70eg).
1) 2)
3) 4)
For the exposure the samples are mounted on a mechanical x – y stage and fixed by vacuum.
In samples alignment is used a 70 nm measurement accuracy dual CCD camera system
(macro/micro). During the exposure the laser sweeps the sample according to the designed mask,
scanning the sample in stripes 200µm wide. Each stripe corresponds to several scans performed by
successively writing pixels from left to right (pixel grid pitch: 200 nm), in the x direction. Finished one
scan, the sample is moved one step in the y-direction and the next scan is done. When the stripe
exposure is complete, the stage moves to the origin in the y-direction and 200µm to the right in the x-
direction, starting the next stripe.
The system existing in INESC’s is placed in a class 10 clean-room and is capable of defining
structures of about 1 µm. The masks exposed were previously converted to the DWL software and
load into the hard disk of the system. Their design is done in the world spread AutoCAD software.
Figure 4.12: Figure of the DWL system.
Figure 4.11: Optical lithography – photo-resist mask creation.
40
4.4. Annealing setup
After completely patterned, it still is needed a magnetic thermal treatment to each MTJ sample
get fully functional. This annealing promotes the crystallization of amorphous materials, and the re-
crystallization of already crystalline structure correcting the defects always existing in their as-
deposited crystalline structure, re-enforcing the magnetic layers magnetization and improving the
exchange magnetic field that fixes the pinned layer. Once the annealing temperature for MTJs is
higher than the blocking temperature of the antiferromagnetic (pinning layer) is necessary to apply a
magnetic field during the annealing which re-sets the exchange field with the pinned layer. There are
two annealing setups at INESC-MN: it was used the system showed below.
Figure 4.13: Picture of the used annealing setup.
The setup consists in a heating system, a removable glass chamber and two large water-
cooled coils around a movable soft iron core. The heat source is a halogen lamp (100W, 12V) inside a
cooper block. The samples are placed on top of the cooper block with the glass chamber positioned
around and sealed, involving samples and heating system. The chamber is pumped by a turbo pump,
reaching pressures of the order of 10-6
Torr, and a magnetic field is applied by the water-cooled
electromagnetic coils. Under the magnetic field, the samples are then heated up to a certain
temperature defined by the user, and then still under the magnetic field, left to cool naturally. This is
the main difference between the two annealing setups: unlike this one this one in the other setup the
samples are heat up without an applied field.
All the samples patterned in the course of this thesis were annealed at 320ºC for 1 hour under
a magnetic field of 4kOe. The temperature cycle, which the samples are submitted to, is as follows:
41
Figure 4.14: Temperature cycle in the annealing process.
42
43
5. Characterization methods – Measurement techniques
[13],[14],[15],[16]
5.1. Vibrating Sample Magnetometer (VSM)
The VSM system (Figure 38) measures the magnetic moment of some material. It allows the
magnetic moment measurement of bulk unpatterned samples as a function of the applied magnetic
field. Because is a characterization method that doesn’t require pattern samples it can be precious by
saving time that would be wasted processing samples with an inappropriate magnetic response.
Two large coils are responsible for the creation of the magnetic field which defines the sample
magnetization. That is mounted onto a quartz rod which is connected to piezoelectric crystal, which
under excitation makes the sample vibrate. In the region between the large coils, near the sample, two
smaller coils were placed. Being a magnetic material the sample creates a magnetic field collected by
the small coils. Because the sample is moving back and forward the magnetic flux crossing the plane
of the inner coils is not constant, and a current, proportional to the variation rate of the flux, will be
induced on them. That current depends naturally on the magnetic moment of the sample.
Figure 5.1: VSM system.
5.2. Manual Transport Measurement Setup
This system used for electrical characterization of patterned samples. It comprises two pairs of
micropositioning probes with TiW needles (10 µm resolution), one voltmeter, two current sources (one
for the sensors biasing and other to create the applied magnetic field), 2 Helmoltz coils and magnifying
lenses.
Transfer curves (curve of Resistance vs external DC magnetic field) were measured in this
setup: curves of device resistance vs applied field. A current source is connected to a top and a
-6.0E-04
-3.0E-04
0.0E+00
3.0E-04
6.0E-04
-6.0E+03 -4.0E+03 -2.0E+03 0.0E+00 2.0E+03 4.0E+03 6.0E+03
M (
e.m
.u)
H (Oe)
VSM unpatterned
sample
T annealing = 320 ˚C
Hex = 1300 Oe
44
bottom contact of a sample, as well as the voltmeter (also connected to top and bottom contacts). It is
through the needles that electrical contact is established. The magnetic field is created by two coils
powered by the second current source, and varies sweeps all the range between -140 and 140 Oe. All
this sources are controlled by a computer with proper software (which also records the data) and a
GPIB communication protocol
The measurements can be done with two or four probes, being preferable the four probes way
whenever is possible.
It is considered that the real voltmeter behaves close to an ideal one: if it hasn’t infinite
impedance however, still should have a very high one, so current passes through him.
With the two probes configuration we have,
and the TMR comes:
In the last equation it’s clear that with two probes the measured TMR is smaller than actual
TMR value.
Because no current flows through the voltmeter, a four probe arrangement allows do
not count with contact resistance in the TMR measurement.
In the present work all the series were measured with a two probes arrangement, which,
because of the series design itself was impossible to measure with four probes.
Figure 5.2: Electrical scheme for measurements with 2 (on the left) and four (the right one) probes.
45
5.3. Profilometer
The prodfilometer system allows to measure the topography of a sample’s surface through a
piezo-resistive sensor. This is a direct measure where a micro-needle (12.5 µm diameter) makes a
physical contact with the surface to study, and sweeps it in a defined range. This system is very useful
to measure thicknesses of films and oxides. In the work contained in this thesis the profilometer was
mainly used to validate the thickness of deposited oxides.
To be able to measure the thickness of deposited material is required the existence of a
boundary between a region that contains the sample film/oxide and another region where it is not
deposited (the latter is a region where the substrate is uncovered at the time of profilometer measure).
There will therefore, in these transition regions a gap that corresponds to the thickness of the film. To
get these “undeposited” areas the substrate is, before the deposition, partially covered with some
material that can be easily removed afterwards (and with it, all the material deposited in top of it will be
removed too).
As the needle travels to the sample surface, the profilometer traces a graph with the height of
the needle. The value for height of the needle is given in each moment for a given reference level.
Usually the profilometer was used in parallel with the ellipsometer to ensure the right
measurement of the oxide thickness.
Substrate
Deposited material Deposited material
Figure 5.3: INESC-MN profilometer system – Allows measure film topography with irregularities higher then 300Å, with a 5Å vertical resolution
Figure 5.4:
46
5.4. Ellipsometer
The AutoEl ellipsometer, like the profilometer was mainly used to measure the thickness of
calibration samples used during the deposition of Al2O3. The principle of operation is based on the
irradiation of the sample’s surface with a collimated beam of monochromatic light (wave length
λ=632.8 nm) from a known angle. The incident light has a controlled state of polarization, and the
difference between this state of polarization and the one of the reflected lights is determined and
quantified in terms of two angles, Δ and Ψ, which are function of the real and imaginary components
of the refraction index. Then, using a numerical algorithm, the ellipsometer returns the refraction index
and the thickness of the layer. It’s important to be aware that this model assumes a perfectly reflective
substrate: this mean that every sample measured in the ellipsometer is grown on top of a Si substrate.
Figure 5.5: Picture of the ellipsometer.
47
6. Microfabrication – Process steps
6.1. Stack deposition
The first step of every new process run is the deposition of the sample stack over a new, clean
substrate. In this work the different material layers that compose a sensor were deposited in the
already presented Nordiko 2000 machine. Afterwards is necessary deposit a passivation layer of
TiWN2 to avoid the oxidation of the top layer, to add some physical protection to the sample stack and
to act as a antireflective layer to the optical lithography. Before the TiWN2 deposition a soft etch is
performed to remove a small thickness of natural oxidized material. This deposition took place in
Nordiko 7000 system.
Figure 6.1: Left - Standard MTJ stack used; Right - 3D model of a bulk MTJ sample after the stack deposition
Ta 50Å
Ru 50Å
CoFeB YÅ
MgO XÅ
CoFeB 30Å
Ru 9Å
CoFe 20Å
MnPt 200Å
Ta 30Å
Ru 180Å
Ta 50Å
Substrate
(Si or glass)
Substrate
MTJ stack
Oxide barrierMgO barrier
MTJ stack deposition
48
6.2. Bottom contact definition
To define the bottom contact of the sensors a lithography, followed by an all stack etching, are
realized. However, before the exposure there are some procedures that must always be met: first of
all, the samples are put in the Vapour Prime oven for 30 minutes at 130ºC, while HDMS is introduced
in the chamber; then the samples are coated in the SVG coating track with a 1.5 µm thick photoresist.
After the coating the samples are exposed and the laser makes an impression of the mask onto the
resist.
Following the lithography the samples are developed in the SVG developing track, in wich the
developmente liquid (usually called developer) is poured onto the sample, with which stays in contact
for 30 seconds. At the end of that time the sample is spread with water to wash the developer,
preventing overdevelopment.
After an optical inspection to confirm that the exposure was coorectly done the samples are
ready to be etched.
The next step is the first ion milling to define the bottom contact, performed either in N3000 or
N3600 system. This is a non selective etch, purely physical phenomenon (linear moment transfer),
both photoresist and stack are etched, but because the photoresist is much thicker (1.5 µm) than the
MTJ structure (around 800 Å) it never be fully etched.
Figure 6.3: Bottom contact definition: resulting structure after the ion milling.
Figure 6.2: Resulting photoresist after the 1st lithography revelation.
1.5 nm
Substrate
MTJ stack
Photoresist -
Photoresist1.5 µm
8 µm
Substrate
1.5 µmIon milling70˚ pan
8 µm
49
Figure 6.4: Picture of bottom contact.
Once the etch is donne th resist strip is required in order to remove the remaining photoresist.
This is donne immersing the samples into a microstrip soluction, which acts like a photoresist solvent.
An ultrassonic mach is also used to help remove the photoresist.
Figure 6.5: Left - Wet bench’s picture: ultrasound machine and hot bath; Right - microstrip solution.
148 µm
Hot bath
Ultrasound
machine
50
Substrate
2 µm
30 µm
6.3. MTJ pillar definition
The bottom contact is defined and is now necessary to define the junction pillar. This is done
again with a lithography followed by an etching. All the steps preceding any lithography have to be
fulfilled.
This second lithography is very important because it has to be well aligned with the previous
defined layer (bottom contact layer). The alignment between two different layers is always done with
the help of alignment marks drown in previously defined layers. Also the development is very
important because of the aspect ratio, which is essential to get a correct response of the sensors, so
an overdevelopment has to be avoided.
The next etching is also crucial because there is a certain thickness of material that has to be
removed, but not all the film layers; so it is vital prevent over-etch: the etching has to be performed till
under the SAF, and stop before reach the substrate (preferable is a good practice stop the etch some
hundreds before the end of the stack). To control the ending point are used calibration sample, which
have only and only a structure corresponding to the part of the MTJ stack that it is supposed to be
removed (it can have more 50Å of material above the SAF to ensure that is level is passed during the
etch). Another precaution to have during this ion milling regards the angle between the sample and the
ion beam: although using a 70º minimizes shadow effect, it implies an important effect of redeposition,
which can short circuit the oxide barrier, so a 40º angle is used to minimize redeposition.
Figure 6.7: Picture of the pillar junction.
Figure 6.6: Pillar junction definition. Left - After the 2nd
lithography; Right – After the 2nd
etch
Substrate
Bottom contact
MTJ pillar
Ion milling40º pan
51
Substrate
Bottom contact
Oxide800Å Al2O3
6.4. Pillar lateral isolation
In the samples microfabrication sequence fallows an oxide deposition. An 800Å of Al2O3 layer
is deposited to isolate the barrier so that the electrons only flow through the junction, avoiding all the
short circuits that would happen at the top contact deposition if the oxide had not been deposited.
After the oxide thickness confirmation in the profilometer and/or in the ellipsometer, the coming step is
the oxide lift off. The samples are kept in hot bath (65ºC) immersed in a microstrip solution used like in
every lift off, and the ultrasound machine is used as often as possible till the lift off become complete.
6.5. Top contact definition – Metallization
A third exposure defines the mask that will define the top contacts. This is exposure is inverted
when compared with the all the orders: this means that, now, after the development is intended to do
not have photoresist in the areas of interest, which are the places where the metal is intended to be
deposited (unlike the previous lithography in which the areas of interest remained covered and
protected).
The metallization is performed in the N7000 system, where 1500 Å of aluminum are deposited
after a prior soft etch. On top of the aluminum a 150 Å layer of TiWN2 is deposited to protect the
sample to the last lithography. A lift off removes the metal outside the contacts.
Figure 6.8: MTJ pillar and bottom contact insulation: after the lift off.
Figure 6.9: Top contact definition.
Substrate
Bottom contact
Top Contact1000Å Al + 150Å TiWN2
52
6.6. Sample insulation
A final lithography is made to define some holes in the contacts. After the development some
small photoresist blocks are on top of the contacts. Then a 1000 Å Al2O3 oxide layer is deposited,
followed by one last oxide lift off which opens small holes in the oxide through which the electrical
contact can be done with the sample.
6.7. Sample dicing
Since that in each substrate more than one die is patterned it is necessary to “cute” the
sample, separating each die and make it an individual module. The dicing is done by an automatic
dicing saw, model Disco DAD312, which can process up to 6 inch diameter substrate and 30.6 mm
thick.
6.8. Annealing
Finally, the samples need a magnetic thermal treatment: they need to be annealed at 320˚C
under a 4 kOe magnetic field during 1 hour, in order to promote the re-crystallization of the as-deposit
materials, re-enforcing the magnetization of the magnetic ones, and improving the TMR of the
samples.
6.9. Encapsulation
The last process step is mount the individual die on a chip-carrier (chip-mounting) and connect
them each other (electrically connect the sample and the chip carrier with very thin low resistance
wires – Wire bonding).
148
µm
Figure 6.10: Picture of the top electrode.
Figure 6.11: Sample patterned and mounted on a chip-carrier
53
6.00E+04
8.00E+04
1.00E+05
1.20E+05
1.40E+05
-150 -110 -70 -30 10 50 90 130
R (
Oh
m)
H (Oe)
ZarMTJ1
TMR = 78.1%Sensibility = 0.426%/OeVbias =1.7601 V
7. Results
During this thesis two generations of samples were patterned: the first generation was
patterned in a Si substrate coated with a 800Å thick SiO3 layer, and the second generation used glass
substrates. For both generations several samples were processed with distinct stacks: different
thicknesses of MgO barrier and CoFeB free layer weere tested.
1st Generation – SiO2 substrate
The used MTJ series have 4 contacts, two at the extremities and two intermediate contacts
which should allow a study in function of the number of individual MTJ.
Each of these three sections (sections ab, bc and cd) of the MTJs array has 120 individual
elements. The graphics below represent measurements of the entire series, all the 360 elements.
ZarMTJ1 ZarMTJ2 ZarMTJ3 ZarMTJ4 ZarMTJ5
Mgo thickness (Å) 17 17 12 12 12
CoFeB thickness (Å) 100 15,5 15,5 30 60
Table 7.1: Variations in the stack of samples with Si substrates.
Figure 7.1: Close view of one MTJ series with the 4 contacts identified as a, b, c and d.
Figure 7.2: ZarMTJ1 transfer curve
54
Figure 7.5: ZarMTJ5 transfer curve.
1.10E+04
1.15E+04
1.20E+04
1.25E+04
1.30E+04
1.35E+04
-150 -100 -50 0 50 100 150
R(O
hm
)
H(Oe)
ZarMTJ2
TMR = 17,04%Sensibility = 0.164%/OeVbias = 1.7173 V
Figure 7.4: ZarMTJ3 transfer curve.
This previous graphics represents the typical transfer curve that is possible to obtain from
each of the 1st generation samples (there are only four graphics because the sample ZarMTJ4 didn’t
make through the patterning process). These TMR values are not great for any of the samples, since,
at least, values around 100% were expected. Looking at samples ZarMTJ1 and ZarMTJ3 graphics we
would say that at least the relative sample behaviour was right, since from the thicker free layer
ZarMTJ1 sample we would expect a higher TMR comparing with the thinner free layer ZarMTJ3,
which should have in turn a higher sensitivity. In fact, these two samples behave just like that when
compared to each other. However the other two graphics show some problems: the one from
ZarMTJ2 sample shows a smaller sensibitility than the sample number 1 even though it had a thinner
free layer; the ZarMTJ5 has by far the worst TMR (it was not expected) and the highest coercivity of
the set.
Figure 7.3: ZarMTJ2 transfer curve.
55
In the graphic below the previous curves of ZarMTJ1 and ZarMTJ3 are overlapped for a
better comparison of their sensibilities.
Figure 7.6: ZarMTJ1 and ZarMTJ3; measurement for 360 elements.
The previous graphic evidences the relative correct behavior of these two samples: as the free
layer thickness decreases the sensibility should increase.
Figure 7.7: The RxA of the sensors in ZarMTJ1 and ZarMTJ2 samples. Each sample has 21 independent series of MTJ, here identified by a number line.
0
0.2
0.4
0.6
0.8
1
-150 -100 -50 0 50 100 150
No
rmal
ize
d T
MR
(%
)
H (Oe)
ZarMTJ3
ZarMTJ1
MgO– 12ÅCoFeB– 15.5ÅTMR = 23.6 %Sensibility = 1.241%/Oe
MgO– 17ÅCoFeB– 100ÅTMR = 78.1%Sensibility= 0.426 %/Oe
56
The RxA of the first two generations (ZarMTJ1 and ZarMTJ2) should be similar since they
have the same MgO thickness. However it is clearly shown in Figure 7.7 that the RxA of the ZarMTJ2
is much smaller. Since these samples were deposited and processed at different moments (not at the
same time or with very few time between them) this means that the reproducibility of the entire
process starting in the deposition (inclusive) through all the patterning protocols until the
measurements it is not reliable. Therefore comparisons between the samples, in order to define a
better stack cannot be done (even the ones already done only would be entirely correct if all the
samples had been processed at the same time).
Although this non reproducibility problem, the study of the behavior of a series based on the
number of elements that compose it would still be possible.
Figure 7.8: ZarMTJ1 - TMR vs number of junctions.
Since it isn’t reasonable to expect that all the junctions in a series are equal, the TMR and
resistance values of one series should be a weighted average of their individual values. With this in
mind, at the first sight the previous figure seems quite nice: a majority of series have more or less the
same TMR value for 120, 240 or 360 individual elements, meaning that they are quite similar through
the array lenght. But complementing this graph with some extra data raised some questions. Looking
at the series number 6:
TMR(%) Rmax (kΩ)
line ab(#120) bc(#120) cd(#120) ab(#120) bc(#120) cd(#120)
6 63.95 51.09 51.11 68.20 49.30 41.10
With these values in the sections ab, bc and cd, when the section ad is measured the TMR
can’t be lower than 51.11% and the resistance should be 158.6 kΩ. Nonetheless the measured
resistance is 92.10 kΩ with 44.62% of TMR (table above).
0
10
20
30
40
50
60
70
80
90
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
TMR
(%
)
line
360 junctions
240 junctions
120 junctions
ZarMTJ1:100Å CoFeB17Å MgOIbias = 20 µA
57
TMR(%) measured Rmax "expected" (kΩ) Rmax measured (kΩ)
line ac(#240) ad(#360) ac(#240) ad(#360) ac(#240) ad(#360)
6 58.9 44.62 117.50 158.60 117.00 92.10
This non expected kind of behaviour can be verified in other series, in other dies and even in
other samples. Because of the decreasing of the resistance relatively to what was expected be this
effect seems like the one obtained with resistors mounted in parallel. In fact this behaviour warned that
there might be some leak of current to the substrates. Subsequent measurements confirmed that
there were current leaks through the oxide that is supposed to insulate the Si substrate. Was then
determined that all the series were this way connected with each others. It was even possible to get
some transfer curves making the contact through pads that should be isolated one from the other.
Figure 7.9: Transfer curve between contacts of two independent series.
Every 1st generation samples had this current leak problem, which obviously affects the
performance of the magnetic tunnel junctions, which were deposited over a silica substrate with 800Å
of SiO2.
2st Generation – Glass substrate
The second sample generation was processed over a glass substrate and the conduction
problem disappeared. Meanwhile, was discovered by the team of INESC researchers that the
deposition rates in the Nordiko2000 system didn’t correspond to the known values. So in order to
regain the control of the thickness of deposited films the deposition times were recalibrated. This new
glass substrate samples were already deposited with the new deposition times.
5.00E+04
6.00E+04
7.00E+04
8.00E+04
9.00E+04
-180 -140 -100 -60 -20 20 60 100 140 180
R (
Oh
m)
H(Oe)
ZarMTJ1 line 12 pad a & line 13 pad dTMR = 36.7%
58
7500
7700
7900
8100
8300
8500
-150 -50 50 150
R (
kO
hm
)
H (Oe)
N2#4
TMR = 3.8%Sensibility = 0.033%/OeVbias = 1.82 V
4.00E+04
4.50E+04
5.00E+04
5.50E+04
6.00E+04
-150 -50 50 150
R (O
hm
)
H (Oe)
N2#1
TMR = 39.5%Sensibility = 0.711%/OeVbias = 1.79 V
16000
18000
20000
22000
24000
-150 -50 50 150
R (k
Oh
m)
H (Oe)
N2#2
TMR =32.0%Sensibility=0.1723%/OeVbias = 1.80 V
N2#1 N2#2 N2#3 N2#4
CoFeB thickness (Å) 15.5 30 100 60
MgO thickness (Å) 12 12 12 12
Table 7.2: Variations in the stack of samples with glass substrates.
Figure 7.10: N2#1 transfer curve.
Figure 7.11: N2#2 transfer curve.
Figure 7.12: N2#4 transfer curve.
59
All these samples were patterned at the same time, so they could be compared with each
other. The sample N2#3 (100 Å CoFeB) was not working at the end of the process. Comparing with
the 1st generation, the sample with 15.5Å of CoFeB showed an improved magnetoristance although
the decrease of the sensibility and the increase of the coercivity. For the sample N2#2 was expected
higher TMR values than N2#1 sample. The sample N2#4 has marginal TMR. The sample with higher
TMR, N2#1 sample, was chosen to further studies.
Once more a study of the behavior of a series based on the number of elements that
compose it was conducted. The results of the sample N2#1 are shown below:
Figure 7.13: N2#1 - TMR vs number of junctions
The TMR values for 120, 240 and 360 individual elements are almost the same, close to what
would be expected from perfect series concerning the likeness of every individual junction. Since
during the patterning process of both ZarMTJ1 and N2#1 nothing occurred that could explain a lesser
uniformity of properly working individual junctions through a series, again can be concluded that was
the conduction path through the substrate in the case of the first generation samples that played a
major influence in the unexpected behavior of ZarMTJ1 (and all the other 1st generation samples), and
not an unidentified cause in the manufacturing process.
0
10
20
30
40
50
0 5 10 15 20
TMR
(%)
Line
360 junctions
240 junctions
120 junctions
sample N2#1Ibias = 40 µA
60
0
10
20
30
40
50
60
70
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
TMR
(%)
V(V)
N2#1line 18 ab (#120)
A study of TMR vs Vbias was conducted (this study required a remake of the program that
controls the manual transport measurement setup).
0
10
20
30
40
50
60
70
-35 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35
TMR
(%)
V(V)
N2#1line 18 ab (#120)
-35
-30
-25
-20
-15
-10
-5
0
5
10
15
20
25
30
35
-0.003 -0.002 -0.001 0 0.001 0.002 0.003V (V
)
I (A)
H = 140 Oe
H = -140 Oe
N2#1line 18 ab (#120)
Figure 7.14: Top – Curves V vs I for maximum and minimum resistance saturation fields; Center – TMR vs I; Bottom – Close look of the near zero bias current.
61
It can be seen that the TMR value has a plateau for currents in the vicinity of I=0 (V=0). In fact
it is possible to apply about 1.5V to the series without any concern about the decrease of TMR. If the
current keeps increasing the TMR starts to decrease. With the used voltmeter wasn’t possible to
measure higher voltages (it has a maximum range limited to the interval between -30 V and 30 V).
Therefore we were not able to determined the value of V1/2 .
This last result shows some of the greatest advantages of using arrays of MTJs:
- with relatively small bias currents high output voltages are achieved;
- the higher robustness of a MTJ array which tolerate a very high potential drop
between its terminals without disruption and with just a small TMR drop.
Regarding the robustness it can be better understand when compared with the result for an
individual MTJ:
Figure 7.15: TMR of a MTJ decreasing with the applied voltage.
The figure above[17]
shows a decreasing of 50% in TMR for an individual junction biased with
350 mV, which is substantially less voltage than what was applied to the MTJ array, which supported
30V without suffering a 50% TMR loss. This feature makes the MTJs series particularly useful and
usable since they support any electrostatic discharge that could occur in their handling, and is justified
by the “distribution” of the total potential drop in the series by each of its individual elements.
62
M
H
H
Functionalized stripe
Nanoparticle
While the work of characterization of the sensors was being done, one set of sensors was sent
to another partner institution in the project with the goal of proceeds real applications measurements.
Thus, sensors of the first generation (ZarMTJ1), were sent. The wheel device already explained was
used.
Figure 7.16: Graph - Response of one sensor under magneto nanoparticles excitation. Scheme - Functionalized stripe with nanoparticles attached, and a vertically magnetized nanoparticle.
During this experiment the particles where vertically magnetized, as shown in the previous
figure. The wheel was not rolling freely but fixed and was pulled over the sensor When the
nanoparticles are trapped in the stripe they have a certain tendency to accumulate themselves at the
edge of the stripes. Thus, the boundaries of the stripes have a higher concentration of the
nanoparticles, and will create a higher magnetic field in these regions of the stripes. Therefore, when
the stripe is pulled over the sensors, considering this nom-uniformity in the vertically magnetized
particles concentration over the stripe, two spikes appears close to the boundaries of the stripe (Figure
7.16). Therefore the shape of the curve is concordant with the prediction made in Chapter 3.
Stripe width
63
8. Conclusions and Future work
During this thesis several samples of MTJs connected in series were patterned. TMRs off about
70% were achieved as well as samples with sensibilities of 1.2%/Oe. It was proved the great
robustness of these devices from which we can get a Voltage output of the order of some tenth of
Volts without occur their disruption.
Being an ongoing work it still requires a fully noise characterization (Sv vs #junctions , Sv vs
Vbias) and the pattern of new samples with improved TMR singnal and Sensibility.
64
65
Bibliography
[1] A. B. Henriques, J. C. Romão, electromagnetismo, IST Press, 2006. [2] S.O. Kasap, Principles of Electronic Materials and Devices, McGraw-Hill, 2nd edition, 2006. [3] R.Waser, Nanoelectronics and information technology: advanced electronic materials and
novel devices,Wiley-VCH, 2nd edition, 2005. [4] A. Fert, et al., Giant Magnetoresistance of (001)Fe/(001)Cr Magnetic Superlattices, Phys. Rev.
Letter 61(21): 2472-2475, 1988. [5] B. Dieny, et al., Giant magnetoresistance in soft ferromagnetic multilayers, Phys. Rev. B 43(1):
1297-1300, 1991. [6] M. Julliere, Tunneling between ferromagnetic films, Phys. Lett. A 54: 225, 1975. [7] P. Grunberg, et al., Enhanced Magnetoresistance in layered magnetic structures with
antiferromagnetic interlayer exchange, Phys. Rev. B 39(7): 4828-4830, 1989. [8] J. C. Slonczewski, Conductance and exchange coupling of two ferromagnets separated by a
tunneling barrier, Phys. Rev. B 39: 6995, 1989. [9] D. Wang, et al., Spin dependent tunneling junctions with reduce Neel coupling, J.Appl.Phys.,
93(10),2002. [10] P. Bruno, C. Chappert, Giant Magnetoresistance of (001)Fe/(001)Cr Magnetic
Superlattices,Phys. Rev. Letter 67: 1602, 1991. [11] P.Wisniowski, et al., Effect of free layer thickness and shape anisotropy on the transfer curves
of MgO magnetic tunnel junctions, J.Appl.Phys., 103, 2008. [12] Manuel João de Moura Dias Mendes, Micromagnetic Simulations of Spin Valve devices,
Senior thesis,Instituto Superior Técnico, 2005. [13] Susana Isabel Pinheiro Cardoso de Freitas, Dual-Stripe GMR and Tunnel Junction Read
Heads and Ion Beam Deposition and Oxidation of Tunnel Junctions, PhD thesis, Instituto Superior Técnico, 2001.
[14] Ricardo Alexandre de Matos Antunes Ferreira, Ion Beam Deposited Magnetic Spin Tunnel
Junctions targeting HDD Read Heads, Non-volatile Memories and Magnetic Field Sensor Applications, PhD thesis, Instituto Superior Técnico, 2008.
[15] Jianguo Wang, Low-resistance tunnel junctions for read head applications, PhD thesis,
Instituto Superior Técnico, 2002. [16] Haohua Li, Spin valve read elements and sensors, PhD thesis, Instituto Superior Técnico,
2004. [17] Filipe Arroyo Cardoso, Design, optimization and integration of magnetoresistive biochips, PhD
thesis, Instituto Superior Técnico 2010. [18] Ricardo Alves Ferreira Costa e Sousa, Magnetic Random Acess Memory (MRAM) based on
Spin Dependent Tunnel Junctions, PhD thesis, Instituto Superior Técnico [19] R. Guerrero, et al., Low frequency noise in arrays of magnetic tunnel junctions connected in
series and parallel, Appl. Phys., Vol. 105, 2009.
66
67
A. Appendix
Figure .A.1: Mask use – Blue lines: top contacts; Yellow lines: define the shape of the contact pathways in the last passivation layer; in the lower right corner: two individual junctions, with a
reference role for comparison.
68
148µm
956µm
Figure A.2: Top: Series detail – Green: bottom contacts; Red: pillar junctions; Blue: top contacts Bottom: The entire array of MTJs, with intermediate contacts;.
Figure A.3: 3D scheme of MTJ’s connected in serious: the current coming from a top contact goes through a junction, in a descending direction until reach the bottom contact; flows in the bottom contact reaching the other junction and crosses it in the upper direction; now the current flows again in a top contact until the next junction and so on. The top contact, the bottom contact and the pillar junction correspond to the blue lines, green lines and red lines in the previous figure
69
B. Appendix
Run Sheet for MTJs Array fabrication: Influenza_ 2010
Process Start: / /2010 Process Complete: / /2010
Substrate: Si wafer / 800 Å SiO2 : ~1 inch2
Machine: Nordiko 2000
Magnetic Tunnel Junction Structure:
Bottom electrode: Ta 50 Å / Ru 180 Å / Ta 30 Å / PtMn 200 Å / CoFe 20 Å / Ru 9 Å / CoFeB 30 Å
Oxide Barrier: MgO x Å
Top electrode: CoFeB y Å / Ru 50 Å / Ta 50
Total height: 619 + x + y Å
Calibration samples: VSM
Top electrode for 2nd etch (glass substrate)
Machine sequences:
Seq. 3: Pre-sputtering of all targets x 2;
Seq. 38: Xianghiong
Seq. 39: FILIPE
Caracterization: VSM after annealed
STEP 1 Tunnel Junction Deposition
70
Step 1: Read Values
B.P:8.6x10-8 Torr
Seq. 3: pre sputtering of all targets x2
B.P:8.4x10-8 Torr
Sequence:
(#/name) Function : Read Values
38/Xianghiong
F18, Ta 50 40 mA | 331V | 10 W | 9.7 sccm | 4.6 mTorr | S4T3 | 100%
F7, Ru 180 40 mA | 302V | 10 W | 7.8 sccm | 5.1 mTorr | S4T2 | 100%
F6, Ta 30 40 mA | 329V | 10 W | 9.8 sccm | 4.6 mTorr | S4T3 | 100%
F5, PtMn 200 30 mA | 292V | - | 9.7 sccm | 4.6 mTorr | S4T3 | 100%
F9, CoFeB 20 F34 R3 B271 | 7.7 sccm |5.2 mTorr | S4T1 | 100%
F3, Ru 9 40 mA | 292V | 10 W | 7.8 sccm | 7.8 mTorr | S4T2 | 100%
F47, CoFeB 30 40 mA | 417V | 10 W | 8.7 sccm |5.2 mTorr | S4T4 | 100%
F94, MgO
Cleaning 2´30’’ F149 R0 B291 | 9.6 sccm |5.1 mTorr | S4T2 | 100%
39/FILIPE
F51, MgO x F129 R0 B272 | 9.5 sccm |8.2 mTorr | S4T5 | 50%
F50, CoFeB y 40 mA | 418V | 10 W | 8.8 sccm |5.1 mTorr | S4T4 | 100%
F4, Ru 50 40 mA | 302V | 10 W | 7.8 sccm | 5.1 mTorr | S4T2 | 100%
F18, Ta 50 40 mA | 330V | 10 W | 9.9 sccm | 4.6 mTorr | S4T3 | 100%
71
Machine: Nordiko 7000
Machine sequence:
Seq.17 – mod.2 – f.9 (1’ soft sputter etch) P=60W/40W, p=3mTorr, 50 sccm Ar
mod 3 – f.19 (150A TiW, 27’’) P=0.5 kW, 3mTorr, 50sccm Ar + 10 sccm N2
Readings – Module 2
Run# Power1 Power2 Ar flux (sccm) Pressure (mTorr)
F59 R8 B131 F40 R1 50.2 3.1
Readings – Module 3
Run# DC Power (kW) Voltage (V) Current (A) Ar flux (sccm) Pressure (mTorr)
0.50 419 1.2 50.5 3.0
Calibration samples: Top electrode for 2nd etch
Machine: Vapour prime; SVG Track; DWL.
Vapour Prime: Program 0 - 30 min.
Coating: Recipe 6/2 - 1.5 μm photoresist
Masks: a)ZarMTJ1 Map: HUGO ( Die size: 8.2x8.2 mm2)
b)Zaragoza_2x30_L1
Alignment mark: none
Energy: 80
Power: 120mW
Focus: -50
Develop : Recipe 6/2 Development time : 1 min
STEP 3 1st Lithography – Main Pillar Definition
STEP 2 MTJ Passivation – 150 Å TiWN2
Easy
Axis
a b
b b
X,Y
72
Optical Inspection:
On an optical microscope, with a green filter to confirm the sample development (filters the wave
lengths which impress the photoresist, allowing to develop the sample a little more if needed).
Sample Comments
All samples Every time that one sample was not perfectly exposed the optical lithography was repeated.
Machine: N3600
Total thickness to etch: 619 + x + y Å (etch rate: ~1 Å/s time: 619 + x + y s)
Base Pressure (Torr): 2.04x10-7 T Cryo (K): 107
Batch: etchjunction
Recipe etch junction stack all
Steps: etch pan 60 deg
cool_down_200s
Assist Gun: 160W 105mA +735V/-350V 12sccm Ar; Assist Neut: 30% subst.rot 60˚ subst.pan
Assist Gun Power (W) V+ (V) I+ (mA) V- (V) I- (mA) Ar Flux (sccm)
Read Values 195 723.5 104.5 344.3 2.6 10.6
STEP 4 1st Ion Milling – Total Structure Etch
73
Machine: Chemical Wetbench: Hot bath; ultrasounds machine.
Hot μ-strip + ultrasonic
Started:_____________ Stoped:_____________
Total Time in hot µ-Strip : 5h Ultrassonic Time : 40 min.
Rinse with IPA + DI water + dry compressed air
Optical Inspection:
Sample Comments
All Samples Some bottom contact remain shorted: problem will be solved in the second ion milling.
Machine: Vapour prime; SVG Track; DWL.
Vapour Prime: Program 0 - 30 min.
Coating: Recipe 6/2 - 1.5 μm photoresist
Masks: a)ZarMTJ2 Map: HUGO ( Die size: 8.2x8.2 mm2)
b)Zaragoza_2x30_L2
Alignment mark: (168, 54) µm
Energy : 80
Power : _120mW
Focus : -70
Develop : Recipe 6/2 Development time : 1 min
STEP 5 Resist Strip
STEP 6 2nd Lithography – Junction Definition
Easy
Axis
a b
b b
X,Y
74
Optical Inspection:
Machine: N3600
Total thickness to etch: 309+x+y Å (etch rate: ~1 Å/s time: 309+x+y s)
Base Pressure (Torr): 2.12x10-7
Batch: etchjunction
Recipe etch junction 60/30
Steps: Etch Junction Top electrode
cool_down_200s
etch pan 30 deg
Assist Gun: 160W 105mA +735V/-350V 12sccm Ar; Assist Neut ; 30% subst.rot 30º subst.pan
Calibration Sample Structure
CoFe 20Å | Ru 9Å | MgO xÅ | CoFeB yÅ | Ru 50Å | Ta 50Å | 150 TiWN2
Assist Gun Power (W) V+ (V) I+ (mA) V- (V) I- (mA) Ar Flux (sccm)
Read Values 195 723.3 104.6 350 2.5 10.2
Wafer samples Etching Turn Time Effect
Etch pan 30˚ 59’
Etch pan 60˚ 158’
Etch pan 30˚ 59’ Calibration sample tranlucid
STEP 7 2nd Ion Milling – Top Electrode and Junction Definition
75
Responsible: Fernando
Machine: UHV2
Dep. Rate ~11.43 Å/s
Deposition
Time
Al2O3 thickness Ar gas flow Pressure Power Source
62’ 700 A 45sccm 200W
Comments: Calibration sample: profilometer and ellipsometer.
Machine: Chemical Wetbench: Hot bath; ultrasounds machine.
Hot μ-strip + ultrasonic
Started:_____________ Stoped:_____________
Total Time in hot μ-strip : 3 days Ultrasonic Time : 40 min.
Rinse with IPA + DI water + dry compressed air
Optical inspection:
STEP 8 Insulating Layer Deposition
STEP 9 Oxide Lift-Off
76
Machine: Vapour prime; SVG Track; DWL.
Vapour Prime: Program 0 - 30 min.
Coating: Recipe 6/2 - 1.5 μm photoresist
Masks: a)ZarMTJ3 Map: HUGO ( Die size: 8.2x8.2 mm2)
b)Zaragoza_2x30_L3
Alignment mark: (168, 54) µm
(168,168) µm
Energy : 80
Power : _120mW
Focus : -50
Develop : Recipe 6/2 Development time : 1 min
Optical Inspection:
Machine: Nordiko 7000
Machine sequence:
Seq.48 (svpad) – mod.2 – f.9 (1’ soft sputter etch) P=60W/40W, p=3mTorr, 50 sccm Ar
mod.4 – f.1 (1500A Al, 40’’) P=2 kW, 3mTorr, 50 sccm Ar
mod 3 – f.19 (150A TiW, 27’’) P=0.5 kW, 3mTorr, 50sccm Ar + 10 sccm N2
STEP 11 Contact Leads Deposition 1500 Å Al98.5Si1.0Cu0.5 / 150 Å TiWN2
STEP 10 3rd Lithography – Top Contact
Easy
Axis
a b
b b
X,Y
77
Readings – Module 2
Run# Power1 Power2 Ar flux (sccm) Pressure (mTorr)
F60 R8 B122 F40 R2 50.2 3.2
Readings – Module 4
Run# DC Power (KW) Voltage (V) Current (A) Ar flux (sccm) Pressure (mTorr)
2 404 5.0 50.5 3.1
Readings – Module 3
Run# DC Power (KW) Voltage (V) Current (A) Gas flux (sccm) Pressure (mTorr)
0.50 417 1.2 50.2 Ar / 10.6 N2 2.9
Machine: Chemical Wetbench: Hot bath; ultrasounds machine.
Hot μ-strip + ultrasonic
Started:_____________ Stoped:_____________
Total Time in hot μ-strip : 1 day Ultrasonic Time : 1 hour
Rinse with IPA + DI water + dry compressed air
Optical inspection:
Comment
This picture shows a localized defected resulting from a over development in the previous lithography.
STEP 12 Metal lift-off
78
Machine: Vapour prime; SVG Track; DWL.
Vapour Prime: Program 0 - 30 min.
Coating: Recipe 6/2 - 1.5 μm photoresist
Masks: a)ZarMTJ4 Map: HUGO ( Die size: 8.2x8.2 mm2)
b)Zaragoza_2x30_L4
Alignment mark: (168, 54) µm
(168,168) µm
(168, 256) µm
Energy : 80
Power : _120mW
Focus : -50
Develop : Recipe 6/2 Development time : 1 min
Optical Inspection:
Responsible: Fernando
Machine: UHV2
Dep. Rate ~11.43 Å/s
Deposition
Time
Al2O3 thickness Ar flow (sccm) Pressure Power Source
1h27’ 1000 A 45 200 W
Comments: Calibration sample: profilometer and ellipsometer.
Machine: Chemical Wetbench: Hot bath; ultrasounds machine.
Hot μ-strip + ultrasonic
Started:_____________ Stoped:_____________
STEP 14 Final oxide deposition
STEP 15 Oxide lift-off ___
STEP 13 4th Lithography – junction contact opening
Easy
Axis
a b
b b
X,Y
79
Total Time in hot μ-strip : 1 day Ultrasonic Time : 30 min.
Rinse with IPA + DI water + dry compressed air
Optical inspection:
Machine: Old annealing setup
Temperature: 320˚C real - 280˚C equipment setpoint
Annealing time: 1h
Applied field: 4 kOe.
STEP 16 Magnetic Anneal