Anda di halaman 1dari 89

3D Magnetic Flux Concentrators with improved efficiency for

Magnetoresistive Sensors

Marinho, Zita

Dissertação para a obtenção de Grau de Mestre em


Engenharia Fı́sica Tecnológica

Júri

Presidente:
Orientador: Prof. Paulo P. Freitas
Vogal: Prof. Susana Cardoso

Outubro 2010
Acknowledgements

I would like to start by thanking Prof. Paulo Freitas for supervising me in this thesis and for giving me the
great opportunity of working at INESC-MN. I am also very grateful to Rui Chaves for all the very helpful
advice and guidance whenever I needed, and also for his patience explaining all the laboratory work. This
thesis would not have been possible without the work developed by Andre Guedes in hybrid sensors. I would
also like to give a special thanks to Prof. Susana Cardoso and Ricardo Ferreira for their help during machines
malfunctioning even during the weekend, and also for their guidance in the laboratory. I would like to show
my gratitude to the clean room engineers, José Bernardo, Fernando Silva and Virginia Soares for all the help
provided. I also thank all the rest of the people working at INESC-MN for the good working environment. I
am indebted to my many of my colleagues that helped me, specially Nuno for his support whenever I needed
the most. Finally I would like to show my deepest gratitude to my family for their support during this five
years.

ii
Resumo

addcontentslinetocsectionResumo
O trabalho efectuado nesta tese foca-se no estudo de guias de fluxo magnético integrados com sensores
magnetoresistivos. A motivação por detrás deste trabalho, advém de melhorias feitas anteriormente, na linha
de investigação de A. Guedes [1] [2], na detecção de campos magnéticos estáticos de baixa amplitude. Guias de
fluxo magnético 3D foram fabricadas a partir de filmes de 7000Å de Co86 Zr5 Nb9 , depositados por pulverização
catódica de partı́culas. Estas estruturas foram optimizadas de modo a alcançar sensibilidades maiores em
estruturas integradas de guias e sensores magnetoresistivos. As guias de fluxo magnético permitiram a
concentração do fluxo magnético ao nı́vel do sensor. Por forma a obter uma melhor resposta do sistema, a
geometria das guias foi melhorada. Foram ainda estudados e comparados diferentes processos de fabricação
de guias, por “lift-off” e por “etch”. O perfil vertical das guias fabricado por “lift-off” foi abondonado em
favor de um perfil 3D, onde a guia converge para o sensor com um declive bem controlado, definido por “ion
milling” a ângulos fixos (20,45,75,85). Foram considerados sensores de válvulas de spin com uma resposta
linear e uma geometria rectângular de 14x3 µm2 com uma sensibilidade inicial dos sensores foi de 0.084%/Oe
(numa estrutura sem guias de fluxo). As guias optimizadas com um perfil de 45o permitiram o aumento
da sensibilidade do sensor até aos 7.4%/Oe e o ganho obtido através dos diferentes concentradores permitiu
chegar a um aumento da sensibilidade dos sensores de 10 a 100 vezes. Esta melhoria não só tornou o sensor
magnetoresistivo mais sensı́vel, mas também contribuiu para a detecção de campos magnéticos de baixa
amplitude no processo integrado com os MEMS, tornando possı́vel uma resolução na ordem dos pico Tesla.

Palavras-chave: concentradores de fluxo magnético, detecção de campos de baixa amplitude, sensores


magnetoresistivos, sensibilidade das válvulas de spin.

iii
Abstract

The work conducted in this thesis focused on the study of magnetic flux concentrators integrated with mag-
netoresistive sensors. The motivation behind this work came from previous work regarding the improvement
of low amplitude static field detection following the research line of A. Guedes [1] [2]. 3D magnetic flux
concentrators based on 7000 Å-thick Co86 Zr5 Nb9 films, were deposited by magnetron sputtering. These
structures were optimized in order to achieve high sensitivity in hybrid flux guide-magnetoresistive struc-
tures. The magnetic flux concentrators, as the name suggests, allowed the concentration of magnetic flux
at the sensor level. An improvement of the design was made [4] in order to achieve a better performance of
the overall system. Different processing approaches for the fabrication of the concentrators were studied and
compared, both by ion milling and lift off processes. The vertical profile of flux guides patterned by lift-off
were replaced by a 3D profile, where the sensor is recessed from the concentrator with a controlled edge slope
defined by ion milling at given angles (20◦ , 45◦ , 60◦ , 75◦ , 85◦ ). Spin valve linear sensors with a 14x3µm2
geometry were designed with initial sensitivity of 0.084%/Oe (without flux concentrators) and optimized flux
guides with 45◦ slope increased the sensitivity up to 7.4%/Oe. The effect of the different concentrators on the
magnetic field gain was studied, and a 10 to 100 fold increase of sensitivity was found when integrating these
concentrators with spin valve sensors. This improvement of the gain factor not only increased the sensor
sensitivity but could also improve the detection limits for low static magnetic fields of the fully integrated
process (including the MEMS), aiming at a resolution in the pico Tesla range.

Keywords: ion beam milling, magnetic flux concentrator, low field detection, magnetoresistive sensors,
spin valve sensitivity

iv
v
Contents

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

Contents vi

List of Tables ix
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

List of Figures xi
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

1 Theorectical background 1
1.1 Magnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Magnetic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Diamagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.2 Paramagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.3 Ferromagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.4 Ferrimagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.5 Antiferromagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.6 Ferromagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Ferromagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 Zeeman energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 Exchange energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.3 Anisotropic energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.4 Demagnetizing energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Magnetorresistive sensors: Spin Valves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Anisotropic Magnetoresistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Giant Magnetoresistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Spin Valves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Interlayer coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Linearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4.1 Soft magnetic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2 Simulations 17
2.1 Magnetic Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.1 Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.2 Gap length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.1.3 Width . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1.4 Asymmetric geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.5 Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

vi
3 Experimental Methods 25
3.1 Thin film deposition and Ion milling process . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1 Magnetron sputtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Ultra High Vacuum 1 (UHV1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Ultra High Vacuum 2 (UHV2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Nordiko 7000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.1.2 Ion beam deposition and Ion milling process . . . . . . . . . . . . . . . . . . . . . . . . 29
Nordiko 3000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Pattern transfer techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.1 Coating tracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.2 DWL exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.3 Developing tracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.4 Lift-off process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.5 Ion milling process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Nordiko 3000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Nordiko 3600 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3 Film characterization methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.1 Ellipsometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.2 Profilometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3.3 Atomic force microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.4 Scanning electron microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.5 Composition analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3.6 Magnetic properties analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.7 Electric properties analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4 Laboratory work 41
4.1 Microfabrication process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.1 Spin valve definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.1.2 Flux guides definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1.3 Ion Milling process control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2 Device Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.1 Spin valve properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.2 CoZrNb magnetic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.3 CoZrNb composition analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.4 Magnetoresistance and sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . 49

5 Conclusions 53

A Fabrication process scheme 57

B CoZrNb volume determination 59

C Flux guide gain analysis 61


C.1 Gain and sensitivity table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
C.2 Type I - transfer curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
C.3 Type II - transfer curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
C.4 Type III - transfer curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
C.5 Type IV - transfer curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

D Process run sheet 67

vii
Bibliography 75

viii
List of Tables

3.1 Deposition conditions for the CoZrNb target on UHV1 . . . . . . . . . . . . . . . . . . . . . . . . 27


3.2 Deposition conditions for the Al2 O3 target on UHV2 . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3 Nordiko 7000 conditions for operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.4 Nordiko 3000 deposition conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Nordiko 3000 ion milling conditions at different angles . . . . . . . . . . . . . . . . . . . . . . . . 33
3.6 Nordiko 3600 ion milling conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.1 CoZrNb Composition Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49


4.2 Spin valve sensor properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 Flux Guide gain analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

B.1 Volume determination of CoZrNb layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

ix
List of Figures

1.1 Zeeman energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4


1.2 Exchange bias effect pinning the ferromagnetic layer of a spin valve . . . . . . . . . . . . . . . . . 6
1.3 Hysteresis curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Domain formation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 Domain wall magnetization change. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Density of states in the vicinity of the Fermi level for a metal and a ferromagnetic material . . . 10
1.7 Electron transport for parallel and antiparallel alignment of the ferromagnet layer’s magnetization 11
1.8 Spin Valve structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.9 Neél interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.10 RKKY interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.1 Magnetic Flux concentrator geometry on FEMM . . . . . . . . . . . . . . . . . . . . . . . . . . . 17


2.2 Magnetic field lines with variable inner width A2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Gain vs. thickness simulated results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Magnetic field lines with variable thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Gain vs. Gap between MFC simulated results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.6 Magnetic field lines with variable gap length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.7 Gain vs. Inner Width simulated results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.8 Magnetic field lines with variable inner width A2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.9 Gain vs. Outer Width simulated results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.10 Magnetic field lines with variable outer width A1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.11 Gain vs. Asymmetric Width simulated results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.12 Asymmetric concentrator geometry. FEMM results with variable A1 . . . . . . . . . . . . . . . . 23
2.13 Gain vs. Length simulated results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.14 Magnetic field lines with varying length L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.1 Magnetron sputtering system scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26


3.2 UHV1 machine image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3 UHV2 machine image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Nordiko 7000 machine schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.5 Ion Beam deposition machine schematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.6 SVG system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.7 Lift-off process scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.8 Ion milling process scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.9 Ellipsometer AutoEl image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.10 Profilometer Desktak image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.11 Atomic force microscope scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.12 Scanning Electron Microscope Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

xi
3.13 VSM device image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.14 Manual measurement setup image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.15 Four probe point system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4.1 Microfabrication integrated structure (MR sensor, MFC, contact leads) . . . . . . . . . . . . . . 41


4.2 Spin Valve structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.3 Spin Valve dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.4 Flux guide concentrator dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.5 Flux guide concentrator and spin valve scheme at gap region . . . . . . . . . . . . . . . . . . . . 44
4.6 Flux concentrator fabrication processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.7 Double angle profile microscope image - Process II - non inverted mask to define the second
tapered profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.8 Sample microscope image of ion milling processed samples with 110W power. . . . . . . . . . . . 45
4.9 AFM gap measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.10 Flux concentrator SEM images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.11 Spin Valve VSM analysis (sv578) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.12 In bulk CoZrNb VSM analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.13 Patterned CoZrNb structures’ geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.14 Patterned CoZrNb VSM analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.15 Spin valve sensor MR-H hysteresis curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.16 Transfer curve for a type II flux guide (45◦ profile) . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.17 Transfer curve for type II flux guides (45◦ , 75◦ , 85◦ profiles) . . . . . . . . . . . . . . . . . . . . . 52

A.1 Spin valve fabrication scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57


A.2 Type I - 2D profile scheme (fabricated by lift-off) . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
A.3 Type II - 3D tapered profile scheme (single angle) . . . . . . . . . . . . . . . . . . . . . . . . . . 57
A.4 Type III - 3D tapered profile scheme (double angle) . . . . . . . . . . . . . . . . . . . . . . . . . 58
A.5 Type IV - 3D tapered profile with vertical stub at sensor level scheme (single angle) . . . . . . . 58

C.1 Properties of flux concentrators: gain and sensitivities obtained for different angles and processes 61
C.2 Type-I transfer curve. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
C.3 Type-II transfer curve (20◦ tapered profile). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
C.4 Type-II transfer curve (45◦ tapered profile). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
C.5 Type-II transfer curve (75◦ tapered profile). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
C.6 Type-II transfer curve (85◦ tapered profile). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
C.7 Type-III transfer curve (85◦ +45◦ tapered profile). . . . . . . . . . . . . . . . . . . . . . . . . . . 65
C.8 Type-III transfer curve (75◦ +20◦ tapered profile). . . . . . . . . . . . . . . . . . . . . . . . . . . 65
C.9 Type-IV transfer curve (45◦ tapered profile with vertical wall at sensor level). . . . . . . . . . . . 66
C.10 Type-IV transfer curve (60◦ tapered profile with vertical wall at sensor level). . . . . . . . . . . . 66

D.1 Process run sheet used for type I flux concentrators . . . . . . . . . . . . . . . . . . . . . . . . . 68


D.2 Process run sheet used for type I flux concentrators . . . . . . . . . . . . . . . . . . . . . . . . . 69
D.3 Process run sheet used for type I flux concentrators . . . . . . . . . . . . . . . . . . . . . . . . . 70
D.4 Process run sheet used for type I flux concentrators . . . . . . . . . . . . . . . . . . . . . . . . . 71
D.5 Process run sheet used for type I flux concentrators . . . . . . . . . . . . . . . . . . . . . . . . . 72
D.6 Process run sheet used for type I flux concentrators . . . . . . . . . . . . . . . . . . . . . . . . . 73
D.7 Process run sheet used for type I flux concentrators . . . . . . . . . . . . . . . . . . . . . . . . . 74

xii
Chapter 1

Theorectical background

Magnetoresistive sensors have been used for several applications, for instance read head sensors, hall effect
sensors, biosensors. This thesis focus on the use of magnetoresistive sensors for low magnetic field detection,
where a spin valve sensor has been used for this purpose. In addition magnetic flux concentrator have been
considered in order to improve the sensitivity of the linear response in the low field range.

1.1 Magnetism

Magnetism is the study of interactions of magnetic fields in matter. Outside of a material, i.e. in vacuum,
the induction B-field and the magnetic H-field are indistinguishable. They only differ in their units and
magnitudes and not in how they vary with location and time. It is only inside a material that the difference
is important. The B-field depends only on currents, both macroscopic and microscopic such as the movement
of electrons around their nuclei, while H depends on the macroscopic currents and a factor that is closely
related to the concept of magnetic charge. Outside matter magnetic field lines never start nor end, they
either close in a loop or extend from and to infinity. Mathematically this phenomena can be described by
the Maxwell equation below:

∇.B = 0 (1.1)
The property of the divergence of a vector field represents how the field ’flows’ outward from a given
point. In this case divergence in null. Such vector fields are called solenoidal vector fields. This property
is called Gauss’s law for magnetism and is equivalent to the statement that there are no magnetic charges
or magnetic monopoles. The electric field, outside matter, on the other hand begins and ends at electrical
charges so that its divergence is non-zero and proportional to the charge density.
However the phenomena inside matter is slightly different. Maxwell’s equation for the divergence of the
magnetic field maintains its validity. In addition in micromagnetism it is considered that all magnetic fields
are quasi static, meaning all the changes in magnetization take a long time and can be considered almost
static, hence its derivative in time can be neglected. Furthermore it is considered that no currents pass
through the material, J=0, and the electric field inside the material is null. Given this considerations, the
Maxwell equations can be written:
∇.B = 0 ∇×H=0 (1.2)
The magnetic H-field differs from B-field, because it treats the magnetic field due to the material differently
from the magnetic field due to external sources. B field lines always form loops around the total current, both
the external ’free currents’ and the internal ’bound currents’, meaning they behave similarly either outside
and inside the material. Whereas H-field re-factors the bound current in terms of ’magnetic charges’. The H

1
field lines loops around, ’free current’, but they begin and end at magnetic charges, near magnetic poles, as
well. Thus the magnetic induction B is given by:
B = µ0 (H + M) (1.3)
Where µ0 represents the magnetic permeability in vacuum, µ0 = 4π ×10−7 N/A2 , and M is the magnetization
of the material.

1.2 Magnetic materials

In nature magnetic materials can be categorized into five main groups:

• Diamagnetic
• Paramagnetic
• Ferromagnetic
• Antiferromagnetic
• Ferrimagnetic

1.2.1 Diamagnets
The ones called diamagnetic are composed of atoms whose orbitals are complete and all electrons are paired.
Therefore each atom has no permanent magnetic moment in the absence of a magnetic field. However in
the presence of an external field the electron’s spinning produce a weak and negative contribution to the
material’s magnetic response, consequently the magnetic susceptibility for these materials is very small and
negative. The orientation of the material’s magnetization opposes the applied magnetic field, i.e. these
materials are repelled by magnetic fields. Although all magnetic materials are affected by this phenomenon,
they are often covered up by other behaviours which are orders of magnitude stronger (paramagnetism or
ferromagnetism).

1.2.2 Paramagnets
Paramagnetic materials in the absence of an applied field also do not possess a net magnetization. The
existing unpaired electrons are randomly oriented due to thermal motion in such a way that the magnetic
moment of each atom is cancelled out by the sum of all the other atoms and the total magnetization is
zero. Nonetheless in the presence of an increasing applied field there is a weak magnetization of the material
because a fraction of the unpaired spins will align with the field in an increasing way, creating a linear
dependence between the magnetic field strength and the total magnetization.

M = χH (1.4)
The magnetic susceptibility χ measures the slope between this relation. In the case of paramagnetic
materials the response is positive and small, whereas for diamagnetic materials the susceptibility is also small
but negative. For these materials the total magnetic induction will be proportional to the H-field:

B = µ0 (1 + χ)H = µH (1.5)
The proportional constant is given by the permeability of the material µ:

µ = µ0 (1 + χ) = µµr (1.6)

2
1.2.3 Ferromagnets

In the particular case of ferromagnets their behaviour cannot be described by the linear relation 1.4, since
they retain a residual magnetization even in the absence of an applied field. The atoms of ferromagnetic
materials have partially paired electrons in non-valence orbitals, and there is a difference in the energy level
available for spin up and spin down electrons, creating an excess of electrons with a certain spin. The overall
effect of these orbitals gives rise to a spontaneous magnetization even at zero applied field. Analogously
to the paramagnetic materials in the presence of an external field the magnetic moments align in the same
direction as the field, thus the magnetic susceptibility is also positive, but some order of magnitude larger.
However the magnetic moment response as a function of the applied field is no longer linear for ferromagnets.
Instead the total magnetization follows an hysteresis curve represented in Figure 1.3. Ferromagnets above a
certain fixed temperature, known as the Curie temperature will start to behave like a paramagnetic material,
because of the competing effect of the thermal vibration of the atoms with the exchange interaction. This
temperature value varies with the material and is significantly higher than room temperature. For instance
materials like Co, Fe, Ni have Curie temperatures of 1388 K, 1043 K and 627 K respectively.

1.2.4 Ferrimagnets

Ferrimagnetic materials have opposing magnetic moments due to the existence of different sublattices, i.e.,
different materials or ions, one moment adds to the field and the other subtracts, but the overall result
leads to a positive magnetic susceptibility. These materials are similar to ferromagnets in that they hold
spontaneous magnetization at zero field below the Curie temperature. Above this temperature they also
have no magnetic order and behave like paramagnets.

1.2.5 Antiferromagnets

Antiferromagnetic materials have a magnetic ordering in which the magnetic moments of consecutive atoms
align in opposite directions. This antiparallel alignment is caused by quantum mechanical exchange forces
discussed in section 1.3.2. In the absence of an applied field the total magnetization is zero. The magnetic
susceptibility is small and positive below a certain temperature, called Néel temperature, which is relatively
larger than room temperature. Above this temperature the material behaves like a paramagnet, where the
entropy wins over the antiferromagnetic tendency to align the atoms.

1.2.6 Ferromagnetism

Ferromagnets can exhibit permanent magnetization in the absence of a magnetic field, since the unpaired
electrons have spontaneous magnetization, however the bulk material may be found with small or no magne-
tization. Ferromagnets tend to have parallel alignment of their neighbouring magnetic dipoles. At a larger
range the tendency for dipoles to anti-align overcomes this exchange interaction, and the material divides
into domains, know as Weiss domains. Ferromagnetic materials spontaneously divide into magnetic domains
in order to minimize the energy configuration. Considering a thermodynamical model, four main terms ap-
pear in the energy contribution for the magnetic properties and domain formation: the Zeeman energy, the
exchange energy, the anisotropic energy and the demagnetizing energy.

3
1.3 Ferromagnetism

1.3.1 Zeeman energy


The external applied field can influence the magnetization of the ferromagnetic material. Considering the
theoretical dipole model where µ represents the dipole vector, the Zeeman energy can be calculated by:

eZ = −µ.Ha = −µHa cos(θ) (1.7)

The momentum µ represent the atom’s momentum. The Zeeman energy corresponds to the energy of
interaction between the dipole rotated θ degrees from the field direction. The minimum of energy will be
reached when the magnetization is fully aligned with applied field Ha . The overall energy can be obtained
for large number of atoms:
Z
EZ = − eZ dV (1.8)
V

This energy term is responsible for the classical property that allows magnetic dipoles to anti-align, since the
magnetization vector from one dipole will tend to align with the field created by other dipole as depicted in
Figure 1.1.

Figure 1.1: Classical tendency for magnetic dipoles to align in opposite directions.

1.3.2 Exchange energy


The ferromagnetic character of the materials is mainly due to the exchange interaction between its atoms.
This interaction is a quantum effect resulting from the spin-spin interaction of two electrons and it is the result
of the fact that the wave function (Ψ) of two indistinguishable particles is subject to exchange symmetry,
that is, the wave function describing two particles must either remain unchanged, in the case of boson with
symmetric wave functions, or change sign upon an exchange, in the case of fermions with antisymmetric wave
functions. The spin-statistics theorem of quantum field theory demands that all particles with half-integer
spin behave as fermions and all particles with integer spin behave as bosons. By the Pauli exclusion principle,
however, two fermions cannot occupy the same state. This means that the overall wave function of a system
must be antisymmetric when two electrons are exchanged.
Each ferromagnetic atom will tend to align with its neighbours due to the spin interaction between
the nearest neighbours. The interaction energy between two neighbour atoms with spin Si and Sj is often
described by the Heisenberg model [6]:
eex = −2Jij Si .Sj (1.9)

4
The energy is proportional to the coupling constant Ji j between the two atomic spins. This constant is given
by the exchange integral:
Z
Jij = Ψ?i (ri )Ψ?j (rj )V (ri , rj )Ψj (rj )Ψi (ri )dri drj (1.10)

Where V stands for the exchange potential. Considering Jij a constant throughout the material. When J is
positive, the exchange energy favours electrons with parallel spins; this is a primary cause of ferromagnetism
[7]. In fact, when the interaction potential is purely due to Coulomb repulsion of electrons , J is always
positive, unless the wave functions do not overlap at all, in which case J is zero. When J is negative, the in-
teraction favours electrons with antiparallel spins, causing antiferromagnetism. Although these consequences
of the exchange interaction are magnetic in nature, the cause is not. It is due primarily to electric repul-
sion and the Pauli exclusion principle. Indeed, in general, the direct magnetic interaction between a pair of
electrons, due to their electron magnetic moments, is negligibly small compared to this electric interaction,
that is, at a small scale the exchange interaction overcomes the classical tendency for magnetic moments to
anti-align. The total exchange energy is the sum over all pairs of nearest neighbours, which can be extended
to the continuum generalization:

JS 2
Z Z X ∂mi
Eex = eex dV = | |2 dV (1.11)
V a i,j=x,y,z
∂rj

Where a is the lattice parameter, mi the magnetization of each atom and rj the distance of the neighbour.
The spin vectors S along with the coupling constant J are considered constant throughout the material. This
energy term will introduce a preference for the atoms to remain aligned with each other. Nonetheless this
interaction only dominates at short distances of the order of the exchange length Lex , that is, the distance
which the magnetization can be considered constant.
r
A
λex = (1.12)
Km

Where A represents the exchange constant A = JS 2 a, Km is the energy density given by:
1
Km = µ0 Ms2 (1.13)
2
For the typical case of permalloy (NiFe), often used as a ferromagnetic layer in spin valves, the saturation
magnetization is Ms = 8 × 105 A/m and the exchange length is about 5.7nm [8]. This length defines the
distance limit which can be considered a single domain. For distances much larger than this value λex
multiple domains arise and the magnetization is no longer along the same direction. This length defines
the range of action of the exchange interaction. In the particular example of the spin valve stack, it is the
exchange force that is responsible for pinning the ferromagnetic layer with a contiguous antiferromagnetic
layer. Figure 1.2 illustrates the exchange bias effect acting on the ferromagnet/antiferromagnet interfaces.
The exchange interaction forces the spins to align thus creating a magnetization in the absence of an
external field. If the intensity of the external field varies and the direction reverses, the magnetic response of
the ferromagnetic material will trace a hysteresis loop. When the intensity of the applied field is swept from
negative values to positive ones a curve describing the magnetization will be traced. This curve will not be
retraced when the applied field is swept in the other direction. In fact, for the same values of applied field
but in opposite directions the magnetization will not be the same. This property is called hysteresis and it
is due to the presence of magnetic domains. This is a very useful property which is explored for magnetic
memories and sensors. Figure 1.3 illustrates a typical hysteresis loop. There are several important points
in this curve. When the applied field (H) is increased there is a motion of the domain walls, enlarging the
domains with an orientation favourable to the applied field. After a certain field value the magnetization of
each domain is completely aligned with the field. At this point we say the material has reached saturation

5
Figure 1.2: Exchange bias effect pinning the ferromagnetic layer of a spin valve for both orientations of the
magnetization.

Figure 1.3: Hysteresis curve. Coercive field (Hc ), saturation magnetization (Ms ), and residual magnetization
(Mr ) are represented.

magnetization (Ms). In a second step the field is decreased until it reaches zero. The magnetization will not
accompany this decrease at the same rate, because there will be a residual magnetization (Mr). This residual
magnetization is what characterizes permanent magnets. Finally the magnetization can be brought back to
zero when a strong enough field is applied in the opposite direction of the first one, thus demagnetizing the
material. This field is known as the coercive field (Hc).

1.3.3 Anisotropic energy


Magnetic anisotropy is the direction dependence of a material’s magnetic properties. Crystalline materials are
magnetically anisotropic because their atomic dipoles orient preferably along a certain axis of the crystalline
structure. A magnetically isotropic material has no preferential direction for its magnetic moment in zero
field, while a magnetically anisotropic material will align its moment with the main crystallographic axis
of the crystalline structure,this direction defines the easy axis of the material. The energy applied to the
material that causes the magnetization rotation out of the easy direction is defined as the magnetocrystalline
anisotropy energy Ek and can be expressed a series expansion of trigonometric functions of the angles that

6
the magnetization vector, K, makes with the easy axis [9]. Considering only one easy axis and taking into
consideration only the first term of the series expansion, the anisotropy energy can be expressed as:
Z
Ek = (K sin2 α)dV (1.14)
V

Where K has the units of an energy density [J/m3 ] and α represents the angle between the magnetization and
the easy axis. Indeed, the Weiss domains of atoms are to some extent coordinated in a certain spontaneous
direction, which indicates that this direction of the easy axis is the lowest energy state configuration. In
addition this term will enable the hysteresis behaviour of ferromagnets. When the magnetization rotates to
directions outside the easy axis, the material offers a certain resistance to this change, meaning that there is
an anisotropy regarding the probability for orienting the magnetic dipoles of the crystalline substance. When
increasing this anisotropic energy the field’s coercivity Hc on the hysteresis curve will also increase.

1.3.4 Demagnetizing energy


The demagnetizing energy Ed is due to the presence of the material. It is related to the interaction associated
with magnetization in its own self-field. The potential energy per volume of magnetization M in an external
field B is given by u = −B.M . In the case where energy is due to the demagnetizing field Hd , the total
energy term is given by: Z
µ0
Ed = − Hd .M dV (1.15)
2 V
This energy term is an intrinsic effect of the own magnetization of the material, hence the factor 1/2.
Considering the demagnetizing field Hd generated by the material’s magnetization M, it is possible to rewrite
the Maxwell equations in 1.2 as:

∇ × Hd = 0 ∇.Hd = −∇.M (1.16)

The fact that the curl of the demagnetizing field is zero gives that the field can be described in terms of a
scalar potential, which can be denominated as demagnetizing potential:

Hd = −∇Vd (1.17)

Considering the analogous case on the polarization and the electric charges, it is possible to compare with the
magnetic case. If we have a material with a certain magnetization M, this magnetization will create effective
magnetic charges on the surface of the material normal to its vector direction. The charges difference on both
surfaces will create a potential, Vd and induce a demagnetizing field, Hd , see figure 4.3. The demagnetizing
potential goes from the negative effective charges to the positive charges, hence having its vector normal to
the surface, whereas its gradient lays in the direction parallel to the surface, and so the same happens to the
direction of the demagnetizing field, which can be deduced from equation 1.17. The effective charge density
can be defined by the Poisson equation:

∇2 Vd = ∇.M = −ρ (1.18)

Where ρ is the effective charge density at the surface normal to the direction of the magnetization of the
material. At the surface interface, where the magnetization goes from M to zero, the boundary conditions
can be derived from the equation 1.18:
∂Vd ∂Vd
|out − |in = −M.n = −σ (1.19)
∂n ∂n
When the magnetization M is normal to the surface the density of magnetic charges reaches its maximum
value, as written by the dot product above, where n is the vector normal to the surface. These magnetic

7
charges arise from the demagnetizing field Hd , which is also maximum when the magnetization is normal
to the surface. However Hd is parallel to the surface of the material as given by equation 1.17, and so the
demagnetizing field will make the magnetization rotate towards a parallel direction, see equation 1.15. The
minimum energy configuration corresponds to the rotation of the magnetization dipoles in order to create a
minimum of magnetic charges, causing the material to be divided into different domains, oriented in opposite
directions.
This energy term is thus responsible for the material to break apart into different magnetic domains.
Although the exchange energy keeps the material with the same magnetic orientation, ferromagnets still have
Weiss domains, because their exchange interaction is a short range one, whereas the demagnetizing interaction
is more significant at larger distances. The exchange and anisotropy interactions are also responsible for the
formation of magnetic domains. The ferromagnetic material divides itself into magnetic domains to reduce
the demagnetizing field, in order to minimize the demagnetizing energy. Actually, in a ferromagnetic material
with no applied field, the existence of several domains that cannot align their magnetic moments parallel to
each other, decreases the magnetostatic energy as can be seen in figure 1.4. If it was possible, the material
would divide itself in as many domains necessary as to completely eliminate the external demagnetizing field.
However because of the exchange interaction this will not happen. Typically within the range of the exchange
length the parallel alignment of the magnetic moments prevails. The exchange and demagnetizing energies
influence the domain size and also the domain walls thickness. Within a domain wall, the magnetization
must rotate from the orientation of the first domain to the orientation of the second. These domain walls do
not correspond to an abrupt change in the spins orientation, indeed they correspond to regions where there
is a gradual rotation of the magnetic moments . Their width is of the order of 100 nm.

Figure 1.4: Domain formation.(a) Single domain. (b) two domains. (c) four domains.

The properties of the domain walls, especially their interaction with defects, but also other domain walls,
determine most of the magnetic properties of ferromagnets. The structure of a domain wall determines how
the magnetization changes from one direction to another. There are two geometric ways of achieving this
goal. The kind of wall to be found in real magnetic materials is the one that has the smallest free energy. In
most bulk materials,the type of wall is the Bloch wall, where the magnetization vector turns bit by bit like
a screw out of the plane containing the magnetization to one side of the Bloch wall. In thin layers, however,
Neél walls will dominate. The reason is that Bloch walls would produce stray fields, while Neél walls can
contain the magnetic flux in the material. Figure 1.5 shows the rotating spins orientation within the wall for
both types of configuration. The anisotropic energy forces the domains to align with the easy axis direction,
thus in between two domains, for instance with anti-parallel alignment, the magnetic dipole will have to
rotate 180◦ . Outside the easy direction the anisotropic energy is not minimum, thus this anisotropic term
makes the wall’s thickness small, whereas the exchange energy makes it larger, because of the tendency for
aligning neighbouring dipoles. The overall balance between these interactions define the wall’s thickness as
given by equation 1.20. p
δ = π A/K (1.20)
Where A is the exchange constant seen in equation 1.12 and K represents the anisotropy constant in equation
1.14.

8
Figure 1.5: Gradual change of the spins orientation within a domain wall.(left) Neél wall. (right) Bloch wall.

1.4 Magnetorresistive sensors: Spin Valves

Magnetoresistive devices are by definition devices in which their electrical resistance changes when an external
magnetic field is applied. This effect is due the Lorentz force which influences the deflection of the charge
carrier’s path, producing changes in the resistance of the material. The magnetoresistance of conventional
materials enables changes in resistance of up to 5%, but materials featuring colossal magnetoresistance (CMR)
may demonstrate resistance changes by orders of magnitude. Changes in resistance higher than 1000 % can be
achieved at room temperature [25] [26]. This large magnetoresistance is commonly accompanied in CMR by
high coercive fields as well as large saturation fields. To measure the magnetoresistance effect the resistance
of the material is measured while varying the applied magnetic field. To calculate the magnitude of the
magnetoresistance effect expression 1.21 is used.

Rmax − Rmin
M.R.[%] = × 100 (1.21)
Rmin

Where MR is the intensity of the magnetoresistance effect in percentage and Rmax and Rmin are the maximum
and minimum values of the resistance in Ohms.

Anisotropic Magnetoresistance

Anisotropic Magnetoresistance (AMR) is the property of a material in which a dependence of electrical


resistance on the angle between the direction of electric current and orientation of magnetic field is observed
[28]. Equation 1.22 describes this dependence in terms of the material resistivity ρ, and the angle θ that the
electrical current does with the magnetization.

ρ = ρ0 + δρcosθ (1.22)

The anisotropic magnetoresistance reaches its minimum when the direction of electrical current is perpendic-
ular to the applied magnetic field and has its maximum when the current flows parallel to the applied field.
The physical origin of the AMR effect is believed to be the spin orbit coupling on the 3d orbitals. The effect
is attributed to a larger probability of s-d scattering of electrons in the direction of magnetic field, which
leads to different scattering cross-sections for parallel and perpendicular orientations [29]. The AMR effect
can be observed in ferromagnetic materials, reaching a few percent at room temperature, and, in bulk alloys
of NiFe and CoFe, it can reach 5%. In thin films, the AMR effect is smaller than in bulk, due to additional
scattering (at grain boundaries, film boundaries, etc.). AMR up to 50% has been observed in some ferromag-
netic uranium compounds [27]. To compensate for the non-linear characteristics and inability to detect the
polarity of a magnetic field, a somewhat more complex structure is used for sensors. It consists of thin films
of NiFe, or permalloy as it is most commonly known, deposited on silicon wafers and patterned as a resistive
strip. For instance, thin films of permalloy (Ni81-83Fe19-17) show an AMR of around 2% which, combined
with its low magnetostriction, successfully enables its use as a read element in MR read heads [30].

9
Giant Magnetoresistance

The giant magnetoresistive (GMR) effect was first observed in 1988 by Fert [31], when measuring the electrical
resistance of a multilayer system, at low temperatures. Room temperature GMR was discovered by Grünberg
[32] with resistance variation of about 1.5% on (Fe/Cr) trilayers. In the studied Fe/Cr/Fe layers was observed
a significant decrease in electrical resistivity of the successive iron layers, when their relative direction changed.
This phenomenon can be explained by the interlayer coupling exchange, which favours the anti-parallel
alignment of consecutive Fe layers. In the absence of an applied field and thanks to the Cr layer, which is
the interlayer coupling exchange layer, the magnetizations of the Fe layers spontaneously align anti-parallel
to each other increasing the electrical resistivity of the multilayer system. In the presence of an external
field, the layers tend to align parallel with the applied field. The layer with magnetization opposite to the
field direction rotates as the magnitude increases. This rotation thus makes the system’s resistivity decrease
until the minimum value is reached, corresponding to the state where all magnetizations are aligned with the
field. The GMR effect is the result of spin dependent conductance in ferromagnetic layers and spin dependent
scattering at interfaces, explained by the “two current model”, originally proposed by Mott [36] [37], and later
extended by Fert [38], [33], and further studies were done by Fert and Valet [34]. According to this model,
spin flips scattering processes usually take place in larger time scales compared with other processes, so this
may be considered the major contributor to the electrical resistance in ferromagnets. In addition it can be
assumed that electrons conserve their spin such that electrons with different spins behave independently. The
real event giving rise to the scattering phenomena can be further explained through spin polarised electron
transport. Since only electrons near the Fermi energy level take part in the transport, most of the current will
flow through the non-magnetic material, which is usually the layer with the lowest resistivity (Cu). Since it is
a non-magnetic (NM) material, the current is said to be spin independent, because it is composed of an equal
number of conductance electrons with spin up and with spin down, which lie close to the Fermi level, see
Figure 1.6. Inside a ferromagnetic layer the electron’s scattering probability will increase for one of the spin
orientations and decrease for the other depending on the magnetization of the ferromagnetic layer. In the

Figure 1.6: Density of states for the spin up and spin down states in the case of a metallic spacer (left), and
in the FM/NM interface (middle) and (right). (middle) shows a preferable spin up orientation in the vicinity
of the Fermi level, whereas (right) show a preference for spin down orientation.

interface of FM/NM materials the density of states for spin orientation near the Fermi level is asymmetric.
Depending on the magnetization of the ferromagnet, one of the spin orientations will outnumber the other.
Since scattering is dominant by collisions between electrons with the same spin, the scattering probability
for one orientation will be enhanced, depending on the ferromagnet magnetization, so the current is said to

10
be spin-polarized.
The “two current model”model can be understood with a simplified analogous interpretation using resis-
tors instead of scattering events. The current is considered as a sum of two independent parts for the spin
projections: the part that is parallel to the ferromagnet layer’s magnetization, being less scattered and the
other is anti-parallel to the magnetization, being more scattered. Figure 1.7 illustrates the scattering event
occurring with parallel and anti-parallel states. The spin up and spin down states are considered as an electric

Figure 1.7: (a) Schematic representation of the spin-valve effect in a trilayer film of two identical ferromagnetic
layers F1 and F2 sandwiching a non-magnetic metal spacer layer M, the current circulating in plane. When
the two magnetic layers are magnetized parallel (lower scheme), the spin-up electrons (spin antiparallel to
the magnetization) can travel through the sandwich nearly unscattered, providing a conductivity short cut
and a low resistance. In contrast, in the anti-parallel case (top scheme) both spin-up and spin-down electrons
undergo collisions in either F1 or F2, giving rise to a higher overall resistance.(b) Effective resistor model
scheme for parallel and antiparallel resistances.

circuit in parallel configuration, with scattering probability depicted as the resistance of each resistor. The
equivalent system resistance is given by ??. When the FM layers are in the anti-parallel state, the system
is in its high resistance state, because the resistance of each spin projection is equal, Rspindown =Rspinup .
Whenever the FM multilayers align parallel with each other the system is in its lowest resistance state, since
Rspindown >Rspinup (in this external field orientation). Therefore for the spins which are strongly scattered
the contribution of the electrons to the current will be small and for the spin’s orientation which is weakly
scattered the electron contribution to the current flow will be large. As we can see in Figure 1.7, for one
of the spin orientations, both the resistances are now small and, for the other, both the resistances are now
big which translates in a larger current flow of one of the spin orientations, and a smaller overall system
resistance.

Spin Valves

One of the best known applications of GMR is the Spin Valve first proposed in 1991, by Dieny [35]. This
arrangement consisted of a set of two ferromagnetic layers separated by a thin non magnetic metallic layer,
and an antiferromagnetic material contiguous to one of the ferromagnetic layers. To observe the GMR effect
between two ferromagnetic layers, there must be the ability to rotate their magnetizations separately. This
is done by pinning the magnetization of one of the ferromagnetic layers, which can therefore be achieved
by exchange coupling with an adjacent antiferromagnetic layer [40] [41] [42]. The exchange bias interaction
is crucial for defining the preferred direction of the pinned layer magnetization. It can be understood as if
a strong magnetic field, the exchange field H ex , acts on the pinned layer. This mechanism was discovered
by Meiklejohn and Bean in 1957 [45]. The pinned layer magnetization can be defined during growth of

11
the antiferromagnetic layer in the presence of an aligning field for the case of top pinned structures (when
the pinned layer is defined above the free layer). Thermal treatment can also change the pinned preferred
magnetization orientation. When the system is heated above the blocking temperature the exchange field
effect disappears. For the antiferromagnetic layer this temperature is considered as the Néel temperature, at
which the antiferromagnet starts to behave as a paramagnet. Above this temperature if an external magnetic
field is applied the magnetization of the spin valve’s layers can be defined. This process is also commonly
referred as thermal annealing and it is used in bottom pinned structures, and also in top pinned structures,
enhancing their response. Figure 4.11 shows an example of the magnetization hysteresis loop when the
magnetization as a function of the applied magnetic field. Often spin valve structure comprises buffer layers
in order to influence the microstructure of the film, its grain size, interface smoothness, and texture. Also a
thin cap layer is used to protect the structure from corrosion and oxidation. In terms of electron transport,
in the considered configuration, the electrical current flows parallel to the plane of the layers, known as a
Current In Plane (CIP) geometry. This current flow and also the resistance of the device can be controlled
by an external applied magnetic field. Figure 1.8 shows a Spin Valve structure, with the correspondent
layer’s magnetization and external applied field. CoFe and NiFe are usuall choices for the ferromagnets for

Figure 1.8: Spin Valve structure. Anti-ferromagnetic layer (AFM) and pinned layers’ magnetization are
represented. Free layer magnetization accompanies the external applied field direction towards the hard axis
of the pinned layer.

their combination of high MR and soft magnetic properties while Cu is usually chosen as the metal. The
second ferromagnet is left free in order to align itself with the external magnetic field. So the applied field
controls the resistance state of the device. In the parallel state the resistance is at its minimum while in the
anti-parallel it is at its maximum. The magnitude of the GMR effect is expressed in the GMR ratio, defined
in equation ??.
R − R⇒ Rspinup Rspindown
GM R = = (1.23)
R⇒ 4Rspinup Rspindown
This expression will be used throughout this thesis as a crucial measurement of the devices in study, where
R⇒ and R are the resistances in the parallel and antiparallel sate, respectively. This expression will be used
throughout this thesis as an important measurement of the devices in study. Typical spin valve structures
only reach a magnetoresistance value of the order of 6% at room temperature [44]. The maximum resistance
obtained was of 20% and is was measured by N. Hasegawa et al. [43], using additional nano-oxide layers next

12
to the ferromagnets. At INESC-MN typical MR values of 10% are obtained in MnIr-based spin valves using
Ta buffers [15].

Interlayer coupling

A pinning mechanism can be used to set the polarization of one of the magnetic layers into a fixed direction
while the direction of the magnetization of the other ferromagnetic layer can freely rotate with the applied
field. There are several mechanisms to set the magnetic layer. During our work we have used a simple 4
layer structure with only a pinning AFM layer, pinned FM, and free FM layer separated by a non magnetic
spacer. Nevertheless more complex structures could have been used, such as a synthetic antiferromagnetic
(SAF) structure, to set the ferromagnetic layer into its pinned state. In the SAF structure the antiferro-
magnetic layer directly pins the adjacent ferromagnet. On the other hand this ferromagnet is pinned to
the another ferromagnet separated by an interlayer spacer. This scheme provides a stronger pinning than
having just an exchange bias material, which could bring advantages when integrated with the magnetic flux
concentrators, since the pinned exchange field would increase allowing a later rotation of the pinned layer
and thus, higher magnetoresistance values in the transfer curves obtained, see Figure ??. The SAF structure
was first introduced in 1993 by Heim and Parkin [49], for spin valve sensors and a couple of years later V.
D. Berg introduced [50] the concept of synthetic antiferromagnetic structures which were composed of two
ferromagnetic layers separated by a non-magnetic layer, thus creating an alternative coupling mechanism to
the simple anti-ferromagnet/ferromagnet coupling. For small values of the external applied field the magne-
tization of the spin valve’s free layer rotates following a hysteresis loop, see Figure 1.2. This curve can be
shifted from the origin due to magnetic interlayer coupling between the ferromagnetic layers [46] [47]. The
parallel and antiparallel state of the free ferromagnetic layer is determined by two types of coupling forces the
Néel Coupling and the RKKY Coupling. The Néel coupling exists in all practical ferromagnetic structures,
such as SAF or spin valve structures.It is associated with the conformal roughness of the ferromagnetic inter-
faces,i.e., the waving profile that inevitably arises during the deposition of the materials. The magnetostatic
interactions between the free poles at the ferromagnetic interfaces next to the non magnetic barrier cause a
ferromagnetic coupling, which is responsible for shifting the off set field to negative values.

Figure 1.9: FM/NM/FM waving profile of the Néel coupling interaction.

The Ruderman-Kittel-Kasuya-Yodsida (RKKY) coupling or indirect oscillatory exchange interaction de-


scribes the ferromagnetic or antiferromagnetic coupling between two ferromagnetic layers separated by a non
magnetic spacer. In fact the interaction oscillates from ferromagnetic to anti-ferromagnetic coupling depend-
ing on spacer thickness as is illustrated in figure 1.10. This coupling effect shifts the offset field to positive
values whenever the coupling is a antiferromagnetic and to negative values when it is a ferromagnetic. In the
SAF scheme the ferromagnetic layers are coupled by RKKY coupling and Néel coupling being that the Néel
coupling effect is several orders of magnitude lower than the RKKY coupling. In a spin valve simple scheme
the spacer thickness is about 22Åso the RKKY effect is almost non existent, while the Néel coupling is orders

13
Figure 1.10: RKKY interaction: Oscillation of ferromagnetic/anti-ferromagnetic coupling with the thickness
of the spacer. The maximums correspond to ferromagnetic coupling whereas the minimums correspond to
antiferromagnetic coupling.

of magnitude stronger. Other coupling effects, such as the direct ferromagnetic coupling, which results from
pinholes in the non magnetic spacer. This coupling effect could destroy the spin valve effect by disabling the
independent rotation of the two ferromagnetic layers. Being a ferromagnetic coupling, the offset field would
then be shifted towards negative values. The demagnetizing field created by the ferromagnetic layers also
induces an antiferromagnetic coupling on the free layer, driving the offset field towards positive values.

Linearity

Spin valves are used in lots of applications including hard disks read heads, low field detection applications
and magnetoresistive sensors. Depending on the application the magnetizations of the free and pinned
layers can be set parallel (hard disks) or perpendicular (MR sensors). In fact, the magnetic response of the
ferromagnetic layers to the applied field is of crucial importance. For instance, a linear response with no
coercivity will be important for magnetoresistive sensors, in particular for this work. There are two main
processes of inducing a linear response. The first is to deposit the spin valves with crossed anisotropies, while
the second one is to control the sensor shape in such a way that the created demagnetizing field would change
the free layer’s magnetization, enhancing the linear response of the output signal. Crossed anisotropies refer
to the orientation of the pinned and free layers magnetization, see spin valve configuration of Figure 1.8.
The pinned layer anisotropy is fixed by exchange coupling with the antiferromagnetic layer, whereas the free
layer’s magnetization is set in the perpendicular direction, known as the hard direction of the pinned layer.
This configuration allows a linear slow rotation of the free layer’s magnetization when an external field is
applied along the pinned direction, see Figure 4.15. In the case where the anisotropies of the free and pinned
layers are set parallel to each other, the induced response is in the form of a sharp transition.

1.4.1 Soft magnetic materials


Soft magnetic materials refer to materials whose magnetic properties make them ideal for magnetic head
cores, toroidal cores, and magnetic shield parts. Their magnetic characteristics consist of high magnetic
permeability, low coercivity near zero magnetostriction, and significant anisotropic magnetoresistance. The
low magnetostriction is critical for industrial applications, where variable stresses in thin films would otherwise
cause a ruinously large variation in magnetic properties. Permalloy is known as the most commonly used soft
magnetic material. This term refers to a Nickel-Iron magnetic alloy. In this work, however, the soft magnetic
material used is not a permalloy, but a CoZrNb metallic film. This choice is the most appropriate since

14
the device was fabricated with a thin film, requiring good adhesion and cohesion properties for such small
dimensions. The purpose of using this type of material is mainly because an increase of the magnetic field
would enhance the detection limit of the MR sensor. The CoZrNb material would concentrate the magnetic
field lines at the sensor level due to the geometric shape of the fabricated structure, designated by magnetic
flux concentrator (MFC). With this structure the sensitivity of the spin valve would increase. Devices with
a 2D geometry have been fabricated [1]. This work aims at achieving higher sensitivities for a 3D tapered
profile obtained by ion milling processes.

15
Chapter 2

Simulations

Magnetic flux concentrators have been used to enhance the field sensitivity of several types of magnetic
sensors, spin valves [11], magnetic tunnel junctions [2] and hall effect sensors [12]. This concentrators are
made of soft ferromagnetic materials, having high relative magnetic permeability, low coercivity and an almost
linear B-H relationship. The magnetic permeability is given by the B-H curve’s slope, equation 1.6. This
quantity is commonly referred to the material’s strength to attract the magnetic field lines. This effect occurs
due to the abrupt change of permeability (from air to the magnetic material), resulting in a bound surface
current distribution that induces an additional magnetic field, ∇H, perturbing the homogeneous external
field. The relative permeability of CoZrNb is high (µr >> 1) so the flux concentrator, as the name suggests,
tends to attract the magnetic field lines towards itself, concentrating the magnetic field at the sensor region,
leading to an increase of the sensor’s magnetic field detection limit with a gain factor of G, as shown in Figure
2.1.
Hsensor
G= (2.1)
Hexternal

This concentration gain factor is defined by the ratio between the magnetic field reaching the sensor region
Hsensor , and the external homogeneous applied field, Hext .

Figure 2.1: FEMM image of magnetic flux concentrator geometry (left). Mesh generated to solve the flux
density distribution(right). Simulated parameters: Length of the MFC (L), outer width (A1 ), inner width at
gap region (A2 ), gap between the two MFCs (gap). The concentration gain factor is measured based on two
field values, one at sensor level (Hsensor ) and the other at an homogeneous external field (Hext ).

17
2.1 Magnetic Simulations

This chapter provides the study of the magnetic flux concentrator’s (MFC) geometric gain, simulated with
varying parameters of the flux concentrators(length, thickness, width). The magnetic simulations were per-
formed using a finite element method magnetics software (FEMM v4.2) [10], which is usually used for solving
2D planar problems in low frequency magnetics and electrostatics. The package consists on a mesh generator,
see Figure 2.1, and several solvers. In order to create a model it is necessary to predefine the geometry and
magnetic conditions of the materials. In particular to simulate the soft magnetic material used to fabricate
the flux concentrators (CoZrNb), it was considered a linear B-H relationship and a relative magnetic perme-
ability of µr =250 in each axis. The magnetic concentration gain factor is a function of several parameters,
which define the flux concentrators, see Figure 2.1, thickness (t), gap between flux concentrators (gap), inner
width at gap site (A2), outer width (A1), length (L) and the case of asymmetric MFC. In this study a
simulation of the impact that this parameters have on the gain factor is conducted to better understand the
appropriate MFC geometry to be fabricated. A typical result of the simulation is shown in Figure 2.2.

Figure 2.2: Magnetic field lines evolution with variable width A2 at gap region.

2.1.1 Thickness

The increase of magnetic field lines concentration as a function of the CoZrNb layer thickness behaves as
the curve represented in Figure 2.3. Due to problems of homogeneity, and good magnetic properties of
the film, the fabricated CoZrNb layer is kept at 0.7µm-thick, with a geometric gain factor of 6.79. When
decreasing the layer thickness the amount of magnetic field lines decreases. Figure 2.4 shows the evolution
of the concentration of magnetic field lines for four different thicknesses (side view).

18
Figure 2.3: Gain vs. thickness simulated results

Figure 2.4: Magnetic field lines evolution with variable layer thickness. FEMM image

2.1.2 Gap length

The gap distance between the magnetic flux concentrators influences greatly the factor of concentration, see
Figure 2.5. When the gap increases the value of the flux concentrated at the sensor level, Hsensor decreases.
This value in the micro-fabricated process was kept to its minimum, from 3.5µm to 4.5µm. The gain factor
G factor in this case is expected to be less than 9. The constraints that prevent a further decrease of this
distance are related to the spin valve sensor’s magnetic properties and the lithography mask alignment, for
further explanation see chapter 4.

19
Figure 2.5: Gain vs. Gap between MFC simulated results

Figure 2.6: Magnetic field lines evolution with variable gap distance between MFCs.

2.1.3 Width

The first study as a function of the MFC width refers to the smaller width (A2) at the gap region, see Figure
2.7. The maximum concentration gain factor is achieved with a 2µm structure, though the values are not
too far apart. The fabricated device was set with a width of 8µm and 14µm, because of the spin valve
sensor dimensions, which were needed to preserve a linear magnetic response (create a demagnetizing field
and keeping the free layer’s easy axis perpendicular to the pinned layer’s). The fabricated values have a gain
factor of about 38% smaller than the maximum value obtained with a 2µm width.
The second varying parameter is the outer width (A1 ). The concentration factor increased with increasing
outer width. The overall improvement in the considered range was of about 33%, see Figure 2.9. Fabricated
flux concentrator widths (A1 ) varied from 30,50,80,100,300, and 500µm, with respectively 27.8, 28.1, 28.5,
28.8, 31.8, and 35.2 gain factor values extrapolated from the fitted equation represented in Figure 2.9.

20
Figure 2.7: Gain vs. Inner Width simulated results

Figure 2.8: Magnetic field lines evolution with variable width A2 at gap region.

Figure 2.9: Gain vs. outer width simulated results

21
Figure 2.10: Magnetic field lines evolution with variable width A1 .

2.1.4 Asymmetric geometry


Asymmetric flux concentrators were simulated 2.11 in order to take into account restrictions of the integrated
scheme [3]. The top flux concentrator has a metallic gate patterned on top of this structure, so the concen-
trator cannot be increased indefinitely. Concentration gain factor, G, is measured at both sides of the MFC,
lef t when it is compared with the external field before the MFC and right, when it is after. Although, the
gain increases in both cases, only the right MFC length changes. The devices fabricated had smaller lengths
600 and 50µm, which corresponds to an extrapolated gain factor of 38.8 (lef t) and 58.2 (right). Figure 2.12
shows the simulated images obtained for different outer widths of the bottom concentrator (right).

Figure 2.11: Gain vs. Asymmetric Width simulated results

22
Figure 2.12: Magnetic field lines evolution with variable width A1 on the right concentrator.

2.1.5 Length
Flux concentrator length increase can be a possible solution to achieve a higher gain factor, see Figure 2.13.
Simulations have been done for values in the 10-100 µm range. Fabricated MFCs have lengths going from 28
to 468 µm, which correspond to concentration gain factors within the interval [4.8,106.3].

Figure 2.13: Gain vs. Length simulated results

23
Figure 2.14: Magnetic field lines with varying length L

The performed simulations were done in order to better understand the involved parameters and their
role on a magnetic level. The study allowed the comprehension of the geometric optimal configuration to be
fabricated. In chapter4 the fabricated devices were chosen in order to achieve a maximum geometric gain,
but also considering the magnetic properties of the materials and fabrication constraints.

24
Chapter 3

Experimental Methods

In this chapter the main fabrication systems needed for the definition of the sensors, contacts and flux
concentrators are described. In addition all the characterization processes and machines used are explained,
both for the electrical and magnetic methods.

3.1 Thin film deposition and Ion milling process

In this work two types of thin film deposition methods were used. The first method consists of a magnetron
sputtering systems, while the second refers to a ion beam deposition systems. In the next sections explanations
of these systems, their functioning principles and differences are explained in detail.

3.1.1 Magnetron sputtering


Magnetron sputtering systems consist of physical deposition of sputtered atoms from a target sample into
the substrate. The sputtering process occurs in a vacuum chamber filled with an inert gas, in which a plasma
is created. When the gas introduced in the vacuum chamber is composed of inert gases such as Ar or Xe, the
material removal from the target occurs mainly due to collisions, and in this case the deposition is said to be
a Physical Vapour Deposition (PVD). On the other hand, when another gas is inserted in the chamber and
reacts chemically with the target surface the deposition is known as a reactive sputtering system. Despite
the fact that the gas is neutral, there are still a few ions moving around the chamber. When a negative
bias voltage is applied to the target, while the shielding surrounding it is grounded, there is a DC potential
difference that attracts the ions of the neutral gas into the target (DC sputtering). Providing that the ions
have enough energy, they will ionize the neutral atoms through collisions before reaching the target. These
ionized atoms will then also be accelerated towards the target, causing new scattering events and more ions
in a cascade fashion. This event will trigger the plasma ignition. Magnetron sputtering systems usually
have an array of permanent magnets placed behind the target with the material to be deposited, see Figure
3.1. The magnets form a magnetic field around the target, that is capable of trapping the electrons created
closer to the surface and confining the ions in the target region. This fact contributes to the increase of
collisions with the neutral atoms and production of more ions, resulting in a more effective plasma ignition
with lower pressure, and also faster removal of material with consequent higher deposition rates. When the
target is made of a non conducting metal, charges accumulate in the material’s surface creating an electric
field shielding the target. The shielding effect prevents low energy ions from reaching the target and removing
the material. In order to avoid this effect, the DC voltage applied to the target is replaced by a RF voltage
(RF sputtering). The RF voltage while negative accelerates the ions towards the target, and when it is

25
positive restores the charge neutrality, by repelling the positive ions from the surface. After the formation of

Figure 3.1: Magnetron sputtering deposition system scheme. Magnets are represented behind the target.
The plasma is confined near the target (coloured region) and the field lines magnetic and electric lines are
specified.

the plasma, ions transfer energy to the system and create heat, either by rearranging the atomic positions
of the target outer layer, creating vibrations, or also due to collisions ionizing the material and transferring
energy into the system. To prevent any damage resulting from over heat a water cooling system is made to
pass behind the target location. When energetic ions collide with the target atoms they will ionize them,
contributing to the release of more electrons to be trapped under the magnetic field. These electrons will
then favour the plasma ionization. If incoming ions are energetic enough to surpass the energy that binds
atoms to the target, they will be responsible for the emission of neutral atoms from the surface material. This
ejected particles will travel unaffected by the electrical or magnetic fields surrounding the target, and will fall
into the substrate placed under the sputtering system, depositing the desired material onto the substrate.

Ultra High Vacuum 1 (UHV1)

UHV1 is a DC sputtering system used to deposit the soft ferromagnetic material CoZrNb, needed for the
definition of magnetic flux concentrators. This machine has only one chamber, that opens to the atmosphere
whenever a new sample is loaded or removed. This fact puts some restriction in the time needed to achieve
pressures within the working limit. To achieve optimum working base pressures of 108 Torr, 8 hours of
pumping are required. There are two pumping systems, a mechanical pump starts working first, reaching
pressures of about 10−6 Torr, then a second system, a turbomolecular pump helps to achieve a higher vacuum,
approximately 10−8 Torr. The samples are placed in a holder containing two strong permanent magnets on
each side, see Figure 3.2(c). The magnetic field created by these magnets has an orientation represented in
Figure 3.2. The magnetic field is essential for the anisotropy definition on the deposited material. A 130 Oe
constant field at the centre of the holder is responsible for setting the material’s easy direction. The target
is placed below the holder, so that the material is deposited through atoms ejected from below into the
substrate surface. The plasma is ignited with 8.5 sccm of Argon gas. A water cooling system passes below
the target to avoid overheating, by cooling down the the machine. Table 3.1 show the deposition conditions
for the CoZrNb target.
The deposition rates were calculated from the deposition of a 7000Å-thick layer, with non-uniformities
of about 500Å, depending on the positioning of the sample within the holder. The material composition is
described in atomic percentage, measured by RBS and PIXE, see section 4.2.3.

26
Figure 3.2: (a) UHV1 machine image. (b) main vacuum chamber door. CoZrNb target location, sample
holder and water cooling system are represented. (c) Sample holder with two permanent magnets located on
both sides of the holder, creating a magnetic field H throughout the area, with the indicated orientation.

material bias voltage power Ar flux pressure deposition rates

Co8 6Zr5 Nb9 at.% 412 VD C 32 W 8.5 sccm 3.0 mTorr 1.19±0.09 Å/s
Table 3.1: Deposition conditions for the CoZrNb target on UHV1

Ultra High Vacuum 2 (UHV2)

UHV2 is an RF sputtering system used to deposit Al2 O3 , used as an insulating layer between Si substrate and
the metallic contacts, and between spin valve (SV) sensors, and CoZrNb flux concentrators. This material
also serves as a protecting passivation layer of the SV sensor. The machine is composed of a single chamber
system, requiring a long time of pumping, approximately 10 hours to reach a working base pressure of 10−7
Torr. Figure 3.3 shows the machine chamber. A six inch diameter target is placed in the top part of the

Figure 3.3: UHV2 machine image.

chamber along with a magnetron of the same size. Below, the samples are placed on the substrate. Deposition
rates are of the order of 0.19Å/s with a 20% inhomogeneity depending on the location of the sample on the
substrate. Table 3.2 shows the deposition conditions for UHV2 with an Al2 O3 target.

27
material power Ar flux pressure deposition rates

Al2 O3 200 WRF 45 sccm 3.0 mTorr 0.19±0.04Å/s


Table 3.2: Deposition conditions for the Al2 O3 target on UHV2

Nordiko 7000

The Nordiko 7000 is an automated sputter system, composed by a central dealer module connected to a
loadlock and four process modules. A robotic arm is placed on the dealer module and allows the transportation
to all four modules and to the loadlock. The modules are separated through gate valves allowing simultaneous

Figure 3.4: Nordiko 7000 machine image. Front view (right) and back view (left) of the machine.

operation of various processes. Depressurization of the loadlock only, allows placing and removal of samples,
with rapid pressure recovering time, of approximately a few minutes. The four modules are designated to
work with 6 inch diameter wafers. The first module is responsible for thermal annealing, the second for
sputter etch, the third for TiW(N2 ) deposition, and the fourth for Al deposition. In this work only modules
2, 3, and 4 were used. Table 3.3 describes the conditions for each module to operate. In module two a soft

module operation power (W) voltage (V) pressure (mTorr) gas flux (sccm) dep. rates (Å/s)

2 soft sputter etch RF1 =70 RF2 =40 105 Ar 50 3.0 -


3 Ti2.5 W50 (N37.5 ) dep. DC=500 431 Ar-50,N2 -10 3.0 5.6
4 Al98.5 Si1 Cu0.5 dep. DC=2000 398 Ar 50 3.0 37.5
Table 3.3: Nordiko 7000 conditions for operations in modules used.

sputter etch is preformed, to remove small amounts of material from the substrate. The soft etch is usually
used before the deposition of TiW(N2 ) and Al in order to remove the material that has been oxidized, ensuring
a better contact between the metal and the substrate. In module three a reactive DC sputter system is used
to deposit TiW(N2 ). This material acts as a protective layer against chemicals, such as micro strip. Ar and
N2 gases are used to ignite the plasma, and help the metal become anti-reflective, when direct lithography is
done. In module four AlSiCu films are deposited as metal contacts to the sensor. Usually a 3000Åthick layer
is used for this purpose.

28
3.1.2 Ion beam deposition and Ion milling process
Ion beam systems are composed of two ion beam guns, a target holder and a substrate table, all disposed in a
’Z’ configuration as shown in the schematic 3.5. These systems make use of highly energetic broad ion beam

Figure 3.5: Ion beam deposition machine schematics

sources, generated by a deposition gun focused onto the target, with the purpose of removing its material
in a non-selective way. The target is usually a dielectric material or grounded electrically if metallic. In
the case of deposition, the material is sputtered from the target by the accelerated ions from the beam. In
the collision process, the energy of the incident ion is transferred to a target atom, and, if high enough, the
atom is sputtered from the target in order to be deposited on the substrate. The deposition gun contains
an RF antenna that supplies the necessary power to ionize the atoms creating the plasma in the vacuum
chamber. A set of three voltage biased grids are used to accelerate and focus the ion beam onto the target.
Because of the significant power used, considerable heating occurs, and so in order to release this heating
and prevent any damage, a water cooling system exists. The system also has a secondary ion beam source
generated in the assist gun, whose role is to provide the films deposited on the substrate with energetic
noble or reactive ions. This can be used to modify the material properties during deposition. Furthermore
two neutraliser guns are used to prevent charge accumulation on the surface of insulating targets, during
depositions or insulating substrates during ion milling. The milling process provides the main dry etch
method used. It is done based on the same principle as the deposition. Focused energetic ions will collide
with the substrate removing its atoms. The major advantage on this kind of systems is that it can reach lower
pressures, around 10−8 Torr, during deposition, which is one order of magnitude lower than the achievable
in magnetron sputtering systems. This occurs because the ions are confined to the beam, so the ion density
in the chamber is significantly lower.

Nordiko 3000

Nordiko 3000 is one of the two ion beam deposition (IBD) systems existent at INESC-MN [14] [15] [16].
This is a fully automated machine comprising a loadlock chamber connected to a dealer chamber with a
robotic transfer arm, and a main vacuum chamber. The loadlock contains 8 slots of 6 inch diameter wafers
and is separated (along with the dealer) from the main chamber through gate valves, allowing a standard
working pressure of 5x10−7 Torr inside. The chamber contains two ion beam guns, two neutralisers, and an

29
hexagonal shaped target, which can rotate in order to let the target with the material to deposit exposed to
the deposition ion beam gun. A shield is placed in front of the other targets to prevent contamination from
other materials during deposition and ion milling. The substrate table has a permanent magnet that creates
a 40 Oe field during deposition, to define the anisotropy of the magnetic deposited layers. The substrate table
also has a shutter that protects the substrate during the guns preparation until all the parameters become
stable, and gives an exact control of etching and deposition times. The table itself rotates, typically at 40% of
30rpm in order to give more uniform deposition and etch processes. It can also be elevated in different angles
between the substrate and the guns (from 0◦ to 90◦ pan). The system works with Ar and Xe gases, and was
used to deposit spin valve sensors as well as performing every ion milling in the microfabrication process.
Table 3.4 illustrates the deposition conditions for each target, and table 3.5, in section 3.2.5 summarizes the
ion milling conditions at different elevation angles.

material deposition rates (Å/s)


Co8 0Fe2 0 0.21
Ni8 0Fe2 0 0.24
Mn7 7Ir2 3 0.18
Ta 0.15
Ru 0.19
Cu 0.28
Table 3.4: Nordiko 3000 deposition conditions. A base pressure of 0.1 mTorr with 3 sccm of Ar flow, was
used in the spin valve deposition.

3.2 Pattern transfer techniques

Patterning methods are essential to define and create microstructures onto thin films, allowing the removal
or deposition of material in specific places of the substrate. These techniques combine photo-lithography, ion
milling and lift-off processes.

3.2.1 Coating tracks


Photo-lithography process involves three major steps: coating of the substrate sample with a photo-resist
(PR) solution, exposure of the PR with a laser writing system, and development of the sample to remove the
unwanted photo-resist.
The coating system comprises a Silicon Valley Group track used to coat the samples with a positive
photo-resist (PFR 7790G 27cP) and afterwards to develop them. The coating track is used to coat samples
with a 1.5µm thick photo-resist layer. The thickness is defined by the rotational speed and time at which the
spinning plate moves. In this case a rotational speed of 2800 rpm during 40 sec is done. After this module, a
heating stage proceeds, allowing the baking of samples, removing the unwanted air molecules trapped in the
photo-resist. Figure 3.6 shows the SVG system used at INESC-MN.

3.2.2 DWL exposure


After sample coating follows the mask exposure. This process is done by direct write laser optical lithography
(DWL), which consists of a 442 nm wavelength HeCd LASER built with the purpose of transferring patterns,
with typical micrometer dimensions onto a photo-resist layer [15]. When exposed to light of certain wave-
lengths the polymeric chains of this material are altered, giving a different property to the exposed material.

30
Figure 3.6: SVG coating and developing system.

Then the material becomes either easier or more difficult to remove, depending on whether it is positive
or negative photo-resist. The laser performs a sweeping of the sample, turning on and off according to a
mask previously designed through software (Autocad). The areas where the laser is turned on, the positive
photo-resist is exposed to light, making its polymeric chains weaker. When developed the PR in these areas
dissolves and disappears. Negative photo-resist would strengthen its chains with light exposure, keeping the
photo-resist when developing.

3.2.3 Developing tracks

The sample is then taken to the developing track where it is initially heated for one minute at 110◦ C. This
process stops uncompleted photo-resist reactions. Then the samples are placed on the developing tracks, and
development liquid (TMA238WA from JSRMicro) is poured onto the sample developing it for approximately
60 seconds. Afterwards, it is sprayed with water to prevent overdeveloping and finally it is dried thanks to
a high speed rotation plate. In the end the defined structures can be observed at the microscope, with an
appropriate filter to make sure the development is complete.

3.2.4 Lift-off process

Another important part of micro-fabrication is to selectively remove or deposit material. To get the desired
pattern on a deposited layer a lift-off technique is required. Before deposition a prior non-inverted mask
(with PR where void spaces will be left after lift-off) needs to be exposed onto a PR layer. In the end the
remaining photo-resist and therefore the excess of material on top of the photo-resist will be removed by a
micro-strip bath (placed in a wet bench). The samples are submersed on this liquid and the photo-resist
starts to degrade until it is completely removed from the sample. Figure 3.7 illustrates the basic steps for
any lift-off process.
Putting the samples in an ultrasonic bath, and keeping the bath temperature at about 60◦ enhances the
removal process. The time needed to remove all the material is dependent on how much time the sample is
kept in the ultrasonic bath, on the areas being removed and on the deposited layer thickness. This process
is used along with ion milling for the definition of flux concentrators, and is also used to define the metallic
contacts to the SV sensor.

31
Figure 3.7: Pattern transfer through lift-off.(a) patterning o f P.R: on the substrate. (b) film deposition. (c)
P.R and film removal after resist strip and ultrasounds procedure.

3.2.5 Ion milling process

Ion milling processes are of utmost importance in this work, since they control the flux concentrators profile.
This is characterized by a physical dry etch where high energy ion bombardment, physically impacts on the
substrate and removes the visible material. This is done at given controlled angles relative to the surface,
20◦ , 45◦ , 60◦ , 75◦ and 85◦ . Figure 3.7 illustrates a schematic of an ion milling process.

Figure 3.8: Ion milling process scheme.(a) photo resist patterning on top of the CoZrNb deposited film. (b)
physical etch of the non protected area. (d) photo resist removal in a resist strip solution. The remain
patterned layer has a defined angle profile of θ degrees with the substrate surface.

The definition of microstructures requires a previous deposition of the desired layer to be patterned. After
mask lithography, the photo-resist protects the areas which will not be bombarded. Those being exposed to
the energetic ions will disappear leaving a delimiting tapered profile, with angles varying according to the
incident ion beam.

Nordiko 3000

Ion milling was performed either in Nordiko 3000 or in Nordiko 3600. This method uses high energy Argon
ions to physically remove material from the stack, at a controllable rate. Afterwards the sample is immersed
in a resist strip solution called microstrip in order to remove the remaining photo-resist. The etch process
by ion beam is not material-selective. Nevertheless, since the thickness of the photo-resist is larger than the
thickness of the stack of thin films (75 times higher than SV stack), the time necessary to remove the desired
quantity of material from the stack will be much smaller than the time necessary to remove all the photo-
resist. However when long milling processes occur, which is the case of CoZrNb layer etch, the thickness
of photo-resist decreases considerably. Nonetheless, the etching rate of the CoZrNb layer is higher than the
photo-resist layer. Table 3.5 summarizes the ion milling conditions at different ion beam incident angles.

32
material angle (◦ ) etch rates (Å/s)
20 0.19
45 1.35
Co86 Zr5 Nb9 60 0.90
75 0.88
85 0.85
SV stack 70 1
Table 3.5: Nordiko 3000 ion milling conditions at different elevation angles. A power supply of 64WRF is
used to feed the plasma beam. The voltage between grids to accelerate the ions is of 488V (V+ ) and -200V
(V− ). The gas flow is about 8 sccm of Ar atoms. The base pressure was of the order of 4x10−6 Torr and the
working pressure reached 2x10−4 Torr. Neutraliser was used with 411V, and 0.1 sccm Ar flux.

Nordiko 3600

Nordiko 3600 is an ion beam deposition system identical to Nordiko 3000, but with the possibility of using 8
inch diameter wafers. This machine has some differences when compared with the previous one. It is com-
posed of three modules, instead of two, separated through gate valves: the main chamber, the loadlock, and
a module of transport were the robotic arm is kept. Since both valves are never open at the same time during
wafer transport, the ability of having lower pressures on the main chamber is improved. Moreover, having
a smaller loadlock makes the depressurization process faster and easier to load and unload the substrates.
This system was mainly used for ion milling of the spin valve stack, since it had slightly better uniformity
than Nordiko 3000. Table 3.6 shows the conditions for the ion milling process with a spin valve stack.

material angle (◦ ) etch rates (Å/s)


SV stack 70 1
Table 3.6: Nordiko 3600 ion milling conditions at 70◦ . A power supply of 204WRF is used to feed the plasma
beams. The voltage between grids to accelerate the ions is of 725V (V+ ) and -345V (V− ). The gas flow is
about 10.2 sccm of Ar atoms. The base pressure was of the order of 5x10−7 Torr and the working pressure
reached 1x10−4 Torr. Neutraliser was used with 340V, and 0.1 sccm Ar flux.

3.3 Film characterization methods

3.3.1 Ellipsometry
Ellipsometry consists of a technique used to determine the refractive index (n) of thin transparent films
(Al2 O3 , SiO2 ) and also measure the film thickness. For this purpose it is used an AutoEl ellipsometer. The
principle of operation involves irradiating a sample with a collimated beam of monochromatic light with
wavelength λ= 632.8 nm, in a fixed and known angle. The differences between the states of polarization of
the incident and reflected beams are then determined and quantified in terms of two quantities: the angles ∆
and Ψ. These angles are function of the real and imaginary components of the refraction index and therefore,
through a numerical model, the ellipsometer returns the values that evaluate the quality (refraction index)
and thickness of the film. Since the model used assumes a substrate that provides total reflection, the films
should be grown on top of Si substrates. The SiO2 sample used to calibrate the system has a refractive index
of 1.473 for 1346 Åthick film.The expected refraction indexes for the materials used are 1.47 for SiO2 and
1.62 for Al2 O3 . Figure 3.9 shows the Auto El Ellipsometer machine.

33
Figure 3.9: Ellipsometer AutoEl image.

3.3.2 Profilometer

A profilometer system allows the measurement of sample topographies. It works based on a piezo-resistive
sensor, which sweeps the sample area for a defined range and detects any changes in topography, usually
steps at the edges of patterned structures. This system is very useful when determining film thickness after
deposition or even ion milling processes, giving information about deposition and etch rates. When using
the profilometer, it is important to calibrate the measurement first. Usually there should be an area with
constant height in the range measured to be used as reference. This system could be complemented with
information read from the ellipsometer, to ensure that the measured thickness is correct, but only for thin
transparent films. This device presents a vertical resolution of 5Åand can be used to measure films with
thicknesses higher than 400Å. Figure 3.10 shows an image of the profilometer device.

Figure 3.10: Profilometer Desktak.

34
3.3.3 Atomic force microscope
Atomic force microscopy (AFM) is a very high-resolution type of scanning probe microscopy. The micro-
scope used was a Digital Instruments Dimension 3000 microscope, located at Instituto Superior Técnico.
This method measures the surface topography on a scale of a few angstroms to hundreds of microns, with
demonstrated resolution on the order of fractions of a nanometer. This technique involves imaging a sample
through the use of a probe, or tip, with a radius of a few nanometers. When the tip is brought into proximity
of a sample surface, forces between the tip and the sample lead to a deflection of the cantilever. The tip is

Figure 3.11: Typical AFM setup. The deflection of the cantilever with a sharp tip is measured be reflecting
a laser beam off the backside of the cantilever while it is scanning over the sample surface.

held several nanometers above the surface using a feedback mechanism that measures surface-tip interactions,
to avoid damage from colliding with the surface. Variations in tip height are recorded while the tip is scanned
repeatedly across the sample, producing a topographic image of the surface. This method was used to control
the 3D tapered profile of the magnetic flux concentrators, see section 4.1.3.

3.3.4 Scanning electron microscope


In the facilities of INESC-MN there are two Scanning Electron Microscope (SEM) Systems, Raith150 and
Hitachi S-2500. The scanning electron microscope (SEM) is a type of microscope that images the sample
surface by scanning it with a high-energy beam of electrons in a raster scan pattern. The electrons interact
with the atoms that make up the sample producing signals that contain information about the sample’s
surface topography. The spatial resolution of the SEM depends on the size of the electron spot, which in
turn depends on both the wavelength of the electrons and the electron-optical system which produces the
scanning beam. The resolution is also limited by the size of the interaction volume, or the extent to which the
material interacts with the electron beam. The spot size and the interaction volume are both large compared
to the distances between atoms, so the resolution of the SEM is not high enough to image individual atoms,
as is possible in the shorter wavelength (i.e. higher energy) transmission electron microscope (TEM). The
SEM has compensating advantages, though, including the ability to image a comparatively large area of the
specimen; the ability to image bulk materials (not just thin films or foils). The instrument’s resolution can
fall somewhere between less than 1 nm and 20 nm. Samples have to be prepared carefully to withstand the
vacuum inside the microscope, and in order to make the samples conductive, they are coated with a very
thin layer of sputtered gold. The sample is then placed inside the microscope’s vacuum column through an
air-tight door, see Figure 3.12. After the air is pumped out of the column, an electron gun (at the top)

35
emits a beam of high energy electrons. This beam travels downward through a series of magnetic lenses
designed to focus the electrons to a very fine spot. Near the bottom, a set of scanning coils moves the focused
beam back and forth across the sample, row by row. As the electron beam hits each spot on the sample,
secondary electrons are ejected from its surface. A detector counts these electrons and sends the signals to an
amplifier. In the end the final image is built up from the number of electrons emitted from each spot on the
sample. The scanning electron microscope has many advantages over traditional microscopes. The SEM has
a large depth of field, which allows more of a specimen to be in focus at one time. The SEM also has much
higher resolution, so closely spaced regions can be magnified at much higher levels. Because the SEM uses
electromagnets rather than lenses, it is possible to have much more control in the degree of magnification.
Figure 3.12 shows a machine schematic of the microscope. Also the two systems existent at INESC-MN are
represented.

Figure 3.12: Scanning Electron Microscope Machines. (a) electron gun schematics. (b) Hitachi S-2500 SEM.
(c) Raith150 SEM.

3.3.5 Composition analysis


The composition of the deposited films is analysed by Rutherford backscattering spectrometry (RBS) [20].
This method determines the structure and composition of materials by measuring the backscattering of a
beam of high energy ions impinging on a sample. All measurements were performed at the ITN-Instituto
de Tecnologia Nuclear (Sacavém, Portugal) using a Van de Graaff particle accelerator. This device contains
a Van de Graaff generator, that creates a potential difference between two electrodes, in which negatively
charged ions (H− ,He− ) are accelerated inside a high voltage terminal. The positive terminal is positioned at
the center of the acceleration tube. A stripper element included in the positive terminal removes electrons
from the ions which pass through, converting He− to He++ and H− to H++ ions. Thus, these start out being
attracted to the terminal, but when they pass through and become positive, the new ions are repelled until
they exit the tube at ground. Ion beams can be obtained with energies up to 3.06 MeV and currents of 1-40
nA. This technique involves silicon surface barrier detectors which can measure the energy of incident ion
through the number of electron-hole pairs generated. The energy loss of a backscattered ions is dependent
on two processes: the energy lost in scattering events with sample nuclei, and the energy lost to small-angle

36
scattering from the sample electrons. The first process is dependent on the scattering cross-section of the
nucleus and thus on its mass and atomic number, providing a means of analysing the sample composition. The
second energy loss process creates a gradual energy loss dependent on the electron density and the distance
traversed in the sample. This process lowers the measured energy of ions which backscatter from nuclei inside
the sample in a continuous manner dependent on the depth of the nuclei, giving information about sample
thickness. In practice, a compositional depth profile can be determined from an RBS measurement. The
elements contained in a sample can be determined from the positions of peaks in the energy spectrum. Depth
can be determined from the width and shifted position of these peaks, and relative concentration from the
peak heights. Although, this technique is sensitive to the mass of the atoms in the film, it is not adequate to
measure the relative contents of elements with similar atomic mass (like Zr and Nb in CoZrNb, or Co and Fe
in CoFe). For these cases, Proton-induced X-ray emission (PIXE), also performed at ITN, is used to measure
the relative proportions of the elements unresolved by RBS. PIXE makes use of the atomic interactions that
occur when a material is exposed to an ion beam, giving off EM radiation of wavelengths in the x-ray part
of the electromagnetic spectrum, specific to an element. The composition profile is analysed for CoZrNb in
4.2.3, and deposited materials in section 3.1.2 with quantities represented in atomic percentage.

3.3.6 Magnetic properties analysis

Magnetic properties measurements of CoZrNb and spin valve stack films were measured using a vibration
sample magnetometer (VSM) device (DMS 1660), [18] [19]. The samples are glued onto a quartz rod with
Apiezon grease, to keep the samples in place while the rod vibrates horizontally at 200Hz. This movement
originates a variable magnetic flux response which induces a potential difference in the pick up coils, placed
at both sides of the vibrating rod, see Figure 3.13(b). This potential difference is proportional to the sample
magnetization. The vibrating rod and the pick up coils are flanked by the poles of an electromagnet that
applies a DC magnetic field that can reach 13 kOe. The response of the sample magnetization is measured
as a function of the variable magnetic field created by the pick up coils. This response is characterized by an
hysteresis curve. This system gives crucial information about the magnetic behaviour of materials, such as
the coercivity and saturation magnetization of CoZrNb films, and interlayer coupling and exchange fields of
spin valve stacks, see section 4.2.

Figure 3.13: (a) VSM device machine. (b) quartz rod and pick up coils image.

37
3.3.7 Electric properties analysis
After microfabrication, spin valve sensor electrical transfer curves were measured using a manual measurement
setup developed at INESC-MN. This process consists of evaluating the electrical resistance of the device as
function of an external magnetic field, applied along the sensor easy axis. This is done by applying a variable
field, in a -140 to +140 Oe range (0.1 Oe of resolution), created by the current passing in a set of two
Helmholtz coils. The coils are powered by a DC current source (±4A). Two micropositioning probes with
tungsten needles with a 10 µm resolution establish the electrical contact with the pads. The sensor current
bias is applied by a Keithley current source and the voltage drop is measured by a voltmeter. The electrical
changes are measured in a two probe arrangement, where the current source and the voltmeter terminals are
connected to the same pad. The probes include a system for a precise positioning and a magnifying lenses
system, used to make contact between the probes and the devices contacts. In order to maintain the the
probes steady while performing the measurements, a vacuum system is used to fixate the probes. All the
components are connected through a GPIB bus to the computer so that the data acquisition and control of
the measurements, applied current and field could be done automatically by a software also developed by
INESC-MN. Figure 3.14 pictures the manual measurements setup.

Figure 3.14: Manual measurement setup image.

To determine the resistivity of a thin film, it is usually measured the sheet resistance first, which can
be done directly using a four point probe system, see Figure 3.15. The four point probe, contains four thin

Figure 3.15: Four probe point system to determine the resistivity of thin films.

collinear wires, (usually tungsten), to make contact with the sample. Current (I), flows through the outer
probes, while the voltage (V), is measured between the two inner probes. When the thickness t is very small,

38
as would be the case for a diffused layer, the resistance sheet (Rs) is often used as the preferred measurement
quantity. Note that Rs is independent of any geometrical dimension and is therefore a function of the material
alone. The significance of the sheet resistance can be more easily seen if we refer to the end-to-end resistance
of a rectangular sample. eq.3.1 shows resistance formula.
l
R=ρ (3.1)
wt
and when the structure is considered square shaped, w=l then eq.3.1 can be written as:
ρ
R= = Rs (3.2)
t
Therefore, Rs may be interpreted as the resistance of a square sample, and for this reason the units of Rs are
taken to be Ohms per square or Ohm/sq. Dimensionally this is the same as Ohm, but this notation serves
as a convenient reminder of the geometrical significance of sheet resistance. So far it has been assumed that
the size of our sample is large compared to the probe spacing so that edge effects could be ignored. This
is usually the case for the bulk resistivity measurement. However, the sheet resistance measurements made
in lab will be made on a strip of the sample. The test area dimensions are not that large compared to the
probe spacing, so in order to get accurate measurements a correction factor for edge (geometry) effects must
be included [48].

39
Chapter 4

Laboratory work

4.1 Microfabrication process

The work carried out in this thesis was based on the improvement of magnetic field sensor sensitivity, following
the research line of A. Guedes. The work presented in [1] comprised the integration of magnetoresistive
(MR) sensors, 2D magnetic flux concentrators (MFCs) and microelectromechanical systems (MEMS). It was
performed with the purpose of achieving magnetic field measurements in the low magnetic field range, for
this purpose it became necessary the study of noise reduction methods in order to achieve field measurements
of the order of nT. The integration of magnetic flux concentrators became advantageous since they allowed
the enhancement of MR sensor sensitivity without any further noise increase. These concentrators were
fabricated by lift-off processes creating a 2D type of concentrator. This chapter provides the study of improved
magnetic flux concentrators with a 3D configuration. The integration with magnetoresistive devices brought
an improvement in sensitivity compared with the previous 2D profile. This systems’ micro fabrication process
consisted of the definition of spin valve (SV) sensors, metallic contacts and magnetic flux concentrators.
Figure 4.1 shows a microfabricated sample of this scheme, where the 3D profile was created by ion milling
at different angles. In addition SV sensors and flux concentrators with different dimensions were fabricated
with the purpose of finding the optimum configuration, bringing out the best results.

Figure 4.1: Microfabrication integrated structure (MR sensor, MFC, contact leads)

41
4.1.1 Spin valve definition

Spin valves were fabricated on top of a Si substrate with <100> crystalline orientation. For electrical isolation
purposes a 500 Å Al2 O3 oxide layer was deposited by magnetron sputtering on Ultra High Vacuum machine
(UHV1), see chapter 3.1.1. The rest of the SV stack was deposited by ion beam deposition on Nordiko 3000
with the structure represented in Fig.4.2, and the deposition rates were of the order of 0.3Å /s. Figure 4.2

Figure 4.2: Top spin valve structure schematic comprising layer composition and thickness in Å .

shows the SV structure, the composition and thickness of each layer. Different colours represent different
components of the sensor. This type of simple structure has two ferromagnetic (pinned and free) layers
separated by a non magnetic spacer layer (Cu). The magnetization of the pinned layer is held fixed in a
certain direction by the antiferromagnetic adjacent layer. This antiferromagnetic layer maintains the pinned
magnetization through a strong exchange interaction, creating a strong local magnetic field, the exchange
field (Hex ). The free and pinned layers magnetizations were defined with crossed anisotropies by applying a
40 Oe alignment field, during deposition. This top pinned configuration was microfabricated by direct laser
lithography and ion milling, at 70◦ . A thermal treatment of the spin valve was not required, since they
were top pinned structures. On top a 50 Å thick TiW(N2 ) passivation layer was deposited by magnetron
sputtering. The SV sensors were made with two different lengths (8 and 14µm), and three different widths (2,
2.5 and 3 µm) [13]. All the possible combinations of these values were tested. The crossed anisotropies were
maintained due to the demagnetizing field created with the SV rectangular shape. Setting the pinned layer
easy axis parallel to the sensor width, induces a magnetic potential difference in the same direction leading
to the creation of a demagnetizing field, Hd . This field is oriented perpendicular to the pinned magnetization
(parallel to SV length). In order to minimize the demagnetizing energy term, the magnetization should remain
perpendicular to the surface,i.e., parallel to the demagnetizing field. This forces the free layer magnetization
to remain oriented along the demagnetizing field direction, when there is no external applied field, Hext .
This situation is depicted in Figure 4.3(a). In the case where an external magnetic field is applied in the
pinned easy axis direction, the system acquires an additional energy term, changing the preferable direction
of the magnetization. The external field is set parallel to the pinned magnetization, so that the free layer
magnetization is forced to rotate from a perpendicular state towards a parallel direction to the field. This
rotation of the free layer varies linearly with the magnetic field, giving the SV sensor a linear response, as
can be seen in Fig.4.3(b).
After the sensor patterning, contact leads were fabricated by laser lithography and lift off, in a 2 contact
geometry. A 3000Å thick AlSiCu films along with a protective layer of 150Å TiW(N2 ) were deposited on
Nordiko 7000. The contact area was of 8x2µ m, 8x2.5µ m, or 8x3µ m, depending on the SV width.

42
Figure 4.3: Spin Valve dimensions. Length of 8 and 14 µm and width of 2, 2.5, 3µm

4.1.2 Flux guides definition


Flux guide concentrators were defined after the spin valve and contact leads patterning, by four different
processes. A 7000 Å CoZrNb thick layer was deposited on UHV1, by magnetron sputtering, with a 130 Oe
magnetizing field, defining the material’s easy axis parallel to the SV sensor largest dimension, see Figure 4.4.
Deposition rates were of the order of 1.19Å/s. The optimal geometry for the flux concentrators was studied
in the previous chapter 2, where a Finite Element Method Magnetics (FEMM) simulation was done. These
simulations allowed to understand how certain fabrication parameters maximized the gain. The thickness
was increased from 2500Å (previous work [1] to 7000Å) to restrictions of the top flux size, because a metal
gate is necessary above this structure, to continue the integration process with the microelectromechanical
system. The gate should be placed as close as possible to the top flux concentrator so that the concentrator
on the cantilever is attracted to this one without letting the flux lines escape. Flux concentrators were defined
with different sizes in order to study which geometry would bring out the best results. Top concentrators
dimensions varied with the outer width (A1 ) between the values: 29, 39, 50µm, the inner width (A2 ): 8,
14µm, and the length (L): 30, 50, 80, 100µm. Because of the fact that no constraints existed below the bottom
concentrator, i.e. no other structure existed, these concentrators were made larger and longer, increasing the
concentration factor. Their dimensions varied with outer width (A1 ) with values of 300, 500µm, (A2 ) with
8, 14µm, and the length (L) with 150, 200, 250, 450, 470µm.

Figure 4.4: Flux guide concentrator dimensions. Flux concentrator easy axis direction is represented, along
with the pinned and free layer magnetization direction in the absence of an external field. The direction of
the external applied field, when existing is depicted below.

Besides the different magnetic flux concentrators (MFCs) dimensions, four different fabrication processes
were taken into account. Figure 4.6 shows a schematic view of the processes in a longitudinal cut of the
gap region. Layer’s materials and their thickness are represented. The first process (Type I) consisted of a
vertical (no taper) profile defined by lift off, that is essentially a 2D profile, Figure 4.6(a). A non inverted

43
Figure 4.5: Flux guide concentrator and spin valve scheme at gap region. Gap between flux concentrator
and sensor is of 1.5µm. Contact leads dimension can be seen, being of 8 µ m in length and 2, 2.5, or 3 µ m
in width depending on the SV width. A2 represents the flux concentrator inner width which is defined with
the same value as the SV length. Its value could be either 8µ m or 14µ m.

Figure 4.6: Flux concentrator 2D and 3D profile at gap region with spin valve sensor. (a) vertical walls
defined by lift-off (Type I). (b) single angle taper (Type II).(c) Double angle taper (Taper III). (d) Single
angle taper terminating in a flux guide stub with vertical walls (Type IV). Distance from the tip of the flux
guide to the spin valve sensor is kept constant 1.5µ m.

lithography mask was patterned on top of the SV sensor’s oxide layer before the 7000Å CoZrNb layer was
deposited. Afterwards the lift off of the photo resist layer defined the flux guide with the desired dimensions.
The detailed microfabrication process scheme for each case is further explained in appendix A. The second
(Type II) and third (Type III) processes have a tapered profiles defined by ion milling with a single and
double angle structure, Fig. 4.6(b)(c) respectively. The CoZrNb layer was deposited on top of a 500Å oxide
layer. Then, an inverted mask was patterned by laser beam lithography in order to define the area of the
flux concentrator that was not to be etched by ion milling. Type III processes had an extra non-inverted
lithography and ion milling step. The region defined by this second lithography is represented in Figure 4.7.
Two small regions at the end of both flux concentrators was left uncovered so that it would be carved in by
the ion milling process, creating the second tapered profile.
Type IV flux guides were defined with a single taper, but terminate in a stub with a vertical wall. The
Type II and III were deposited on top of an extra 500Å thick oxide layer (located beneath the SV sensor).
The type IV concentrator was defined on top of an extra 1000Å oxide layer, to allow a higher ion milling
stopping margin. Ion milling was done, on Nordiko 3000, at defined angles relative to the sample surface,
namely, 20◦ , 45◦ , 60◦ , 75◦ , and 85◦ . Every fabrication process step is illustrated in appendix A.

4.1.3 Ion Milling process control


Initially, the ion milling process was done with double source power supply (110W). This allowed faster etch
rates, reaching 2.45Å /s at 45◦ . However, the plasma beam ejected very energetic ions onto the substrate,
creating holes in the sample surface, Figure 4.8. To avoid this issue and to better control the stopping margin,
the power supply was decreased to about half of its original value (64W).
Type II and III ion milling process was done with the help of a thin stripe of glass to prevent over etching

44
Figure 4.7: Double angle profile microscope image (Process II) 100x amplification. Non inverted mask to
define the second tapered profile. The SV sensor is protected by photo resist not to be over etched with the
second ion milling.

Figure 4.8: Sample microscope image of ion milling processed samples with 110W power. Holes in the surface
can be seen with an amplification of 100x.

of the CoZrNb layer in some areas. The ion milling system allowed a uniformity within about 50Å , but
the metal deposition, on UHV1, had a non-uniformity of about 500Å . This made the etch stopping criteria
difficult to find, since the oxide layer protecting the sensor was made of only 500Å . Type IV flux concentrators
had an extra 1000Å oxide layer, making this process easier to achieve and more reproducible. To control
the tapered profile, measurements of fabricated concentrators were measured by Atomic force Microscopy
(AFM). The nominal dimensions of the lithography mask used to define these structures were 3.5µ m for
the gap between flux concentrators (gap), and 2µ m for the inner width (A2 ). The ion milling was done at
an angle of 45◦ . The nominal values represented differed from the measured AFM samples within less than
10%. the gap distance measured was of 3.516µm, differing 0.5% of the setting value, the inner width of the
flux guide (A2 ) had a value of 2.2µm, differing in 10% from the nominal value. The angle measured for this
structure was of 40◦ , differing 11% from the 45◦ setting ion milling angle. Angle profile was also observed
by scanning electron microscopy (SEM) using both an electron beam SEM (Raith150) and a SEM (S-2500).
The angle measured by Raith150 was of about 55◦ with an error of 22% compared with the nominal angle
45◦ , and for the SEM(S-2500) the angle measured was of 50.2◦ with an 12% error. Figure 4.10 represents a
view of the flux concentrator tip, without SV sensor. The angle and layer dimensions were calculated using
the scale dimension.

45
Figure 4.9: 3D construction of AFM measurements - 25µm scan size, 0.3 Hz scan rate, 128 samples

Figure 4.10: SEM image of flux concentrator tip at gap region, without SV sensor. The tapered profile angles
are represented. On the left hand side a SEM S-2500 image is pictured. On the right hand side it is a SEM
Raith150 picture.

4.2 Device Characterization

4.2.1 Spin valve properties


Spins valve magnetic properties were measured by Vibration Sample Magnetometry (VSM). In bulk ma-
terial measurements were considered, see Figure 4.11. These measurements showed an exchange field of
He x=270 Oe, coercivity Hc =13.2 Oe,and a ferromagnetic Néel coupling field of Hf =15.1 Oe. The spin
valve used in the hysteresis loop analysis was a top pinned spin valve with following structure in [Å ]: Si
substrate/Al2 O3 [500]/Ta[20]/NiFe[25]/CoFe[27]/Cu[22]/CoFe[25]/MnIr[60]/Ta[20]/TiW(N2 )[50], represented
in Fig.4.2. The magnetic field is applied parallel to the layers surface, and parallel to the exchange bias field
(Hex ). In the large field range, for values below 270 Oe the pinned layer has its magnetization fixed by the
antiferromagnet, while above Hex the pinned layer reverses its magnetization. In the small field range, it is
the free layer that has its magnetization reversed, whereas the pinned is kept unchanged. The Néel coupling
field Hf is defined as the shift of the curve relative to the origin, produced by the interlayer coupling. The
coercive field Hc is defined as the magnetic field required to reduce the free layer magnetization from zero to
the saturation state. The in bulk electrical properties were measured with a DC four-point probe setup, see
section 3.3.7, and showed a giant magnetoresistive (GMR) signal of 7.5% and a resistance of 220Ω.

4.2.2 CoZrNb magnetic properties


The magnetic properties of the CoZrNb layer were studied using a vibration sample magnetometer (VSM).
A 7000Å thick layer was deposited on top of a Si and 500Å oxide layer. Figure C.1 shows the magnetization
m vs. H field curve in the easy and hard axis direction.
In the first case it is possible to see the step like behaviour of the magnetization vector oriented parallel to

46
4
e a s y a x is s v 5 7 8
3 h a r d a x is s v 5 7 8

2
1 H a r d a x is :
e m u )
H c
= 1 3 .2 O e
0
H f
= 1 5 .1 O e
-4

-1
H = 2 7 0 O e
(1 0

e x
-2
M

-3
-4
-5
-1 5 0 0 -1 0 0 0 -5 0 0 0 5 0 0 1 0 0 0 1 5 0 0
H (O e )

Figure 4.11: Experimental curves of the magnetic moment versus the applied magnetic field for a top pinned
in bulk spin valve. Exchange (Hex ), offset (H0 ), and coercive (Hc ) fields are represented as well as the free
(FL) and pinned (PL) layers magnetizations along the cycle.

C o 8 6
Z r5 N b 9
7 0 0 0 A la y e r
4 0
e a s y a x is c z n
3 0 h a r d a x is c z n

2 0 E a s y a x is :
H c = 0 .2 O e
1 0 H a r d a x is :
m s a t = 2 7 .2 m e m u
H c = 0 .2 O e
(m e m u )

M s a t = 1 6 1 1 k A /m
0 H k = 1 1 .5 O e
m s a t = 2 7 .2 m e m u
-1 0
M

M s a t = 1 6 1 3 k A /m
Χm = 1 7 5 8
-2 0
µr = 1 7 5 9
-3 0

-1 0 0 -8 0 -6 0 -4 0 -2 0 0 2 0 4 0 6 0 8 0 1 0 0
H (O e )

Figure 4.12: In bulk CoZrNb VSM analysis

the material’s easy direction, showing a coercive field of about Hc =0.2Oe, and total saturation magnetization
of Msat = 1611 kA/m (msat /V). When an external field is applied in a perpendicular direction regarding the
material’s easy axis, the magnetization rotates linearly towards this hard direction. Therefore the hysteresis
loop has a linear response, being the magnetic susceptibility χm given by the linear slope (m[memu]/V[cm3 ]),

47
its value was of χm =1758, and the relative magnetic permeability µm =1759, see 1.6. The coercive field is
Hc =0.2Oe and the anisotropic field Hk =11.5Oe. The total saturation magnetization is practically the same
(Msat = 1613 kA/m). The CoZrNb material is then characterized as a high moment, soft ferromagnet, with a
linear and hysteresis free response, justifying its use as a magnetic flux concentrator. Patterned samples were
processed on a Si substrate with a 500Å oxide layer in order to study the influence of the shape on the CoZrNb
magnetic properties. The samples were processed side by side in a 5x5 matrix. Each flux concentrator had
a micrometer structure in order to provide enough signal for the magnetic measurement. Figure 4.13 shows
a scheme of the patterned structures’ geometry.

Figure 4.13: Patterned CoZrNb structure. Geometry defined with inner width (A2 ), outer width A1 , and
length L. 25 samples were processed side by side in 5x5 matrix.

Two different sets of patterned samples were processed with also 7000Å of CoZrNb. Figure 4.14 shows
data for both patterned flux concentrators with vertical walls (Type I - no taper) defined by lift off, and with
45◦ tip taper (Type II) defined by ion milling.

2 .0 h a r d a x is lift o ff 9
h a r d a x is e tc h 9
1 .5

1 .0 H a r d a x is b y e tc h :
H c = 0 .2 5 O e
0 .5 H k = 2 4 .6 O e
(m e m u )

m s a t = 1 .9 2 m e m u H a r d a x is b y lif t- o f f :
0 .0
M s a t = 1 0 6 .6 k A /m H c = 0 .3 1 O e
-0 .5 Χm = 2 8 4 H k = 2 7 .5 O e
µr = m s a t = 1 .8 0 m e m u
M

2 8 5
-1 .0 M re m = 1 0 0 k A /m
-1 .5 Χm = 2 3 3 .5
µr = 2 3 4 .5
-2 .0

-1 2 0 -9 0 -6 0 -3 0 0 3 0 6 0 9 0 1 2 0
H (O e )

Figure 4.14: Patterned CoZrNb VSM analysis

Patterned samples defined with type I and II (45◦ taper) profiles have similar characteristics: coercive
(Hc ), and anisotropic (Hk ) fields, magnetization in the saturated state (Msat ), and even their magnetic
susceptibility (χm ). Nonetheless, type II samples had a slightly better response, higher susceptibility and
lower coercivity, demonstrating that this kind of fabrication process is more adequate for flux concentrators.

48
In comparison with in bulk samples, the magnetic susceptibility decreased by a factor of 6.2 for type II
concentrators, and 7.5 for type I. The coercivity, and anisotropic field increased, but this change was not
much. Anisotropic fields remained twice the value found for in bulk samples, and the saturation magnetization
had an 8-fold decrease in type II structures and 15-fold in type I. In general, magnetic properties for patterned
samples showed a worse response than in bulk material. These changes, however, are understandable since
the patterned case has a lower volume of soft magnetic material, and patterned structures.

4.2.3 CoZrNb composition analysis


Table 4.1 shows the composition analysis of a 7000Å -thick deposited CoZrNb layer. the metal was deposited
on top of a Si substrate with a 500Å -thick oxide layer. The atomic percentage of each element (Co,Zr,Nb)
was measured by Rutherford backscattering spectroscopy (RBS) and proton induced X-ray emission (PIXE).
The composition was found to be Co86 Zr4 Nb9 , in atomic percentage.

CoZrNb composition analysis


Co Zr Nb
mass density 454 ± 32 µg/cm2 43.4 ± 3 µg/cm2 71.2 ± 5 µg/cm2
15 2
atomic density 4639 ± 325 10 at./cm 286 ± 20 1015 at./cm2 461 ± 32 1015 at./cm2
fraction 82.13 ± 0.02 % 5.31 ± 0.01 % 8.56 ± 0.01 %
atomic weight 58.93 g/ mol 91.22 g/ mol 92.91 g/ mol
Table 4.1: Sample analized by RBS and PIXE. The conditions of the radiation used were: a 2050M eV H +
beam emitted through a Mylar250µm filter and a collimator of 5 mm into the sample, at an incident angle
of 22.5◦ and detected at 47.5◦ . Measurements done by Instituto Tecnológico e Nuclear - Unidade de Física
e Aceleradores in Sacavém, Portugal.

4.2.4 Magnetoresistance and sensitivity analysis


After fabrication the magnetoresistance hysteresis curves were obtained using a DC two-point probe with
two Helmholtz coils creating a varying magnetic field in order to measure the electrical changes. Samples
with fabricated spin valve sensors and contact leads were measured. Six different sizes of SV were considered,
having two different lengths (8, 14µm) and three different widths (2, 2.5, 3µm). A comparative study of
these different structures was carried out, regarding the sensor sensitivity, the maximum MR obtained in the
hysteresis curve, and the resistance of the parallel state, i.e., when the free layer’s magnetization vector is
aligned with the magnetization of the pinned layer. Fig. 4.15 shows the magnetoresistance variation of a
3x8µm2 spin valve.
Different sensors were characterized and their sensitivity and MR were analysed. For each type of sensor
dimension about 4 to 6 sensors were measured, allowing the calculation of their standard deviation. It can
be observed the fact that with increasing length and width the sensitivity improves, so the best results were
usually achieved with 3x14 µm2 sensor dimension. Table B.1 summarizes these results.
Despite the differences in geometry of the fabricated flux concentrators, the sensor sensitivity increased in
every case independently of size and process. For process (type I) the sensitivities obtained were on average
around 0.8%/Oe. For the remain processes the values differ depending on the angle profile. The best results
are summarized in Table4.3
About 300 sensors were measured in total, among this values only the best ones were considered. The
best results were usually achieved with sensor sizes 3x14µm2 and flux guides with 470µm length and 300µm
or 500µm outer width. In particular for the considered angles 75◦ type(II), 75◦ +20◦ and 85◦ +45◦ (type III),
and 45◦ and 60◦ (type IV) the sensor used was a 3x14µm2 SV, and the respective sensitivity without flux

49
7
S V s e n s o r - ( 3 x 1 4 ) µm 2
6 R pp = 1 6 9 Ω
I b ia s = 2 0 µA
5 H c = 1 .2 5 A /m

4 S = 0 .0 8 4 % /O e
M .R . (% ) 3

-1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0
H (O e )

Figure 4.15: Spin valve sensor MR-H hysteresis curve. SV sensor had a 3x14µm2 dimension. Its properties:
parallel resistance state (Rpp ), coercive field (Hc ), and sensitivity (S), were measured with a bias current
(Ibias ).

Spin Valve sensor properties


area 2x8µm 2.5x8µm 3x8µm 2x14µm 2.5x14µm 3x14µm
S[%/Oe] 0.064±0.007 0.084±0.007 0.088±0.007 0.067±0.009 0.079±0.004 0.089±0.005
M.R.[%] 6.84±0.26 6.73±0.14 6.73±0.16 7.03±0.06 6.88±0.21 6.88±0.23
Rpp [Ω] 146.5±23.3 111.9±9.4 96.7±8.2 210.5±39.9 173.6±26.3 148.5±6.0
Table 4.2: Spin valve sensor properties: Sensitivity(S), magnetoresistance (M.R.), parallel resistance (Rpp ).

concentrator was of 0.084%/Oe. For the case of 45◦ (type I), where the best result was obtained with a
2x8µm2 sensor, with initial sensitivity 0.073%/Oe (without concentrator). Flux concentrators were defined
with 6µm gap distance between them. Among the four different processes the type I was the easiest to
reproduce however the results were not as good as the other three processes defined by ion milling. The
optimum results were obtained in the case where flux guides are patterned using process II and IV. The
dependence of sensitivity on the ion milling angle was studied, and a systematic increase for lower angle is
observed. However, for the 20◦ (type II) profile the sensor’s gain was relatively low (only 11.67), despite
the fact that it is still 1.5 times better than the lift off case (type I). For larger angle profiles the gain was
also higher than the lift off process, being about 7 times higher for 75◦ (type II) and 3 times for 60◦ (type
IV). Compared with the reference values without flux guides these tapered profiles showed an increase in
sensitivity of 63 and 33 times respectively. Independently of the ion milling process type (II,III, or IV),
these results were better than the lift off process (type I). Moreover, in all cases the maximum gain is always
achieved with a 45◦ profile. The optimum value is obtained for process type II, leading to gains of 101. This
value represents a major achievement when compared with process I, as it is 9.6 times higher. The optimal
angle profile MFC and sensor were measured and the hysteresis curve is represented in Fig.4.16 for the 45◦
(type II) profile.
It is important to note that, for negative external field values, lower than -3Oe, the sensor’s pinned layer
magnetization vector starts to rotate, leading to a decrease in the magnetoresistance. The SV exchange field

50
Flux Guide Gain Analysis

process Type I Type II Type III Type IV


profile [◦ ] Lift off θ1 =45 θ1 =75 θ1 =75 θ2 =20 θ1 =85 θ2 =45 θ1 =45 θ1 =60
A1 [µm] 81 81 81 81 81 81 81
top MFC A2 [µm] 14 8 14 14 14 14 14
L [µm] 49 49 49 49 29 49 49
A1 [µm] 300 300 500 500 500 300 300
bottom MFC A2 [µm] 14 8 14 14 14 14 14
L [µm] 450 470 470 470 148 450 450
S [%/Oe] 1.1 7.4 5.2 1.2 5.8 4.2 2.7
Gain (FG/no FG) 13.3 101.2 52.9 11.7 59.5 49.7 32.8
ratio (IM/LO) 9.6 6.8 1.5 7.6 3.8 2.5
Table 4.3: Summary of the sensor transfer curve results, showing the ion milling angles, sensitivity(S), flux
concentrator gain (Gain), and ratio between sensitivities (ratio) for lift off process (LO) type I, and ion
milling profiles (IM), type II, III and IV. The flux gain was compared with the same sensor geometry,having
sensitivities without flux guides of 0.084%/Oe. The MFC dimensions varied as represented, where A1 stands
for the outer width, A2 the inner width, and L the concentrator’s length

7 Without FG:
Rpp = 157 Ohm
6 Ibias = 10 uA
Hc = 0.59 Oe
5 S = 0.098 %/Oe
45 deg FG:
M.R. (%)

4 Rpp = 135 Ohm


Ibias = 20 uA
3 Hc = 0.69 Oe
S = 7.38 %/Oe
2 Gain = 101

0
-100 -80 -60 -40 -20 0 20 40 60 80 100
H (Oe)

Figure 4.16: Transfer curve for a type II flux guide (45◦ profile) and a 2x8µm2 SV sensor, with a sensitivity
of 7.4%/Oe. The gap si 3.5µm. SV sensors without flux guide had a sensitivity of 0.084%/Oe, and the
corresponding gain with this flux structure was 101.

measured represented in Figure 4.11, is around 300 Oe. Since the pinned layer rotates in the integrated
scheme around 3 Oe, this fact indicates that the flux gain for this structure is close to 100. Fig. 4.17 shows
the evolution of the sensor sensitivity with the ion milling angle profile, at low field values. GMR vs. H field
transfer curve was obtained for 45◦ , 75◦ , 85◦ (type II), profile. The slope increases with decreasing angle being
maximum, for 45◦ , and similar with type I process for 85◦ . The sensor and flux concentrator dimensions were
the same for the 45◦ and 75◦ , but for the 85◦ profile the sensor was 3x14µm2 . The concentrator structure
was the same as the one use for (type I). Gap distances between concentrators were of 6µm.

51
6 .4
w ith o u t F G - S = 0 .0 9 8 % /O e
L ift- o ff (ty p e I) - S = 0 .7 7 % /O e
5 .6
8 5 d e g (ty p e II) - S = 0 .7 1 % /O e
7 5 d e g (ty p e II) - S = 5 .1 9 % /O e
4 .8 4 5 d e g (ty p e II) - S = 7 .3 8 % /O e

4 .0
M .R . (% )

3 .2

2 .4

1 .6

0 .8

0 .0
-7 .5 -6 .0 -4 .5 -3 .0 -1 .5 0 .0 1 .5 3 .0 4 .5 6 .0 7 .5
H (O e )

Figure 4.17: Transfer curve for type II flux guides (45◦ , 75◦ , 85◦ profiles) and type I (lift-off) profiles.Sensor
dimensions were 3x14µm2 in every case, except for the 45◦ (type II) profile, where the sensor was 2x8µm2 .
The gap between concentrators is 3.5µm in every profile. Sensitivities are represented in the LEGENDA.
The corresponding gain relative to the sensor (3x14µm2 ) sensitivity without concentrators was of 7.86 (type
I), 7.24 (85◦ type II), and 52.96 (75◦ type II). For the 45◦ tapered profile a gain of 101.1 was achieved in
comparison with the sensor (2x8µm2 )sensitivity 0.073 %/Oe used without the flux concentrator fabrication.

52
Chapter 5

Conclusions

The aim of this work was to increase the sensitivity of linear spin valve sensors. To do so, magnetic flux
concentrators were incorporated with the sensor in an integrated scheme. In particular this work shows how
3D fabricated concentrators can increase the concentration factor of magnetic flux lines, and consequently
the sensor sensitivity. Magnetoresistive sensors have sensitivities of about 0.084%/Oe (spin valves). With
the integration of magnetic flux concentrators these values were increased 10 to 100 fold depending on their
geometry. Four different fabrication processes of concentrators were performed for comparison reasons. The
geometry in plane (parallel to the substrate surface) varied, but for all processes in the same way, changing
only the perpendicular profile at the edges of the concentrator (at the sensor level). Starting with a 2D
geometry (type I) fabricated by only lift off processes. Its profile near the sensor had a vertical wall without
defining a concentrating shape towards the sensor. The sensitivity gain reached with this structure varied
from 7.8 to 13. The other three profiles were obtained with ion milling processes creating a tapered shape
of the flux concentrator towards the sensor. The second geometry (type II) was achieved solely by dry etch
with a single angle profile, whereas the third profile (type III) was defined by etch with two different angles,
creating two tapered shapes. The maximum gain obtained with these structures was of 101, for type II at 45◦ ,
and of 59.5, for type III at 85◦ + 45◦ . The respective sensitivities measured were of 7.4%/Oe and 5.8%/Oe.
The last profile (type IV) was defined with a single taper by etch but terminated in a stub with a vertical
wall defined by lift off. The maximum sensitivity gains obtained were of 49.7 for a 45◦ taper angle. Overall
more than 500 sensors were measured with different processes and geometries. The best results were obtained
for concentrators having an asymmetric geometry with large dimensions, being 500µm in outer width and
468µm in length (bottom concentrator). The sensor with highest sensitivity was fabricated with a 14x3µm2
shape.
In chapter 2 magnetic simulations were done to understand the best geometry to be fabricated, such that
the concentration of the magnetic flux lines would be maximum at the sensor level. Several parameters
were considered, and it was concluded that the concentration factor increased with the soft magnetic layer
thickness, length and outer width.
Chapter 4 provided the microfabrication process information, all the relevant details about the devices, sensors
and flux concentrators, and also presented the experimental results. The dimensions of the structures and
the electrical and magnetic properties of the materials were explained (the hysteresis curves of the CoZrNb
layer and spin valve stack). In addition the transfer curves of the best obtained results were shown.
This work focused only on the optimization of the magnetic flux concentrators. Further work could be done
at the sensor level, either by improving the initial spin valve sensitivity, or considering other types of sensors,
such as a SAF structure for the spin valve. This last improvement could bring the magnetoresistance values
up for the integrated scheme, towards the initial magnetoresistance obtained without concentrators. This
could happen because the pinned layer magnetization would be fixed in a more effective way. In addition,

53
if optimized MEMS resonators or micro cantilever structures were considered, the oscillating response of the
complete system could have even better results. Nevertheless, the improvement carried out in this thesis
allowed a 10 fold increase in sensitivity, compared with previously fabricated devices. The final integrated
system with the microelectromechanical cantilever would also be improved, reaching a detection limit in the
pico Tesla range.

54
55
Appendix A

Fabrication process scheme

Figure A.1: Spin valve fabrication scheme.

Figure A.2: Type I - 2D profile scheme (fabricated by lift-off)

Figure A.3: Type II - 3D tapered profile scheme (single angle)

57
Figure A.4: Type III - 3D tapered profile scheme (double angle)

Figure A.5: Type IV - 3D tapered profile with vertical stub at sensor level scheme (single angle)

58
Appendix B

CoZrNb volume determination

Volume of CZN layer determination


in bulk patterned
mass [ g ] 0.038 ± 0.001 0.041 ± 0.001
area [ cm2 ] 0.241 ± 0.026 0.257 ± 0.027
thickness [ A ] 7000 ± 400 7000 ± 300
V [ um3 ] 16.9 ± 0.005 18.0 ± 0.006

Sample density determination: Si substrate


mass [ g ] 0.986 ± 0.001
area [ cm2 ] 6.25 ± 0.50
density [ g/cm2 ] 0.158 ± 0.013
Table B.1: Volume determination of CoZrNb sample layer

59
Appendix C

Flux guide gain analysis

C.1 Gain and sensitivity table

Figure C.1: Properties of flux concentrators: gain and sensitivities obtained for different angles and processes

61
C.2 Type I - transfer curve

8
N o F lu x G u id e :
7 R p p
= 1 5 7 O h m
Ib ia s
= 1 0 u A
6
H c = 0 .6 O e
5 S = 0 .0 9 8 % /O e
(T y p e I) F G :
4 R pp = 1 8 8 O h m
M .R . (% )

Ib ia s
= 2 0 u A
3
H c = 0 .4 O e
2 S = 0 .7 7 % /O e
G a in = 7 .8 3
1

-1
-1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0
H (O e )

Figure C.2: Type-I transfer curve. Concentrator deposited and fabricated by lift-off.

62
C.3 Type II - transfer curve

Figure C.3: Type-II transfer curve (single angle taper). Concentrator deposited and fabricated by ion milling
with a 20◦ profile.

7 Without FG:
Rpp = 157 Ohm
6 Ibias = 10 uA
Hc = 0.59 Oe
5 S = 0.098 %/Oe
45 deg FG:
M.R. (%)

4 Rpp = 135 Ohm


Ibias = 20 uA
3 Hc = 0.69 Oe
S = 7.38 %/Oe
2 Gain = 101

0
-100 -80 -60 -40 -20 0 20 40 60 80 100
H (Oe)
Figure C.4: Type-II transfer curve (single angle taper). Concentrator deposited and fabricated by ion milling
with a 45◦ profile.

63
8
N o F lu x G u id e :
7 R p p
= 1 5 7 O h m
Ib ia s
= 1 0 u A
6
H c = 0 .5 9 O e
5 S = 0 .0 9 8 % /O e
o
T y p e II F G 7 5 :
4 R pp = 1 5 5 O h m
M .R . (% ) Ib = 2 0 u A
ia s
3
H c = 1 .1 1 O e
2 S = 5 .1 9 % /O e
G a in = 5 2 .9
1

-1
-1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0
H (O e )

Figure C.5: Type-II transfer curve (single angle taper). Concentrator deposited and fabricated by ion milling
with a 75◦ profile.

8
N o F lu x G u id e :
7 R p p
= 1 5 7 O h m
Ib ia s
= 1 0 u A
6
H c = 0 .5 9 O e
S = 0 .0 9 8 % /O e
5 o
T y p e II F G 8 5 :
4 R pp = 1 5 5 O h m
M .R . (% )

Ib ia s
= 2 0 u A
3
H c = 0 .2 O e
2 S = 0 .7 1 % /O e
G a in = 7 .2 2
1

-1
-1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0
H (O e )

Figure C.6: Type-II transfer curve (single angle taper). Concentrator deposited and fabricated by ion milling
with a 85◦ profile.

64
C.4 Type III - transfer curve

8
N o F lu x G u id e :
7 R p p
= 1 5 7 O h m
Ib ia s
= 1 0 u A
6
H c = 0 .5 9 O e
5 S = 0 .0 9 8 % /O e
o o
T y p e III 8 5 + 4 5 :
4 R pp = 3 7 8 O h m
M .R . (% )

Ib ia s
= 2 0 u A
3
H c = 1 .0 3 O e
2 S = 5 .8 4 % /O e
G a in = 5 9 .5
1

-1
-1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0
H (O e )

Figure C.7: Type-III transfer curve (double angle taper). Concentrator deposited and fabricated by ion
milling with a 85◦ and a 45◦ double angle profile.

8
N o F lu x G u id e :
7 R p p
= 1 5 7 O h m
Ib ia s
= 1 0 u A
6
H c = 0 .5 9 O e
5 S = 0 .0 9 8 % /O e
o o
T y p e III 7 5 + 2 0 :
4 R pp = 1 5 6 O h m
M .R . (% )

Ib ia s
= 2 0 u A
3
H c = 1 .6 9 O e
2 S = 1 .1 5 % /O e
G a in = 1 1 .7
1

-1
-1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0
H (O e )

Figure C.8: Type-III transfer curve (double angle taper). Concentrator deposited and fabricated by ion
milling with a 75◦ and a 20◦ double angle profile.

65
C.5 Type IV - transfer curve

Figure C.9: Type-IV transfer curve (single angle taper). Concentrator deposited and fabricated by ion milling
with a 45◦ profile, with vertical wall defined by lift-off at sensor level.

W ith o u t F lu x G u id e :
6
R p p
= 1 6 9 O h m
Ib ia s
= 2 0 u A
5
H c = 1 .2 5 O e
S = 0 .0 8 4 % /O e
o
4 T y p e IV F G 6 0 :
R pp = 1 6 9 O h m
M .R . (% )

3 Ib ia s
= 2 0 u A
H c = 1 .8 7 O e
S = 2 .7 4 % /O e
2
G a in = 3 2 .8

0
-1 0 0 -8 0 -6 0 -4 0 -2 0 0 2 0 4 0 6 0 8 0 1 0 0
H (O e )

Figure C.10: Type-IV transfer curve (single angle taper). Concentrator deposited and fabricated by ion
milling with a 60◦ profile, with vertical wall defined by lift-off at sensor level.

66
Appendix D

Process run sheet

67
!"#$%&''($ %%./%'012%#134$5%%"64%()%5$65&75%&6%(3%518597"9$5%
%
!
)$ %*+,-.$/0.,*0*1234$
%
!
"#$%&'& %(18597"9$%":135393&6% /"9$% %
% ;$5<&65380$% %
(3=$>%?%@%36%A"'$7%
+&69>%5BCDE%FG.HHHI% % % %

!7&<$793$5>%J;K%DLCM%N$2%O.CH%P$%%
(971Q917$>%R"%SCT%G3+$%SCT%U&+$SDT%U1%SST%U&+$%SHT%J6V7%@WT%R"%SH%
%
RX<$>%9&<%<366$4%
%
%
%
5$ %,24$6*-7.$8.92421234$
%
%
%
"#$%&!& U&"936#% % %
% % %
J"QY36$>% (),%97"QZ5%
%
U&"936#>%
RX<$>% !&5393B$%<Y&9&7$5359%!+;DDEH,%SQ!%!"#$%&'()*+,-(.$ %
;$Q3<$>% !7&#7"[%@\S%
!7$KY&B$6>% ("[<0$%Y$"936#%'&7%@H5%"9%WC]U%
U&&0%/&A6>% .H5% %
(<36636#>% -H5%"9%SCHH%7<[%'&7%F^LC%_[%<Y&9&7$5359I%
`"Z36#>% @H5%"9%WC]U%
%
%
%
%
59
"#$%&(& ^ %a39&#7"<YX%K%b2<&517$%&'%()% % %
% % %
J"QY36$>% /ca%

b6$7#X% WC% +&Q15% KCH% +36"0%&''5$9% 2dX% KWL@% .% %

% % % % %
J"<%6"[$% ()ce++b;% J"5Z%6"[$%% +,()-f5B%

+3$04% ^DCHH2%^DCHH%%%F1[I%%%%%%%F$"QY%43$I% 53=$% ^DHHH2^DHHH%F1[I%


%
%
% U&[[$695%
P<93Q"0%V65<$Q93&6%% %
%
%
%
%

Figure D.1: Process run sheet used for type I flux concentrators(1/7).
!"#$%!%&'%!!
()*+,%-% 68
%
"#$%&'& .$/$0&123#% % %
% % %
%
4"5623$7% (),%89"5:;%
%
.$/$0&123#7%
<=1$7% >0825%-!
?$521$7% !9&#9"@%ABC%
!9$D6&/$37% ("@10$%6$"823#%'&9%AE;%"8%FFEGH%
H&&0%.&I37% -E;%
.$/$0&123#7% AE;%%
H0$"323#7% ?23;23#%I286%J$2&32;$J%I"8$9%"3J%;123%J9=23#%'&9%KE;%
%
<&8"0%J$/$0&123#%82@$7% AE;%
%
% %
4259&;5&1$%L3;1$582&3% %
%
%
%
%
"#$%&(& M856%N=%L&3%420023#% % %
% % %
%
<&8"0%8&%$8567% KEE%%>% <&8"0%$85623#%82@$7% ϰϬϬ͛͛%
% % L38$9/"0%$8567% ϮϬϬ͛͛нϮϬϬ͛͛%
H&&0%J&I3%238$9/"07% ϮϬϬ͛͛%
4"5623$7% LO.%ʹ%P&9J2:&%-EEE%
M856%?"8$7% Q%%%F%%%%>B;%RSEG%

O"8567% M856%(123%)"0/$%%

%
T"'$9%?$521$7% M856%
!9&5$;;7% M856%(123%)"0/$%
T"'$9%U7% S%
O";$%!9$;;V9$%W<&99X7% -YZ[%FEDA%
.$1&;282&3%!9$;;V9$%W<&99X7% FY\[FEDK%
%
>;;2;8%,V3%
!&I$9% )*% L*% )D% LD% ,";%+0V[% %
<$;8%U% %
WT?+X% W)X% W@>X% W)X% W@>X% W;55@X%]>9^%
(/_S\% AK% KZZ% -CY_% F\-Y-% CYF% SY\% %
%
%
)% L% ,";%+0V[% %
P$V89"02`$9%
W)X% W@>X% W;55@X%]>9^% %
% KFFYA% CEYK% EYF%%%%%%%%%%% %
%
%
?&8"82&3% !"3%
(VN;89"8$% %
WaX% WJ$#X%
%
($8%)"0V$;% KE% SE% %
!
!
! Figure D.2: Process run sheet used for type I flux concentrators(2/7).
!
!

69
!"#$%!%&'%!!
()*+,%-%
!" !#$%&'%(")*+,$,%,#$"
%
%
"#$%&'& .&"/01#% % %
% % %
2"3401$5% (),%/6"378%
%
.&"/01#5%
9:;$5% !&80/0<$%;4&/&6$808/%!+=>>?@,%A3!%!"#$%&'()*+,-(.$ .&BB$1/8%
=$30;$5% !6&#6"B%CDA%
!6$E4&<$15% ("B;F$%4$"/01#%'&6%C@8%"/%GHI.%
.&&F%J&K15% -@8% %
%
(;01101#5% L@8%"/%AH@@%6;B%'&6%MNOH%PB%;4&/&6$808/Q%
R"701#5% C@8%"/%GHI.%
%
%
%
1S
"#$%&(& A %T0/&#6";4:%E%3&1/"3/8%$U;&8V6$% % %
% % %
2"3401$5% JWT%
8/
EH@% YF0#1B$1/%B"678% N % ENCG% EHL%
X1$6#:% GH% +&3V8% 1S
% MVBQ% A % % %
% Z60#01%&''8$/%MVBQ% U% :% % %
+01"F%&''8$/% U% :% EN>COC% EHN%
%
%
2";%1"B$% (<K"''$6% 2"87%1"B$% +,()L[;"S8%

+0$FS% N>H@@UN>H@@%%%%%%MVBQ%%%%%M$"34%S0$Q% J0B$180&18% %N>@@@%%%U%N>@@@%%%%%%%%%%%MVBQ%


%
%
% .&BB$1/8%
Z;/03"F%\18;$3/0&1%% %
%
%
%
%
%
"#$%&)& J$<$F&;01#% % %
%
2"3401$5% (),%/6"378%
%
J$<$F&;01#5%
9:;$5% YF/03%-$
=$30;$5% !6&#6"B%CDA%
!6$E4&<$15% ("B;F$%4$"/01#%'&6%C@8%"/%NN@I.%
.&&F%J&K15% -@8%
J$<$F&;01#5% C@8%%
.F$"101#5% =01801#%K0/4%S$0&108$S%K"/$6%"1S%8;01%S6:01#%'&6%L@8%
%
9&/"F%S$<$F&;01#%/0B$5% C@8%
%
% %
2036&83&;$%\18;$3/0&1% %
%
!

Figure D.3: Process run sheet used for type I flux concentrators(3/7). !"#$%!%&'%!!
()*+,%-% 70
"#$%&'& ./%"01%2%3456%1$7&8393&0% % %
%%%%%%%%%%%%%%%%%%%%%%% % %
2:3;<0$88%9&%1$7&839=% -??%0@%./%*%AB%0@%23456% %%
>";:30$=% CDEF5&G13<&%H???%
>&1I/$8=% 6J-JK% %

($LI$0;$=% KH%M8N7"1O%
F%^ŽĨƚƐƉƵƚƚĞƌĞƚĐŚ;ϲϬ͛͛ͬϲϬнϰϬt;Z&ͿͬϱϬƐĐĐŵƌͬϯŵdŽƌƌ%
% HPAF%>&1I/$%6%+I0;93&0%Q%% R$"1=%
%
%
% HP6F>&1I/$%K%+I0;93&0%A%% F%DĞƚĂůůŝnjĂƚŝŽŶ;ϭ͛ϮϬ͛͛ͬϮŬtͬϱϬƐĐĐŵƌͬϯŵdŽGG%
R$"1=%
% %
% % %
HP-F>&1I/$%-%+I0;93&0%AQ% FdŝtͺƉƌŽƚĞĐƚŝǀĞͺůĂLJĞƌ;Ϯϳ͛͛ͬϱϬϬtͬϱϬ;ƌͿнϭϬ;E6O%8;;@%S%-@2&GG%
%
%
"#$%&(& ./I@30I@S234M56O%/3'9&''% % %
%%%%%%%%%%%%%%%%%%%%%%% % %
23@$%30%@3;G&8937%6??A%T"9:=% K:%%%%%%%%%%%%%%% 2$@7$G"9IG$=% UBVW%
23@$%30%I/9G"8&03;%T"9:=% B*A6%@30% 2$@7$G"9IG$=%%
XLI37@$09=% 4$9%T$0;:% % %
% R308$%Y39:%C!.%"01%EC%Y"9$G%
%
%
%
"#$%&!)& R$8389"0;$%"01%>R%"0"/Z838% % %
%%%%%%%%%%%%%%%%%%%%%% % %
!
8"@7/$% E3$% W&/I@0% [30$% CT3"8% (% >R% R77% 4% [% R8L%
%% %% %% %% I.% \S]$% \% ]:@% I@% I@% ]:@%
A% A% 6% B% A?% ?P?U6% UPAQ% 6A?% 6% ^% B6PB%
A% A% U% B% A?% ?P?^U% UPBA% AU6% 6PB% AK% 6^PQ%
A% A% H% B% A?% ?PA??% UPBB% AK-% -% AK% -?PU%
A% 6% B% B% A?% ?P?H6% HP?-% 6-6% 6% AK% --PA%
A% 6% U% B% A?% ?P?^?% HP?A% A^Q% 6PB% AK% --P^%
A% 6% H% B% A?% ?PA??% UPQ-% AKH% -% AK% -APB%
A% -% A% A% A?% ?P?Q?% UPHB% A-B% 6PB% ^% K6P6%
A% -% A% 6% A?% ?P?^U% HP?-% A-?% 6PB% ^% K?PU%
A% -% A% -% A?% ?P?^^% UP^^% A-A% 6PB% ^% K?PQ%
A% -% A% K% A?% ?P?^^% UPQ% A-A% 6PB% ^% K?PQ%
A% -% A% B% A?% ?P?^-% UPHA% A?B% 6PB% ^% -6P^%
A% -% A% U% A?% ?P?Q?% UPHU% A?-% 6PB% ^% -6P6%
A% -% A% H% A?% ?P?^B% UPHB% A?K% 6PB% ^% -6PB%
A% -% A% ^% A?% ?PQ?K% UPH-% A?B% 6PB% ^% -6P^%
A% -% A% A-% A?% ?P?^^% UPQB% A-6% 6PB% ^% KAP-%
A% -% 6% A-% A?% ?P?UU% UPQQ% AUH% 6% ^% KAP^%
A% -% -% A-% A?% ?P?^Q% UPQ-% A6Q% 6PB% ^% K?P-%
A% -% K% A-% A?% ?P?QB% UPQA% AA-% -% ^% K6PK%
A% -% B% B% A?% ?P?UB% HP?6% 6AQ% 6% AK% -AP-%
A% -% B% A-% A?% ?P?HQ% UPQH% AB6% 6% AK% 6APH%
A% -% U% B% A?% ?P?^B% UPQ^% A^U% 6PB% AK% --P6%
A% -% U% A-% A?% ?P?^B% UP^Q% A--% 6PB% AK% 6-P^%
A% -% H% B% A?% ?P?Q^% UPQB% AKH% -% AK% -APB%
A% -% H% A-% A?% ?PA??% UP^% AAA% -% AK% 6-P^%
!
!
Figure D.4: Process run sheet used for type I flux concentrators(4/7).

!"#$%!%&'%!!
71
()*+,%-%
!" #$%&"'%()*+""!*,(-(.(/-"
%
#$%&'!!' ./01$%1$2&3040&5%ʹ%26&4$7408$%9":$6% % %
%%%%%%%%%%%%%%%%%%%%%%% % %
!
;<07=5$33%4&%1$2&304>% %%@AA%%%%%%%%B%&'%B9C.-% ;&4"9%1$2&3040&5%
?"7<05$>% DE)C% 40F$>ϭŚϮ͛%
G$2&3040&5%
H"4$>IIJK-BLF05%%
% % %
%
% ?"7<05$% ;<07=5$33%
G$2&3040&5%&'%4$34%3"F29$%4&%7&546&9%B9C.-%4<07=5$33%% 26&'09&F$4$6% %%%%%%%@AMB%*N%IAAB%

%
#$%&'!"' ;"6#$4%79$"505#%OP&Q-R6-ST01% % %
%%%%%%%%%%%%%%%%%%%%%%% % %
!6$N%(2U44$605#>%
;&4"9%40F$%&'%79$"505#>% ϭŚϭϬ͛% % G&5$%
?"7<05$>% DE)I% P&&905#%V"4$6%.5% .=%
W"47<>%% % % % ;?!%9&X%32$$1% &=%
% % %
% % % %
H$"1% !&X$6GP% ,"3%+9U/% !6$33U6$% )% Z%% %
)"9U$3% OVY% O377FY% OF;&66Y% O)Y% OFBY% ;0F$%
OB6Y%
34"64[%% -C% MJ@% -JA% K\A% ]M% ϭϭŚϭϳ͛%
% % % % % % %
$51[% % % % % % ϭϮŚϮϲ͛%
%
#$%&'!(' ?$4"9%1$2&3040&5%OP&Q-R6-ST01% % %
%%%%%%%%%%%%%%%%%%%%%%% % %
;<07=5$33%4&%1$2&304>% AJM%UF%&'%P&Q-R6-ST0% % G&5$%
?"7<05$>% DE)I% P&&905#%X"4$6%.5% .=%
W"47<>% % ;?!%9&X%32$$1% .=%
;&4"9%1$2&3040&5%40F$>% ϭŚϯϳ͛% ;$34%3"F29$% .=%
^340F"4$1%40F$>% I<-]͛C@͛͛;Λϭ͘CIBL3Y% ^"3:%"/03%1$'05040&5% .=%
G$2&3040&5%6"4$>% %%%IJIQ%%%%%%%%%%BL3% %
% %
% %
,"3%+9U/%
H$"1% !&X$6GP% !6$33U6$% )% Z%
O377FY% ;0F$%
)"9U$3% OVY% OF;&66Y% O)Y% OFBY%
_B6`%
34"64[% -C% \JA% -JA% K@@% MC% ϭϬŚϮϳ͛%
% -C% MJQ% CJQ% KC\% M]% ϭϬŚϰϱ͛%
% -C% MJ\% -JA% KI-% \I% ϭϭŚϯϬ͛%
% % % % % % %
$51[% % % % % % ϭϮŚϬϰ͛%
%
;$34%("F29$%a0'4%ʹ%.''>%%79$"5%X04<%"7$4&5$% % %
%
;&4"9%a0'4N.''%;0F$>% I<%
;&4"9%U946"N3&U513%40F$>% @F05%
%
%
% ?"7<05$% ;<07=5$33%
G$2&3040&5%&'%4$34%3"F29$%4&%7&546&9%P&Q-R6-ST0%% 26&'09&F$4$6% ]\M]L]KCML]@AAL]M\KL%
MCAM%B%

!"#$%!%&'%"!
()*+,%-% Figure D.5: Process run sheet used for type I flux concentrators(5/7).
72
%
!"# $%&'#(&)*+#!+,)-).)/-#0123#
%
%
"#$%&!'& .&"/01#% % %
% % %
2"3401$5% (),%/6"378%
%
.&"/01#5%
9:;$5% !&80/0<$%;4&/&6$808/%!+=>>?@,%A3!%!"#$%&'()*+,-(.$ %
=$30;$5% !6&#6"B%CDA%
!6$E4&<$15% ("B;F$%4$"/01#%'&6%C@8%"/%GHI.%
.&&F%J&K15% -@8% %
(;01101#5% L@8%"/%AH@@%6;B%'&6%MNOH%PB%;4&/&6$808/Q%
R"701#5% C@8%"/%GHI.%
%
%
"#$%&!(& JST%UV;&8W6$%&'%+FWV%#W0X$8%% % %
% % %
2"3401$5% JST%
8/
EH@% YF0#1B$1/%B"678% N % ENCG% EHL%
U1$6#:% GH% +&3W8% 1X
% MWBQ% A % % %
% Z60#01%&''8$/%MWBQ% V% :% EGOC% -%
+01"F%&''8$/% V% :% EN>COC% EHN%
%
%
2";%1"B$% ()SY++U=% 2"87%1"B$% +,()L['#GH%

+0$FX% N>H@@VN>H@@%%%%MWBQ%%%%%M$"34%X0$Q% J0B$180&18% %N>@@@%%%V%N>@@@%%%%%%%%%%%MWBQ%


%
%
% .&BB$1/8%
Z;/03"F%\18;$3/0&1%% %
%
%
%
"#$%&!)& J$<$F&;01#% % %
% % %
2"3401$5% (),%/6"378%
J$<$F&;01#5%
9:;$5% YF/03%-$
=$30;$5% !6&#6"B%CDA%
!6$E4&<$15% ("B;F$%4$"/01#%'&6%C@8%"/%NN@I.%
.&&F%J&K15% -@8%
J$<$F&;01#5% C@8%%
.F$"101#5% =01801#%K0/4%X$0&108$X%K"/$6%"1X%8;01%X6:01#%'&6%L@8%
%
9&/"F%X$<$F&;01#%/0B$5% C@8%
%
% %
%
2036&83&;$%\18;$3/0&1%
%
%
%
%
Figure D.6: Process run sheet used for type I flux concentrators(6/7).
!"#$%!%&'%!!
()*+,%-%
73
%
"#$%&!'& !.$/0$"123#%4"567$% % %
% % %
!.$/0$"123#%4"567$%&3%8$9$7&6$.%1.":;% % % %
% % % %
?$"123#%125$<%% %%ϯϬϬ͛͛%%%%%%%%%%%ΛϭϭϬȗ%
!.&#."5<% -=>%
@&&7%8&A3<% ϯϬ͛͛%
%
"#$%&!(& B1:0%CD%E&3%F27723#% % %
% % %
ϴϲϬϬ͛͛Λϴϱ͕ϴϬϬϬ͛͛ΛHKL%
G&1"7%1&%$1:0<% HIII%%J% G&1"7%$1:023#%125$<%
ϲϬϬϬ͛͛Λϰϱ%
% % E31$.9"7%$1:0<% %
@&&7%8&A3%231$.9"7<% O&1%8&3$%
F":023$<% EMN%ʹ%O&.82;&%
B1:0%P"1$<% Q%%%%%>R-K%STK8$#%%%%%%%%%J=4%%

M"1:0<% B1:0%@UO%

%
V"'$.%P$:26$<% B1:0%@UO%TK%6"3%
!.&:$44<% B1:0%@UO%TK%6"3%
V"'$.%W<% X%
M"4$%!.$44Y.$%ZG&..[<% \R-]>I/\%
N$6&4212&3%!.$44Y.$%ZG&..[<% X]>I/^%
%
J44241%,Y3%
!&A$.% )*% E*% )/% E/% ,"4%+7Y]% %
G$41%W% %
ZVP+[% Z)[% Z5J[% Z)[% Z5J[% Z4::5[%fJ.g%
TK6"3% \K% ^TT% -X% >_-% X% HR_% %
HK6"3% \K% ^THRT% -X% >_-R-% >RT% HR_% %
^K6"3% \K% ^THRT% ->RK% >_-R-% >RT% HR_% %
%
%
%
)% E% ,"4%+7Y]% %
O$Y1."72h$.% P&1"12&3% !"3%
Z)[% Z5J[% Z4::5[%fJ.g% % (YC41."1$%
Zd[% Z8$#[%
TK6"3% XIR^% ^>>R\% IR>%%%%%%%%%%% %
% ($1%)"7Y$4% ^I% TKLHKL^K%
HK6"3% XIR^% ^>>R_% IR>%
^K6"3% XIR^% ^>>R\% IR>% %
%
%
"#$%&!)& G&1"7%B1:023#%?$2#01% % %
% % %
G$41%("567$%:7$"3<%%% % %
F$"4Y.$8%0$2#01<%%%%%%%%%%%%%%%%%%%%%%%%J%
B1:0%."1$<%%%%%>R>_%%J=4%S^K`%%a%SHK`aSTK`%

@&55$314%
% %

%
"#$%&*+& P$4241"3:$%"38%FP%"3"7D424% % %
%%%%%%%%%%%%%%%%%%%%% % %
!
4"567$% N2$% @&7Y53% b23$% EC2"4% (% FP% P66% V% b% P4c% ,"23%
%% %% %% %% YJ% d=e$% d% e05% Y5% Y5% e05% %%
TK6"3% X% H% \% XI% IRH>% ^R^T% >KK% -% >^% --RX% HRXX%
HK6"3% -% H% \% XI% KR>_% \R-_% >KK% -% >^% --RX% KXRT\%
^K6"3% -% X% H% XI% HR-T% KRXK% >-K% X% T% --RT% >I>RX^%
%

Figure D.7: Process run sheet used for type I flux concentrators(7/7).
!"#$%!%&'%!!
()*+,%-%
74
Bibliography

[1] A. Guedes , J. M. Almeida , S. Cardoso , R. Ferreira , and P. P. Freitas, INESC-MN, Improving


Magnetic Field Detection Limits of Spin Valve Sensors Using Magnetic Flux Guide Concentrators ,
IEEE Transactions on magnetics, Vol. 43, no. 6, 2007.

[2] A. Guedes , Samadhan B. Patil , Piotr Wisniowski , Virginia Chu , João P. Conde, and P. P. Freitas,
Hybrid Magnetic Tunnel Junction - MEMS High Frequency Field Modulator for 1/f Noise Suppression
, IEE Transactions on magnetic, Vol. 44, no. 11, 2008.

[3] A. Guedes, S.B. Patil, S. Cardoso, V. Chu, J.P. Conde, and P.P.Freitas, Hybrid magnetoresistive
microelectromechanical devices for static field modulation and sensor 1/f noise cancellation, Journal of
Applied Physics 103,07E924, 2008.

[4] P. Drljacca, F. Vincent, P. Besse, and R. Popovic, Design of planar magnetic concentrator for high
sensitivity Hall devices, Sens. Actuators A, vol. 97-98, pp. 10-14, 2002.

[5] C. Fermon, M. Pannetier-Lecoeur, N. Biziere, and B. Cousin, Optimizing GMR sensor for low and high
frequencies applications, Sens. Actuators A, vol. 129, pp. 203-206, 2006.

[6] Charles Kittel, Introduction to Solid State Physics, John Wiley and Sons Inc, 1996.

[7] J. H. van Vleck, The Theory of Electric and Magnetic Susceptibilities, London: Oxford University
Press, chapter XII, section 76, 1932.

[8] M. Mendes, Micromagnetic Simulations of Spin Valve devices, senior thesis IST, 2005

[9] L. D. Landau and E. M. Lifshitz, Dynamics of Continuous Media, section 40, Pergamon Press, 1981.

[10] Finite Element Method Magnetics (femm) v4.2 - magnetostatic tutorial,


http://www.femm.info/wiki/HomePage

[11] A. Guedes,J.M. Almeida,S. Cardoso, R. Ferreira nd P.P. Freitas, Improving Magnetic Field Detection
Limits of Spin Valve Sensors using Magnetic flux guide concentrators, IEEE transactions on magnetics,
43, 2376-2378, 2007.

[12] P.M. Drljacca, F. Vincent, P.A. Besse, R.S. Popović, Design of planar magnetic concentrators for high
sensitivity Hall devices, Sensors and Actuators:A. Physical, 97,10-14, 2002.

[13] C. Fermon, M. Pannetier-Lecoeur, N. Biziere, and B. Cousin, Sens. and Actuators A, 129, 203-206,
(2006)

[14] Susana C. Freitas, Dual-stripe GMR and Tunnel junction read heads and ion beam deposition and
oxidation of tunnel junctions, PhD thesis, Lisbon, IST, 2001

[15] Ricardo A. Ferreira, Ion beam deposited magnetic spin tunnel junctions targeting HDD read heads,
non-volatile memories and magnetic field sensor applications, PhD thesis, Lisbon IST, 2008

75
[16] S. Cardoso et al, Ion Beam Deposition and Oxidation of Spin-Dependent Tunnel Junctions, IEEE
Transactions on Magnetics, 35, 2952, 1999.

[17] Susana C. Freitas, Micro and Nanofabrication Techniques (Handouts), Instituto Superior Técnico,
Lisboa 2008.

[18] S. Foner, Versatile and sensitive vibrating-sample magnetometer, Rev. Sci. Instrum., vol 30, no.7,
pp.548-557, July 1959.

[19] J.C. Mallinson, Magnetometer coils and reciprocity, J. Appl. Phys., vol 37, pp.2514-2515, 1966.

[20] L. C. Feldman and J. W. Mayer, Fundamentals of surface and thin films analysis, North-Holland, 1986.

[21] M.D.Mendes, Micromagnetic Simulations of Spin Valve devices, PhD thesis, Lisbon IST, 2005.

[22] Nuno G. Oliveira, Spin-Valve Heads for Tape Applications, PhD thesis, Lisbon IST, 2005.

[23] M.J. Almeida, Low resistance Magnetic Tunnel Junctions, Nanofabrication of sub-100nm devices based
on low resistance magnetic tunnel junctions, MSc thesis, Lisbon IST, 2009.

[24] I.C.Perdigão, Optimization of magnetic tunnel junction linear response for sensor applications - Simu-
lation and experimental, MSc thesis, Lisbon IST, 2009.

[25] S. Jin, Colossal Magnetoresistance in La-Ca-Mn-O, MEtal sand materials 5, No 6, 533, 1999.

[26] T. Nakajima et al, 1000% colossal magnetoresistance at room temperature in the A-site ordered per-
ovskite manganites, Sm1-xLax+yBa1-yMn2O6, J. Appl. Phys., 98, 046108, 2005.

[27] P. Wiśniewski, Giant anisotropic magnetoresistance and magnetothermopower in cubic 3:4 uranium
pnictides, 2007.

[28] I. Genish, Paramagnetic anisotropic magnetoresistance in thin films of SrRuO3 , Journal of Applied
Physics 95: 6681, 2004.

[29] T. R. Mcguire and R. I. Potter, Anisotropic Magnetoresistance in Ferromagnetic 3d Alloys.

[30] A.A. Veloso, PhD thesis, Advanced spin-valves for high density disk and tape drive storage applications,
Lisbon IST, 2001.

[31] M.N. Baibich, J.M. Broto, A. Van Dau Fert, et al., Giant Magnetoresistance of (001)Fe—(001)Cr
magnetic superlattices, Phys. Rev. Lett., 61,2472-2472, 1988.

[32] Binasch, P. Grünberg, F. Saurenbach, W. Zinn, Enhanced magnetoresistance in layered magnetic struc-
tures with antiferromagnetic interlayer exchange, Phys. Rev. B 39 (7), 4828-4830, 1989.

[33] A. Fert and I. A. Campbell, Phys. Rev. Lett. 21, 1190 - 1192, 1968.

[34] T. Valet, A. Fert, Phys. Rev. B 48, 7099 - 7113, 1993.

[35] B. Dieny, V. S. Speriosu, S. S. P. Parkin, B. A. Gurney, D. R. Wilhoit, and D. Mauri, Giant magne-
toresistance in soft ferromagnetic multilayers, Phys. Rev. B, 43(1), 1297-1300, 1991.

[36] Mott, N. F., A Discussion of the Transition Metals on the Basis of Quantum Mechanics. Proc. Phys.
Soc. 47, 571, 1935.

[37] Mott, N.F., The Electrical Conductivity of Transition Metals. Proc. Roy. Soc. A 153, 699., 1936.

[38] A. Fert et al, Electric resistivity of ferromagnetic nickel and iron based alloys, J. Phys. F 6, 849., 1976.

76
[39] B. Dieny, Giant magnetoresistance in spin-valve multilayers, Journal of Magnetism and Magnetic Ma-
terials 136(3): 335-359, 1994.

[40] A. P. Malozemoff, Random-field model of exchange anisotropy at rough ferromagnetic antiferromagnetic


interfaces, Phys. Rev. B 35(7), 3679-3682, 1987.

[41] A. P. Malozemoff, Mechanisms of exchange anisotropy, J. Appl. Phys. 63(8), 3874-3879, 1988.

[42] H. Sakakima, M. Satomi, Y. Sugita, and Y. Kawawake, Mechanism of the exchange-bias field in ferro-
magnetic and antiferromagnetic bilayers, J. Magn. Magn. Mater. 210, L20, 2000.

[43] N. Hasegawa et al, Nano-oxide-layer specular spin valve heads with synthetic pinned layer:Head per-
formance and reliability, J. Appl. Phys. 91, No 10, 8774, 2002.

[44] A. Fert, The present and the future of spintronics, Thin Solid Films 517(1): 2-5, 2008.

[45] W. H. Meiklejohn, C.P. Bean, New magnetic anisotropy, PR, 105, 904-904, 1957.

[46] J.C.Kools, T.Rijks,A.M.De Veirman, R.Coehoorn, On the ferromagnetic interlayer coupling in


exchange-biasedspin-valve multilayers, IEEE transactions on magnetics, 31, 3918-3920, 1995.

[47] J.L.Leal,M.H. Kryder, Interlayer coupling in spin valve structures, IEEE transactions on magnetics,
32, 4642-4644, 1996.

[48] L.G. Valdes,I.R.E Proc., Measurements of Sheet Resistivity with the Four-Point Probe, BSTJ, 37, p.
711, 1958.

[49] D. E. Helm, et al, Magnetoresistive spin valve sensor with improved pinned ferromagnetic layer and
magnetic recording system using the sensor, US Patent 5465185, 1995.

[50] H. A. M. Van den Berg, et al., GMR sensor scheme with artificial antiferromagnetic subsystem, IEEE
Transactions on Magnetics, 32, 4624, 1996.

77

Anda mungkin juga menyukai