Magnetodynamics in Spin Valves and Magnetic Tunnel Junctions with Perpendicular and Tilted Anisotropies

QUANG TUAN LE

Doctoral Thesis in Physics School of Information and Communication Technology KTH Royal Institute of Technology Stockholm, Sweden 2016 KTH Royal Institute of Technology School of Information and Communication Technology TRITA-ICT 2016:21 SE-164 40 Kista ISBN 978-91-7729-072-8 SWEDEN

Akademisk avhandling som med tillstånd av Kungl Tekniska högskolan framlägges till offentlig granskning för avläggande av teknologie doktorsexamen i Fysik fredagen den 30 sep 2016 klockan 13.00 i Sal C, Electrum, Kungl Tekniska högskolan, Kistagången 16, Kista.

Tryck: Universitetsservice US-AB

Abstract

Spin-torque transfer (STT) effects have brought spintronics ever closer to practical electronic applications, such as MRAM and active broadband microwave spin-torque oscillator (STO), and have emerged as an increasingly attractive field of research in spin dynamics. Utilizing materials with perpendicular magnetic anisotropy (PMA) in such applications offers several great advantages such as low-current, low-field operation combined with high thermal stability. The exchange coupling that a PMA thin film exerts on an adjacent in-plane magnetic anisotropy (IMA) layer can tilt the IMA magnetization direction out of plane, thus creating a stack with an effective tilted magnetic anisotropy. The tilt angle can be engineered via both intrinsic material parameters, such as the PMA and the saturation magnetization, and extrinsic parameters, such as the layer thicknesses. STOs can be fabricated in one of a number of forms—as a nanocontact opening on a mesa from a deposited pseudospin-valve (PSV) structure, or as a nanopillar etching from magnetic tunneling junction (MTJ)—composed of highly reproducible PMA or predetermined tilted magnetic anisotropy layers. All-perpendicular CoFeB MTJ STOs showed high-frequency microwave generation with extremely high current tunability, all achieved at low applied biases. Spin-torque ferromagnetic resonance (ST-FMR) measurements and analysis revealed the bias dependence of spin-torque components, thus promise great potential for direct gate-voltage controlled STOs. In all-perpendicular PSV STOs, magnetic droplets were observed underneath the nanocontact area at a low drive current and low applied field. Furthermore, preliminary results for microwave auto-oscillation and droplet solitons were obtained from tilted-polarizer PSV STOs. These are promising and would be worth investigating in further studies of STT driven spin dynamics.

Keywords: perpendicular magnetic anisotropy, tilted-polarizer, spintronics, STT, MTJ, pseudospin-valve, STO, ST-FMR, magnonics, magnetic droplet, droplet nucleation, droplet annihilation.

iii

Sammanfattning

Effekter av spinnvridmoment (STT) har fört spinntroniken allt närmare praktiska elektroniska tillämpningar, såsom MRAM och den spinntroniska mikrovågsoscillatorn (STO), och har blivit ett allt mer attraktivt forskningsområde inom spinndynamik. Användning av material med vinkelrät magnetisk anisotropi (PMA) i sådana tillämpningar erbjuder flera stora fördelar, såsom låg strömförbrukning och funktion vid låga fält i kombination med hög termisk stabilitet. Den utbyteskoppling (”exchange bias”) en PMA-tunnfilm utövar på ett intilliggande skikt med magnetisk anisotropi i planet (IMA) kan få IMA-magnetiseringsriktningen att vridas ut ur planet, vilket ger en materialstack med en effektivt sett lutande magnetisk anisotropi. Lutningsvinkeln kan manipuleras med både inre materialparametrar, såsom PMA och mättningsmagnetisering, och yttre parametrar, såsom skikttjocklekarna. STO:er kan tillverkas som flera olika typer - som en nanokontaktsöppning på en s.k. mesa av en deponerad pseudospinnventilstruktur (PSV) eller som en nanotråd etsad ur en magnetisk tunnlingsövergång (MTJ) – och bestå av mycket reproducerbar PMA eller av skikt med på förhand bestämt lutning av dess magnetiska anisotropi. MTJ-STO:er av CoFeB med helt vinkelrät anisotropi visar högfrekvent mikrovågsgenerering med extremt stort frekvensomfång hos strömstyrningen, detta vid låg biasering. Mätning och analys av spinnvridmoments-ferromagnetisk resonans (ST-FMR) avslöjade ett biasberoende hos spinnvridmomentskomponenter, vilket indikerar en stor potential för direkt gate-spänningsstyrda STO:er. I helt vinkelräta PSV-STO:er observerades magnetiska droppar under nanokontaktområdet vid låg drivström och lågt pålagt fält. Dessutom erhölls preliminära resultat av mikrovågssjälvsvängning och av s.k. ”droplet solitons” hos PSV-STO:er med lutande polarisator. Dessa är lovande och skulle vara värda att undersökas i ytterligare studier av STT-driven spinndynamik.

Nyckelord: vinkelrät magnetisk anisotropi, lutande polarisator, spinntronik, STT, MTJ, pseudospinnventil, STO, ST-FMR, magnonics, magnetisk droppe, nukleering av droppe, anihilering av droppe.

Acknowledgement

I am deeply grateful to my supervisor, Prof. Johan Åkerman, who was the one to give me the opportunity, to aid me getting it up and running; and showed your precious support all along to complete my projected research objectives. Also I would like to thank you very much, Johan, Maria, Hanna, Silas and Pontus, for creating and sharing joyful moments, which my family and I will never forget. Massive! To get this thesis printed, I would like to specifically thank Dr. Sun-Jae Chung. There have been many times my focus appeared to drift away and my motivation dissipated, every time you showed your tireless eagerness and passion to science that motivated and helped me getting a grip and pushing towards. Without you and your wholehearted efforts, I could not have it done. Magnificent! It is little wonder I could reach this significant milestone, given the knowledge and experience I benefited from the graduate Majid Mohseni and Sohrab Sani, as well as my current colleagues: Anders Eklund, Fatjon Qejvanaj, Hamid Mazraati, Sheng Jiang and Amir Banuazizi in Applied Spintronics group at KTH. A brief time working at NanOsc AB with Johan Persson and Fredrik Magnusson was valuable, too, and I am really appreciated. Influential! I have also got generous helps, in many ways, from scores of friends: Reza Nikpars, Magnus Lindberg, Himanshu Kataria, Cecillia Aronsson and Helena Strömberg (Acreo, IRnova, Ascatron); Christian Ridder, Per-Erik Hellström, Gabriel Roupillard, Yong-bin Wang, Tingsu Chen, Konstantinos Garidis, Ganesh Jayakumar, Arash Salemi, Hossein Elahipanah and Ahmad Abedin (EKT, ICT-KTH); Philipp Dürrenfeld, Afshin Houshang and Martina Ahlberg (Götenborgs Universitet); and technical supports from the Electrum staff, to work out my device processing inside KTH’s Electrum Fab and Chalmers’ MC2 Fab. Handy! The working environment is really enjoyable, thanks to Madeleine Printzsköld – my dedicated department administrator to simplify the related paperwork – and others along the corridor: Mats Göthelid, Sergei Popov, Ilya Sychugov, Markus Soldemo, Tobias Övergaard, Milad Yazdi, Roodabeh Afrasiabi, Federico Pevere, Shaozheng Ji, Yashar Hormozan, Miao Zhang, Viktor Jonsson and many others. Warm!

My sincere and lasting thanks are to all my beloved: Ngoc Anh, Ha Vy, Hoai Nam, my parents and parents-in-law, my brothers and sisters. This is the work I want to dedicate after your tremendous sacrifice and marvelous patience along the road. Your love, encouragement and tolerance invaluably helped to stiffen my back, harden my mind, and keep the strands of my life and work together. This little book marked the end of one chapter so a new chapter of life begins, but I always believe that I can count on you. Permanent!

Tuan Le

Stockholm – August 08, 2016

Table of Contents

Abstract ...... iii Acknowledgement ...... v Table of Contents...... vii List of Publications/Manuscripts ...... ix List of Acronyms and Symbols ...... xi

1. General introduction and thesis outline ...... 1 2. Methodology ...... 5 2.1 Sputtering ...... 5 2.2 Chemical vapor deposition ...... 6 2.3 Evaluation of magnetic properties ...... 7 2.4 Device fabrication procedure ...... 8 2.5 Device measurement ...... 12 2.5.1 dc characterization ...... 12 2.5.2 Microwave measurement ...... 13 2.5.3 ST-FMR measurement and analytical model ...... 13

3. Perpendicular magnetic anisotropic thin films...... 17 4. Spin dynamics in all-perpendicular MTJ STOs ...... 21 4.1 Development of all-perpendicular MTJs ...... 21 4.2 Spin dynamics in all-perpendicular MTJ STOs ...... 26 4.2.1 Device fabrication and dc characterization ...... 26 4.2.2 Magnetodynamic modes in p-MTJ nanodevices ...... 30 4.2.3 Auto-oscillation characteristics ...... 32 4.2.4 Bias-dependent spin torque components ...... 37

5. Magnetic droplets in all-perpendicular PSV STOs ...... 47 5.1 Application of PMA multilayers in spin valve devices ...... 47 5.2 Conceptual magnetic droplet solitons ...... 48 5.3 Results and discussion ...... 52

6. Tilted polarizer spin valve STOs ...... 57 6.1 Development of tilted magnetic anisotropy materials ...... 57 6.2 Experimental results ...... 61 vii

Conclusions and future work ...... 67 Bibliography ...... 69 List of Figures and Table ...... 81 Appendix: Papers and Manuscripts ...... 83

viii

List of Publications/Manuscripts

A. List of appended papers

I. “Depth dependent magnetization profiles of hybrid exchange springs”, Physical Review Applied, 2, 04414 (2014). T. N. Anh Nguyen, R. Knut, V. Fallahi, S. Chung, S. M. Mohseni, Q. Tuan Le, O. Karis, S. Peredkov, R. K. Dumas, C. W. Miller, and J. Åkerman. Author’s contribution: sample preparation.

II. “Investigation on the tunability of spin configuration inside exchange coupled spring of hard/soft magnets”, IEEE Transactions on Magnetics, 50 (6), 2004906 (2014). T. N. Anh Nguyen, V. Fallahi, Q. Tuan Le, S. Chung, S. R. Sani, R. K. Dumas, Casey W. Miller, and J. Åkerman. Author’s contribution: sample preparation.

III. “Free- and reference-layer magnetization modes vs. in-plane magnetic field in a magnetic tunnel junction with perpendicular magnetic easy axis”, Physical Review B (2016) (accepted). H. Mazraati, Q. Tuan Le, Ahmad A. Awad, S. Chung, S. Ikeda, H. Ohno, and J. Åkerman. Author’s contribution: measurements, analyses and revising.

IV. “Ultra-high frequency tunability of low-current spin torque oscillators based on perpendicular magnetic tunnel junctions”, manuscript ready for Applied Physics Letter (2016). Q. Tuan Le, A. Eklund, A. Banuazizi, S. Chung V. Fallahi, T. N. Anh Nguyen, M. Yamanouchi, Eli C. I. Enobio, S. Ikeda, H. Ohno, and J. Åkerman. Author’s contribution: measurements, analyses and writing.

V. “Bias dependence of parallel spin torque in all perpendicular magnetic tunnel junction nanodevices”, manuscript in preparation for Applied Physics Letters (2016). Q. Tuan Le, H. Mazraati, S. Chung, T. N. Anh Nguyen, and J. Åkerman. Author’s contribution: measurements, analyses and writing.

VI. “Magnetic droplet solitons in all-perpendicular nano-contact spin torque oscillators”, manuscript in preparation for Nature Nanotechnology (2016). S. Chung, Q. Tuan Le, M. Ahlberg, A. Awad, H. Mazraati, A. Houshang, S. Jiang, T. N. Anh Nguyen, R. Dumas, and J. Åkerman. Author’s contribution: device fabrication, measurements, analyses and writing.

B. List of published papers not included in this thesis

I. “CoFeB based spin Hall nano-oscillators”, M. Ranjbar, P. Dürrenfeld, M. Haidar, E. Iacocca, M. Balinskiy, Q. Tuan Le, M. Fazlali, A. Houshang, A. Awad, R. K. Dumas, and J. Åkerman, IEEE Magnetics Letters 5, 3000504 (2014)

II. “Role of Boron diffusion in CoFeB/MgO magnetic tunnel junctions”, S. Mukherjee, R. Knut, S. M. Mohseni, T. N. Anh Nguyen, S. Chung, Q. Tuan Le, J. Åkerman, J. Persson, A. Sahoo, A. Hazarika, B. Pal, S. Thiess, M. Gorgoi, P. S. Anil Kumar, W. Drube, O. Karis, and D. D. Sarma, Phys. Rev. B. 91, 085311 (2015).

III. “Thick Double-Biased IrMn/NiFe/IrMn Planar Hall Effect Bridge Sensors”, F. Qejvanaj, M. Zubair, A. Persson, S. M. Mohseni, V. Fallahi, S. R. Sani, S. Chung, Q. Tuan Le, F. Magnusson, and J. Åkerman, IEEE Trans. Magn. 50 (11), 1–4 (2015)

List of Acronyms and Symbols

PMA perpendicular magnetic anisotropy

IMA in-plane magnetic anisotropy

P (AP) parallel (antiparallel)

PSV pseudospin-valve

GMR giant magnetoresistance

TMR tunneling magnetoresistance

p-MTJ all perpendicular magnetic tunneling junction

STO spin-torque oscillator

ST-FMR spin-torque ferromagnetic resonance

STT spin transfer torque

M (M) magnetization (vector)

H (H) magnetic field (vector)

HK magnetic anisotropy field

R resistance

ML multilayer

θ angle

τ// parallel spin-torque component

τ⊥ perpendicular spin-torque component

∂ partial derivative

Å ångström (= 0.1 nanometer)

Chapter 1

General introduction and thesis outline

Spintronics is a field of research whose central focus lies in the manipulation of spin-up and spin-down populations of electrons in solid materials. Consideration of the attachment of spin to the electron leads to the key topics in spintronics: spin transport and spin dynamics. Active polarization of spins has become the most popular means of manipulation aimed at generating an inequality in the numbers of spin-up and spin-down electrons, and thus creating a spin current that can flow even when the net charge current is equal to zero. A typical way to polarize electron spins is to use a ferromagnetic electrode (called a spin polarizer or a spin filter, among other names) in electrical contact with a medium where spin-polarized electrons then accumulate. The accumulation rate strongly depends on the relaxation, or flip, of spins, which tends to drive the polarization back towards the equilibrium state. In the early days, the transport properties of polarized spins were mostly observed in tunneling measurements carried out on various ferromagnet/insulator/superconductor and ferromagnet/insulator/ ferromagnet systems. These studies resulted in Julliere’s formula [1] modeling the tunneling magnetoresistance ratio (TMR):

= = = (1.1) ∆푅 푅↑↓ − 푅↑↑ 퐺↑↑ − 퐺↑↓ 푇푀푅 푅↑↑ 푅↑↑ 퐺↑↓

1. INTRODUCTION

which quantitatively measures the change in a system’s conductance G (or conversely resistance R) between the parallel (G↑↑, R↑↑) and antiparallel (G↑↓, R↑↓) states of magnetization in the two ferromagnetic regions. This prediction led to another important discovery — the spin-valve effect: by enforcing the misalignment of the magnetization in two separated ferromagnetic regions of a system, the resistance could be altered accordingly. The effect was then illustrated experimentally in multilayered structures, whose originators, A. Fert and P. Grünberg, received the 2007 Nobel Prize in Physics for their contribution to the revolution in information data-storage capacity [2][3]. A stark difference with these spin-valve structures, however, is that the spin-polarized electrons are transported via direct conduction, rather than tunneling, from one region to the other through a conductive nonmagnetic buffer; the change in resistance is then referred to as the giant magnetoresistance ratio (GMR, instead of the TMR. Inversely, as independently proposed by Berger [4] and Slonczewski [5], a current of spin-polarized electrons flowing into a ferromagnetic electrode can transfer their spin angular momenta, thus causing a change in the orientation of the receiver’s magnetization, a phenomenon known as the spin torque transfer (STT) effect. Realizations of STT-induced magnetization reversal, or switching, have been successfully made, opening the door to the development of a new type of computer memory, nonvolatile STT Magnetoresistive Random-Access Memory (STT-MRAM) [6, 7, 8, 9, 10]. In general, the motion of magnetic moments in a ferromagnet under the active interaction with its surrounding environment, without taking any effects of manipulation into account, is described by the well-known Landau–Lifshitz–Gilbert (LLG) equation [11][12]:

= × × (1.2) 푑푴 푑푴 훾푴 푯푒푓푓 − 훼 푴 푑푡 푑푡 The first term represents the Larmor precession of the ferromagnet magnetization vector M in the effective magnetic field Heff, which is the vector sum of the external field, the coupling field, the anisotropy field, etc.; the second term indicates that such a precession would eventually be damped out with the moment fully aligned with the field; γ and α are the gyromagnetic ratio and the Gilbert damping factor, respectively.

1. INTRODUCTION

If an electric current is driven through a trilayer structure of ferromagnet

1/nonmagnetic/ferromagnet 2 (F1/N/F2) the dynamics is now altered by the presence of the STT effect. The validity of the LLG equation can then only be maintained by additional terms representing the STT effect [13][14]. The modified LLG equation, known as the LLG-Slonczewski (LLGS) equation, becomes:

= × × 푑푴ퟐ 푑푴ퟐ 훾푴ퟐ 푯푒푓푓 − 훼 푴ퟐ (1.3) 푑푡 푑푡 + × ( × ) + × 2 2 ℏ ℏ 퐽푐 훽푆푇푇푴ퟐ 푴ퟏ 푴ퟐ 퐽푐 훽퐹퐿푇푴ퟏ 푴ퟐ The two additiona푒 l terms explicitly describes푒 the role played by the current in either enhancing or canceling the damping term, determined by the critical current density Jc. The scalar coefficients in these terms, βSTT and

βFLT, represent the two origins of transferred torques: the STT and the field- like torques (FLT), respectively. Further details on the STT theory can be found in Ref. [14]. Under a specific set of conditions, the transferred torques can sustain a steady precession, leading to oscillatory changes in the resistance. Consequently, the voltage between the ferromagnetic electrodes oscillates and a microwave frequency signal is generated. The term spin-torque transfer oscillator (STO) is given to devices operating in this fashion, though the name spin-torque transfer nano-oscillator (STNO) perhaps illustrates more explicitly the nanosized scale on which the events of precession occur [15]–[17]. The engineering objectives to which this thesis is devoted concern the realization and evaluation of STO operation in various magnetic structures, in which the manipulation of magnetic anisotropy is the most important point. This work builds on both internal achievements and external collaboration. Nanoscale devices were successfully fabricated from the deposited wafers of tilted, orthogonal, and perpendicular spin-valve structures; while highly reliable perpendicular magnetic tunnel junction devices were acquired through collaboration. Effort was also devoted to arriving at a partial understanding of the mechanism of spin dynamics in the structures under study.

1. INTRODUCTION

Apart from the major achievements presented here in the form of scientific articles, ready or in preparation for submission, the main body of this thesis is devoted to proving detailed descriptions of various aspects of the work, ranging from technical preparations to the data acquisition from measurements. The thesis is divided into several chapters: Chapter 2 presents the full process for fabricating the final measurable devices, and gives details of the techniques and instruments used. The descriptions of the measurements conducted to characterize the response of the device being tested are also given within the context of applied experimental conditions. Chapter 3 shows the significant role of perpendicular magnetic anisotropy thin films, particularly in spintronics. The merits of all-perpendicular MTJ STOs are then illustrated and discussed generally in Chapter 4. The unique characteristics of low-field and zero-field magnetic droplets in all- perpendicular spin-valve STOs are presented in Chapter 5. In a similar manner, Chapter 6 presents the observations on spin dynamics from yet another type of spin-valve STOs based on a tilted exchange spring polarizer. Finally, several conclusions are drawn, together with a brief outlook of a possible future expansion based on the work conducted so far. Six manuscripts related to the thesis objectives appear at the end of the thesis.

Chapter 2

Methodology

In this chapter, the entire fabrication process from thin film deposition to the final device will be described. The included information of processing recipes and measurement conditions will help the STO device fabrication and characterization to be reproduced at other facilities.

2.1 Sputtering

Sputtering is a thin film physical vapor deposition (PVD) technique in which a target is bombarded by highly energized ions and releases its atoms or molecules which then land onto a substrate. The bombarding ions are generated from “plasmarized” gas or gaseous mixture that is flowing constantly into the sputtering chamber and is accelerated by an electric field. The ion energy is therefore defined by setting the pressure of the chamber and the voltage applied to the target holder. For material deposition in this thesis, the gas used in sputtering deposition was argon, as it is nonreactive and is expected to produce a film composition similar to that of the target. However, in general, the gas or gases could also be reactive ones such as oxygen if the intention is to deposit an oxide thin film of the target material. The multilayered stack deposition used in this thesis was carried out on an AJA Phase II-300 ultra-high vacuum sputtering system with confocal holders supporting 7 targets. The multilayered structures could therefore be

2. METHODOLOGY

deposited without disruption under a stable sputtering environment. Furthermore, a load-lock chamber attached to the main sputtering chamber also prevented the high vacuum state from breaking down even when a substrate was introduced to, or withdrawn from, the main chamber. The uniformity of the deposited films on a full 4-inch wafer was about ± 2%. An illustration of the confocal sputtering technique is shown in Figure 2.1.

Figure 2.1: Schematic diagram of confocal sputtering technique

2.2 Chemical vapor deposition

Silicon dioxide (SiO2) is commonly used to insulate fabricated devices and protect them from electrical short-circuits. In the present work, a reliable

SiO2 layer was deposited by plasma-enhanced chemical vapor deposition (PE-CVD) method, instead of the sputtering technique previously described. In PE-CVD, a gaseous composition of chemical agents is introduced into an evacuated chamber. The product of the chemical reactions between them is in the solid phase and is deposited onto the substrate surface sitting inside

2.3. Evaluation of magnetic properties

the chamber. SiO2 is deposited in this manner over the whole surface of the full 4-inch wafer as the product of the following chemical reaction:

( ) + 2 ( ) (2.1) 4 2 = (푆푖퐻 푔푎푠) + 2 (푁 푂) 푔푎푠+ 2 ( )

푆푖푂2 푠표푙푖푑 퐻2 푔푎푠 푁2 푔푎푠 While it is best to deposit SiO2 on a solid surface at as high as 300 °C, this temperature is particularly close to the Curie point of Ni-based ferromagnetic alloys and may induce the undesired change in magnetic properties of the structure being insulated. Through careful in-house calibration, however, a 30-nm thick SiO2 layer was reliably deposited at 180 °C without undermining the critical magnetic properties, such as the giant magnetoresistance ratio. The thickness reproducibility of SiO2 deposition is of importance here; this will be seen later in terms of device processing, when the etching through it to make contact openings needs to be conducted precisely. Both under- and over-etching are likely to affect the performance of the final device.

2.3 Evaluation of magnetic properties

Since device processing is normally costly and time-consuming, it is important to optimize the material performance before the final deposition of multilayered structures onto the wafer, such as the determination of the magnetic easy axis. This magnetic anisotropy behavior is strongly influenced by the interface between layers and is sometimes driven by the crystalline orientation; these, in turn, are affected by the deposition rate, the heating and cooling treatment during and after deposition, and other effects. The magnetic anisotropy of a magnetic structure, whether single-layer or multilayer, can be determined by various magnetometric techniques—for instance, by using an alternating gradient magnetometer (AGM), as was the case here. AGM measurements are capable of rapidly producing the magnetic hysteresis loops of a magnetic structure in both principal directions of the applied magnetic field: in-plane (IP) and perpendicular (or out-of- plane, OOP) to the sample surface. Figure 2.2 shows the examples of IP and OOP hysteresis curves of a) an ideal in-plane magnetic anisotropic film with zero IP saturation field and high OOP saturation field, and b) an ideal 7

2. METHODOLOGY

Figure 2.2: Illustrations of ideal hysteresis curves of (a) in-plane magnetic anisotropy and (b) perpendicular magnetic anisotropy ferromagnetic thin films perpendicular magnetic anisotropic film with zero OOP saturation field and high IP saturation field. From these curves, the magnetic easy axis can be determined, as can the values of the saturation fields, Hsat, corresponding to the structure. While both ideal cases show zero coercivity, in reality, coercive field is different from zero and these curves become the hysteresis loops, as can be seen in the coming chapters. A further calibration and more detailed sample preparation can also allow measurement of the saturated magnetization, MS, of a given certain magnetic layer.

2.4 Device fabrication procedure

The fabrication of electrically measurable devices consists of fundamental steps common in CMOS technology, such as lithography, etching and metallization. The fabrication yield is therefore affected by the reproducibility of various processing steps. After verifying the magnetic properties of the materials, as previously described, the multilayered structure is then deposited on a standard commercial 4-inch thermally oxidized SiO2(1000 nm)/Si wafer, for the sake of compatibility between the various tools used in the whole processing procedure. The steps of the fabrication routine are illustrated in Figures 2.3– 5, and described in what follows.

2.4. Device fabrication

A spin-coating station automatically handles the sequential jobs of wafer loading, photoresist coating, and soft baking. Typically, a layer of 1.0 – 1.2 micrometer thick photoresist is applied on the stacked wafer, stabilized by brief baking at 90 °C. The automatic run plus the standard photoresist ensure a high reproducibility of the coated layer. Afterward, the coated wafer is subjected to ultraviolet (UV) light exposure. Owing to the ability of UV light to crosslink the molecules of positive-tone photoresist, the exposed photoresist turns to polymer and becomes soluble in a specific solvent, while the unexposed portion remains, due to low solubility. This selective illumination is assisted by a physical mask with predefined patterns (Fig. 2.3a). The wafer exposure is carried out automatically by the XLS stepper, whose resolution is about 700 nm. This stepper, equipped with a reticle (mask carrier), possesses the advantage of auto-alignment both on the global (wafer) and local (dies) scale. With its ability of demagnification by 5 times (meaning that it is capable of scaling the mask pattern to one-fifth of its size), microsized patterns can be written on the stack wafer with ease using a low-cost mask, while still maintaining high reproducibility. The wafer exposed by the stepper is then baked briefly at 110 °C to enhance the contrast between the illuminated and masked areas in terms of solubility during the developing step. Finally, the photolithography process is wrapped up by a hardening step on remaining coated areas of the wafer in an oven or furnace at 110 °C, which makes it sufficiently solid to tolerate the subsequent etching step. Along the entire processing routine, this photolithography is applied in several major stages, including mesa etching, contact opening, alignment key etching, and electrode pad molding. Photolithography, however, is limited in producing more confined patterns than its tool resolution, while STO fabrication typically requires either design of the contact or pillar down to sub-100 nm in device dimensions. E-beam lithography is therefore acquired through the collaboration with the MC2 foundry, a sister clean room at Chalmers University, where the job is carried out on an electron beam lithography system (JEOL JBX-9300FS). With the exception of the final step of metallization, every lithography step is followed by etching of the unmasked area on the wafer. Various etching methods are applied depending on the criteria of the etched materials and pattern scales. In order to fabricate microsized mesas, argon-

2. METHODOLOGY

Figure 2.3: Fabrication process of spin valve mesas from deposited stack based ion milling is sufficient; however, reactive etching is required for more selective etching requirements. The general principle of ion milling is quite similar to what happens in the sputtering event described at the beginning of this chapter. That is, inert argon ions are energized and bombard the targeted wafer, thus removing the uncovered portion of the multilayered stack while leaving a hardened photoresist masking area relatively intact—provided that the photoresist mask is thick enough to endure the bombardment (Fig. 2.3b). The etching by-products and photoresist mask are then removed by common resist remover enhanced with gentle ultrasonic, leaving only micro-size mesas of spin-valve stack on the wafer (Fig. 2.3c). In order to protect the mesa side walls from oxidation, as well as from possible electrical short-circuiting, a 30-nm-thick SiO2 layer is deposited using the PE-CVD technique (Fig. 2.4).

Figure 2.4: Short-circuit and oxidized protection of fabricated mesas

In the other steps, such as contact and alignment key openings—where an unmasked portion of the insulating SiO2 layer needs to be etched away— another etching technique is applied: the highly selective reactive ion etching (RIE). In SiO2 RIE, the etchant gaseous mixture of CHF3 and Ar is introduced into the etch chamber and initiates a chain of chemical reactions with exposed SiO2 surface:

2.4. Device fabrication

+ + (2.2) 3 4 퐶퐻 퐹 +푝푙푎푠푚푎 → 퐶+퐹 퐻퐹

푆푖푂2 퐻퐹 → 푆푖퐹4 퐻2푂 The SiO2 is thus transformed into gaseous products which are vented out. The etching parameters—such as the etchant ratio, flow rate, applied bias, and so on—are carefully calibrated to enhance vertical etching (that is, etching normal to the wafer surface), which is critical in preventing the SiO2 area underneath the mask from being damaged. The RIE recipe used for device fabrication in the present work is capable of etching a nano-contact opening 30 nm deep without significantly widening its nominal 50 nm diameter. The routine steps taken to open the STO contacts through the SiO2 insulator are illustrated in Figure 2.5.

Figure 2.5: Contact opening process including lithography and reactive ion etching

In order to obtain as high a fabrication yield as possible, inspections are carried out repeatedly throughout the procedure, ranging from simple techniques, such as optical microscopy, to highly accurate ones, like SEM. Further details of these steps are omitted within the context of this thesis. The final STO is illustrated as the 3D sketch in Figure 2.6 whose the electrode configuration matches with the commercial ground-signal-ground probe later used in both dc and microwave measurement.

2. METHODOLOGY

Figure 2.6: 3D sketch of a spin-valve STO: SiO2-insulated stack mesa with two ground contacts at its sides and signal nanocontact at the center

2.5 Device measurement

2.5.1 dc characterization

As briefly mentioned in the thesis introduction, magnetoresistance ratio is a quantitative measure of the resistance change of a magnetic multilayer structure caused by the reconfiguration of magnetization in its different ferromagnetic layers. To measure that ratio in magnetic tunneling junction (MTJ) and spin- valve (SV) devices composed of a relatively weaker (free) ferromagnetic layer and a stronger (fixed) one, the device electrodes are probed with an electrical connection to a combo of dc current supplier and nanovoltmeter under the application of a magnetic field generated by a pair of Hemholtz coils. By tuning the field strength, polarity, and angle, the resistance versus field curves (R–H loops) are obtained. Generally, the maximum change of resistance indicates the re-alignment of magnetization in free and fixed layers from parallel (P) to antiparallel (AP) or vice versa. It is worth being noticed that the dc current should be essentially small to avoid the significant excitation through spin torque effects.

2.5. Device characterization

From the R–H loops, several important parameters can be extracted: TMR or GMR ratio, anisotropy orientation (magnetic easy axis), and even the exchange coupling strength one ferromagnetic layer exerts on its counterpart. These parameters will contribute considerably in obtaining the intrinsic magnetic parameters from numerical calculations or simulation.

2.5.2 Microwave measurement

In order to observe and analyze the operational performance of an STO device, a broadband spectrum analyzer that records the microwave signals generated by the device is added to the above dc measurement setup. The connection is supported by a circuit consisted of a low-noise amplifier (LNA) to enhance the generated signals from device and a bias-tee to independently introduce the driving current and pick out the signals. Conventionally, there are two modes of a microwave measurement. In the field-sweep mode, the applied magnetic field increases or decreases either in single or both polarities and with fine step while the drive dc current being kept constant, to study the field-dependent behavior of the STO operation. On the contrary, in the current-sweep mode, the dc current scans up and down under a constant applied field to obtain the bias-dependent response of the device. The STO performance characteristics, such as peak frequency, spectral linewidth and generated microwave power, are then extracted from the Lorentzian fittings of the computer recorded data.

2.5.3 ST-FMR measurement and analytical model

In conventional ferromagnetic resonance (FMR) measurements, a low amplitude radio frequency (rf) field is applied to excite the oscillation of the magnetic moment in a ferromagnetic thin film under the resonance condition of frequency: f field = fosc, at which the effective damping is reduced drastically. In spin-torque FMR (ST-FMR), however, a mixture of both rf and dc current is introduced in place of an rf field to drive the excitations of the magnetic moments in the free layer of an STO device under the application of a dc magnetic field. Details of ST-FMR measurement setup can be found in Figure 2.7.

2. METHODOLOGY

Figure 2.7: Scheme of the ST-FMR measurement setup

The output dc mixed voltage from ST-FMR measurement is detected from the device being measured, and is broadly known to be a result of the spin-diode effect [18]. This output voltage becomes much larger at resonance, and hence generates a voltage spectrum over the frequency sweeping band, corresponding to each set of external conditions included the current amplitude and polarity, the field strength and angle, the input rf power, and so on.

The output voltage (Vout) spectrum from the ST-FMR measurement can be fitted using the model derived from the LLGS equation:

4( ) = 4( ) + 푓푟푒푠 − 푓 ∆푓 푉표푢푡 푉푎푠푦푚 2 2 (2.3) 푓푟푒푠 − 푓 ∆푓 + 4( )2 + ∆푓 푉푠푦푚 2 2 푓푟푒푠 − 푓 ∆푓 where f, fres, and ∆f are input signal frequency, resonant frequency, and spectral linewidth, respectively. The overall output is therefore split into two fundamental components: the asymmetric (Vasym) and symmetric (Vsym). Each term is analyzed in the light of the effects that originate from, or are

2.5. Device characterization

induced by, the STT and FLT effects which account for the last two terms of the LLGS equation of motion – Eq. (1.3).

Chapter 3

Perpendicular magnetic anisotropic thin films

Magnetic surface anisotropy was theoretically predicted by L. Néel in 1954 [19], as arising from the reduced symmetry of surface atoms of Co, Ni, Fe and their alloys. A few decades after this prediction, magnetic surface anisotropy was quantitatively measured using the ferromagnetic resonance method in Ni(111) and Fe(100), as well as in the alloys NiFe(100) and NiCo(100) [20]. More interestingly, the opposite signs shown in the measured anisotropy energy density allowed out-of-plane and in-plane magnetic surface anisotropies to be realized and distinguished from each other. Research into perpendicular magnetic anisotropy (PMA) materials proved essential for improving magnetic and magneto-optic recording density during the 1980s, and led to great expectations that such magnetic surface anisotropy exists at the interface of ferromagnetic/non-magnetic bilayers as long as they are considered as reduced symmetry, or in other words, the bilayer individual thicknesses are in the range of a few atoms [21]–[26]. Experimentally, PMA was first discovered in [Co/Pd] multilayer (ML) (a bilayer in square brackets represents the basis of a multilayer deposited by multiple repetitions; this convention is used throughout this thesis), particularly when the Co layer thickness (tCo) was less than 8 Å and the Pd thickness (tPd) was sufficiently thin to keep Co and Pd lattice mismatching in balance. The results obtained by Draaisma et al. [23] in [Co/Pd] and [Fe/Pd] MLs showed that PMA in the former was easily

3. PMA THIN FILMS

retained for tCo < 7.2 Å, but was quite small in the latter to have any significant contribution. Furthermore, a phenomenological relationship was built upon the experimental data to describe the Co thickness-dependent effective anisotropy Keff of the ML formulated in terms of the bulk, or volume, anisotropy Kv which contains shape, magneto-crystalline and magneto-elastic anisotropies; and interfacial anisotropy Ki:

2 = + (3.1) 퐾푖 퐾푒푓푓 퐾푣 푡퐶표 The perpendicular magnetic anisotropy was also reported in several Co- based MLs such as [Co/Pt] and [Co/Au]. In [Co/Pt] MLs prepared under similar conditions as the [Co/Pd] MLs [25], the perpendicular anisotropy was rather small due to the low values of coercivity and remanent magnetization. However, it was quite interesting that the PMA was retained even at tCo > 14 Å which is significantly thicker than the critical Co thickness of 8 Å needed to maintain PMA in [Co/Pd] MLs, and was independent of the Co thickness, perhaps owing to the fact that the diffusion occurred at the Co/Pt interface and hence diluted the Néel surface anisotropy. The interface diffusion also helped to explain the similar behavior of PMA in [Co/Au] MLs [26]. As it happened, there existed a critical thickness for both layers constituting the interface, beyond which the PMA vanished, and an inverse proportional relationship between Keff and tCo was also observed in most cases, suggesting that the anisotropy contribution of magnetostriction caused by lattice mismatching strain might become significant, depending on layer thickness. For example, PMA became weaker in [Co(2 Å)/Pd] MLs when tPd exceeded 9 Å [27]. This complication, however, creates room to manipulate the PMA of a multilayer in terms of individual thickness, thickness ratio, and crystallinity. More recently, studies of sputtered [Co/Ni] and [Co/Pd] MLs [28][29] gave further detailed insights into PMA and damping constants in the relation with material parameters by utilizing time-resolved magneto- optical Kerr effect spectrum (TR-MOKE) measurements. In both cases, where the respective thickness ratios tCo/tNi = 0.4 and tCo/tPd = 0.55 were kept constant while the individual thicknesses were varied, the values of the Gilbert damping constant α were found to be independent of the anisotropy

3. PMA THIN FILMS

field, and hence of PMA. Meanwhile, the PMA was seen to be strengthened when the bilayer periods tCo + tNi and tCo + tPd decreased. Although the PMA multilayers, with their advantages of high density and thermal stability, have greatly benefited the development of recording media, it has only recently been recognized as a rich source for spin dynamics research, thus motivating another part of the current thesis. The details of the spin dynamics studied in various spin-valve structures based on PMA multilayers will be presented in Chapters 5 and 6 below.

Chapter 4

Spin dynamics in all-perpendicular MTJ STOs

4.1 Development of all-perpendicular MTJs

The wave nature of electrons gives rise to the tunneling phenomenon which can occur through a sufficiently thin insulating buffer between two electrical conductors. If these two electrodes are made of ferromagnetic materials to form a magnetic tunneling junction (MTJ), and as long as the electron spins are conserved during tunneling, the MTJ resistance R can be related to the relative angle θrel between the magnetization of the two electrodes [30]:

( ) = 1 + × (4.1) 2 2 휃푟푒푙 푅 휃푟푒푙 푅↑↑ � 푇푀푅 푠푖푛 � where R↑↑ is the resistance in the parallel state; and TMR is the tunneling magnetoresistance ratio. At a microscopic level, the resistance change is related to the number of electrons in different states, or in other words, to the electron density of states N at the Fermi level EF. By splitting the electron density of states into two bands corresponding to spin-up (↑) and spin-down (↓) electrons, the spin polarization factor P of conduction electrons can be defined as [31][32]:

4. ALL PERPENDICULAR MTJ STO

( ) ( ) = (4.2) ( ) + ( ) 푁↑ 퐸퐹 − 푁↓ 퐸퐹 푃 푁↑ 퐸퐹 푁↓ 퐸퐹 Julliere [1] derived the relationship between TMR and the spin polarization factors:

2 = 1 (4.3) 푃1푃2 푇푀푅 − 푃1푃2 where P1 and P2 are the polarization factors of two corresponding ferromagnetic electrodes; then demonstrated this experimentally in a Fe/Ge/Fe junction. The measured TMR was 14%, while the calculated value taken from the polarization factors of Fe and Co reported previously was 26%. The difference was said to be due to spin relaxation at the ferromagnetic/nonmagnetic interface. That is, in an ideal medium where the value of P is unity and without considering any spin flipping events, the TMR could have been infinity. In reality, an extremely high TMR of 1000% or more has been demonstrated in several MgO-based MTJs measured at low temperature; for example, a TMR of 1144% was recorded in

CoFeB/MgO/CoFeB at 5 K [33] and of 1995% in full-Heusler Co2MnSi/MgO/

Co2MnSi at 4.2 K [34]. The development and improvement over time of MTJs, illustrated in Figure 4.1, produced a dramatic increase in TMR in MgO-based junctions, even though they were discovered almost a decade after Al2O3-based structures. The cause behind the extremely high TMR in MgO-based MTJs was first predicted theoretically in 2001 [35][36]; it was argued that the crystalline matching between MgO(001) and bcc Fe(001) or bcc Co(001) dramatically enhances the polarization factor and, consequently, the TMR [37]. The prediction was then accompanied by a surge in discoveries of several systems, including CoFeB(001)/MgO(001) [38]–[40] and Co(001)/MgO(001) [41][42]. MTJs with NiFe free layers also exhibit a significant improvement compared to Al2O3-based structures [43]. Despite its great promise of high TMR, the strict rule of crystalline matching represents a formidable challenge for thin film deposition of MTJs. Without a preferably oriented crystal substrate, for instance, the sputtering

4.1. All perpendicular MTJs

deposition technique is typically incapable of producing well-oriented thin films of Co(001), Fe(001), or their alloys, thus greatly undermined their potential from commercial perspective. It is because the easy axis of a magnetic thin film generally prefers to align in the plane of the thin film. This tendency originates from the nature of free-energy minimization, as the magnetostatic energy makes the out-of-plane alignment very unfavorable. The effect of the magnetostatic energy can be described as a so-called demagnetizing field acting on the magnetization, and being proportional to the perpendicular component of the magnetization but with opposite sign. Realizing a perpendicular easy axis of the ferromagnetic layer therefore requires a sufficient force against such a demagnetizing tendency, if the demand for high TMR perpendicular anisotropy junction structures in applications is to be satisfied. Accordingly, tremendous efforts have been put into pursuing better material performance to fulfill that purpose. A variety of PMA material systems has been investigated thoroughly and is very briefly reviewed here. The first group of materials was based on rare- earth elements, mainly Gd, Dy and Tb [44]–[46]. MTJs based on (Gd,Tb)FeCo/FeCo/Oxide/FeCo/(Gd,Tb)FeCo were reported to show rather high TMR of up to 74% [47]–[50]. However, putting aside the disadvantages of rare-earth-based materials, such as high commodity cost and cross chemical contamination, the high resistance-area product (RA value) has rendered them unfavorable candidates for MRAM application. On the other hand, classes of multilayered [Co/Pd] or [Co/Pt] have attracted tremendous interest because PMA could be achieved with relative ease and they offered a relatively low RA value. Unfortunately, their values of TMR were generally less than 10% at room temperature [51][52], so low that they were not the preferable choices for MTJ applications. Meanwhile, by inserting CoFeB between the [(CoFe)/Pd] multilayer and the MgO buffer on both sides of the multilayer, the TMR was drastically improved from a few percent to 91%; yet again, the RA value was too high to yield a practical solution [53]. In short, despite that PMA was comfortably attainable in these material systems, they did not satisfy the specific criteria desired for the most attractive application of MTJs – MRAM – particularly in the context of its continuingly confined pitch size, high thermal stability and low onset switching current. A breakthrough was achieved when Ikeda et al. [54] reported in 2010 on the perpendicular anisotropy of CoFeB/MgO interface which, interestingly, was developed from the in-plane structures that have been known for quite a long time. A systematic exploration of structures based on CoFeB/MgO and its reverse order MgO/CoFeB, as well as the symmetric MTJ structure of 23

4. ALL PERPENDICULAR MTJ STO

Ta/CoFeB/MgO/CoFeB/Ta, was conducted with different thicknesses of

CoFeB layers (tCoFeB) and annealing temperatures. The hysteresis loops showed a clear perpendicular easy axis at tCoFeB = 1.3 nm and in-plane anisotropy at 2.0 nm, both annealed at the same temperature of 300 °C. The anisotropy field extracted from the FMR measurements showed a change from negative sign (in-plane) to positive sign (perpendicular) as tCoFeB decreased. This evidently presented the possibility to tune the CoFeB thickness to attain the desired magnetic anisotropy. It was equally important that the decisive factor in tuning the easy axis alignment is located at the metal/oxide interface, leading to the conclusion that an additional noble metal buffer between ferromagnetic layers and the oxide spacer is in fact unnecessary as believed so far. Figure 4.1 shows the robust developmental history of MgO-based MTJs (Refs. [33][37][39][41][55]–[60]) in comparison with a rather modest development of alumina-based MTJs in terms of tunneling magnetoresistance ratio (Refs. [61]–[69]).

Figure 4.1: Tunnel magnetoresistance ratio (TMR) with trend lines

4.1. All perpendicular MTJs

A brief history of the MTJ applications that have attracted the most interest is presented as sources of motivation for one of the objectives this thesis is devoted to. Firstly, magnetic recording and storage of data is widely recognized as the most effective and capable method in the information technology sector; and one of its most important benchmark is storage density. The density, in turns, strongly depends on the size of the magnetic domain in which one bit of data is stored and on the size of the head needed to read precisely that bit. While the intensive research and development activities of materials for storage media have delivered a drastic increase in density, i.e. the great decrease in the storage domain size, this aspect was outpacing the development of reading heads to the extent that floppy disk technology has maintained its popularity for two decades, until the turn of the twenty-first century when it reached a capacity of 120 MB. The discovery of MTJs and their successful implementation into reading head technology and production in 2005 by Seagate eventually introduced 400 GB drives to the markets, marking a true revolution. Another important application of MTJs is MRAM, one of the major emerging memory technologies, as it combines the advantages of SRAM’s fast performance, DRAM’s high density, and flash memory’s nonvolatility. With the discovery of the STT effect, which allows switching to be controlled by an electric current alone, the challenges of high power consumption and unintended switching errors the available field-induced switching MRAM has been encountering can be overcome. The previous introduced perpendicular anisotropy CoFeB/MgO MTJs promise yet another significant step towards the realization of high density STT-MRAM technology. With such advantages, the introduction of perpendicular anisotropy CoFeB/MgO-based MTJs (p-MTJs) to the spin-torque oscillation performance and spin dynamics studies is therefore the inspiration of a sizeable part of the present work, whose results will be presented in the next section.

4. ALL PERPENDICULAR MTJ STO

4.2 Spin dynamics in all-perpendicular MTJ STOs

4.2.1 Device fabrication and dc characterization

Since the p-MTJ devices used for measurements in the present work were acquired through our collaboration with Prof. Hideo Ohno’s group at Tohoku University, Sendai, Japan, without any extra processing, a description of device materials, treatment, and fabrication is briefly summarized below; further details can be found in the relevant references [54][70].

An MTJ stack of Co20Fe60B20(0.9)/MgO(1)/Co20Fe60B20(1.6) was deposited on a sapphire wafer using the sputtering technique, buffered by a Ta(5)/Ru(5)/Ta(5) seed layer and protected by a Ta(5)/Ru(5) cap layer (numbers in parentheses are in nanometers). The thinner (bottom) CoFeB layer is defined as the fixed layer and the thicker (top) one as the free layer. These CoFeB thicknesses were selected on the basis of the PMA tuning in the CoFeB/MgO system, which showed that, for thicknesses below 1.3 nm, CoFeB retains a rather strong PMA owing to the dominant interfacial

Figure 4.2: Entire as-fabricated wafer; one die with numbers in nanometer as nanopillar diameter; and a device with top (TC) and bottom (BC) contacts

4.2. Spin dynamics in p-MTJ STOs

anisotropy mentioned in the Chapter 3. The deposited wafer was then annealed at 300 °C in a vacuum for 1 hour with a magnetic field of 1 T applied perpendicularly to the wafer plane in order to predefine the easy axis in both the top and bottom CoFeB layers. E-beam lithography and ion milling were combined to pattern the stack into round-shaped nanosized pillars on the wafer, and this was followed by the conventional metallization step to fabricate the Au/Cr electrical contact pads. Within the scope of this thesis, 100 nm diameter p-MTJ devices were investigated and presented. Images of the as-fabricated wafer and one device are illustrated in Figure 4.2. The measurement of magnetoresistance effects was carried out under the applied field scanning from -3000 to +3000 Oe (-0.3 to +0.3 T) at various field angles relative to the stack plane. In order to probe the device resistance, it was necessary for a small amount of dc current (about 2 μA) to be introduced, so as to prevent the STT effects from triggering. A combo of a dc current source (Keithley 6221) and a nanovoltmeter (Keithley 2182A) was used to send the current to and pick out the voltage from the device. This pure field-induced switching at the field angle of 90° (out-of-plane, or OOP, field) turned over the maximum TMR for all devices inspected, hence confirmed that the easy axis of both CoFeB layers is normal to the stack plane. Due to the unavoidable damages during nanosized device fabrication processes TMR varies between 60% and 120% for most of the 100 nm diameter devices. Despite that, the general behaviors of magnetodynamics in all devices being investigated are similar. The resistance-area product (RA value) is also derived from this measurement. The parallel state RA values are in the range from 4 to 8 Ω.μm2, relatively low and suitable for MTJ switching applications, thus promising a low onset current which in turn would be preferable for STO applications. Figure 4.3a shows a resistance- field (R–H) OOP major loop of a typical device. Another aspect was investigated by obtaining minor R–H loops within a reduced field range, typically from -1500 to +1500 Oe in our measurements. As illustrated in Figure 4.3b where an R–H OOP minor loop is plotted, there is a shift to the right of the zero-field line. In other words, an extra amount of field is required to compensate for the exchange coupling field that the fixed layer exerts on the free layer (referred to as the stray field, Hstray). Below the field angle of 20°, however, no such a magnetization reversal was identified, even though the resistance monotonically increases with the increase of the applied field in both polarities within the same scanning range, again, pointing to the strong PMA of CoFeB layers.

4. ALL PERPENDICULAR MTJ STO

Figure 4.3: Resistance vs. applied field ( R–H) OOP major (a) and minor (b) loops

4.2. Spin dynamics in p-MTJ STOs

Figure 4.4: Resistance (red circle) and converted relative angle (blue diamond) of 100 nm MTJ device measured in in-plane filed

With the interests of dynamic responses in the in-plane (0° field angle – IP) applied field, the device resistance was measured with a small amount of dc current to avoid the STT effect as much as possible under the IP field sweeping from 0 to 14 kOe. Figure 4.4, showing the field-dependent resistance, demonstrates that the device resistance increases monotonically with the applied field up to about 5 kOe, where it attains a maximum and then steadily decreases, even as the field continues to increase further, finally saturating at a value close to the beginning zero-field resistance value; this indicates that the parallel state of the magnetization in the two ferromagnetic layers is retained, even though, at this high-field regime, the magnetizations of both CoFeB layers are aligned no longer along their original perpendicular easy axis but close to the applied IP field direction. The relative angle, θrel – converted by Eq. (4.1) – is also plotted against the applied field. Qualitatively, it can be said that the MTJ resistance approaches its peak when θrel between the two magnetizations is maximum. This happens when the magnetization of the free layer almost lies in the wafer plane, due to the overcoming of the anisotropy field by the applied field. The decrease in the resistance observed afterwards reflects the fact that the fixed layer is also

4. ALL PERPENDICULAR MTJ STO

gradually forced towards the IP direction, which results in the decrease of

θrel.

4.2.2 Magnetodynamic modes in p-MTJ nanodevices

This subsection is to briefly introduce the possible magnetodynamic modes in 100 nm p-MTJ nanopillars under the applied IP field, supporting by a 3D micromagnetic simulation based on a cross-platform GPU-accelerated micromagnetic simulation program, MuMax3 [71]. The detailed description and discussion of experimental and simulated results can be found in Paper III. In order to perform the simulation, a device was characterized using dc magnetoresistance and ST-FMR measurements, described previously in Chapter 2. From the IP field-dependent resistance data, the relative angle,

θrel, between the magnetizations of the free and fixed layers is calculated using Eq. (4.1) and plotted as data points in Figure 4.5a. ST-FMR data were analyzed following the principle mentioned in Chapter 2 from which the resonant frequency, fres, and dc output, Vout, are mapped within the applied field range of 0 – 12 kOe as being seen in Figure 4.5b. Material parameters were finely tuned during the performance of the micromagnetic simulation until the best matching between simulated and experimental data was obtained. The simulated relative angle, θrel, is also plotted in Figure 4.5a as the red curve, showing an excellent agreement with the data obtained from the magnetoresistance measurement. From the simulation, the values of the magnetization angles in both the free and fixed

CoFeB layers are also extracted and plotted as θfree and θfixed, respectively, together. θfree shows a steep decrease with the increase of the applied field and approaches the IP direction at about 5 kOe; meanwhile, the magnetization of the fixed layer is gradually driven towards the wafer plane and aligned at about 20° for the field as high as 12 kOe. Simultaneously, a Fast Fourier Transform (FFT) of the time-varying transverse component of the averaged magnetization of each CoFeB layer in a modeled p-MTJ device [72], with the inclusion of the in-plane Oersted field generated by the rf bias [73], revealed different modes of excitation in each layer, as well as their power density, plotted against the applied field in Figure 4.5c.

4.2. Spin dynamics in p-MTJ STOs

The modeling results matches very well with the field-dependent behavior of the resonant frequencies extracted from ST-FMR measurement; thus shows that there exist several modes of magnetodynamics within both the free and fixed layers of the p-MTJ device. The oscillation is excited in the fixed layer as soon as the applied field passed 3.5 kOe, together with the appearance of an extra mode (Mode 2b) inside the free layer. Three strongest modes in the free layer (Mode 1, Mode 2 and Mode 3) and the fundamental excitation of the fixed layer (Mode 1 (fixed layer)) are shown in Figure 4.5d, with their respective simulated results. While the field-dependent tendency of every presented mode is similar between ST-FMR measurement and micromagnetic simulation, the best match is seen in the low field regime for the fundamental mode (the strongest – Mode 1) of the free layer and in the high field regime for Mode 1 of the fixed layer.

Figure 4.5: a) Experimental θrel and simulated θrel, θfree, θfixed; b) ST-FMR output voltage mapping; c) FFT spectral power mapping; and d) Experimental and simulated resonant frequency of different magnetodynamic modes in 100 nm p-MTJ nanopillar 31

4. ALL PERPENDICULAR MTJ STO

Figure 4.6: Amplitude (row a), phase (row b) and real part (row c) of magnetodynamic modes in p-MTJ nanopillar model at 2 kOe and 4 kOe

The spatial profiles of the excited modes are then constructed from the FFT for each ferromagnetic layer of the simulated p-MTJ nanopillar [74] and shown in Figure 4.6, representing two field regimes: low (2 kOe) and high (4 kOe). Due to the finite radius of simulated elements [75], the quantization of the in-plane wave vector results in dipolar confined spin wave modes presented in Figure 4.5c. At low field, there are only excitation modes of the free layer, of which the three strongest are presented. The fundamental Mode 1 is a radial mode with the resonance amplitude concentrating in the center of the pillar, while both Mode 2 and Mode 3 are azimuthal modes with strong coupling with the excitation induced by the Oersted field [76]. In the high field regime (> 3.5 kOe), the magnetizations of both CoFeB layers are driven gradually towards the in-plane direction. Mode 1 of the free layer is localized further to the edge of the pillar due to the symmetry breaking, so are Mode 2 and Mode 3. The appearance of Mode 2b in this field regime is due to the spin waves that are localized both horizontally and vertically in the modeling disk. Another radial mode, labeled Mode 1 (fixed layer), is also seen with its strong amplitude (frequency) and its symmetry is sustained thanks to its stronger perpendicular anisotropy field.

4.2.3 Auto-oscillation characteristics

In order to characterize the MTJ microwave performance, another setup whose layout is sketched in Figure 4.7, was used. The setup was equipped

4.2. Spin dynamics in p-MTJ STOs

Figure 4.7: MTJ device structure and microwave measurement setup with a high-field electromagnet (up to 14 kOe) and a spectrum analyzer (R&S FSU 20 Hz – 67 GHz). The device to be measured was placed under a ground-signal microwave probe connected to a bias-tee that provides both a direct drive current and the effective transmission of the generated STO signal to a broadband low-noise amplifier (LNA) prior to a spectral analyzer. A computer with a General Purpose Interface Bus (GPIB) card and built-in LabView programmable software allowed full control of measurement parameters and collection of spectrum data. Figure 4.8 presents the power color mapping of the current-sweep oscillation frequency at various applied fields. Clearly the low-field and high- field regimes whose boundary lies somewhere around 3.5 kOe, can be separated. In the low-field regime, the oscillation signal shows red-shifting behavior—that is, the oscillation peak frequency (fosc) decreases when the applied dc bias goes up from its onset value of about -0.02 mA (c.f. Paper IV) to -0.35 mA. Obviously, the activation of STT effect beyond the onset current is the cause that excites the oscillation of magnetic moments in the free layer. The number of spin-polarized electrons increases, expected as the consequence of the increasing dc current, results in the surging in the oscillation signal power, which can be evidently seen from the color gradient in Figure 4.8.

In the regime of high applied fields, however, fosc experiences a blue- shifting trend, in which the frequency goes up with the increase in dc current. The contrasting tendencies of the current-dependent oscillation

4. ALL PERPENDICULAR MTJ STO

Figure 4.8: Power color mapping of spectral signal generated by MTJ STOs in the applied field range from low (1 kOe) to high (7 kOe) regimes 34

4.2. Spin dynamics in p-MTJ STOs

frequency in these two regimes of the magnetic field likely originate from the response of the fixed layer magnetization to an external field strong enough to break down its perpendicular anisotropy, and results in a tilt precession axis from its original perpendicular direction. In other words, as the precession axis tilts further towards the film plane, the precession angle becomes smaller; the oscillation thus goes blue-shifting. In this scenario of high applied field, the fixed layer is not absolutely “fixed” as presumed, and the mechanism of STT-induced oscillation becomes complex and passes beyond the scope of the present work. The spectra presented above are fitted by using the popular Lorentzian peak-fitting formula (see full text in Paper IV), with the peak frequency as one of the output parameters characterizing the microwave oscillation generated by the p-MTJ STO device being measured. Figure 4.9, in which the extracted peak frequency is plotted against the biased current at various applied magnetic fields, again illustrates the current-dependent behaviors of the STO operations observed in Figure 4.8: the red-shifting is dictated by the trend in the field region below 3.5 kOe and the blue-shifting is imposed by the trend in the field regime of 4 kOe and beyond. More interestingly from the application point of view, the linear frequency-bias relationship over almost the entire ranges of applied current and external magnetic field offers a huge advantage in terms of current (voltage) fine tuning, which is critical in

Figure 4.9: Current-dependence of generated oscillation frequency, measured in in-plane field 35

4. ALL PERPENDICULAR MTJ STO

realizing the potential of p-MTJ STO operations. An example of the extremely high current tunability of 100 nm p-MTJ STO devices can be found in Paper IV. It is also shown there that the onset current—a lower threshold from which the STT effect can excite the oscillation in a device, extrapolated from the integrated power-bias relation—can be as low as about -20 μA, or 2.77 × 105 A/cm2 in terms of current density, which is about one order of magnitude lower than that reported in in-plane MTJ devices. On the other hand, the spectra obtained from measurements in field-sweep mode are also fitted to extract the characteristic peak frequency and are plotted against the applied field at different bias currents in Figure 4.10. At currents below -0.35 mA, the frequency shifts in an identical manner: first going red- shift to a minimum then turning towards linear blue-shift as the applied field increases. However, for the increases in current from -0.05 to -0.30 mA, the field where the minimum frequency reaches drifts from 3.5 kOe downwards to 2.5 kOe. When the current is higher than -0.30 mA, the oscillation frequency shows a slight change within the field region below 2 kOe and then increases linearly afterwards, similar to the blue-shift behavior of frequency at smaller currents. It is also noticeable that the manners of the low-field red-shifting and high-field blue-shifting of the oscillation frequency in the device in the field-sweep mode are completely consistent with the performance in the current-sweep mode previously described.

Figure 4.10: Shift tendency of oscillation frequency as function of applied in- plane field

4.2. Spin dynamics in p-MTJ STOs

4.2.4 Bias-dependent spin torque components

ST-FMR measurements were carried out both below the onset and at high dc currents with the intention of illustrating the STT effect on the FMR spectra, as described in Chapter 2. The total output voltage expressed in the form of its asymmetric and symmetric components was introduced in Eq. (2.3). A rather low rf power (-22 dBm) was introduced to the device during the ST- FMR measurements in order to minimize the intervention of possible extra STT effect. From the fitting, are extracted the asymmetric and symmetric components (Vasym and Vsym), as well as the resonance frequency fres and linewidth ∆f of the FMR spectra. After subtracting from the offset, two typical sets of the overall spectra (black dots), and their asymmetric (red line) and symmetric (blue line) components, are plotted together in Figure 4.11, corresponding to the 2 kOe and 4 kOe applied fields, representing the low-field and high-field regimes described in the discussion of spectrum analyzer results; the biased current goes from well below onset (-2 μA) to a rather high value of -0.25 mA. Data at a current of -2 μA are treated approximately as pure FMR signals. In the low-field region, the tendency of both the asymmetric and symmetric voltages in relation to the applied current is typically distinguishable. Below the onset value, it is clear that only the asymmetric component exists and overlaps the total output (Fig. 4.11a). However, beyond the onset, where the STT effect exerts its influence on the dynamics of STO performance, the contribution of the symmetric component to the output voltage becomes more and more dominant, to the point that, at -250 μA, it completely overpowers the asymmetric component (see Fig. 4.11c–d). Nevertheless, in the high-field regime, the developments of the asymmetric and symmetric voltages are not in the same manners, even though the tendency towards the increasing influence of STT effect over that of FLT effect with the increase in biased current is still indisputable (Fig. 4.11e–h). Noticeably, the field-dependent FMR frequency behavior shows a great similarity of the oscillation frequency measured by spectrum analyzer, shown in the previous section of this chapter, regardless of how much dc current is applied, as shown in Figure 4.12. The frequency curves at each applied current show the red-shifting characteristics until a turning point where the minimum frequency is reached and then there is a rise upwards in the blue-shifting manner. However, the field corresponding to the minimum point, varying from 2.75 kOe to 3.25 kOe, is current-dependent: the larger

4. ALL PERPENDICULAR MTJ STO

Figure 4.11: Typical ST-FMR spectra with fitting symmetric and asymmetric components at different dc biases in low-field (left column) and high-field regimes 38

4.2. Spin dynamics in p-MTJ STOs

the dc current, the lower the value it is reached at. The high-field branches of the frequency curves, meanwhile, show a very slight dependence on the dc current, even though their wholesale linear field dependence is obvious and in contrast to the low-field branches. Again, similarly to the spectrum analyzer data from microwave measurements, the spin dynamics in this high-field regime are heavily dominated by field-induced effects. On the other hand, the field-dependent linewidth behavior extracted from the ST-FMR spectra looks more interesting. While there is no distinctive difference within the range of applied dc current, as being seen in Figure 4.13, the field dependence of the linewidth shows a rise from about 250 MHz to 400 MHz before decreasing down again to 250 MHz as the applied magnetic field ranges from 1 to about 3.5 kOe. For higher fields up to 5 kOe, however, the linewidth grows dramatically to 800 MHz. Due to the weak signal generated, the data obtained from fittings of spectra at the field higher than 5 kOe are deemed irrelevant; the apparent maxima between 4.5 and 5 kOe are therefore open to further analysis based on better quality of measurement setup in the future.

Figure 4.12: IP field-dependent resonance frequency of p-MTJ STOs at different strengths of drive dc current

4. ALL PERPENDICULAR MTJ STO

Figure 4.13: Field-dependent linewidth of FMR spectra

In order to analyze the contribution of the electric field across the junction—particularly the torques transferred by the electron spins—the field-dependent behaviors of both Vasym and Vsym, the magnitudes of the two components of FMR output, are plotted in Figure 4.14 at different applied dc currents. At each applied dc current, the magnitude of the asymmetric component monotonically increases as the applied field increases to about 3.5 kOe, where a maximum change of 45 μV is observed. A further increasing of the field, however, shows the reverse trend of monotonic decrease in the asymmetric magnitude. Remarkably, the enormous change in dc current— from -2 to -250 μA—does not seem to affect either the minimum point in

Figure 4.14(top) or the value of Vasym in the high-field regime (> 3.5 kOe). In the low-field regime, while there is little change in Vasym between -2 and -50 μA, it does decrease significantly at currents of -150 and -250 μA, reflecting that the STT effect actually represses the contribution from the FLT effect.

On the other hand, the changes in Vsym magnitude are plotted in Figure 4.14(bottom). Overall, this shows an increase in magnitude of Vsym up to a maximum as the field increases, as observed similarly in Figure 4.10 for oscillation frequencies and Figure 4.12 for resonant frequencies. The increase in Vsym also becomes clear when the current increases from-50 μA 40

4.2. Spin dynamics in p-MTJ STOs

Figure 4.14: Field dependence of extracted values of asymmetric (top) and symmetric (bottom) output voltage components from ST-FMR measurements

4. ALL PERPENDICULAR MTJ STO

to -250 μA, with the maximum value correspondingly changing from about

45 μV to 250 μV, an order of magnitude stronger than that of Vasym, in the low region of applied fields. By contrast, particularly for the field beyond 4 kOe, Vsym is suppressed to just 10 μV and becomes independent of the applied current—illustrating the very weak STT effect.

The contrast between Vasym and Vsym in terms of both the field-dependent and current-dependent behaviors becomes obvious when drawn together in

Figure 4.15, particularly for Ibias = -2 μA (pure FMR) and Ibias = -250 μA (strong STT effect).

Figure 4.15: Comparison of field-output component relations at near zero- bias (square symbols) and high bias (circle symbols) Since it has been well established that the output of a homodyne FMR (which is ST-FMR in principles) measurement, represented by the asymmetric component Vasym, can be described as a function of the cant angle of the magnetization in the free layer, θfree, e.g. in Ref. [77], to characterize the contribution of field-like torque effects (the first term) and electric-field induced effects (the second term):

4.2. Spin dynamics in p-MTJ STOs

휕휏⊥ 2 −푉푎푠푦푚 ∝ 1 푐표푠 휃푟푒푙 (4.4) ∆푓 휕푉푏푖푎푠 휕퐻퐾⊥ − 푐표푠 휃푟푒푙 푠푖푛 휃푓푟푒푒 푐표푠 휃푓푟푒푒 ∆푓 휕푉푏푖푎푠 where τ⊥ , HK⊥ and ∆f are, respectively, the field-like torque (FLT) (to be distinguished from the term τ// of spin-transfer torque (STT)), the effective perpendicular anisotropy field and the FMR spectrum linewidth. Consistently, the elevation-angle dependence of the signal intensity in FMR measurement reported in Ref. [78] also presumes a similar relation:

1 (4.5) 휕퐻퐾⊥ 2 푉푎푠푦푚 ∝ − 푠푖푛 휃푓푟푒푒 푐표푠 휃푓푟푒푒 ∆푓 휕푉푏푖푎푠 in the case of the field applied perpendicularly to the film plane—that is, along the magnetic easy axis and the fixed layer’s magnetization was also kept at 90° to the film plane.

Assuming that the fixed layer is absolutely “fixed” so that θrel ≈ θfree; the second term of Eq. (4.4) is then similar to Eq. (4.5). It is also worth noticing that the symbols and angle definition were transformed here to be consistent with those used in this thesis. The reported measurements in both of the above groups confirmed the dominant effect induced by the electric field by pointing out that the maximum of Vasym occurred at the angle θfree = 35°, where the term (cosθ)2sinθ is maximized, instead of at 0° which corresponds to the maximum of the term (cosθ)2, as shown in Figure 3c in Ref. [77] and Figure 3b in Ref. [78]. The reported consistency of the characteristic angle at which the electric- field effect displayed its maximal contribution to the FMR signal, fundamentally promotes the idea of using a similar recipe to investigate the electric-field effect on the STT-induced dynamics of p-MTJ STOs. Determining such a relation would have potential in using gate voltage to manipulate the STO performance.

4. ALL PERPENDICULAR MTJ STO

Accordingly, both the asymmetric and symmetric components of the output, Vasym and Vsym, from the ST–FMR spectrum are split in the terminology of Eq. (4.4): field-like-torque effects (∂τ⊥/∂V), spin-transfer torque effects (∂τ///∂V), and electric field-induced effects (∂H K ⊥ /∂V):

휕휏⊥ 2 푉푎푠푦푚∆푓 ∝ − 푐표푠 휃푟푒푙 (4.6) + 휕푉 휕퐻퐾⊥ 2 � � 푠푖푛 휃푓푟푒푒 푐표푠 휃푓푟푒푒 휕푉 푎푠푦푚 and

휕휏 2 푉푠푦푚∆푓 ∝ − 푐표푠 휃푟푒푙 (4.7) + 휕푉 휕퐻퐾⊥ 2 � � 푠푖푛 휃푓푟푒푒 푐표푠 휃푓푟푒푒 휕푉 푠푦푚

Equipped with Eqs. (4.6) and (4.7), the characteristic factors ∂τ///∂V,

∂τ⊥/∂V, (∂HK⊥/∂V)sym, and (∂HK⊥/∂V)asym are extracted from the fitting, taking into account the fact that the relative angle θrel is valid to the point where the magnetization of the fixed layer only slightly tilts away from its easy axis—in other words, within the low regime of the applied magnetic fields. θrel is approximately converted from the IP field R–H curve in Figure 4.4. As one example, Figure 4.16 – where the asymmetric and symmetric components of the output mixed voltage obtained at the dc current of -50 μA are plotted against the relative angle illustrates together with the fitting curves, it shows that the fits are excellent at the fields below 3 kOe. Indeed, the minima in both component curves indicate the respective angle of about 30°, quite close to the theoretical value of 35° with the deviation presumably originating from the accuracy of the simplified resistance to angle conversion. This clearly points to the dominance of the electric-field effects.

4.2. Spin dynamics in p-MTJ STOs

Figure 4.16: Extracted data and fitting with Eqs. 4.6–7 of V∆f vs. relative angle curves for both symmetric (red-circle) and asymmetric (blue-square) ST-FMR output components

Table 4.1 summarizes the extracted factors for several values of dc currents. While there is a little fluctuation, the representative of the FLT effect, ∂τ⊥/∂V, is generally independent of the applied dc current. The electric field effect, however, is experienced in a rather different way. By breaking this effect into two terms, each is assigned to asymmetric and symmetric components: the term (∂HK⊥/∂V)asym is then more or less constant with the applied current, but the other term, (∂HK⊥/∂V)sym, is dramatically enhanced by an order of magnitude. Similarly, there is a drastic increase in the magnitude of the term caused by STT effect, ∂τ///∂V. In short, the data in the table confirm that the electric-field effect continues to make the dominant contribution. Further increases in the dc current tip the balance toward the contribution of STT-induced effect, but still not exceed that from the electric field. Further details and discussion can be found in Paper V attached to this thesis.

4. ALL PERPENDICULAR MTJ STO

Table 4.1: Bias dependence of parallel torque, perpendicular torque and perpendicular anisotropy obtained from ST-FMR measurement and fitting in p-MTJ STOs

Bias ∂τ /∂V (∂H /∂V) ∂τ /∂V (∂H /∂V) (μA) // K⊥ sym ⊥ K⊥ asym

-2 -5 -5 1.44×10 -3.96×10

-5 -5 -5 -5 -50 0.83×10 -4.62×10 1.06×10 -3.92×10

-5 -5 -5 -5 -150 -29.64×10 -13.51×10 0.82×10 -3.13×10

-5 -5 -5 -5 -250 -44.80×10 -62.47×10 2.30×10 -1.05×10

Chapter 5

Magnetic droplet solitons in all- perpendicular pseudospin-valve STOs

5.1 Application of PMA multilayers in spin valve devices

In-plane NiFe thin film possesses low damping and thus offers a reliable choice of free layer material for conventional spin-valve stacks. Consequently, spintronic devices, such as STOs have employed it extensively [79]–[94]. However, there remain obstacles to the full realization of its promising potentials, including the requirements of high onset current and high external applied field. Naturally, the precession axis of an in-plane magnetic anisotropy (IMA) free layer tends to stay within the film plane to minimize its free energy, resulting in a small angle mode of precession. While the small precession angle may offer the advantage of high microwave frequency generation (up to 65 GHz, according to Ref. [80]), the output power is insufficiently low, which technically undermines the prospective of STOs. A study of comparability in the switching modes of three different spin- valve structures—IMA polarizer/IMA free layer, tilted polarizer/PMA free layer, and PMA polarizer/PMA free layer—by point-contact measurements resulted in a current-field (I–H) switching diagram [95]. The STT efficiency extracted from the diagrams was highest in the tilted polarizer/PMA free layer configuration and lowest in the all-in-plane one. Micromagnetic simulations for free layer magnetization dynamics in these configurations

5. DROPLET SOLITONS IN P-PSV STO

also supported the experimental data by illustrating that the switching current density and switching time were lowest and shortest, respectively, in the tilted polarizer structure; by contrast, the in-plane structure is much more coercive (with switching times 3–6 times slower, as well as being one order of magnitude higher in terms of the current density required). On the other hand, the resistance of nanopillar spin-valves with PMA polarizer and PMA free layer was also investigated under an external sweeping field at different currents [96], from which an I–H resistance phase diagram was built with clear switching boundaries. In order to match the experimental diagram, several macrospin models were tried and the non-uniaxial model presented the best fit, taking into account the hypothesis of symmetry breaking. It was concluded that, the misalignment of the external field, the slightly canting of the easy axis, or the strong second order anisotropy, are most likely responsible for this symmetry breaking in real spin-valve structures. With such developments of understanding about PMA materials, including the details in Chapter 3, gives rise to an advanced alternative of materials for PSV STO in the context of larger precession angles—that is, higher output power and proven high thermal stability on account of the well-defined out-of-plane easy axis.

5.2 Conceptual magnetic droplet solitons

It has been theoretically predicted for a long time that the interaction of excitations in magnetic media—magnons—can lead to the formation of magnon-bound states, so-called magnon drops or magnetic droplet solitons, provided that the local density of magnons is sufficiently large in a ferromagnetic thin film [97]. This formation of magnon drops requires a balance between the mechanisms of relaxation (dispersion) and nucleation (nonlinearity). While this theory predicted a continuous family of droplet solitons under a given set of external influences via relaxation–nucleation cycles, stable dissipative droplet solitons are much fewer in population, as their steady-state dynamics requires an additional balance between gain and loss of magnons. In a nanoscale ferromagnet with perpendicular magnetic anisotropy, where the magnon loss due to damping is significant, the gain, however, can be generated by the injection of anti-damping through STT effects, which makes this an attractive medium for studying droplet behaviors. Accordingly, further conditions were taken into account in order to approach the real system of point-contact spin valves included the

5.2. Magnetic droplet solitons

addition of the current-induced Oersted field, the canted applied field, and a possible tilt angle of PMA fixed layer [98][99]. Consequently, the derivation of the LLGS equation (Eq. (1.3)) led to several significant conclusions regarding droplet behaviors: the droplet forms locally with reversed magnetization under the contact, and is encapsulated by a large-angle precession boundary; this precession frequency, so-called droplet frequency, must be much lower than the FMR frequency of the extended film and have a weak current-dependent tendency at high drive currents; propagation, droplet drift instability and hysteresis could also be expected [98][100]. Furthermore, droplet propagation in a realistic ferromagnetic nanoobject was demonstrated by the full micromagnetic simulation for the ability of using only an external field to sustain, accelerate, and control [99]. Following these interesting theoretical predictions and simulations, droplets were observed experimentally, for the first time, in a nanocontact PSV STO composed of PMA free layer and IMA fixed layer—that is, an orthogonal magnetic configuration [101][102]. The field-sweep measurements of the generated STO frequency showed the conventional blue-shifting characteristics of FMR-like auto-oscillation until a critical value of field (which depends on the applied current); the frequency then dramatically dropped by 10 GHz and the generated power jumped by a factor of approximately 40. The magnetoresistance ratio also showed a jump and a change in slope sign when the field reached that critical value, thus gave the evidence of a region whose magnetization was reversed. These experimental data pointed to the existence of a magnetic droplet, which was further verified by micromagnetic simulation which illustrated several behaviors— such as droplet drifting, spinning, and breathing—through the manipulation of external conditions, such as the applied field and current. Further experimental details of droplet nucleation, collapse/annihilation, and drift instabilities were presented with well-defined states under different conditions of the applied field and drive current [103]–[108]. At a perpendicular field sufficient to saturate the original IMA polarizer of an orthogonal STO, the field-dependent trend of magnetoresistance ratio (MR) went back to the monotonic red-shift, prior to nucleation of the observed droplet, indicating that the reversed core had switched to fully align with its surroundings. Using resistance step-like behavior as an indication of droplet nucleation, an investigation over a wide range of perpendicular fields applied to orthogonal STOs with an IMA polarizer disclosed the different tendencies of nucleation current against the applied field: inversely proportional in the low-field regime and linear in the high-field regime. A time-dependent 49

5. DROPLET SOLITONS IN P-PSV STO

resistance measurement by real-time oscilloscope sampling also illustrated drift instability around the droplet collapse field. For the purpose of signal- to-noise ratio improvement, low temperature (4.2 K) measurements of resistance were conducted on the NiFe fixed layer-[Co/Ni]×6 free layer nanocontact STO to detect both droplet nucleation and annihilation from the MR vs. current hysteresis loops under a constant applied field [105]. The change in resistance after the nucleation confirmed the fully reversed magnetization nature of the stabilized droplet previously predicted. The annihilation current was generally lower than the nucleation current, which might suggest the contribution of the Oersted field to balancing between damping loss and STT-induced gain. With the introduction of asymmetric parameters into the supporting micromagnetic simulation, the appearance of low frequency (< 1 GHz) noise was believed to be due to the instability of droplet boundaries [106]. Interestingly, the droplet frequency showed a clear blue-shift with the increasing applied current. However, the nucleation current was extremely high in this system, and therefore curtailed much of its potential in technical applications due to the high probability of thermal- induced events. Alternatively, comprehensive maps of the magnetic droplet nucleation I–H boundary were constructed for various nanocontact diameters of STO devices fabricated from Co–[Co/Ni] multilayer orthogonal PSVs and showed nucleation currents as low as -5 mA—much lower than the reported value of -25 mA in NiFe–[Co/Ni] stacks [107]. The remarkably different I–H nucleation conditions, the inverse proportionality at low fields and the linearity at high fields, smoothly met each other at a well-defined minimum within an intermediate field region. Better still, a direct and accurate determination of the STT asymmetry could be derived from the relations of current and field with the material parameters at the minima of the nucleation boundary, thus allowing the optimization of material and device engineering for applications. Unfortunately, high fields required to nucleate droplets in these IMA polarizer/PMA free layer PSVs limited further studies of other predicted properties of magnetic droplet solitons. Consequently, the perpendicular component of the PSV polarizer is expected to reduce the applied field required for stable droplets, as well as lower the nucleation current; this inspires the study of magnetic droplet solitons and their dynamics in all-perpendicular PSV nanocontact (NC) devices.

A stack of [Co(0.5)/Pd(1)]×5-Co(0.5)/Cu(6)/[Co(0.32)/Ni(0.97)]×4Co(0.3) was deposited on a Si/SiO2 wafer with the seed and capped layers as Ta(4)/Cu(11)/Ta(4) and Cu(2)/Pd(2), respectively. The numbers in 50

5.2. Magnetic droplet solitons

parentheses indicate the thickness in nanometers. The deposited wafer was then processed into 8 μm × 16 μm rectangular mesa devices with the NCs located at the centers, see Figure 5.1 for an example, through the fabrication process described in Chapter 2.

Figure 5.1: Mesa device of all-perpendicular PSV stack with contacts. The double green arrow implies that the external field H can be applied in either polarities

An electromagnet was placed on a motorized rotational stage. A KEPCO power supply controlled the field strength and polarity via its current. The device to be measured was positioned within the magnet’s pole gap, thus permitting field angular measurements without requiring disconnection, which is essential if the tiny resistance change is to be detected. A ground– signal–ground microwave probe was used to connect the device’s electrodes to a setup consisting of a combo of dc current source (Keithley 6221), a nanovoltmeter (Keithley 2182A) to obtain the resistance, and a spectrum analyzer (0—26.5 GHz R & S FSQ26) to record the generated microwave spectrum, with a bias-tee to separate dc and microwave components so that a single probe was sufficient for simultaneous measurements. The spectrum signal-to-noise ratio was enhanced using a broadband (0—40 GHz) LNA before recording. The measured results of a 60-nm diameter STO are

5. DROPLET SOLITONS IN P-PSV STO

presented here as an example. The results obtained from a 70-nm device are discussed in details in the appended Paper VI.

5.3 Results and discussion

The results of field-sweep magnetoresistance (MR) effect measurements at various currents are shown in Figure 5.2 and clearly illustrate the current- induced change of the switching behaviors. For each current value investigated, the initiation of the Co/Pd fixed layer switching occurs at constant field: +2000 Oe in the forward-sweeping direction and -2000 Oe in the backward direction; however, the switching field of the Co/Ni free layer changes hyperbolically with the current past -3 mA, as indicated by the small arrows in Figure 5.2. The minus sign of the dc current represents the electrons flowing from the free layer to the fixed layer of the PSV stack. Up to -3 mA, the conventional field-induced switching is maintained, without any change in the switching field of the Co/Ni layer (about +1200 Oe). The MR ratio of about 1% is also retained. At -4 mA, however, a step appears during

Figure 5.2: Field-sweep magnetoresistance ratio at different applied currents (an offset of 1.25 has been added between the curves for visibility) 52

5.3. Results

the switching of the Co/Ni free layer, indicating the existence of a nanomagnetic object—such as a magnetic droplet—underneath the NC, where the density of current penetrating from the fixed layer to the free layer is sufficiently high to switch the free layer magnetization partially through STT effects. Further increases in the applied field in both forward and backward directions leads to the annihilation of the nanomagnetic object— that is, full switching of the free layer (the antiparallel (AP) state), before the fixed layer also switches and the MR goes back to zero. A similar mechanism is observed for currents up to -6 mA. For currents of -7 mA and higher, it is interesting that a second transition step was recorded during the switching of the Co/Pd fixed layer. More remarkably, the nucleation field of the first object and the annihilation field of the second object are coincident, suggesting that the two objects are in fact of a similar nature and are driven by current-induced STT effects. The appearance of a step-like transition during switching is one of the key indicators of the magnetic droplet nucleation and annihilation; thus further evidences of dynamics are worth searching for to confirm the observations. The MR results and the low-frequency microwave signal for the out-of- o plane applied field (θH = 90 ) are plotted together in Figure 5.3 for two values of dc current: -6 and -7 mA. At -6 mA and with backward field sweeping (that is, from positive to negative), the parallel state is retained until +1000 Oe, when the Co/Ni free layer begins switching. However, it does not switch directly to the AP state, as happens at -3 mA and below. Instead, an intermediate state, indicated by a step-like transition in the MR curve due to a partial reversal of magnetization directly beneath the NC, is well accompanied by the appearance of a low-frequency noise band of less than 1 GHz. This and the previous observation of the MR current-dependent behavior are the two key pieces of evidence for droplet nucleation, as previously predicted and reported [101][105]–[107]. Nevertheless, the collinear configuration of magnetization of the two layers and of the applied field results in the nonappearance of both the droplet and the FMR-like frequencies within the spectrum analyzer’s operational range. A flat step around zero-field without any low-frequency noise reflects the existence of a different, nondynamic nanomagnetic object, which could be a subject of further research. Further increasing in the applied field leads to the full switching of the entire free layer to the AP state. When the field reaches - 2000 Oe—sufficient to switch the entire fixed layer—the MR value drops sharply to zero, indicating the parallel state configuration. The observation is

5. DROPLET SOLITONS IN P-PSV STO

similar when the field-sweeping direction is opposite, except for the sign of switching fields, as illustrated in the left column of Figure 5.3. With a slight increasing in the current to -7 mA, further interesting results were observed. Although no big difference is seen before the AP configuration is attained—except that the field strength corresponding to the nucleation of droplet (Co/Ni switching field) changes to +2000 Oe—the MR curve shows another step-like transition when the fixed layer begins switching to -2000 Oe. The accompanying low-frequency noise shows that a second droplet nucleation has occurred as a direct consequence of the strong current-induced spin torque. Once again, the right column of Figure 5.3 shows that the overall behaviors are identical for both directions of field- sweeping. A threshold current could be suggested between -6 and -7 mA in this 60- nm device, as well as between -3.6 and -4.6 in the 70-nm device described in Paper V, in order to distinguish the single and double nucleation of the magnetic droplets in these all-perpendicular PSV STNOs. A magnetoresistance phase diagram was constructed by normalizing the measured resistance change to the maximum change; it is thus equal to 0 in the parallel (P) state and 1 in the AP state, for the current range of -1 to -10

Figure 5.3: Field dependence of magnetoresistance (white dots) and low- frequency behaviors of 60-nm all-perpendicular STO at -6 mA (left) and -7 mA (right) for both backward (top) and forward (bottom) directions of field sweeping 54

5.3. Results

Figure 5.4: Magnetoresistance mapping against current and field coordination illustrates the states existing in an all-perpendicular PSV STO for both backward (left) and forward (right) directions of field sweeping mA and the field scanned from -0.5 to +0.5 T (-5000 to +5000 Oe) in the forward and backward directions. As Figure 5.4 illustrates, one is essentially the mirror image of the other, similar to the MR curves described above. At low applied currents (≤ -3 mA), the boundaries of the AP state are almost independent of the current, revealing the dominant field-induced switching mechanism within this range of current. A small abrupt discrepancy at about -3 mA, above which the AP state is retained over a wider field range, indicates the activation of STT effects. Meanwhile, the “droplet” phase, defined as intermediate values of 0 < MR < 1, makes almost linear borders with the P state, except for a slight deviation at a current somewhere between -6 and -7 mA—not a coincidence with the difference observed in Figure 5.2. The linearity is of importance in that the droplet nucleation is under full control by current–field manipulation. It should also be noted that the “droplet”–P state boundary slopes are of similar magnitude on both sides of the region containing antiparallel phase. Further details and discussions can be found in the appended Paper VI.

Chapter 6

Tilted polarizer spin valve STOs

6.1 Development of tilted magnetic anisotropy materials

STOs were introduced with a great potential to replace the rather bulky conventional oscillators used to generate broadband microwave signals in telecommunications, thanks largely to their compact, CMOS-compatible designs. The requirements of the external magnetic field and the large driving current, however, significantly undermine the competiveness of STOs in practical applications. Accordingly, tremendous efforts have been spent on pursuing solutions to overcome those obstacles. Amongst them, new material designs have become increasingly attractive, particularly in the light of the proven advantages of perpendicular magnetic anisotropy (PMA) thin films, presented in previous chapters. Tilted anisotropy materials have been expected to inherit the advantages of low onset currents, low- to zero- applied field, high thermal stability and scalability from PMA materials in spintronic applications such as STT-MRAM and STOs. Moreover, this family of materials also allows the tunability of STO generated frequency by the means of tuning the tilt angle and damping constant [109]–[112] (see full text in Paper II). Apart from the structures composed of all PMA ferromagnetic layers mentioned in Chapters 4 and 5, the macrospin and micromagnetic simulations of microwave generation in STOs whose fixed layer magnetization tilts at an angle, delivered spectacular results [113]–[115]. Even without the presence of an external applied field, the in-plane

6. TILTED POLARIZER-PSV STO

anisotropy free layer could be driven to precession and generate ultra-high frequency microwaves of up to 29 GHz within a decent current-tunable range. The calculated MR also pointed to a high yield of output power. Several candidates for such a tilted polarizer were suggested based on the availability of strong magnetocrystalline anisotropy materials: D022–MnGa (36° tilt angle) [116], L10–FePt(111) (36°), and L10–FePt(101) (45°) [117][118]. Unfortunately, these materials suffered from two major setbacks: unfavorable fabrication conditions (high annealing temperature and selective substrate) and discrete tilt angles. Meanwhile, the competition between a PMA (bottom) layer and an IMA (top) layer in the same stack, through sufficient exchange coupling strength, was believed to result in the inclination of the total magnetic moment of the IMA layer. This type of systems was first realized in PMA [Co/Pd] multilayer—NiFe structures [119], where a gradual transition is seen from dominant PMA in 3 nm-thick NiFe to comparable PMA and IMA at tNiFe = 10 nm. Moreover, micromagnetic calculations for tNiFe varying from 4 to 8 nm revealed quantitatively that the equilibrium angle of the NiFe layer

Figure 6.1: Normalized remanence and calculated tilt angle as functions of the stack thickness (left – courtesy of [119]); and magnetization angles of NiFe and Co/Pd layers extracted from two-spin model’s fitting (right – courtesy of [120]) 58

6.1. Tilted magnetic anisotrpy

magnetization changes from 0° (i.e. PMA) to 60°, limited only by the multi- domain formation in the thicker NiFe (Figure 6.1–left). Obviously, the tunability of the tilt angle through the NiFe layer thickness gives an extra important degree of freedom to maneuver the magnetic anisotropy of such systems and, consequently, becomes a significant candidate for the polarizer in PSV STOs. Alternatively, exchange coupling—the main force for tuning the magnetization in the NiFe layer in the above structures, was manipulated further by inserting a nonmagnetic exchange breaker between PMA and IMA, such as in the [Co/Pd] multilayer-Pt-NiFe structures [120]. By altering Pt thickness between 1.3 and 3.3 nm, the calculated NiFe magnetization canting angle could be tuned from 35° to 75°. The calculations came from the magnetization perpendicular component measured by the sophisticated vector vibrating sample magnetometry (V-VSM) technique. This study also showed a slight tilt angle of the underneath PMA [Co/Pd] multilayer, evidently illustrated the dominant role of Ruderman–Kittel–Kasuya– Yoshida (RKKY)-type exchange coupling (Figure 6.1–right). The micromagnetic calculations are thus still in need for quantitatively determining the tilt angle—an important parameter in defining the performance of STOs [121]. In an effort to measure the tilt angle experimentally, a deep-resolved study of spin orientation by means of the x-ray magnetic circular dichroism (XMCD) technique was carried out on such hybrid exchange springs series— for example, [Co/Pd]/Co/Ni/[Co/Pd]–CoFeB [122] (see full text in Paper I).

It showed that, while the top layer with sufficient thickness (tCoFeB > 1 nm) had an apparent in-plane anisotropy, the underlying structure was definitely PMA, regardless of the position of the inserted Co/Ni bilayer. The coupling between these two different magnetic anisotropies therefore led to the inclination of the CoFeB layer’s magnetization, similar to what was observed in [Co/Pd]-NiFe systems. Based on the detection of x-ray absorption of Fe, representing CoFeB, and Ni, the indicator of the lower PMA stack, the XMCD measurement revealed the tilt angle of the Fe and Ni magnetic moments at various Ni positions and different CoFeB thicknesses. While the Ni magnetic moments stayed firmly out of plane in a deep position for any thickness of CoFeB, the angle of the Fe moments changed dramatically from 8° to 73°, largely on account of the enhancement of the in-plane anisotropy of CoFeB with the increase in tCoFeB. In the sample with the thickest CoFeB layer (1.75 nm), however, the Ni angle was inversely proportional to its position—an expected response to the exchange coupling exerted by CoFeB—while the Fe angle changed in opposite way from 43° to 73°. The micromagnetic

6. TILTED POLARIZER-PSV STO

simulation demonstrated the uniform tilt angle of CoFeB magnetization, which varied from 0° (PMA) for tCoFeB < 1 nm to 69° at tCoFeB = 1.75 nm. On one hand, these results illustrated similar tilt characteristics observed in [Co/Pd]–NiFe systems. On the other hand, the fact that the tilt angle calculated by micromagnetic simulation differs from the measured tilt angle

Figure 6.2: (top) Measured and fitted tilt angles for Fe and Ni against CoFeB thickness and Ni depth, and (bottom) micromagnetic calculation of local magnetization angle for different Co/Ni locations and CoFeB thickness (courtesy of Ref. [122] (Paper I)) 60

6.1. Tilted magnetic anisotrpy

of Fe by only a small margin validated the simulation model for the purpose of determining the tilt angle in the exchange coupled hybrid structure. These observations are also illustrated in Figure 6.2. Such profound developments of tilted anisotropy materials within the group inspired the studies of spin dynamics in devices based on those materials. In this thesis, [Co(0.5)/Pd(1)]×5/Co(3) was used as tilted polarizer for PSV structures. The in-plane anisotropy thicker Co(3) layer in the place of the PMA thinner Co(0.5) is the only material modification from all- perpendicular PSV described previously. Nanocontact (NC) devices of this structure were fabricated similar to ones of all-perpendicular PSV studied in Chapter 5. Identical setup and conditions for dc and microwave measurements used in Chapter 5 were also applied to these tilted-polarizer PSV STOs. Here the results from a 70-nm NC diameter STO are presented.

6.2 Experimental results

Alternating gradient magnetometry (AGM) measurements were performed on a cleaved 5 × 5 mm2 tilted-polarizer PSV sample. The M–H loop (Figure 6.3a) shows the switching of the magnetization in the [Co/Ni] ML free layer at about +400 Oe for the field scanning from negative to positive and -400 Oe for the reverse scan; the interlayer exchange coupling by the stacked fixed layer on [Co/Ni] ML is thus insignificant in this extended stack. Further increases in the applied field ultimately saturated the whole stack at ±1100 Oe, which is the switching field of the [Co/Pd] ML. Between the two switching fields, however, there was a gradual change of magnetization with the field, reproducing what was observed in [119][120][122] and described in Section 6.1: the 3 nm-thick IMA Co layer inserted on top of the PMA [Co/Pd] ML was tilted away from its in-plane easy axis. Figure 6.3b shows the magnetoresistance (MR) major loop from dc characterization at low dc current of the tilted-polarizer STO, for the field direction normal (θH = 90°) to the stack plane, illustrating a window for an antiparallel (AP) state between +700 and +1400 Oe (-700 and -1400 Oe in the backward sweeping direction), corresponding to the switching of [Co/Ni] ML and [Co/Pd] ML, respectively; however, the effect on switching by the tilted Co layer is rather unclear. The switching fields of both MLs in a patterned device are higher than the values obtained from the M–H loop of

6. TILTED POLARIZER-PSV STO

the extended stack, in agreement with observations on the PMA-based patterned devices [123][124].

At θH = 30° from the stack plane, the increase in the switching fields of [Co/Ni] and [Co/Pd] MLs led to an AP window between +(-) 1100 Oe and +(- ) 2800 Oe in the forward (backward) sweeping direction. Taking into

Figure 6.3: OOP hysteresis loop of the deposited tilted-polarizer PSV stack (a); and magnetoresistance major loops of the fabricated STOs for (b) perpendicular and (c) 30°applied fields

6.2. Results

account that the perpendicular component (= Hsin30°, or half) of the applied field is responsible for the switching of these PMA MLs, it can be said that the coercivity of the [Co/Pd] ML is independent of the field angle but that of [Co/Ni] ML is reduced. Moreover, the negative slope observed during the AP state—that is, the decrease in R↑↓ with the increase in the applied field—is likely attributed to the IMA Co layer between the two MLs in the presence of the strong in-plane component of H (Figure 6.3c). The slope angle is 30°, but further angular investigation is required to clarify whether this has its origin in the tilt angle of Co layer or the applied field angle, or is merely a coincidence. We now turn to the microwave measurements. Field-sweep spectra and MR curve of the tilted-polarizer STO for an applied current of -11 mA are plotted together in Figure 6.4. For the field applied normal to the device plane, a similar transition during the switching of the [Co/Ni] ML free layer was observed as in the 60-nm NC diameter all-perpendicular PSV STO at -6

Figure 6.4: Field-dependent resistance and microwave behavior of the tilted- polarizer PSV STO in perpendicular (left) and 30o (right) applied fields at -11 mA

6. TILTED POLARIZER-PSV STO

mA presented in Section 5.2, even though a similarly clear transition step was not seen. The measurement of low-frequency noise was also achieved, but its power was rather weak. In contrast to the all-perpendicular devices, for θH = 90°, there are sharp signals at FMR frequency along the entire range of the applied field. Obviously, the tilted anisotropy of the Co layer in this PSV structure contributes significantly to the magnetic non-collinearity between the free and fixed layers and hence the excitation of the auto- oscillation mode. The coexistence of both droplet soliton and auto- oscillation, however, is impossible in a single layer, so if one exists in the [Co/Ni] ML, the other must be in the [Co/Pd] ML, thus giving rise to complicated dynamics that will require further time and effort to clarify in the future. On the other hand, for θH = 30°, the characteristic noise power of the droplet was much more pronounced than when θH = 90°, and the droplet frequency also became more visible on account of its higher power. On the [Co/Ni] ML switching side—that is, the first switching along the field sweeping direction—the coexistence of the droplet and auto-oscillation was similar to the complex dynamics in the perpendicular field. In striking contrast, on the other side of MR switching, the combination of i) frequency falling from FMR-like to droplet-like, ii) MR slope change, and iii) low- frequency noise, indicates the nucleation of the droplet solitons, previously observed in PSV STOs based on orthogonal configurations of IMA free layer– PMA fixed layer in our group [101][107]. This means that, at this field angle, the magnetic configuration, with the support of strong in-plane field component and the IMA nature of the Co layer, became more rigid, resulting in more limpid dynamics on this side of the switching loop. It is worth noticing that, at this relatively large applied dc current of -11 mA, a clear negative slope was observed during the AP state for both field angles, unlike the small current MR loop in the perpendicular field. Unlike the all- perpendicular PSV STOs, however, there was no sequence of droplet nucleation observed in tilted-polarizer PSV devices. The phase diagrams of magnetoresistance in Figure 6.5 for the 70-nm NC tilted-polarizer PSV STO at two field angles, θH = 90° and 30°, were drawn up in the same was as Figure 5.4 (left) for the all-perpendicular PSV STO— that is, in the backward field sweeping direction from +5000 to -5000 Oe. Within the dc current range of -1 to -13 mA, the perpendicular field diagram shows a stable AP state whose MR = 1 has vertical boundaries, while the “droplet” state (0 < MR < 1) borders linearly with the P state (MR = 0). This is almost similar to what was observed in Chapter 5. By contrast, when the field is applied 30o from the device plane, the right boundary of the AP state

6.2. Results

Figure 6.5: Map of the resistance states: Effect of the field angle θH (left: perpendicular to; right: 30° from the plane) on the nucleation of droplet soliton in the 70-nm NC tilted-polarizer PSV STO. undergoes a jump between -4 and -5 mA and then further broadens when the current increases; the left boundary however retains its independence of the current. The P state also expands upwards with higher current at the expense of the “droplet” state, particularly in the negative field wing where the “droplet” state is reduced to a tiny region between -11 and -13 mA and its MR is only a little larger than 0. Due to the complexity of the dynamics in this tilted–polarizer PSV structure with three different components—Co/Ni ML, thick Co layer and Co/Pd ML—are competing with each other in the presence of a strong in-plane field component, further effort is required for better understanding.

Conclusions and future work

The ultimate goal of this thesis is to demonstrate microwave signal generation, and investigate the microwave performance, of spin-torque oscillators (STOs) based on perpendicular magnetic anisotropy (PMA) materials, in order to illustrate their merits for both practical applications and for a deeper understanding of spin dynamics. Various pseudo spin-valve (PSV) structures based on PMA multilayers of Co/Pd and Co/Ni were deposited with different magnetic configurations – both all-perpendicular and tilted polarizer – from which nanocontact STO devices were fabricated. On the other hand, PMA CoFeB was used in both the free layer and polarizer to produce p-MTJ devices. CoFeB p-MTJ nanopillar STOs were studied in details. A 100-nm diameter device generated an STT-driven microwave oscillation whose frequency (15 GHz) is much higher than other reported configurations (usually about 5 GHz) and required low current to activate STT effects (the onset current is 20 μA). Remarkably, the current tunability of the oscillation frequency is extremely high (4.4 GHz.mA-1), and thus needs much less power to reach broadband microwave performance. The linear bias dependence of the STT torque component with the slope of 3 A-1 suggests the huge potential of gate control, and may thus have a large impact on the scaling of devices. Nucleation and annihilation of magnetic droplets were empirically observed in all-perpendicular PSV nanocontact STOs for the first time. Most remarkably, these excitation mechanisms were obtained at much lower applied fields (1000–2000 Oe) and drive current (3–6 mA) than previously observed. Additionally, a sequence of droplet nucleation was performed purely by manipulating the applied field scanning direction. The low-field and low-current operating conditions greatly simplify the experimental requirements for studying droplets, hence opening up for using a wider range of experimental tools, such as magnetic force microscopy and scanning transmission X-ray microscopy. Another type of PSV nanocontact STOs was fabricated with a tilted anisotropy fixed layer (polarizer) based on PMA Co/Pd multilayer and IMA Co layer. Microwave characterization showed the effects of the Co layer on the dynamics of uniform magnetization and localized droplets at different angles of the applied field. In particular, the conventional droplet was

Conclusion

present with a dc current of -11 mA and a 30o-from-plane applied field. With further efforts to manipulate the material designs and measurement conditions, this family of unique tilted anisotropy PSVs would deliver better understanding of such dynamics.

Bibliography

[1] M. Julliere, “Tunneling between ferromagnetic films”, Phys. Lett. A, vol. 54, no. 3, pp. 225–226, 1975. [2] M. N. Baibich, J. M. Broto, A. Fert, F. N. Van Dau, F. Petroff, P. Etienne, G. Creuzet, A. Friederich, and J. Chazelas, “Giant Magnetoresistance of (001)Fe/(001)Cr Magnetic Superlattices”, Phys. Rev. Lett., vol. 61, no. 21, pp. 2472–2475, 1988. [3] G. Binasch, P. Grundberg, P. Saurenbach, and W. Zinn, “Enhanced magnetoresistance in layered magnetic structures with antiferromagnetic interlayer exchange”, Phys. Rev. B, vol. 39, no. 7, pp. 4828–4830, 1989. [4] L. Berger, “Emission of spin waves by a magnetic multilayer traversed by a current”, Phys. Rev. B, vol. 54, no. 13, pp. 9353–9358, Oct. 1996. [5] J. Slonczewski, “Excitation of spin waves by an electric current”, J. Magn. Magn. Mater., vol. 195, no. 2, pp. 261–268, 1999. [6] J. Åkerman, “Applied physics. Toward a universal memory”, Science, vol. 308, p. 508, 2005. [7] B. Engel, J. Åkerman, B. Butcher, R. Dave, and M, “A 4-Mb toggle MRAM based on a novel bit and switching method”, IEEE Trans. Magn., vol. 41, no. 1, pp. 132–136, Jan. 2005. [8] E. Chen, D. Apalkov, Z. Diao, A. Driskill-Smith, D. Druist, D. Lottis, V. Nikitin, X. Tang, S. Watts, S. Wang, S. A. Wolf, A. W. Ghosh, J. W. Lu, S. J. Poon, M. Stan, W. H. Butler, S. Gupta, C. K. A. Mewes, T. Mewes, and P. B. Visscher, “Advances and future prospects of spin-transfer torque random access memory”, IEEE Trans. Magn., vol. 46, no. 6, pp. 1873–1878, 2010. [9] M. Durlam, D. Addie, J. Åkerman, B. Butcher, P. Brown, J. Chan, M. DeHerrera, B. N. Engel, B. Feil, G. Grynkewich, J. Janesky, M. Johnson, K. Kyler, J. Molla, J. Martin, K. Nagel, J. Ren, N. D. Rizzo, T. Rodriguez, L. Savtchenko, J. Salter, J. M. Slaughter, K. Smith, J. J. Sun, M. Lien, K. Papworth, P. Shah, W. Qin, R. Williams, L. Wise, and S. Tehrani, “A 0.18 μ m 4Mb toggling MRAM”, IEEE IEDM, no. August 2016, pp. 4–7, 2003. [10] G. Grynkewich, J. Åkerman, P. Brown, B. Butcher, R. W. Dave, M. DeHerrera, M. Durlam, B. N. Engel, J. Janesky, S. Pietambaram, N. D. Rizzo, J. M. Slaughter, K. Smith, J. J. Sun, and S. Tehrani, “Nonvolatile magnetoresistive random-access memory based on magnetic tunnel

Bibliography

junctions”, MRS Bull., vol. 29, no. 11, pp. 818–821+809–813, Nov. 2004. [11] S. Iida, “The difference between gilbert’s and landau-lifshitz's equations”, J. Phys. Chem. Solids, vol. 24, no. 5, pp. 625–630, 1963. [12] L. Landau and E. Lifshits, “on the Theory of the Dispersion of Magnetic Permeability in Ferromagnetic Bodies”, Phys. Zeitsch. der Sow., vol. 169, no. 14, pp. 14–22, 1935. [13] J. Slonczewski, “Current-driven excitation of magnetic multilayers”, J. Magn. Magn. Mater., vol. 159, no. 1–2, pp. L1–L7, 1996. [14] S. Zhang, P. M. Levy, and A. Fert, “Mechanisms of Spin-Polarized Current-Driven Magnetization Switching”, Phys. Rev. Lett., vol. 88, no. 23, p. 236601, 2002. [15] M. E. Flatte, “Spintronics”, IEEE Trans. Electron Devices, vol. 54, no. 5, pp. 907–920, 2007. [16] P. Villard, U. Ebels, D. Houssameddine, J. Katine, D. Mauri, B. Delaet, P. Vincent, M. C. Cyrille, B. Viala, J. P. Michel, J. Prouvée, and F. Badets, “A GHz spintronic-based RF oscillator”, IEEE J. Solid-State Circuits, vol. 45, no. 1, pp. 214–223, 2010. [17] P. Vincent, M.-C. Cyrille, B. Viala, B. Delaet, J. P. Michel, P. Villard, J. Prouvee, D. Houssamedine, U. Ebels, J. A. Katine, D. Mauri, S. Florez, O. Ozatay, L. Folks, B. D. Terris, and F. Badets, “A GHz spintronic- based RF oscillator”, 2009 IEEE Int. Solid-State Circuits Conf. - Dig. Tech. Pap., pp. 196–197,197a, Feb. 2009. [18] A. Tulapurkar, Y. Suzuki, A. Fukushima, H. Kubota, H. Maehara, K. Tsunekawa, D. D. Djayaprawira, N. Watanabe, and S. Yuasa, “Spin- torque diode effect in magnetic tunnel junctions.”, Nature, vol. 438, no. 7066, pp. 339–42, Nov. 2005. [19] L. Néel, “Anisotropie magnétique superficielle et surstructures d’orientation”, J. Phys. Radium, vol. 15, no. 4, pp. 225–239, 1954. [20] U. Gradmann, “Magnetic Surface anisotropies”, J. Magn. Magn. Mater., vol. 57, pp. 733–736, 1986. [21] F. J. A. den Broeder, H. C. Donkersloot, H. J. G. Draaisma, and W. J. M. de Jonge, “Magnetic properties and structure of Pd/Co and Pd/Fe multilayers”, J. Appl. Phys., vol. 61, no. 8, pp. 4317–4319, 1987. [22] H. J. G. Draaisma, F. J. A. den Broeder, and W. J. M. de Jonge, “Perpendicular anisotropy in Pd/Co multilayers”, J. Appl. Phys., vol. 63, no. 8, p. 3479, 1988. [23] H. J. G. Draaisma, W. J. M. de Jonge, and F. J. A. den Broeder, “Magnetic interface anisotropy in Pd/Co and Pd/Fe multilayers”, J. 70

Bibliography

Magn. Magn. Mater., vol. 66, no. 3, pp. 351–355, 1987. [24] P. F. Carcia, A. D. Meinhaldt, and A. Suna, “Perpendicular magnetic anisotropy in Pd/Co thin film layered structures”, Appl. Phys. Lett., vol. 47, no. 2, pp. 178–180, 1985. [25] P. F. Carcia, “Perpendicular magnetic anisotropy in Pd/Co and Pt/Co thin-film layered structures”, J. Appl. Phys., vol. 63, no. 10, p. 5066, 1988. [26] F. J. A. den Broeder, D. Kuiper, A. P. van de Mosselaer, and W. Hoving, “Perpendicular Magnetic Anisotropy of Co-Au Multilayers Induced by Interface Sharpening”, Phys. Rev. Lett., vol. 60, no. 26, pp. 2769–2772, 1988. [27] D. G. Stinson and S.-C. Shin, “Magnetization and anisotropy of Co/Pd multilayer thin films”, J. Appl. Phys., vol. 67, no. 9, p. 4459, 1990. [28] T. Kato, Y. Matsumoto, S. Okamoto, N. Kikuchi, O. Kitakami, N. Nishizawa, S. Tsunashima, and S. Iwata, “Time-resolved magnetization dynamics and damping constant of sputtered Co/Ni multilayers”, IEEE Trans. Magn., vol. 47, no. 10, pp. 3036–3039, 2011. [29] T. Kato, Y. Matsumoto, S. Kashima, S. Okamoto, N. Kikuchi, S. Iwata, O. Kitakami, and S. Tsunashima, “Perpendicular anisotropy and gilbert damping in sputtered Co/Pd multilayers”, IEEE Trans. Magn., vol. 48, no. 11, pp. 3288–3291, 2012. [30] M. Frankowski, M. Czapkiewicz, W. Skowroński, and T. Stobiecki, “Micromagnetic model for studies on Magnetic Tunnel Junction switching dynamics, including local current density”, Phys. B Condens. Matter, vol. 435, pp. 105–108, Feb. 2014. [31] R. J. Soulen, J. M. Byers, M. S. Osofsky, B. Nadgorny, T. Ambrose, S. F. Cheng, P. R. Broussard, C. T. Tanaka, J. Nowak, J. S. Moodera, A. Barry, and J. M. D. Coey, “Measuring the spin polarization of a metal with a superconducting point contact”, Science (80-. )., vol. 282, no. 5386, pp. 85–88, 1998. [32] D. J. Monsma and S. S. P. Parkin, “Temporal evolution of spin- polarization in ferromagnetic tunnel junctions Temporal evolution of spin-polarization in ferromagnetic tunnel junctions”, Appl. Phys. Lett., vol. 77, no. 6, pp. 883–885, 2000. [33] S. Ikeda, J. Hayakawa, Y. Ashizawa, Y. M. Lee, K. Miura, H. Hasegawa, M. Tsunoda, F. Matsukura, S. Ikeda, J. Hayakawa, Y. Ashizawa, Y. M. Lee, K. Miura, and H. Hasegawa, “Tunnel magnetoresistance of 604 % at 300 K by suppression of Ta diffusion in Co Fe B∕Mg O∕Co Fe B pseudospin-valves annealed at high temperature Tunnel magnetoresistance of 604 % at 300 K by suppression of Ta diffusion in CoFeB/MgO/CoFeB

Bibliography

pseudo-spin-valv”, Appl. Phys. Lett., vol. 93, p. 082508, 2008. [34] H. Liu, Y. Honda, T. Taira, K. Matsuda, M. Arita, T. Uemura, H. Liu, Y. Honda, T. Taira, K. Matsuda, and M. Arita, “Giant tunneling magnetoresistance in epitaxial Co 2 MnSi / MgO / Co 2 MnSi magnetic tunnel junctions by half-metallicity of Co 2 MnSi and coherent tunneling”, Appl. Phys. Lett., vol. 101, p. 132418, 2012. [35] W. H. Butler, X. Zhang, and T. C. Schulthess, “Spin-dependent tunneling conductance of Fe/MgO/Fe sandwiches”, Phys. Rev. B, vol. 63, pp. 0544161–12, 2001. [36] J. Mathon and A. Umerski, “Theory of tunneling magnetoresistance of an epitaxial Fe/MgO/Fe(001) junction”, Phys. Rev. B, vol. 63, pp. 2204031–4, 2001. [37] S. Yuasa, T. Nagahama, A. Fukushima, Y. Suzuki, and K. Ando, “Giant room-temperature magnetoresistance in single-crystal Fe/MgO/Fe magnetic tunnel junctions”, Nat. Mater., vol. 3, pp. 868–871, 2004. [38] S. Yuasa, Y. Suzuki, T. Katayama, and K. Ando, “Characterization of growth and crystallization processes in CoFeB∕MgO∕CoFeB magnetic tunnel junction structure by reflective high-energy electron diffraction”, Appl. Phys. Lett., vol. 87, no. 24, p. 242503, 2005. [39] D. D. Djayaprawira, K. Tsunekawa, M. Nagai, H. Maehara, S. Yamagata, N. Watanabe, S. Yuasa, Y. Suzuki, and K. Ando, “230% room- temperature magnetoresistance in CoFeB∕MgO∕CoFeB magnetic tunnel junctions”, Appl. Phys. Lett., vol. 86, no. 9, p. 092502, 2005. [40] K. Tsunekawa, D. D. Djayaprawira, M. Nagai, H. Maehara, S. Yamagata, N. Watanabe, S. Yuasa, Y. Suzuki, and K. Ando, “Giant tunneling magnetoresistance effect in low-resistance CoFeB∕MgO(001)∕CoFeB magnetic tunnel junctions for read-head applications”, Appl. Phys. Lett., vol. 87, no. 7, p. 072503, 2005. [41] S. Yuasa, A. Fukushima, H. Kubota, Y. Suzuki, and K. Ando, “Giant tunneling magnetoresistance up to 410% at room temperature in fully epitaxial Co∕MgO∕Co magnetic tunnel junctions with bcc Co(001) electrodes”, Appl. Phys. Lett., vol. 89, no. 4, p. 042505, 2006. [42] S. Yuasa, T. Katayama, T. Nagahama, A. Fukushima, H. Kubota, Y. Suzuki, and K. Ando, “Giant tunneling magnetoresistance in fully epitaxial body-centered-cubic Co/MgO/Fe magnetic tunnel junctions”, Appl. Phys. Lett., vol. 87, no. 22, pp. 1–3, 2005. [43] R. W. Dave, G. Steiner, J. M. Slaughter, J. J. Sun, B. Craigo, S. Pietambaram, K. Smith, G. Grynkewich, M. DeHerrera, J. Åkerman, and S. Tehrani, “MgO-based tunnel junction material for high-speed toggle magnetic random access memory”, IEEE Trans. Magn., vol. 42,

Bibliography

no. 8, pp. 1935–1939, 2006. [44] R. Sbiaa, H. Le Gall, and J. M. Desvignes, “Magnetization reorientation in ferrimagnetic GdFeCo/DyFeCo double layers”, Phys. Rev. B, vol. 57, no. 13, pp. 7887–7891, 1998. [45] C. Lee, L. Ye, J. Lee, W. Chen, C. Huang, G. Chern, and T. Wu, “Ultrathin ( Gd , Tb ) -FeCo Films With Perpendicular Magnetic Anisotropy”, IEEE Trans. Mag., vol. 45, no. 10, pp. 3808–3811, 2009. [46] T. Hatori, H. Ohmori, M. Tada, and S. Nakagawa, “MTJ Elements With MgO Barrier Using RE-TM Amorphous Layers for Perpendicular MRAM”, IEEE Trans. Magn., vol. 43, no. 6, pp. 2331–2333, 2007. [47] N. Nishimura, T. Hirai, A. Koganei, T. Ikeda, K. Okano, Y. Sekiguchi, and Y. Osada, “Magnetic tunnel junction device with perpendicular magnetization films for high-density magnetic random access memory”, J. Appl. Phys., vol. 91, no. 8, pp. 5246–5249, 2002. [48] M. Nakayama, T. Kai, N. Shimomura, M. Amano, E. Kitagawa, T. Nagase, M. Yoshikawa, T. Kishi, S. Ikegawa, and H. Yoda, “Spin transfer switching in TbCoFe∕CoFeB∕MgO∕CoFeB∕TbCoFe magnetic tunnel junctions with perpendicular magnetic anisotropy”, J. Appl. Phys., vol. 103, no. 7, pp. 07A710–07A710–3, 2008. [49] A. C. Cabrera, C. Chang, C. Hsu, M. Weng, C. C. Chen, C. T. Chao, and J. C. Wu, “Perpendicular Magnetic Tunneling Junction With Double Barrier Layers for MRAM Application”, IEEE Trans. Magn., vol. 43, no. 2, pp. 914–916, 2007. [50] H. Ohmori, T. Hatori, and S. Nakagawa, “Perpendicular magnetic tunnel junction with tunneling magnetoresistance ratio of 64% using MgO (100) barrier layer prepared at room temperature”, J. Appl. Phys., vol. 103, no. 7, pp. 2006–2009, 2008. [51] Y. J. Chang, A. Canizo-Cabrera, V. Garcia-Vazquez, Y. H. Chang, and T. H. Wu, “Perpendicular magnetic tunnel junctions with synthetic antiferromagnetic pinned layers based on [Co/Pd] multilayers”, J. Appl. Phys., vol. 113, p. 17B909, 2013. [52] Y. Wang, W. X. Wang, H. X. Wei, B. S. Zhang, W. S. Zhan, and X. F. Han, “Effect of annealing on the magnetic tunnel junction with Co/Pt perpendicular anisotropy ferromagnetic multilayers”, J. Appl. Phys., vol. 107, no. 9, p. 09C711, 2010. [53] K. Mizunuma, S. Ikeda, H. Yamamoto, H. D. Gan, K. Miura, H. Hasegawa, J. Hayakawa, K. Ito, F. Matsukura, and H. Ohno, “CoFeB Inserted Perpendicular Magnetic Tunnel Junctions with CoFe/Pd Multilayers for High Tunnel Magnetoresistance Ratio”, Jpn. J. Appl. Phys., vol. 49, no. 4, pp. 04DM04–1–4, 2010.

Bibliography

[54] S. Ikeda, K. Miura, H. Yamamoto, K. Mizunuma, H. D. Gan, M. Endo, S. Kanai, J. Hayakawa, F. Matsukura, and H. Ohno, “A perpendicular- anisotropy CoFeB-MgO magnetic tunnel junction.”, Nat. Mater., vol. 9, no. 9, pp. 721–4, Sep. 2010. [55] M. Bowen, V. Cros, F. Petroff, A. Fert, C. Martı́nez Boubeta, J. L. Costa- Krämer, J. V. Anguita, A. Cebollada, F. Briones, J. M. de Teresa, L. Morellón, M. R. Ibarra, F. Güell, F. Peiró, and A. Cornet, “Large magnetoresistance in Fe/MgO/FeCo(001) epitaxial tunnel junctions on GaAs(001)”, Appl. Phys. Lett., vol. 79, no. 11, p. 1655, 2001. [56] J. Faure-Vincent, C. Tiusan, E. Jouguelet, F. Canet, M. Sajieddine, C. Bellouard, E. Popova, M. Hehn, F. Montaigne, and A. Schuhl, “High tunnel magnetoresistance in epitaxial Fe/MgO/Fe tunnel junctions”, Appl. Phys. Lett., vol. 82, no. 25, p. 4507, 2003. [57] S. S. P. Parkin, C. Kaiser, A. Panchula, P. M. Rice, B. Hughes, M. Samant, and S.-H. Yang, “Giant tunnelling magnetoresistance at room temperature with MgO (100) tunnel barriers”, Nat. Mater., vol. 3, no. 12, pp. 862–867, 2004. [58] S. Tacchi, F. Montoncello, M. Madami, G. Gubbiotti, G. Carlotti, L. Giovannini, R. Zivieri, F. Nizzoli, S. Jain, A. O. Adeyeye, and N. Singh, “Band Diagram of Spin Waves in a Two-Dimensional Magnonic Crystal”, Phys. Rev. Lett., vol. 107, no. 12, p. 127204, Sep. 2011. [59] S. Ikeda, J. Hayakawa, Y. M. Lee, R. Sasaki, T. Meguro, F. Matsukura, and H. Ohno, “Dependence of tunnel magnetoresistance in MgO based magnetic tunnel junctions on Ar pressure during MgO sputtering”, Jpn. J. Appl. Phys., vol. 44, no. 48, pp. L1442–L1445, 2005. [60] J. Hayakawa, S. Ikeda, Y. M. Lee, F. Matsukura, and H. Ohno, “Effect of high annealing temperature on giant tunnel magnetoresistance ratio of CoFeB∕MgO∕CoFeB magnetic tunnel junctions”, Appl. Phys. Lett., vol. 89, no. 23, p. 232510, 2006. [61] J. S. Moodera, L. R. Kinder, T. M. Wong, and R. Meservey, “Large magnetoresistance at room temperature in ferromagnetic thin film tunnel junctions”, Phys. Rev. Lett., vol. 74, no. 16, pp. 3273–3276, 1995. [62] T. Miyazaki and N. Tezuka, “Giant magnetic tunneling effect in Fe/Al2O3/Fe junction”, J. Magn. Magn. Mater., vol. 139, no. 3, pp. L231–L234, 1995. [63] J. S. Moodera, E. F. Gallagher, K. Robinson, and J. Nowak, “Optimum tunnel barrier in ferromagnetic–insulator–ferromagnetic tunneling structures”, Appl. Phys. Lett., vol. 70, no. 22, p. 3050, 1997. [64] R. C. Sousa, J. J. Sun, V. Soares, P. P. Freitas, A. Kling, M. F. da Silva, and J. C. Soares, “Large tunneling magnetoresistance enhancement by

Bibliography

thermal anneal”, Appl. Phys. Lett., vol. 73, no. 22, pp. 3288–3290, 1998. [65] S. Cardoso, V. Gehanno, R. Ferreira, and P. P. Freitas, “Ion beam deposition and oxidation of spin-dependent tunnel junctions”, IEEE Trans. Magn., vol. 35, no. 5, pp. 2952–2954, 1999. [66] J. H. Yu, H. M. Lee, Y. Ando, and T. Miyazaki, “Electron transport properties in magnetic tunnel junctions with epitaxial NiFe (111) ferromagnetic bottom electrodes”, Appl. Phys. Lett., vol. 82, no. 26, p. 4735, 2003. [67] D. Wang, C. Nordman, J. M. Daughton, Z. Qian, and J. Fink, “70% TMR at room temperature for SDT sandwich junctions with CoFeB as free and reference layers”, IEEE Trans. Magn., vol. 40, no. 4 II, pp. 2269– 2271, 2004. [68] Y. Sakuraba, M. Hattori, M. Oogane, Y. Ando, H. Kato, A. Sakuma, T. Miyazaki, and H. Kubota, “Giant tunneling magnetoresistance in Co[sub 2]MnSi∕Al–O∕Co[sub 2]MnSi magnetic tunnel junctions”, Appl. Phys. Lett., vol. 88, no. 2006, p. 192508, 2006. [69] H. X. Wei, Q. H. Qin, M. Ma, R. Sharif, and X. F. Han, “80% tunneling magnetoresistance at room temperature for thin Al–O barrier magnetic tunnel junction with CoFeB as free and reference layers”, J. Appl. Phys., vol. 101, no. 9, p. 09B501, 2007. [70] K. Mizunuma, M. Yamanouchi, H. Sato, S. Ikeda, S. Kanai, F. Matsukura, and H. Ohno, “Size Dependence of Magnetic Properties of Nanoscale CoFeB–MgO Magnetic Tunnel Junctions with Perpendicular Magnetic Easy Axis Observed by Ferromagnetic Resonance”, Appl. Phys. Express, vol. 6, no. 6, p. 063002, May 2013. [71] A. Vansteenkiste, J. Leliaert, M. Dvornik, M. Helsen, F. Garcia-Sanchez, and B. Van Waeyenberge, “The design and verification of MuMax3”, AIP Adv., vol. 4, no. 10, pp. 0–22, 2014. [72] K. S. Lee, D. S. Han, and S. K. Kim, “Physical origin and generic control of magnonic band gaps of dipole-exchange spin waves in width- modulated nanostrip waveguides”, Phys. Rev. Lett., vol. 102, no. 12, pp. 1–4, 2009. [73] M. Fazlali, M. Dvornik, E. Iacocca, P. Dürrenfeld, M. Haidar, J. Åkerman, and R. K. Dumas, “Homodyne-detected ferromagnetic resonance of in-plane magnetized nanocontacts: Composite spin-wave resonances and their excitation mechanism”, Phys. Rev. B, vol. 93, no. 13, p. 134427, 2016. [74] R. D. McMichael and M. D. Stiles, “Magnetic normal modes of nanoelements”, J. Appl. Phys., vol. 97, no. 10, 2005.

Bibliography

[75] G. N. Kakazei, P. E. Wigen, K. Y. Guslienko, V. Novosad, A. N. Slavin, V. O. Golub, N. A. Lesnik, and Y. Otani, “Spin-wave spectra of perpendicularly magnetized circular submicron dot arrays”, Appl. Phys. Lett., vol. 85, no. 3, pp. 443–445, 2004. [76] V. Naletov, G. de Loubens, G. Albuquerque, S. Borlenghi, V. Cros, G. Faini, J. Grollier, H. Hurdequint, N. Locatelli, B. Pigeau, A. Slavin, V. Tiberkevich, C. Ulysse, T. Valet, and O. Klein, “Identification and selection rules of the spin-wave eigenmodes in a normally magnetized nanopillar”, Phys. Rev. B, vol. 84, no. 22, pp. 1–23, Dec. 2011. [77] S. Kanai, M. Gajek, D. C. Worledge, F. Matsukura, and H. Ohno, “Electric field-induced ferromagnetic resonance in a CoFeB/MgO magnetic tunnel junction under dc bias voltages”, Appl. Phys. Lett., vol. 105, no. 24, p. 242409, 2014. [78] T. Nozaki, Y. Shiota, S. Miwa, S. Murakami, F. Bonell, S. Ishibashi, H. Kubota, K. Yakushiji, T. Saruya, A. Fukushima, S. Yuasa, T. Shinjo, and Y. Suzuki, “Electric-field-induced ferromagnetic resonance excitation in an ultrathin ferromagnetic metal layer”, Nat. Phys., vol. 8, no. 6, pp. 492–497, Apr. 2012. [79] J. A. Katine and E. E. Fullerton, “Device implications of spin-transfer torques”, J. Magn. Magn. Mater., vol. 320, pp. 1217–1226, 2008. [80] S. Bonetti, P. Muduli, F. Mancoff, and J. Åkerman, “Spin torque oscillator frequency versus magnetic field angle: The prospect of operation beyond 65 GHz”, Appl. Phys. Lett., vol. 94, no. 10, p. 102507, 2009. [81] M. Madami, S. Bonetti, G. Consolo, S. Tacchi, G. Carlotti, G. Gubbiotti, F. B. Mancoff, M. A. Yar, and J. Åkerman, “Direct observation of a propagating spin wave induced by spin-transfer torque.”, Nat. Nanotechnol., vol. 6, no. 10, pp. 635–8, Oct. 2011. [82] S. I. Kiselev, J. C. Sankey, I. N. Krivorotov, N. C. Emley, R. J. Schoelkopf, R. A. Buhrman, and D. C. Ralph, “Microwave oscillations of a nanomagnet driven by a spin-polarized current.”, Nature, vol. 425, no. 6956, pp. 380–3, Sep. 2003. [83] I. N. Krivorotov, N. C. Emley, J. C. Sankey, S. I. Kiselev, D. C. Ralph, and R. A. Buhrman, “Time-domain measurements of nanomagnet dynamics driven by spin-transfer torques.”, Science, vol. 307, no. 5707, pp. 228–31, Jan. 2005. [84] W. Rippard, M. Pufall, S. Kaka, T. Silva, S. Russek, and J. Katine, “Injection Locking and Phase Control of Spin Transfer Nano- oscillators”, Phys. Rev. Lett., vol. 95, no. 6, p. 067203, Aug. 2005. [85] S. Bonetti, V. Tiberkevich, G. Consolo, G. Finocchio, P. Muduli, F.

Bibliography

Mancoff, A. Slavin, and J. Åkerman, “Experimental Evidence of Self- Localized and Propagating Spin Wave Modes in Obliquely Magnetized Current-Driven Nanocontacts”, Phys. Rev. Lett., vol. 105, no. 21, p. 217204, Nov. 2010. [86] P. K. Muduli, Y. Pogoryelov, S. Bonetti, G. Consolo, F. Mancoff, and J. Åkerman, “Nonlinear frequency and amplitude modulation of a nanocontact-based spin-torque oscillator”, Phys. Rev. B, vol. 81, no. 14, p. 140408, Apr. 2010. [87] R. K. Dumas, E. Iacocca, S. Bonetti, S. R. Sani, S. M. Mohseni, A. Eklund, J. Persson, O. Heinonen, and J. Åkerman, “Spin-wave-mode coexistence on the nanoscale: A consequence of the oersted-field- induced asymmetric energy landscape”, Phys. Rev. Lett., vol. 110, no. 25, pp. 1–5, 2013. [88] S. Sani, J. Persson, S. M. Mohseni, Y. Pogoryelov, P. K. Muduli, A. Eklund, G. Malm, M. Käll, A. Dmitriev, and J. Åkerman, “Mutually synchronized bottom-up multi-nanocontact spin-torque oscillators.”, Nat. Commun., vol. 4, p. 2731, Jan. 2013. [89] Y. Pogoryelov, P. K. Muduli, S. Bonetti, E. Iacocca, F. Mancoff, and J. Åkerman, “Frequency modulation of spin torque oscillator pairs”, Appl. Phys. Lett., vol. 98, no. 19, p. 192501, 2011. [90] S. Bonetti, V. Puliafito, G. Consolo, V. Tiberkevich, A. Slavin, and J. Åkerman, “Power and linewidth of propagating and localized modes in nanocontact spin-torque oscillators”, Phys. Rev. B, vol. 85, no. 17, pp. 1–8, May 2012. [91] Y. Pogoryelov, P. K. Muduli, S. Bonetti, F. Mancoff, and J. Åkerman, “Spin-torque oscillator linewidth narrowing under current modulation”, Appl. Phys. Lett., vol. 98, no. 19, p. 192506, 2011. [92] R. K. Dumas, S. R. Sani, S. M. Mohseni, E. Iacocca, Y. Pogoryelov, P. K. Muduli, S. Chung, P. Dürrenfeld, and J. Åkerman, “Recent Advances in Nanocontact Spin-Torque Oscillators”, IEEE Trans. Magn., vol. 50, no. 6, p. 4100107, Jun. 2014. [93] A. Eklund, S. Bonetti, S. R. Sani, S. Majid Mohseni, J. Persson, S. Chung, S. Amir Hossein Banuazizi, E. Iacocca, M. Östling, J. Åkerman, and B. Gunnar Malm, “Dependence of the colored frequency noise in spin torque oscillators on current and magnetic field”, Appl. Phys. Lett., vol. 104, no. 9, p. 092405, Mar. 2014. [94] M. Madami, E. Iacocca, S. Sani, G. Gubbiotti, S. Tacchi, R. K. Dumas, J. Åkerman, and G. Carlotti, “Propagating spin waves excited by spin- transfer torque: A combined electrical and optical study”, Phys. Rev. B, vol. 92, no. 2, p. 024403, 2015.

Bibliography

[95] U. Roy, H. Seinige, F. Ferdousi, J. Mantey, M. Tsoi, and S. K. Banerjee, “Spin-transfer-torque switching in spin valve structures with perpendicular, canted, and in-plane magnetic anisotropies”, J. Appl. Phys., vol. 111, no. 7, p. 07C913, 2012. [96] S. Le Gall, J. Cucchiara, M. Gottwald, C. Berthelot, C. H. Lambert, Y. Henry, D. Bedau, D. B. Gopman, H. Liu, A. D. Kent, J. Z. Sun, W. Lin, D. Ravelosona, J. A. Katine, E. E. Fullerton, and S. Mangin, “State diagram of nanopillar spin valves with perpendicular magnetic anisotropy”, Phys. Rev. B - Condens. Matter Mater. Phys., vol. 86, no. 1, pp. 1–16, 2012. [97] B. A. Ivanov and A. M. Kosevich, “Bound states of a large number of magnons in a ferromagnet with a single-ion anisotropy”, Zh. Ursp. Teor. Fiz., vol. 72, pp. 2000–2015, 1977. [98] M. Hoefer, T. Silva, and M. Keller, “Theory for a dissipative droplet soliton excited by a spin torque nanocontact”, Phys. Rev. B, vol. 82, no. 5, pp. 1–14, Aug. 2010. [99] M. Hoefer, M. Sommacal, and T. Silva, “Propagation and control of nanoscale magnetic-droplet solitons”, Phys. Rev. B, vol. 85, no. 214433, pp. 1–7, 2012. [100] D. Xiao, Y. Liu, Y. Zhou, S. M. Mohseni, S. Chung, and J. Åkerman, “Merging droplets in double nanocontact spin torque oscillators”, Phys. Rev. B - Condens. Matter Mater. Phys., vol. 93, no. 9, pp. 1–5, 2016. [101] S. M. Mohseni, S. R. Sani, J. Persson, T. N. A. Nguyen, S. Chung, Y. Pogoryelov, P. K. Muduli, E. Iacocca, A. Eklund, R. K. Dumas, S. Bonetti, A. Deac, M. A. Hoefer, and J. Åkerman, “Spin Torque- Generated Magnetic Droplet Solitons”, Science, vol. 339, no. 6125, pp. 1295–1298, 2013. [102] S. M. Mohseni, S. R. Sani, J. Persson, T. N. Anh Nguyen, S. Chung, Y. Pogoryelov, and J. Åkerman, “High frequency operation of a spin- torque oscillator at low field”, Phys. Status Solidi - Rapid Res. Lett., vol. 5, no. 12, pp. 432–434, 2011. [103] S. Chung, S. M. Mohseni, S. R. Sani, E. Iacocca, R. K. Dumas, T. N. Anh Nguyen, Y. Pogoryelov, P. K. Muduli, A. Eklund, M. Hoefer, and J. Åkerman, “Spin transfer torque generated magnetic droplet solitons (invited)”, J. Appl. Phys., vol. 115, no. 17, pp. 0–6, 2014. [104] S. Chung, S. Majid Mohseni, A. Eklund, P. Dürrenfeld, M. Ranjbar, S. R. Sani, T. N. Anh Nguyen, R. K. Dumas, and J. Åkerman, “Magnetic droplet solitons in orthogonal spin valves”, Low Temp. Phys., vol. 41, no. 10, pp. 833–837, 2015. [105] F. Macià, D. Backes, and A. D. Kent, “Stable Magnetic Droplet Solitons

Bibliography

in Spin Transfer Nanocontacts”, Nat. Nanotechnol., vol. 9, no. 12, pp. 1– 19, 2014. [106] S. Lendínez, N. Statuto, D. Backes, A. D. Kent, and F. Macià, “Observation of droplet soliton drift resonances in a spin-transfer- torque nanocontact to a ferromagnetic thin film”, Phys. Rev. B, vol. 92, no. 17, p. 174426, 2015. [107] S. Chung, A. Eklund, E. Iacocca, S. M. Mohseni, S. R. Sani, L. Bookman, M. A. Hoefer, R. K. Dumas, and J. Åkerman, “Magnetic droplet nucleation boundary in orthogonal spin-torque nano-oscillators”, Nat. Commun., vol. 7, p. 11209, 2016. [108] S. M. Mohseni, S. R. Sani, R. K. Dumas, J. Persson, T. N. Anh Nguyen, S. Chung, Y. Pogoryelov, P. K. Muduli, E. Iacocca, A. Eklund, and J. Åkerman, “Magnetic droplet solitons in orthogonal nano-contact spin torque oscillators”, Phys. B Condens. Matter, vol. 435, pp. 84–87, 2014. [109] S. Tacchi, T. N. A. Nguyen, G. Gubbiotti, M. Madami, G. Carlotti, M. G. Pini, A. Rettori, V. Fallahi, R. K. Dumas, and J. Åkerman, “[Co/Pd]- CoFeB exchange spring magnets with tunable gap of spin wave excitations”, J. Phys. D. Appl. Phys., vol. 47, no. 49, p. 495004, 2014. [110] T. N. Anh Nguyen, N. Benatmane, V. Fallahi, Y. Fang, S. M. Mohseni, R. K. Dumas, and J. Åkerman, “[Co/Pd]4–Co–Pd–NiFe spring magnets with highly tunable and uniform magnetization tilt angles”, J. Magn. Magn. Mater., vol. 324, pp. 3929–3932, 2012. [111] T. Nguyen, V. Fallahi, Q. Tuan Le, S. Chung, S. R. Sani, R. K. Dumas, C. W. Miller, and J. Åkerman, “Investigation of the Tunability of the Spin Configuration Inside Exchange Coupled Springs of Hard/Soft Magnets”, IEEE Trans. Magn., vol. 50, no. 6, p. 2004906, 2014. [112] C. Banerjee, S. Pal, M. Ahlberg, T. N. A. Nguyen, J. Åkerman, and A. Barman, “All-Optical Study of Tunable Ultrafast Spin Dynamics in [Co/Pd]-NiFe Systems: The Role of Spin-Twist Structure on Gilbert Damping”, RCS Adv., vol. 6, p. 80168, 2016. [113] Y. Zhou, C. L. Zha, S. Bonetti, J. Persson, and J. Åkerman, “Spin-torque oscillator with tilted fixed layer magnetization”, Appl. Phys. Lett., vol. 92, no. 26, p. 262508, 2008. [114] Y. Zhou, C. L. Zha, S. Bonetti, J. Persson, and J. Åkerman, “Microwave generation of tilted-polarizer spin torque oscillator”, J. Appl. Phys., vol. 105, no. 7, pp. 1–4, 2009. [115] Y. Zhou, H. Zhang, Y. Liu, and J. Åkerman, “Macrospin and micromagnetic studies of tilted polarizer spin-torque nano-oscillators”, J. Appl. Phys., vol. 112, no. 6, p. 063903, 2012. [116] C. L. Zha, R. K. Dumas, J. Persson, S. M. Mohseni, J. Nogues, and J. 79

Bibliography

Åkerman, “Pseudo Spin Valves Using a (1 1 2)-Textured D022 Mn2.3−2.4Ga Fixed Layer”, IEEE Magn. Lett., vol. 1, p. 2500104, 2010. [117] C. L. Zha, “Study of Pseudo Spin Valves Based on 1 0 (111)-Oriented FePt and FePtCu Fixed Layer With Tilted Magnetocrystalline Anisotropy”, IEEE Trans. Magn., vol. 45, no. 10, pp. 3491–3494, 2009. [118] C. L. Zha, J. Persson, S. Bonetti, Y. Y. Fang, and J. Åkerman, “Pseudo spin valves based on L 10 (111)-oriented FePt fixed layers with tilted anisotropy”, Appl. Phys. Lett., vol. 94, no. 16, pp. 0–3, 2009. [119] T. N. A. Nguyen, Y. Fang, V. Fallahi, N. Benatmane, S. M. Mohseni, R. K. Dumas, and J. Åkerman, “[Co/Pd]-NiFe exchange springs with tunable magnetization tilt angle”, Appl. Phys. Lett., vol. 98, no. 17, p. 172502, 2011. [120] C. Morrison, J. J. Miles, T. N. Anh Nguyen, Y. Fang, R. K. Dumas, J. Åkerman, and T. Thomson, “Exchange coupling in hybrid anisotropy magnetic multilayers quantified by vector magnetometry”, J. Appl. Phys., vol. 117, no. 17, p. 17B526, 2015. [121] S. Tacchi, T. N. A. Nguyen, G. Carlotti, G. Gubbiotti, M. Madami, R. K. Dumas, J. W. Lau, J. Åkerman, A. Rettori, and M. G. Pini, “Spin wave excitations in exchange-coupled [Co/Pd]-NiFe films with tunable tilting of the magnetization”, Phys. Rev. B - Condens. Matter Mater. Phys., vol. 87, no. 14, pp. 2–6, 2013. [122] T. N. A. Nguyen, R. Knut, V. Fallahi, S. Chung, Q. Tuan Le, S. M. Mohseni, O. Karis, S. Peredkov, R. K. Dumas, C. W. Miller, and J. Åkerman, “Depth-Dependent Magnetization Profiles of Hybrid Exchange Springs”, Phys. Rev. Appl., vol. 2, no. 4, p. 044014, 2014. [123] S. Mangin, D. Ravelosona, J. A. Katine, M. J. Carey, B. D. Terris, and E. E. Fullerton, “Current-induced magnetization reversal in nanopillars with perpendicular anisotropy”, Nat. Mater., vol. 5, no. 3, pp. 210–215, Feb. 2006. [124] G. Hu, T. Thomson, C. T. Rettner, S. Raoux, and B. D. Terris, “Magnetization reversal in CoPd nanostructures and films”, J. Appl. Phys., vol. 97, no. 10, pp. 18–21, 2005.

List of Figures and Table

Figure 2.1: Confocal sputtering technique ...... 6 Figure 2.2: Magenic hysteresis curves ...... 8 Figure 2.3: Fabrication process of spin valve mesas ...... 10 Figure 2.4: Oxidation and short-circuit protection of mesas ...... 10 Figure 2.5: Contact opening process ...... 11 Figure 2.6: 3D sketch of a spin-valve STO ...... 12 Figure 2.7: Scheme of the ST-FMR measurement setup ...... 14 Figure 4.1: Tunnel magnetoresistance ratio (TMR) ...... 24 Figure 4.2: p-MTJ wafer and device ...... 26 Figure 4.3: R–H OOP major and minor loops ...... 28 Figure 4.4: Resistance and relative angle of p-MTJ device ...... 29 Figure 4.5: Magnetization misalignment and dynamic modes in p-MTJ ..... 31 Figure 4.6: Parameters of magnetodynamic modes in p-MTJ ...... 32 Figure 4.7: MTJ device structure and microwave measurement setup ...... 33 Figure 4.8: Spectra generated by MTJ STOs ...... 34 Figure 4.9: Current-dependence of oscillation frequency ...... 35 Figure 4.10: Shift tendency of oscillation frequency ...... 36 Figure 4.11: Components of ST-FMR spectra ...... 38 Figure 4.12: FMR frequency of p-MTJ STOs ...... 39 Figure 4.13: Linewidth of FMR spectra ...... 40 Figure 4.14: Output voltage components from ST-FMR measurements ...... 41 Figure 4.15: Comparison of field-output component relations ...... 42 Figure 4.16: ST-FMR output components versus relative angle curves ...... 45 Figure 5.1: Mesa device of all-perpendicular PSV stack ...... 51 Figure 5.2: Field-sweep magnetoresistance ratio ...... 52 81

Figures and Table

Figure 5.3: Magnetoresistance and low-f noise of p-PSV STO ...... 54 Figure 5.4: Magnetoresistance mapping in p-PSV STO ...... 55 Figure 6.1: Tunable tilt angle of Co/Pd-NiFe structure...... 58 Figure 6.2: Tunable tilt angle of Co/Pd-Co-Ni-CoFeB structure ...... 60 Figure 6.3: Tilted polarizer PSV hysteresis loop and magnetoresistance .... 62 Figure 6.4: Resistance and microwave properties of tilted PSV STOs ...... 63 Figure 6.5: Map of the resistance states in the tilted PSV STO...... 65

Table 4.1: Bias dependence of torque conponents in p-MTJ STOs ...... 46

Appendix: Papers and Manuscripts