Study of nanostructured hard ferrites and their coupling with a soft layer

FACULTAD DE CIENCIAS Departamento de Física Aplicada Study of nanostructured hard ferrites and their coupling with a soft layer Memoria presentada para optar al grado de Doctora en Ciencias Físicas por Guiomar Delgado Soria Directores: Dr. Juan de la Figuera Bayón Dr. Adrián Quesada Michelena Madrid, Junio 2021 Acknowledgements Comenzando esta tesis doctoral quiero agradecer a todas las personas que me han acompañado a lo largo de este bonito camino. En primer lugar me gustaría dar las gracias a mis directores de tesis Juan y Adrián. Gracias por confar en mí para realizar esta investigación, ofrecerme vuestra ayuda incondicional, motivarme en los momentos más duros y sobre todo, por transmitirme tantos conocimientos. Ha sido un placer aprender de vosotros durante todos estos años. Quisiera expresar también mi gratitud a mi tutora Pilar por su apoyo y disponibilidad cada vez que lo he necesitado. En segundo lugar y no por ello menos importante, quisiera dar las gracias a Pepe. Ha sido un orgullo trabajar con un investigador que me ha aportado tanta sabiduría, conocimientos, templanza, valores, confanza... En defnitiva "mi padre científco". No quisiera olvidarme de Jose Emilio, quien además de enseñarme, siempre ha tenido buenas palabras hacia mí. Doy las gracias a todos los compañeros del grupo "Surfmoss" con los que he tenido la suerte de trabajar y con los que he compartido muy buenos ratos: María, Ania, Miguel, Eva, Edu, Victor, Fernando, Guille y Alejandro. En especial quiero destacar a Ania y a María, quienes comenzaron ayudándome y alegrando mi día a día en el laboratorio y acabaron formando parte de mi vida. Dentro del Instituto Rocasolano me gustaría agradecer al grupo de Láseres por su simpatía y colaboración con nuestro grupo así como a Juan Dávalos por su ayuda con el manejo del sputtering y su apreciado sentido del humor. Quisiera también mencionar a mis compañeros del Instituto Cerámica y Vidrio, Ceci y Jesús. Gracias por la predisposición siempre a colaborar con cualquier estudio y vuestra acogida en mis visitas por allí. A Cesar y Jorge del Instituto de Ciencias de Materiales y a Santiago del Instituto de la Materia por su valiosa contribución a esta investigación. A Sandra por nuestro viaje en coche a Barcelona y las horas de análisis en el Sincrotrón ALBA. En este punto quisiera destacar la inestimable ayuda de Lucia y Michael en la línea de luz CIRCE de ALBA, sin los cuales no hubiera sido posible la adquisición de muchos de los resultados de esta tesis. I want to thank all people who support me in my stay at Jerzy Haber Institute of catalysis and surface chemistry in Krakow. I felt really comfortable in your research group. Thank you again for bringing me the opportunity to amplify my goals in another country. i Acknowledgements Un agradecimiento muy especial para mí es el dirigido a mi familia. A mis padres, a mis hermanos y a mis primos. A todos vosotros muchísimas gracias por estar conmigo en cada etapa que he afrontado, reforzarme en mis decisiones y darme vuestro animo. Me siento muy afortunada de pertenecer a la familia que tengo. Doy las gracias a mis amigas que han estado junto a mí desde mi más tierna juventud: Mari Vega, Sandra, Cristina, Nerea, Elsa y Alba. Gracias a todas por brindarme con vuestra amistad, vuestras risas y comprensión y vuestra capacidad por hacer que cualquier momento sea inolvidable. Y también quisiera agradecer a aquellas amigas que conocí en la facultad y que desde entonces no se han separado de mi lado: Mónica, Marina e Isa. Gracias por tanto chicas. Y a ti Fran, gracias por vivir conmigo cada momento, haber compartido mis alegrías y haber sido mi inquebrantable apoyo ante las difcultades. Gracias por ser mi vaso medio lleno. ii Resumen En los últimos 100 años, los imanes permanentes han desempeñado un papel fundamental en el desarrollo de múltiples campos de innovación tecnológica. Dichos materiales han sido utilizados principalmente para su aplicación en motores y generadores ya que permiten transformar la energía eléctrica en mecánica y viceversa. Además otros usos a destacar son como medios de grabación, componentes de dispositivos de microondas, radiofrecuencias y magneto-ópticos. Sin embargo, en la actualidad, los mejores imanes permanentes están compuestos por una considerable proporción de tierras raras. Las tierras raras presentan dos grandes problemas: su extracción provoca una elevado daño para el medio ambiente y tanto dicha extracción como su separación es controlada por China. Para evitar estos inconvenientes, se están dedicando esfuerzos a desarrollar nuevos imanes permanentes que no contengan tierras raras. En este contexto, esta investigación estudia óxidos magnéticamente duros que puedan sustituir a los imanes que incluyen tierras raras. Concretamente, el oxido estudiado en profundidad en la tesis ha sido la hexaferrita de estroncio (SrFe12O19, SFO). Esta ferrita hexagonal se ha convertido desde su descubrimiento a mediados del siglo XX en un material de gran importancia comercial y tecnológica gracias primordialmente a su alta anisotropía magnetocristalina unido a su bajo coste. No obstante, la hexaferrita de estroncio pese a su alto campo coercitivo presenta unos valores de imanación remanente moderados, lo que provoca unos valores del producto energético por debajo de los alcanzados en imanes permanentes con tierras raras. Una estrategia para mejorar las propiedades magnéticas de este material será su acoplamiento con un material magnéticamente blando. Esta combinación posibilita, en las condiciones apropiadas, que el blando aumente la imanación sin disminuir de forma signifcativa el campo coercitivo aportado por el material magnéticamente duro. De esta forma se obtiene un mayor producto energético. Por lo tanto, un punto importante en esta investigación será comprender el acoplamiento magnético en la interfaz de dos materiales con coercitividades sustancialmente diferentes. Este es un problema científco continuo y sutil subyacente al desarrollo de futuros dispositivos espintrónicos/nanomagnéticos e imanes permanentes avanzados. Es importante señalar que estos materiales constituyen, además, sistemas muy interesantes para comprender la inversión de imanación colectiva e individual. Su comportamiento colectivo depende de las propiedades magnéticas de las capas individuales, así como de las interacciones dominantes entre ellas: acoplamiento de intercambio directo y/o interacciones magnetostáticas. La primera etapa de la tesis se centra en entender las propiedades estructurales y magnéticas de SrFe12O19 y expone la caracterización de plaquetas de este compuesto mediante distintas técnicas microscópicas y espectroscópicas. Un resultado novedoso de esta sección fue obtener por primera vez su espectro de absorción de rayos X. Buscando mejorar sus propiedades magnéticas, se investigó su acoplamiento magnético con un material magnéticamente blando (cobalto) crecido por epitaxia de haces moleculares (MBE). Dicho estudio fue llevado a cabo en el microscopio de fotoemisión de electrones (PEEM) por medio de iii Resumen dicroismo magnético circular de rayos X (XMCD) en el sincrotrón ALBA. Esta técnica permitió la determinación de los dominios magnéticos de cada capa. Paralelamente, simulaciones de micromagnetismo fueron realizadas para entender el comportamiento magnético observado en los resultados experimentales. Los análisis evidenciaron plaquetas de SFO de cientos de nanómetros con imanación prefencial perpendicular al plano. El acoplamiento en el sistema plaqueta-metal reveló una falta de acoplo magnético proveniente de la competición entre la anisotropía magnetocristalina de la plaqueta con la anisotropía de forma de la capa de cobalto. Para evitar dicha competición y promover el acoplamiento magnético entre ambos compuestos, la segunda etapa de la tesis consiste en el crecimiento de láminas delgadas de SFO por pulverización catódica con la orientación magnética preferencialmente en el plano para la posterior deposición del metal por MBE. Inicialmente se comprobó el efecto del calentamiento en la formación de la fase cristalina de las láminas delgadas y se determinaron los parámetros involucrados en modifcar el eje de fácil imanación de cada muestra. Para ello se estudió la composición, estructura y magnetismo de estas láminas en base a distintas técnicas de caracterización como la difracción de rayos X, espectroscopía Raman y Mössbauer. Nuevamente el estudio del acoplamiento magnético con la capa magnéticamente blanda fue analizado por PEEM-XMCD. En este experimento se apreció un acoplo estructural en el sistema bicapa. Completando esta investigación, se ha estudiado la ferrita de cobalto (CoFe2O4, CFO) debido a sus notables propiedades como una alta constante de anisotropía magnetocristalina y una gran constante de magnetostricción. Al igual que la ferrita anteriormente comentada, este compuesto es magnéticamente duro. En esta tercera sección se han presentado y discutido láminas delgadas de CFO crecidas mediante MBE. Estas muestras se han caracterizado en un sistema de ultra-alto vacío con técnicas de microscopía y espectroscopía in-situ con especial énfasis en la discusión de la espectroscopia Mössbauer. La variación en el espesor y el calentamiento con y sin oxígeno promueve cambios en la estequiometría de la fase crecida y en consecuencia en sus propiedades. Finalmente, para comprender el origen del acoplamiento tipo “muelle”, observado en un sistema experimental formado por una lámina fna de CoFe2O4 y una capa de aleación Fe-Co, se llevaron a cabo simulaciones micromagnéticas. Estas simulaciones apoyaron como mecanismo dominante en el comportamiento magnético de la bicapa, la propagación de paredes de dominios en la fase blanda en contraste con lo predicho por los modelos teóricos, los cuales no tienen en cuenta este efecto. iv Abstract In the last 100 years, permanent magnets have played a key role in advancing many felds of technological innovation. These materials have been used mainly for their application in motors and generators since they allow the transformation of electrical energy into mechanical energy and vice versa. Other uses include recording media, microwave, radiofrequency and magneto-optical device components. However, at present, the best permanent magnets are composed of a considerable proportion of rare earths. Rare earths present two significant problems: their extraction causes great environmental damage, and China controls both extraction and separation. To avoid these handicaps, e˙orts are devoted to develop new permanent magnets that do not contain rare earths. In this context, these research studies magnetic materials based on hard oxides that can replace magnets that include rare earths. Specifcally, the oxide studied in depth in the thesis has been strontium hexaferrite (SrFe12O19, SFO). Since its discovery in the mid-twentieth century, this hexagonal ferrite has become a material of great commercial and technological importance thanks primarily to its high magnetocrystalline anisotropy coupled with its low cost. However, despite its high coercive feld, strontium hexaferrite presents a moderate remanent magnetization, which results in energy product values below those achieved in rare earth permanent magnets. A strategy to improve the material magnetic properties would be coupling it with a magnetically soft material. Under the appropriate conditions, this combination allows the soft material to increase the magnetization of the system without signifcantly reducing the coercive feld provided by the magnetically hard material. Thus, a higher energy product is obtained. Hence, an important point in this research will be to understand the magnetic coupling at the interphase of two materials with substantially di˙erent coercivities. This is an ongoing and subtle scientifc problem underlying the development of future spintronic/nanomagnetic devices and advanced permanent magnets. Importantly, these materials are also very interesting systems for understanding collective and individual magnetization reversal. Their collective behaviour depends on the magnetic properties of the individual layers and the dominant interactions between them: direct exchange coupling and/or magnetostatic interactions. The frst stage of the thesis focuses on understanding the structural and magnetic properties of SrFe12O19 and exposes the characterization of this compound in platelets form by di˙erent microscopic and spectroscopic techniques. A novel result of this section was to obtain for the frst time its X-ray absorption spectrum. The magnetic coupling with a magnetically soft material (cobalt) grown by molecular beam epitaxy (MBE) was investigated. This study was carried out in the photoemission electron microscope (PEEM) by means of X-ray circular magnetic dichroism (XMCD) on the ALBA synchrotron. This technique allowed the determination of the magnetic domains of each layer. In parallel, micromagnetism simulations were performed to understand the magnetic behavior observed in the experimental results. The analysis revealed SFO platelets of hundreds of nanometers in size with magnetization preferential normal to the platelet plane. The platelet-metal system v Abstract revealed a lack of magnetic coupling due to competition between the magnetocrystalline anisotropy of the platelet with the shape anisotropy of the cobalt layer. To avoid such competition and promote magnetic coupling between the two compounds, the second stage of the thesis consists of growing SFO thin flms by sputtering with the magnetic orientation preferentially in-plane for subsequent metal deposition by MBE. Initially, the annealing e˙ect on the formation of the crystalline phase of the thin flms was tested, and the parameters involved in modifying the easy axis magnetization of each sample were determined. The composition, structure and magnetism of these flms were studied using di˙erent characterization techniques such as X-ray di˙raction, Raman and Mössbauer spectroscopy. Again the study of the magnetic coupling with the magnetically soft layer was analyzed by PEEM-XMCD. In this experiment, a structural coupling was observed in the bilayer system. Completing this research, cobalt ferrite (CoFe2O4, CFO) has been studied due to its remarkable properties such as a high magnetocrystalline anisotropy and large magnetostriction constant. Like the ferrite discussed above, this compound is magnetically hard and is used in permanent magnet applications. In this third section, MBE-grown CFO thin flms have been presented and discussed. These samples have been characterized in an ultra-high vacuum system with microscopy and in-situ spectroscopy techniques with particular emphasis on the discussion of Mössbauer spectroscopy. The variation in thickness and heating with and without oxygen promotes changes in the stoichiometry of the grown phase and consequently to its properties. Finally, to comprehend the origin of the spring-magnet coupling observed in an experimental system consisting of a CoFe2O4 thin flm and a iron-cobalt alloy layer, micromagnetic simulations were carried out. These simulations supported the propagation of domain walls in the soft phase as the dominant mechanism for the bilayer magnetic behaviour of the bilayer, in contrast to theoretical models, which do not take into account this e˙ect. vi Contents Acknowledgements i Resumen iii Abstract v List of Figures ix List of Tables xiii 1 Preface 1 2 Experimental details 5 2.1 Depositiontechniques .............................. 5 2.1.1 Radio-frequency Magnetron Sputtering . . . . . . . . . . . . . . . . . 5 2.1.2 Molecular Beam Epitaxy . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 Characterization methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2.1 Mössbauer spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.2 X-ray absorption spectroscopy (XAS) . . . . . . . . . . . . . . . . . 19 2.2.3 X-ray magnetic circular dichroism (XMCD) . . . . . . . . . . . . . . 21 2.2.4 LEEM/PEEM microscope . . . . . . . . . . . . . . . . . . . . . . . . 24 2.3 Micromagnetic simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3 SrFe12O19 29 3.1 Strontium ferrite structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.2 Magnetic order in strontium ferrite . . . . . . . . . . . . . . . . . . . . . . . 31 3.2.1 Intrinsic magnetic properties . . . . . . . . . . . . . . . . . . . . . . 35 4 SrFe12O19 platelets 39 4.1 Introduction.................................... 39 4.2 Synthesis by hydrothermal method . . . . . . . . . . . . . . . . . . . . . . . 39 4.3 Morphological, structural and compositional characterization . . . . . . . . 40 4.4 Magnetic characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.5 Conclusions.................................... 57 5 Magnetic interactions in magnetic single domain platelet with cobalt overlayer 59 5.1 Introduction.................................... 59 5.2 Cobalt....................................... 60 5.3 XAS-XMCD characterization from a single domain platelet . . . . . . . . . 60 5.4 Growth and characterization of cobalt overlayer . . . . . . . . . . . . . . . . 62 5.5 Magnetic domains in the bilayer system . . . . . . . . . . . . . . . . . . . . 63 5.6 Energetic contributions to the magnetic response from cobalt domains . . . 64 vii Contents 5.7 Micromagnetic simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 5.8 Conclusions.................................... 69 6 SrFe12O19 thin flms 71 6.1 Introduction.................................... 71 6.2 SFOtarget .................................... 72 6.3 Annealing e˙ect in SFO thin flms formation . . . . . . . . . . . . . . . . . . 73 6.3.1 Compositional and structural characterization . . . . . . . . . . . . . 74 6.3.2 Magnetic characterization . . . . . . . . . . . . . . . . . . . . . . . . 77 6.4 Thinflms..................................... 79 6.4.1 Characterization of crystallinity and composition . . . . . . . . . . . 80 6.4.2 Magnetic characterization . . . . . . . . . . . . . . . . . . . . . . . . 82 6.5 Infuence of thickness and power sputtering on the magnetic behaviour of thinflms ..................................... 85 6.6 Interaction between strontium hexaferrite thin flm with cobalt overlayer . . 88 6.7 Conclusions.................................... 94 7 CoFe2O4 ultra-thin flms 95 7.1 Introduction.................................... 95 7.2 Cobalt ferrite: Structure and magnetic properties . . . . . . . . . . . . . . . 96 7.3 Growth of the cobalt ferrite thin flms . . . . . . . . . . . . . . . . . . . . . 98 7.4 20nmthinflm.................................. 99 7.4.1 Compositional, structural and morphological characterization . . . . 99 7.4.2 Magnetic characterization . . . . . . . . . . . . . . . . . . . . . . . . 102 7.5 5nmthinflm...................................107 7.5.1 Compositional, structural and morphological characterization . . . . 107 7.5.2 Magnetic characterization . . . . . . . . . . . . . . . . . . . . . . . . 109 7.6 Conclusions....................................113 8 Magnetic interactions in CoFe2O4/FeCo bilayer thin flms 115 8.1 Introduction....................................115 8.2 Experimental background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 8.3 Micromagnetic simulations of CoFe2O4/FeCo system . . . . . . . . . . . . . 119 8.3.1 Exchange sti˙ness of soft layer . . . . . . . . . . . . . . . . . . . . . 119 8.3.2 Interlayer exchange coupling .......................120 8.3.3 Saturation magnetization of soft layer . . . . . . . . . . . . . . . . . 121 8.3.4 Simulating multiand single-domain confgurations . . . . . . . . . . 122 8.4 Conclusions....................................124 9 General conclusions 125 10 Conclusiones generales 129 A List of acronyms and abbreviations 133 B List of publications 137 Bibliography 139 viii Hard magnetic material Soft magnetic material Coupled bilayer system M H Figure 1.2: Magnetization curve vs applied magnetic feld (“hysteresis loop”) corresponding to a hard magnetic material (black curve), soft magnetic material (red curve) and rigid coupling between both (blue curve). In the rigid coupling regime, both layers are fully exchange-coupled, and the spins of the magnetically soft layer are aligned with those of the hard layer. The magnetization reversal co-occurs for the whole system. Bilayer magnetic properties arise from averaging the magnetic properties provided by both materials. The regime of a partial coupling or spring-magnet refers to a rigid exchange-coupling of the soft phase spins with those of the hard one at the interface, but those soft spins that are further from the interface are not coupled. This causes that for small magnetic felds, the uncoupled spins of the soft layer reverse their magnetization, while the spins that are coupled to those of the hard layer will need greater magnetic felds for reversing them, fgure 1.3. This type of coupling also implies an improvement in the bilayer’s magnetic properties since the partially coupled soft phase enables an increase in the system magnetization in the remanent state. Figure 1.3: Illustration of an exchange-spring state in a hard-magnetic/soft-magnetic bilayer. Image modifed from Ref. [30]. 3 1 Preface In the third regime, we consider that the bilayer is not exchange-coupled. In this case, the magnetization reversal is independent in each layer. However, dipole interactions might promote the alignment of the spins between both layers and, thus, might improve the system magnetic properties. Therefore, in the context of the use of ferrites as an alternative to permanent magnets composed of rare earths as well as the magnetization improvement of these oxides by coupling magnetically soft layers, this thesis tries to address the following points: •To determine the structural and magnetic properties of several hard ferrites (strontium hexaferrite and cobalt ferrite) in platelets and thin flms form. •To explore the magnetic nature of the coupling established between a hard ferrite layer and magnetically soft layer (cobalt and iron-cobalt). To achieve these objectives, I present below the outline followed during the thesis. Chapter 2. This chapter includes the growth and characterization techniques used through the thesis with a description of the systems where these techniques were employed. Chapter 3. Here, the structure of strontium hexaferrite and its magnetic properties are discussed to provide context for the following chapters. Chapter 4. This part is dedicated to the characterization of the structural and magnetic properties of SFO platelets grown through hydrothermal synthesis. Di˙erent microscopic and spectroscopic techniques with emphasis on Mössbauer spectroscopy and X-ray soft absorption were used. Chapter 5. This chapter focuses on the growth of a soft cobalt layer on the SFO platelets and their magnetic interaction. The study was carried out by X-ray absorption techniques together with micromagnetic simulations. Chapter 6. In this chapter, strontium hexaferrite thin flms are grown on Si(100) by radio frequency magnetron sputtering followed by annealing. The growth is optimized to provide for in-plane magnetization in the flms. Finally, the coupling with a magnetic cobalt layer is studied. Chapter 7. Here, cobalt ferrite ultra-thin flms on Pt(111) are grown by molecular beam epitaxy. The characterization of the flms was carried out in-situ by several methods, including Mössbauer spectroscopy. Chapter 8. In this part, micromagnetic simulations are performed to understand the experiments on the coupling between a cobalt ferrite thin flm with a thin cobalt-iron layer. Chapter 9. This chapter summarizes the main results obtained in this thesis. 4 2 Experimental details This chapter presents the deposition techniques employed to grow the oxide flms and metal layers studies in this thesis, followed by some of the most relevant characterization techniques applied to them. These growth and characterization techniques were carried out mainly in the laboratory "Ramón-Gancedo" at the Instituto de Química Física "Rocasolano" (CSIC), located in Madrid (Spain) and the LEEM-PEEM system at the CIRCE beamline of the ALBA Synchrotron Light Facility, located in Barcelona (Spain). Additionally, the cobalt ferrite ultrathin flms growth and characterization experiments were performed at the Jerzy Haber Institute of Catalysis and Surface Chemistry located in Krakow (Poland). The sample preparation for each experiment has been described in each corresponding chapter. Further, micromagnetics simulations have been performed to understand the magnetic behaviour of the experimental systems. 2.1 Deposition techniques The deposition techniques used in this research have been radio-frequency magnetron sputtering and molecular beam epitaxy. Both have been used to grow oxide or metal flms on substrates or flms. These techniques are included in the so-called physical vapour deposition processes (PVD). PVD comprises atomistic deposition processes in which a material of interest in solid form evaporates to atoms and is transported in the form of vapour through a vacuum or low-pressure gaseous (or plasma) environment to the substrate, where it condenses [36]. 2.1.1 Radio-frequency Magnetron Sputtering The radio-frequency (RF) magnetron sputtering technique is a type of deposition technique widely used for the growth of thin flms, coatings, and multilayers [37, 38, 39]. This technique is performed in a high vacuum chamber. The growth is based on stripping atoms from a target material by charged particles bombardment and the subsequent deposition of the ejected atoms on a substrate. The experimental system consists of vacuum pumps (rotary and turbomolecular pumps), vacuum chamber, gas trigger/gas supply subsystem, cathode with a built-in magnetron, target, a power source and substrate. The deposition process is described as follow: Initially, we must start from a vacuum in the chamber. The vacuum pumps reach a pressure of 10−6 mbar. Next, gas is introduced into the chamber. Typically, the gasses used are inert as argon since they do not react chemically with the deposition material and thus 5 2 Experimental details do not a˙ect the deposited flm’s composition. This is called non-reactive sputtering. The material’s target to grow is located in the cathode, where a negative electronic potential is applied, causing a plasma or glow discharge. This creates positive ions in the gas, which sputter the surface of the flm negatively biased. Due to the exchange of momentum between the gas ions and the atoms on the target surface, the latter’s extraction occurs. A magnet (magnetron) behind the cathode creates a transversal magnetic feld that traps the secondary electrons generated in the target to avoid possible e˙ects to the substrate as an increase of temperature or damage radiation. The magnetron also allows for faster deposition rates because an increase of negative charge in the target promotes the collision with the gas atoms. Finally, the atoms that arise from the target are deposited on the substrate located in the anode forming a flm. Radiofrequency in sputtering is used as the source of electrical potential, mainly for the growth of insulating materials and oxides. Unlike the growth using direct current (DC), RF allows to avoid the charge build-up on the target surface and continue the process for this type of materials [40, 41]. A schematic cycle illustration of the process is shown in fgure 2.1. Cathode Anode Ar flow Power Supplied - RF Vacuum Atoms from target material Ar neutralized Ar ionized Plasma Target Thin film Substrate Magnetron Figure 2.1: Diagram of the RF magnetron sputtering process. Furthermore, in other cases, the deposition process can be carried out in the presence of gases such as N2 and O2 either in place of or in addition to argon or other inert gas. The non-inert gas can interact chemically with the target giving rise to other subspecies such as nitrides and oxides in the deposited samples (reactive sputtering). Commonly, a reactive gas is usually combined with inert gas to obtain a given stoichiometry in the case of oxides. 6 2.1 Deposition techniques It also infuences the characteristics of the deposited flm, such as providing a preferential orientation. Many sputtering systems allow the substrate to be heated to facilitate the di˙usion of the deposited atoms and promote the crystalline phase’s growth. RF magnetron sputtering deposition was used for the growth of thin flms of SFO. It is important to mention that although a target with the stoichiometric composition (SrFe12O19) was used, the deposited thin flms do not present the same stoichiometry as the target. In our case, to achieve thin flms of SFO, a post-annealing step in air was required. In RF magnetron sputtering deposition, the deposited thin flm is infuenced by the deposition parameters such as oxygen fow ratio, sputtering power, base pressure, working pressure, deposition time, substrate-target distance, and substrate temperature [42]. In chapter 6, we will observe how the change in the sputtering power and post-annealing treatment a˙ects the magnetic and structural properties of the strontium hexaferrite thin flms, keeping the rest of the parameters fxed. The deposition rate for this method changes according to the values set in each parameter involved. As an example, for the SFO thin flm grown with sputtering power of 140 W at room temperature, with a base pressure of 1× 10−6 mbar, working pressure of 7 × 10−3 mbar, a target-substrate distance 60 mm, Ar/O2 ratio of 2%, the deposition rate is 5.3 nm per minute. The magnetron sputtering system used in the laboratory is shown in fgure 2.2. To achieve the vacuum conditions, a rotary device is required to reach a pressure of 10−2 mbar, and a turbopump then works to reach 10−6 mbar. The argon and oxygen fow are computercontrolled, and the source settings are programmed in an 1500 W power supply. Figure 2.2: Magnetron sputtering system. 7 2 Experimental details 2.1.2 Molecular Beam Epitaxy Molecular beam epitaxy (MBE) is a technique that consists of the evaporation of atoms of a certain element (cobalt, nickel, iron) that are deposited onto a substrate. In many cases, the substrate’s sample holder allows heating to increase the di˙usion of the atoms on the substrate surface. The deposition process begins with the heating of a tungsten (W) flament, in which electrons are generated. Applying a high voltage (HV), these electrons from the W flament are accelerated against a rod of the desired metal (electron bombardment). This produces a heating of the rod, causing the evaporation of the atoms. Finally, the atoms encounter a substrate on which they are deposited [43, 44]. Molecular beam Substrate Thin film UHV chamber Metal rods Shutter W filament Figure 2.3: Diagram of the molecular beam epitaxy. MBE technology is used for the growth of single-crystal thin flms, quantum wells, superlattices and similar structures. This is a deposition method that allows a precise growth of the samples and that in addition to producing high-quality layers, good control of the thickness, doping, and composition of the samples is achieved [45, 41]. In order to deposit a particular element, the deposition rate of such element is previously calibrated by adjusting the heating power. It should be noted that this method is carried out in an ultra-high vacuum to avoid contamination from residual gas in the growth chamber. This deposition technique was applied for the experiments of cobalt metal overlayers on oxide SFO flms (Chapter 5 and 6). For this purpose, an evaporator with a cobalt metallic rod was used and the settings established were HV = 2 kV, (IF il) = 2.45 A and (Iemiss) = 14 mA. These parameters values typically give one monolayer’s growth rate per four minutes (0.06 nm/min). 8 2.2 Characterization methods In comparison with RF magnetron sputtering, the deposition rates for this technique are much lower. This provides precise control of the growth process by MBE. Moreover, we are able to observe using the LEEM/PEEM microscope at CIRCE beamline in ALBA synchrotron how the metal flm grows, while in the RF magnetron sputtering system, it is not possible to monitor the growth. Also, another di˙erence between both techniques is that the previous one is performed in a high vacuum (pressure close to 10−6) while molecular beam epitaxy is carried out in an ultra-high vacuum (pressure close to 10−9). Furthermore, the MBE technique can be used to grow oxides by depositing the metal atoms in an oxidizing background gas. This combination is called oxygen assisted-MBE (O-MBE). O-MBE will be used in cobalt ferrite growth (chapter 7), where the typical oxygen pressure is 10−6 mbar. The molecular beam epitaxy technique was carried out in the ultra-high vacuum system of the CIRCE light line, ALBA Synchrotron, and the Jerzy Haber institute’s ultra-high vacuum system. 2.2 Characterization methods Throughout the thesis, di˙erent analysis techniques have been applied that have helped characterize the samples under study, determining their morphology, chemical composition, structural and magnetic properties. All characterization techniques used in this research, the location where they were performed and the personnel involved in each of them are described below: •Transmission electron microscopy (TEM) [46] was performed at Josef Stephan Institute (Ljubljana, Slovenia), at the latter location in collaboration with Dra. Petra Jenus. •Transmission X-ray Microscopy (TXM) [47] was performed at the MISTRAL beamline of the ALBA Synchrotron Light Facility in collaboration with Dra. Eva Pereiro. •X-ray di˙raction (XRD) [48] was performed at Instituto de Ciencia y Tecnología de Polímeros (CSIC) in collaboration with Pedro Gonzalez. •X-ray photoelectron spectroscopy (XPS) [49] was performed at the Instituto de Química Física "Rocasolano" (CSIC). •Mössbauer spectroscopy was performed at the Instituto de Química Física "Rocasolano" (CSIC) and Jerzy Haber Institute of Catalysis and Surface Chemistry (Krakow, Poland), at the latter location in collaboration with Dra. Nika Spiridis and Dr. Jozef Korecki. •Vibrating-sample magnetometer (VSM) [50, 51] was performed at the Instituto de Química Física "Rocasolano" (CSIC) and at Josef Stephan Institute (Ljubljana, Slovenia), in collaboration with Dra. Petra Jenus. 9 2 Experimental details •X-ray absorption spectroscopy (XAS), X-ray magnetic circular dichroism (XMCD) and Photoemission electron microscopy (PEEM) were performed at the CIRCE beamline of the ALBA Synchrotron Light Facility in collaboration with Dra. Lucia Aballe and Dr. Michael Foerster. •Rutherford backscattering spectrometry (RBS) [52] was performed at the Centre for Micro Analysis of Materials (CMAM, Universidad autónoma de Madrid) in collaboration with Patricia Galán. •Raman spectroscopy [53, 54] was performed at Instituto de Estructura de la Materia (CSIC) in collaboration with Dr. Santiago Sánchez. •Atomic force microscopy (AFM) [55, 56] was performed at the Instituto de Química Física "Rocasolano" (CSIC). •Auger spectroscopy (AES) [57] was performed at Jerzy Haber Institute of Catalysis and Surface Chemistry (Krakow, Poland) in collaboration with Dra. Nika Spiridis and Dr. Jozef Korecki. •Low-energy electron di˙raction (LEED) [58] was performed at Jerzy Haber Institute of Catalysis and Surface Chemistry (Krakow, Poland) in collaboration with Dra. Nika Spiridis and Dr. Jozef Korecki. •Scanning Tunneling Microscopy (STM) [59, 52] was performed at Jerzy Haber Institute of Catalysis and Surface Chemistry (Krakow, Poland) in collaboration with Dra. Nika Spiridis and Dr. Jozef Korecki. In particular, the following section will explain in detail the characterization techniques most relevant in this thesis: Mössbauer spectroscopy, XAS, XMCD and PEEM. As we will see in later chapters, these techniques have been crucial in understanding the results obtained in this thesis. 10 2.2 Characterization methods 2.2.1 Mössbauer spectroscopy Mössbauer spectroscopy is a powerful technique that has been used throughout this research for the characterization of both SFO platelets and thin flms of cobalt ferrite and SFO. The main reason is that it allows us to characterize iron compounds providing information about the chemical, structural and magnetic state of the iron cations in a particular phase. This spectroscopy is based on the Mössbauer e˙ect, that is, on recoilless resonant absorption of a γ quanta by atomic nuclei [60, 61, 62, 63, 64]. Let us assume that a nucleus in an excited state emits a γ ray to return to its ground state. If the nucleus is isolated, it will recoil to maintain the momentum and energy conservation, therefore the gamma ray will have an energy given by: E = Eγ − ER (2.1) where Eγ is the energy of the nuclear transition and ER is the recoil energy, E2 γ ER = (2.2) 2Mc2 being Mis the mass of the nucleus and cthe speed of light. If in the vicinity of that nucleus there is a second isolated one of the same type, the nuclear resonant absorption will not occur since (i) the emitted γ ray will not have the required energy and (ii) this absorber nucleus will also recoil. Thus, in the case of isolated nuclei, the emission and absorption bands are separated by twice the recoil energy, and the overlap of the bands is not possible. In the case of 57Fe, the recoil energy for the 14.4 keV nuclear transition is about 106 times higher than the linewidth of the emission and absorption bands what is really a huge amount. However, if the nuclei belong to solids, the recoil energy will be shared by all the other atoms in the crystal, and the recoil energy will be negligible, providing the emission of the gamma ray will not a˙ect the vibrational state of the atoms in the lattice signifcantly. The fraction of γ rays that are emitted and absorbed without energy loss by nuclear recoil is called recoilless free fraction (or f factor). This factor that defnes the probability of observing a Mössbauer event can be written, within the Debye model of the solid, can be written as: ( " #)  2 3 ER 2π2 T f = exp − 1 + (2.3) 2 KB θB 3 θB where θB is Debye temperature, Tcorresponds to sample temperature and KB is the Boltzmann constant. 11 2 Experimental details Therefore, having a large fvalue implies having moderate energy of the nuclear transition and a large Debye temperature compared with the temperature of the nucleus. Additionally, a favourable isotopic abundance is also needed. As a consequence of this, only a few transitions of a small number of isotopes are appropriate to observe the Mössbauer e˙ect. In practical terms, observing the Mössbauer e˙ect at room temperature is only feasible for 57Fe, 119Sn, 121Sb and 151Eu. As far as this work is concerned, we will consider in the following only the 14.4 keV transition of 57Fe. The parent radioactive nucleus of 57Fe is 57Co (fgure 2.4). 57Co decays by electron capture into a metastable 57Fe state (I = 5/2). As the deexcitation scheme indicates, the I = 5/2 excited state can decay directly to the ground state (I = 1/2) by emitting a γ quantum of 136.3 KeV or indirectly through an intermediate excited state (I = 3/2) by emitting a γ quantum of 122 KeV that in its turn decays to the ground state through the emission of γ radiation of 14.4 KeV which is the adequate to observe Mössbauer e˙ect. 57Co electron capture Excited state Ground state Excited state Ground state Source Sample g X-ray g Auger and internal conversion electrons 57Fe 136.3 keV 5/2 3/2 1/2 14.4 keV Figure 2.4: Symbolic illustration of the source (decay of 57Co to 57Fe) and the absorber in a Mössbauer experiment. However, even if the conditions mentioned above to observe the Mössbauer e˙ect are fulflled, this does not guarantee its occurrence. The reason is that the interaction of the atoms in the solid with its environment (hyperfne interactions) modifes the nuclear levels. Therefore if the Mössbauer nuclei in the emitter and the absorber are not in identical environments, the resonance will be destroyed. In Mössbauer spectroscopy, in order to reestablish the resonant absorption, the source is moved with respect to the absorber, which remains stationary. In this way, the frequency (energy) of the emitted photons is modifed by the Doppler e˙ect to match the di˙erence in energy of the nuclear levels in the emitter and the absorber and achieve the nuclear resonant absorption. Therefore, a Mössbauer spectrum is the representation of the number of counts collected in a detector as a function of the velocity applied to the emitter (source) with respect to the absorber (sample). Mössbauer spectroscopy allows the quantifcation of the hyperfne interactions through the determination of several hyperfne parameters. The main hyperfne interactions that can be observed in the Mössbauer spectrum are the electric monopole, the electric quadrupole 12 2.2 Characterization methods 2.2.2 X-ray absorption spectroscopy (XAS) This spectroscopy allows obtaining information about the crystal structure and chemical composition of the sample. This technique has been used to characterize SFO platelets, thin flms and the metallic layers on top of them (chapter 4, 5 and 6). When an atom absorbs an X-ray photon, an electron from the inner atomic shell (K, L or M) absorbs the photon’s energy and is excited to an unoccupied valence state. This electron leaves a hole in the inner shell that is flled with another electron from another upper shell, which can be de-excited by photon emission or by the escape of an electron of the outer shell. This electron is called Auger. Figure 2.9 depicts the process. On the way of the Auger electron to the surface, it causes a cascade of secondary electrons to escape from the sample [70]. K L MNucleus Auger electron Photoelectron X-ray photon X-ray photon Figure 2.9: Interaction process when a photon illuminates an atom and the processes it triggers. Thus, the spectrum resulting from the technique corresponds to the photons or electrons emitted from the sample when the X-ray beam interacts with the matter vs the incident beam energy. This spectrum identifes the electronic transitions of a particular element. The electronic transitions provide information about the atom, such as its oxidation state or its crystalline feld. Typically, the method used to measure X-ray absorption is the total electron yield detection (TEY). This mode is based on collecting all the electrons that escape from the sample, both those produced by photoemission as well as Auger electrons and secondary ones being these latter electrons that contribute the most to the signal [71]. It should be noted that for an electron spectroscopy to work, the electrons have to exit the sample. The average distance an electron can travel in a solid between inelastic collisions is called the inelastic mean free path of electrons (IMFP). The "universal curve" represents the trend of IMPF in function of the energy for many elements (fgure 2.10) [72]. The XAS TEY signal corresponds mostly to the secondary electrons, which correspond to those 19 2 Experimental details with low kinetic energy. So, this technique supports the sample’s characterization with high surface sensitivity (order to nanometers of depth), although this surface sensitivity is typically lower than for XPS. Electron Kinetic Energy (eV) Mean Free Path (nm) 10 5 1 0.5 0.3 2 5 10 50 100 500 1000 2000 Au Au Au Au Au Au Au Al Ag Ag Ag Ag Ag Ag Ag Ag Ag Ag Mo Mo Mo W Be CC Be C W P Be Be Ni Be Fe Be 10 5 1 0.5 0.3 2 5 10 50 100 500 1000 2000 Au Au Au Au Au Au Au Al Ag Ag Ag Ag Ag Ag Ag Ag Ag Ag Mo Mo Mo W Be CC Be C W P Be Be Ni Be Fe Be Secondary electrons (XAS analysis) Photoelectrons Figure 2.10: Universal curve plot of electron ineslatic mean free path of various elements. Reprinted from Ref [72]. Specifcally, in this thesis, we have studied XAS spectra at the L2,3 edge transitions to characterize the iron and cobalt atoms’ properties in our samples. These transitions correspond to electrons excited from the 2p core levels to the unoccupied 3d electronic states (holes). Due to the spin-orbit splitting produced in the 2p core shells, the spectrum recorded shows two characteristic absorption peaks, L3 and L2, which correspond to the initial states 2p3/2 and 2p1/2 core shells electrons, respectively. In addition, the XAS spectrum intensity is directly correlated with the number of holes in the 3d core shell. Figure 2.11 shows typical XAS spectra for iron at the edge L2,3 in di˙erent compounds. Figure 2.11: XAS spectra at the Fe L2,3 edges of metal iron and iron oxides reference compounds. Extracted from Ref. [73] 20 2.2 Characterization methods 2.2.3 X-ray magnetic circular dichroism (XMCD) X-ray magnetic circular dichroism (XMCD) is the di˙erence in absorption that a ferro or ferrimagnetic material undergoes when irradiated with X-rays with leftand rightcircularly-polarized light [70, 74]. This absorption provides detailed information on the electronic and magnetic structure of an element. For example, the oxidation state, the crystal symmetry and the spin and orbital magnetic moments can be determined. As we will see throughout the thesis, the elements characterized by XMCD have been iron and cobalt. It is well known that the magnetic properties of 3d transition metals are mainly determined by their "d" valence electrons [75, 76]. The spin magnetic moment arises from the di˙erence between the number of spin-up and spin-down electrons. This imbalance is equivalent to the di˙erence in the spin-up and spin-down holes states. Considering that the intensity of the XAS spectra is proportional to the 3d holes, the magnetic moment of spin can be obtained as a result of the XAS spectra with di˙erent polarizations. This is the principle of XMCD phenomenon, which can explain as: 1. We have an atom of a transition metal whose 2p core-shell is divided into two levels, j = 3/2 (L3 edge) and j = 1/2 (L2 edge), where spin and orbit are coupled parallel and antiparallel, respectively. 2. Electrons are excited with a preferential direction of spin up (or down) depending on the circular polarization of the illuminating X-ray beam. As the light is polarized in two opposite directions, photoelectrons with opposite spin directions are obtained (fgure 2.12a). Note that from the 2p3/2 level X-rays with positive helicity (µ+) excite 62.5% spin up electrons and those with negative helicity (µ−) excite 37.5% spin up electrons, while from the 2p1/2 level the opposite happens. 3. These spin polarized excited photoelectrons are directed to the unoccupied states of the 3d valence band. As it is not possible to change the spin direction in an electric dipole transition, polarized excited electrons from the 2p shell (for instance, with spin-up confguration) can only occupy spin-up 3d hole bands. The spin-split valence shell acts as a detector for the spin of the excited photoelectron. Thus, the XMCD e˙ect depends on the relative orientation between the photons’ helicity and the sample magnetization direction. The XMCD signal scales with the scalar product between those two quantities [77, 78, 79]. For maximum dichroism e˙ect, the photon spin direction needs to be aligned with the sample magnetization direction (fgure 2.12b). Conventionally, the XMCD spectrum is obtained as the di˙erence between the two XAS spectra with the circular polarization vector parallel and antiparallel to the external magnetic feld applied to the sample. Specifcally, XAS spectra are measured in total electron yield using a positive (µ+) and negative (µ−) helicity light and for both magnetic felds applied (M+ , M−), normalizing by the incoming X-ray beam intensity. The resulting XMCD spectra are thus: 21 2 Experimental details XMCD(M+) = XAS(M+ , µ +) − XAS(M+ , µ −) XMCD(M−) = XAS(M− , µ −) − XAS(M− , µ +) The fnal XMCD spectrum is the average of the XMCD (M+) and XMCD (M−) spectra in order to remove the non-magnetic interaction [80]. This method was used in the SFO platelets spectra acquired in the BOREAS beamline [81] at the ALBA synchrotron (chapter 4). 770 780 790 800 810 -4 -2 0 0 2 4 6 8 XMCD (arb.units) Photon energy (eV) XAS (arb. units) μ+ 2p1/2 L3L3 L2 L2 2p3/2 μ+ μμL3 L2 3d band EF μ- - μ+ a) b) Core level L3 L2 Energy H M≠0 Figure 2.12: a)The XMCD e˙ect illustrated for the L-edge absorption in cobalt metaL. b) Top: XAS spectra at Co L-edge at di˙erent X-ray polarization. Botton: XMCD spectrum from the di˙erence between the XAS spectra. Figure reprinted from Ref. [71]. Another method used to acquire XAS and XMCD spectra in this research has been using the X-PEEM microscope at CIRCE beamline. This method will be explained in the LEEMPEEM microscope section. Sum Rules An important aspect of the XAS and XMCD spectra is that they can be used to estimate the spin and the orbital part of the magnetic moment. The sum rules [82] relates the integrated intensities on the absorption edges of these spectra with the spin and orbital moments. The sum rules correspond to the following equations: 4q morb = − Nh (2.7) 3r 22 2.2 Characterization methods   6p − 4q 7 hTzi mspin = − Nh 1+( ) (2.8) r 2 hSzi where morb and mspin are the orbital and spin magnetic moments in units of µB /atom, respectively. Nh in the number of empty 3d states (holes) of the specifc transition metal. hT zi is the expectation value of the magnetic dipole operator, and hSzi is the half of mspin in Hartree (atomic units). It has to be assumed that hT zi is zero or must be known from other experiments or theoretically approximated [70]. The values of p,qand rare the integrals of the absorption edges intensities: Z �  − p = µ + − µ dw (2.9) L3 Z �  q = µ + − µ− dw (2.10) L3+L2 Z �  + − r =1 µ + µ dw (2.11) 2 L3+L2 0 5 10 15 20 25 690 700 710 720 730 740 750 -2 -1 0 1 XAS intensity (norm.)XMCD (a.u) L3 L2 0 m+ + mr ∫m+ + mPhoton energy (eV) 120 80 40 0 Integral XAS (a.u.) p q m+ - m- ∫m+ - m--2 -1 0 Integral XMCD (a.u.) m+ - mm+ - mBackground Figure 2.13: L2,3 XAS (upper panel) and XMCD (lower panel) spectra of Fe from strontium ferrite platelets sample. The integrals from XAS and XMCD spectra are shown indicating p, q and r. 23 2 Experimental details 2.2.4 LEEM/PEEM microscope Many of the experiments performed during the research presented in this thesis have been carried out at the ALBA synchrotron. A synchrotron is a powerful X-ray source that, together with the analysis techniques available in the beamline1 , it has made possible a complete characterization of the samples. Synchrotron light is electromagnetic radiation emitted by charged particles, moving at a relativistic speed in an accelerator when their trajectory deviates on a single curved trajectory. The three types of magnet structures carry out the generation and defection of the charged particles: bending magnets, undulators, and wigglers [83, 84]. The CIRCE beamline allows doing low-energy electron microscopy (LEEM), and photoemission electron microscopy (PEEM) [74]. The latter, combined with X-rays from synchrotron light (X-PEEM) and X-ray absorption and dichroism techniques (XAS and XMCD, respectively), are the methods used in this thesis. The system works at ultra-high vacuum (pressure ∼ 10−9 mbar). Figure 2.14 represents the LEEM-PEEM instrument located in the CIRCE beamline. Sample Electron gun Illumination column Objective Analyzer Beam separator Objective Imaging column Screen Image ScreenScreenScreen X-ray beam Figure 2.14: Schematic representation of the LEEM-PEEM instrument of the CIRCE beamline. Modifed image of the poster displayed on the ALBA synchrotron website. The image shown corresponds to the characterization of SFO platelets by photoemission electron microscopy. A LEEM provides images of the sample surface at a feld of view of several microns with 10 nm lateral resolution in real-time imaging the electrons refected from the sample [85, 77]. For this, the following mechanism occurs: •First, a high-energy electron beam (up to 20 keV) is generated in the illumination 1It is an experimental area of the synchrotron, which is made up of a set of equipment that directs the X-rays to the sample and allows analysis through di˙erent characterization techniques according to the incorporated instrumentation. 24 2.2 Characterization methods column. The beam enters into a beam separator or splitter, which turns the beam towards the sample. This is generated by a magnetic feld in a prism defector. •The electrons are decelerated before reaching the sample due to an electrical potential applied between the objective lens and the sample. The energy at which they reach the sample is between 1-100 eV. The objective lens again accelerates the electrons refected from the sample. Also, a di˙raction pattern from the surface forms at the back focal plane of the objective lens. •The electron beam passes back through the separator that directs it to the imaging column. This part is composed of a set of lenses that magnifes either a real space image formed by the refected electrons or their di˙raction pattern. •Finally, the electron beam hits a multi-channel plate detector (MCP). This multichannel plates amplifes the signal and retransmits it to a fuorescent screen to obtain images in real-time on a CCD camera. PEEM also provides high spatial resolution (20 nm) real-time surface images and works similarly to LEEM. The di˙erence between both is that the resulting image, instead of being formed with the electrons refected in the sample, is formed by the photoelectrons from the sample surface due to the interaction with ultraviolet radiation or X-rays (XPEEM). The synchrotron supplies this radiation. The mechanism by which the image is obtained follows the same steps as for LEEM except that since an electron source is not used, the illumination column is not necessary. The sample illuminated with X-ray radiation produces photoelectrons. These photoelectrons are accelerated in the objective lens to form an image. The electron beam will pass through di˙erent electromagnetic lenses (image column) being fnally detected on a CCD camera. In addition, there is an energy analyser that can select the energy of the electrons that form the fnal image. This is done to reduce chromatic aberrations in the image as well as to select an energy range for performing photoemission spectroscopy. The LEEM microscope can provide information on the morphology of the sample surface as well as its structure. However, as it depends on the refection of electrons at the surface, a sample that is crystalline and not rough is required. This drawback does not a˙ect PEEM. Further, as it has been previously advanced, most of the absorption and dichroism spectra in this thesis have been obtained using X-PEEM. Following this method, it is essential to mention that the spectra’s measurement is performed in remanence. This fact is because in PEEM, applying a magnetic feld to magnetize the sample in a specifc direction a˙ects the electrons’ path emitted from the sample. Likewise, for the XAS and XMCD spectra obtention, a pair of XAS images were acquired with opposing X-ray helicities, measuring the spatially resolved emission of secondary electrons at low kinetic energy. The two images are then added and subtracted pixel by pixel to provide the averaged XAS image and the XMCD image, respectively. When the full spectrum is required, a stack of images is made for each energy. Thus, the XAS spectrum is obtained from the averaged intensities of the pair of XAS images as a function of the 25 2 Experimental details photon energy and the XMCD spectrum of the di˙erence of intensities of the pair of XAS images scanning the photon energy. XMCD images show an asymmetry contrast according to the maximum or minimum intensity recorded in the XMCD spectrum at a specifc energy. Such contrast reveals the magnetic signal in the sample. Indeed, the dichroic contrast is proportional to the local magnetization along the x-ray beam direction. So, to determine the easy axis of magnetization in the sample or the vector magnetization pattern of its magnetic domains, three XMCD images corresponding to the sample rotated with respect to the direction of the X-ray beam at di˙erent azimuthal angles are compared. For instance, see fgure 4.12 from chapter 4, where the magnetization direction was determined of magnetic domains from SFO platelets following the procedure described above. Therefore, the XMCD-PEEM combination makes it possible to determine specifc element magnetic information in acquiring images at the submicron scale. This method allows observing the magnetic contrast in a particular area of the surface [78, 79]. This dichroic e˙ect will be seen in the following chapters of the thesis about the SFO platelets, SFO thin flms and bilayer systems. Figure 2.15 shows the LEEM-PEEM system of the CIRCE beamline. It consists of an entrance chamber (blue box), a preparation chamber (green box) and the main chamber (red box) where the microscope is located. The entrance chamber incorporates a parking to store samples that allow a quick transfer without losing the samples’ vacuum conditions. The preparation chamber is equipped with di˙erent evaporators for MBE growth and gas inlet and ion gun to prepare the sample surface. In addition to the LEEM-PEEM microscope, the main chamber allows monitoring of in-situ growth also with the inclusion of evaporators. The sample inside the main chamber can be rotated with respect to the X-ray beam. In addition, the sample holder used allows the sample to be heated up to 2000 K in both the preparation and main chamber. The temperature is measured by a W/Re thermocouple attached to the sample holder. Figure 2.15: CIRCE beamline. 26 2.3 Micromagnetic simulations 2.3 Micromagnetic simulations Throughout this thesis, a tool that has been frequently used to understand experimental systems has been micromagnetics simulation. These simulations have been used to reproduce and predict the magnetic behaviour observed in SFO platelets and flms, as well as the SFO/Co and CFO/FeCo bilayer systems also introduced in this research. The program used to perform these simulations is MuMax3 [86]. It calculates the space and time-dependent magnetization dynamics in nano to micro-sized ferromagnets using a fnite di˙erence discretization. Thus, the sample is divided into a 2D or 3D grid of orthorhombic cells. The simulation cell dimensions are established in function to reproduce the magnetic features accurately. The parameters involved in the simulation are: saturation magnetization (Ms), exchange sti˙ness constant (As) and uniaxial and cubic magnetocrystalline anisotropy constant(Ku and Kc, respectively). All of them are characteristic factors of the material under study. It is important to keep in mind that volumetric quantities such as magnetization and e˙ective feld are referenced within each cell, while coupling parameters such as exchange strength are referenced at the interface between two cells. Therefore, a simulation represents the magnetization moment evolution per cell and its interaction with the magnetization moment from neighboring cells according to the parameters set. The micromagnetic simulations that give rise to these magnetization changes are defned by the Landau-Lifshitz-Gilbert (LLG) equation [87]. The LLG equation gives temporal evolution of the magnetization towards to system’s equilibrium state, and it is expressed as: d~ M−γ γα ~ M× (~ M× ~ Heff ) (2.12) ~ M× ~ Heff −= 1 + α2 (1 + α2)dt Ms where α is the dimensionless damping coeÿcient, γ is the electron gyromagnetic ratio, ~ Heff is the e˙ective magnetic feld and ~ Mis the magnetization. The LLG equation is composed of two terms: •The frst term ∼ − ~ M× eff is related to the magnetic precession around the e˙ective ~ H magnetic feld. •The second term ∼ − × (~ M× ~ Heff ) is refers to the damping torque which induces ~ M the transverse relaxation towards to e˙ective feld direction. It causes minimizing energy, driving the system into an equilibrium state. This state is reached when magnetic moment and applied feld are aligned corresponding to the condition ~ M× ~ Heff = 0 [88]. The fgure 2.16 represents the torques involving in the movement of the magnetization respect with a ~ Heff and depicts the magnetization dynamics. 27 2 Experimental details M Heff - M x Heff - M x (M x Heff)Heff M a) b) Figure 2.16: Time evolution of a single magnetic moment as described by the LLG equation. (a) Torques a˙ecting the magnetization vector ~ M(precessional and damped motions). (b) Resulting motion due to the precessional and damping torques. Furthermore, it should be noted that the e˙ective magnetic feld is made up of several contributions which the most important are: Externally applied feld (Hext): this contribution arises from the interaction between the applied magnetic feld with the magnetic moments from the sample. This is also known as Zeeman term. The applied feld drives the magnetization motion, trying to orient the latter to the same direction as Hext. Magnetostatic or demagnetization feld (Hdemag): this term is the feld created by the magnetization from itself material, and it corresponds to dipole-dipole interactions. The demagnetization feld is responsible for the magnetic domain formation since that reduces the energy associated with the sample magnetization. Exchange feld (Hexch): this is due to the exchange interaction between the magnetic moments. Such interaction produces the alignment of the magnetic moments of the neighbouring atoms. As it will be seen in the following chapters, when considering di˙erent materials, we can rescale the exchange interaction with a factor to parametrise the coupling between them. Magneto-crystalline anisotropy feld (Hanis): this feld describes the preference of the magnetization to point along a crystallographic direction depending on the crystalline structure of the material. That is, it is easier to align the magnetization parallelly to particular crystallographic axes. This contribution occurs as a result of the spin-orbit interaction. In the micromagnetic simulations presented in this thesis, we have performed energy minimization of each initial confguration. In some cases, the initial confguration has been obtained directly from the experiment, while in others, either a uniform or a random magnetization was used. MuMax3 performs the energy minimization disabling the precession term so that the e˙ective feld points towards decreasing energy. As a fnal step, it minimizes the magnitude of the torque instead of the energy since close to equilibrium; the torque should decrease monotonically and is less noisy than the energy. 28 3.2 Magnetic order in strontium ferrite b) a) O Sr Fe (2b) Fe (2a) Fe (4f2) Fe (4f1) Fe (12k) MSrFe12O19 OII SrII FeIII Figure 3.6: a) Cross section view of the M ferrite (SrFe12O19) structure in which the vertical lines are threefold symmetry axes. Picture modifcated from [13]. b) 3D view of the M unit cell. 3.2.1 Intrinsic magnetic properties It is the magnetic structure in terms of sublattices and their mutual orientation that governs magnetic behavior, which in turn is described in terms of intrinsic and material properties (saturation magnetization, magnetocrystalline anisotropy and coercive force [99, 100, 101, 102, 78, 103]). •Saturation magnetization, Ms: it is the maximum magnetic moment per unit of volume. As it was studied in the previous section, the Ms value for strontium hexaferrite is derived from the spin confguration of the sublattices contributing with 20 µB per unit formula. However, this value ideally only occurs in a compound with its fully ordered magnetic moments, high crystallinity, and a temperature of 0 K. The temperature increase reduces the saturation magnetization of the compound (fgure 3.7). For instance, for barium hexaferrite (isostructural compound), the theoretical approximation (20 µB ) was reached in a polycrystalline sample at liquid hydrogen temperature (20 K), and an external magnetic feld of 26000 Oe [13]. The data reported in the literature for the saturation magnetization of strontium ferrite single crystal range between 20.4 µB and 16.8 µB [91, 104]. 35 3 SrFe12O19 Figure 3.7: Saturation magnetization of BFO (1), SFO (2) and PFO (3) as a function of temperature. Reprinted from Ref [90]. Saturation magnetization is the governing parameter of the magnetostatic energy, Em. This energy defnes the conventional dipole-dipole interaction between magnetic moments, located at the atomic positions in the lattice: 1 Em = µ0V NMs 2 (3.2) 2 where µ0 is the permeability of the vacuum, Nis the demagnetizing factor for the easy axis and Vthe sample volume. This energy is directly related to the shape of the sample (shape anisotropy). •Magnetocrystalline anisotropy: it describes the preference for magnetization to be oriented in a certain crystallographic direction. Specifcally, for M-type ferrites, it corresponds to an uniaxial anisotropy with preferential magnetization along the hexagonal c-axis. The mathematical expression for the energy, Ea, is given by: Ea = K1 sin2 φ + K2 sin4 φ + K3 sin6 φ... (3.3) where φ is the angle between the magnetization and the crystallographic easy axis and K1 is the so-called frst order anisotropy constant. Higher order constants (K2, K3) are negligibly small. The value of this constant for SFO is K1 = 3.5 × 105 Jm−3 [105]. Also, the temperature a˙ects this parameter. This fact can be understood in terms of dependence upon spontaneous magnetization. The K1 should change with the temperature proportionally to ∼ M3 . The reason is that thermal agitation can s produce local variations of the direction magnetization vector and hence, changes in the energy [106]. An important factor related to the previous ones is the anisotropy feld strength, HA. This parameter corresponds to the maximum internal reverse feld needed to switch the magnetization in the direction perpendicular to the preferred axis: 36 3.2 Magnetic order in strontium ferrite Ha = 2K1/Ms (3.4) It represents the upper limit for the coercive force (see below). •Coercive force: it can be defned as the external feld necessary to demagnetise a magnet completely. In general terms, it can be said that in the case of ferrite magnets, such as sintered or bonded compacts of SFO powders, this parameter arises from the magnetic behaviour of single domain particles with uniaxial magnetocrystalline anisotropy. To understand this, we need to take into account several considerations involved in the coercive force value. Critical diameter: It is the size of a particle below which a single domain state can be energetically achieved, and thus, the magnetization reversal proceeds by coherent rotation: 9 σw Dc = (3.5) 2π M2 s where σw is the specifc domain wall energy. Inside this wall, the spin direction gradually changes from one preferred direction to another one. σw can be expressed as: σw = 4(AK)1/2 (3.6) being Athe exchange sti˙ness coeÿcient which correspond to the energy associated with the (anti)parallel coupling of the ionic moments, exchange energy (Ee).Ais an intrinsic parameter for each material (ASF O = 6 × 10−12Jm−1). The critical diameter for strontium hexaferrite is close to 1 µm, so the coercive force for a particle of smaller diameter than this Dc can be written as: Hc = Ha − NMs (3.7) It is important to note that depending on the shape of the particle, the demagnetizing factor changes. The latter ranges from 0 (for needles) to 1 (for thin plates). For platelet-shaped M ferrite crystals, N ranges from 0.6 to 0.9. The coercivity values are high, and, as we have seen, they are a function of shape and size. For particles with a higher size than Dc, coercitivity values become smaller due to the formation of transient domains and magnetostatic interactions. The presence of domain walls facilitates the reversal of magnetization compared to the coherent rotation process, and this causes the coercive feld to drop [22, 28, 107]. 37 4 SrFe12O19 platelets 4.1 Introduction Strontium hexaferrite has been a highly studied compound since its discovery. This is due to its magnetic properties [91, 108] that make it a good material to manufacture permanent magnets [109, 7, 28]. It has been grown by several methods: chemical coprecipitation [110, 111], solid-state synthesis [112], conventional ceramic process [113], sol-gel [114, 115], ball milling [116] and hydrothermal route [117, 118, 119, 120, 121, 122]. Among them, hydrothermally synthesis method has provided very homogeneous and high purity powders. Such powders in most of the cases presented platelet-shaped particles which have their crystal c-axis perpendicular to platelet plane. This phenomenon can be explained by the fact that the crystal growth rate for M-type ferrites is much faster in the basal plane than in the c-axis direction [13]. SFO single crystal platelets are of great scientifc interest because the magnetocrystalline anisotropy is located in the c-axis, producing an anisotropic magnetic material. As it will be studied in this section, the intrinsic properties of the material, both morphological and magnetic, can be a˙ected to a large extent by the shape and size of the material particles. The results presented in this chapter have been published in: •G. D. Soria, P. Jenus, J. F. Marco, A. Mandziak, M. Sanchez-Arenillas, F. Moutinho, J. E. Prieto, P. Prieto, J. Cerdá, C. Tejera-Centeno, S. Gallego, M. Foerster, L. Aballe, M. Valvidares, H. B. Vasili, E. Pereiro, A. Quesada, and J. de la Figuera, “Strontium hexaferrite platelets: a comprehensive soft X-ray absorption and Mössbauer spectroscopy study", Scientifc Reports, vol. 9, no. 1, p. 11777, 2019. 4.2 Synthesis by hydrothermal method The platelets were provided by Dr. Petra Jenus from the Josef Stephan Institute in Ljubljana, Slovenia, and they were grown by hydrothermal synthesis. In this method aqueous solutions containing the appropriate metal ions prepared from strontium (II) nitrate (Sr(NO3)2, 99% +, Across Organics) and iron (III) nitrate nonahydrate (Fe(NO3)3 ×9H2O, 98%, Carlo Erba Reagents) salts are mixed. The exact metal concentration of the reagents is determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES) [123, 124]. To the aqueous Sr2+ and Fe3+ containing solution with a Sr2+/Fe3+ ratio of 1/6, a sodium hydroxide (NaOH, Alfa Aesar, 98%) aqueous solution is added at room temperature so that the fnal NO−/OH− ratio is 1 2 . The mixture is then put into a stainless3 steel autoclave and keeps in an oven until a temperature of 503 K is reached for a holding 39 4 SrFe12O19 platelets time of 15 min, and immediately after that, the heating is turned o˙, and the autoclave is cooled to room temperature. 4.3 Morphological, structural and compositional characterization In order to check the morphology and crystallinity of the powders, they were frst examined by transmission electron microscopy (TEM). This technique generates an image of the sample using a beam of high-energy electrons transmitted through it. TEM images of the SFO powders (fgure 4.1) revealed a few small particles, below 100 nm, and a majority of platelets with lateral dimensions around 1 µm and thickness of a few tens of nanometers. The platelets present a hexagonal shape [91] and observable lattice fringes at higher magnifcations, which suggests good crystallinity. The high degree of crystallinity was supported by selected area electron di˙raction (SAED) measurements. This method provides di˙raction patterns acquired from a zone of a single platelet. For the identifcation of the SAED pattern, the reference di˙raction pattern for SrFe12O19, simulated with SingleCrystalTM using the 69022 ICSD fle, was used [125]. The SAED pattern is compatible with the SrFe12O19 crystal planes being oriented in the [001] direction. I.e., the single particle has its basal plane up (fgure 4.1 in the upper right panel). Figure 4.1: Left: TEM image of platelets. Bottom right: TEM image corresponding to a zoom of the selected area. Top right: di˙raction pattern from SAED analysis. The structure has an orientation in the [001] plane. Continuing with the morphological characterization of platelets, images were taken from Transmission X-ray Microscopy (TXM). In this case, the image is acquired by measuring the transmitted X-rays through the sample. The microscope is composed of both condenser and objective lenses. X-ray radiation comes to the elliptical glass capillary condenser, which focuses the light on the sample. An objective Fresnel Zone plate (FZP) collects the transmitted signal. This FZP is rotationally symmetric di˙ractive grating, composed of radially decreasing width rings. This lens is used to form an image of the feld of view on the detector (direct illumination CCD camera) [47, 126]. The measurements were 40 4.3 Morphological, structural and compositional characterization carried out at the MISTRAL beamline of the ALBA synchrotron [127]. They were taken with a Fresnel zone plate of 25 nm. TXM images were acquired in absorption contrast mode at the Fe L3 and L2 edges, fgure 4.2. Each panel corresponds to the same region rotated with respect to the Y-axis. The images were taken at -50◦ , -30◦ , 0◦ , 30◦ and 50◦ . Again, platelets present a hexagonal shape with lateral sizes several orders of magnitude higher than the thickness. Furthermore, platelets appear to form aggregates, and they are randomly oriented with respect to each other within the aggregates. Additionally, this technique allows determining the magnetic contrast of the platelets by the di˙erence in the absorbance signal obtained at the Fe L3 and L2 edges. However, platelet stacking complicates correctly distinguish the magnetic signal from each one. Figure 4.2: TXM images show the absorption contrast of the same cluster of SFO particles, at the angles: -50◦ , -30◦ , 0◦ , 30◦ , 50◦ , respectively. E˙ective pixel size = 10 nm. The structural analysis of the sample was also performed by X-ray di˙raction. Figure 4.3 shows the XRD di˙raction pattern of the SFO platelets and the reference di˙raction pattern for SrFe12O19 simulated with Crystal Di˙ractTM using the 69022 ICSD fle (reference from [125]). All the refections found on the platelet are indexed to crystallographic planes belonging to the reference hexagonal structure of SrFe12O19. Di˙erent families of planes are detected due to the random orientation of the platelets relative to the incident X-ray beam. No additional peaks were observed, indicating the absence of other phases. However, the width of the peaks indicates that the crystallite size is not too large. The average size of crystallites can be estimated using the Scherrer formula [128]. The value obtained was 50 ± 2 nm. This crystallite size is much smaller than that revealed by the TEM and TXM images which are often hundreds of nanometers wide. The most likely explanation could be that the platelets appear at di˙erent orientations relative to the X-ray beam, so even for large micron sized platelets, most of them would appear much smaller. The unit cell lattice parameters were calculated, obtaining a = 0.588 nm and c = 2.308 nm. Both are in good agreement with values reported in literature [91, 125]. X-ray photoelectron spectroscopy (XPS) was used to confrm the presence of strontium and iron species expected for the SFO structure. The energy scale was referenced to the binding energy (BE) of the C 1s core level of the adventitious contamination layer, which was set at 284.6 eV. The Sr 3d spectrum (BE Sr 3d5/2 = 132.5 eV and 3d3/2 = 134.3 eV) revealed the existence of Sr2+ cations, fgure 4.4a. Figure 4.4b presents the corresponding Fe 2p spectrum. The binding energy of the Fe 2p3/2 core level is 710.0 eV, and the presence of a weak shake up satellite at 8.9 eV above the main Fe 2p3/2 line indicates that Fe is in the Fe3+ oxidation state as expected for the SFO structure. 41 4 SrFe12O19 platelets 20 30 40 50 60 0 50 100 150 200 Intensity (a.u.) 2 theta (°) reference synthesized (006) (110) (008) (107) (114) (116) (205) (206) (203) (108) (201) (200) (2011) (304) (0014) (217) Figure 4.3: XRD di˙raction patterns from the SFO powder, together with the reference pattern (bottom one). 710720730740 Binding energy (eV) Intensity (arb. units) Fe 2p3/2 Fe 2p1/2 Fe (III) 2p3/2 satellite Fe (III) 2p1/2 satellite 130132134 136 138 Binding energy (eV) Intensity (arb. units) a) b) Sr 3d3/2 Sr 3d5/2 Figure 4.4: Sr 3d and Fe 2p core level X-ray photoelectron spectra recorded from SrFe12O19 platelets. 42 4.3 Morphological, structural and compositional characterization Mössbauer spectroscopy, as it has been explained in the techniques section, can provide information about the oxidation state and coordination of the iron cations. The room temperature Mössbauer spectrum of strontium hexaferrite is quite complex as it contains fve overlapping sextets each one associated to a specifc Fe3+ site within the SFO structure. Because of this complexity, there is some dispersion in the hyperfne parameters which characterize these sextets [129]. In this section, we have measured Mössbauer data at 298 K (RT) and 26 K (LT) from a commercial pure powder of SFO in order to have a reference to guide the ft of the data recorded from the platelets. These two reference spectra are shown in fgure 4.5a and 4.5b, respectively. The corresponding hyperfne parameters obtained from the ft to a sum of sextets having Lorentzian-shaped lines are collected in Table 7.4. The obtained hyperfne parameters and spectral areas for the RT measurement are all in reasonable agreement with those reported previously for this material [130, 131, 132]. The 26 K spectrum has a very di˙erent shape resulting from the increase in both the hyperfne magnetic feld values (whose temperature variation is not the same for each of the di˙erent sites [132]) and the isomer shifts of the various contributions. Similarly to that described in Ref. [132], this spectrum has been ftted to fve sextets (Figure 4.5b), the results being also in reasonable agreement with these literature values. The RT Mössbauer spectrum recorded from the SFO platelets is depicted in fgure 4.5d. Compared to that of the standard sample, this spectrum shows much broader lines (about twice broader than in the SFO reference) and resembles other spectra recorded from SFO samples made by hydrothermal synthesis [131, 132]. This might be related to the occurrence of disorder, or structural inhomogeneities in the material [130]. The broadening of some of the lines in the X-ray di˙raction data (see fgure 4.3) is compatible with this interpretation. An explanation of the line broadening considering the occurrence of superparamagnetic e˙ects could be discarded as the lateral dimensions of the platelets are in the micrometer range. The spectrum fts well considering a 3:2:1:1:2:3 area ratio for the lines of all the sextets, indicating that the platelets in the sample are randomly oriented. In any case, the hyperfne parameters obtained from the ft are very similar to those of the SFO standard sample (Table 7.4). It is interesting to compare this spectrum with the RT ILEEMS spectrum recorded from the platelets (fgure 4.5c). In an ILEEMS spectrum, the surface contributions are enhanced [68], therefore the di˙erences, if any, between the ILEEMS and transmission spectra have to be due to structural/compositional changes in the surface with respect to the bulk. Inspection of fgures 4.5c and 4.5d shows clear di˙erences between these two spectra, particularly in the outermost lines. The ft to the ILEEMS spectra shows a considerable increase of the intensity of the area of the sextet corresponding to the tetrahedral 4f1 sites, which almost doubles respect to that shown in the transmission spectrum (Table 7.4). Therefore, the results point out a higher concentration of tetrahedral sites at the surface. In this respect is also very interesting to compare the 26K spectrum recorded from the platelets (fgure 4.5e) with that recorded at that temperature from the SFO standard (fgure 4.5b). Again, there is a large di˙erence between these two spectra: the spectrum of the platelets is less asymmetric, and the outer lines are broader. It must be recalled that there is no unique ft to this low temperature platelets spectrum. It is clear that the lines are much narrower than in the RT spectrum and, therefore, that the strong overlap between the di˙erent contributions complicates the ft. In the case of the spectrum of the SFO standard, this diÿculty is mitigated at some extent because the outer sextets are relatively well distinguished from that corresponding to the most populated 12k site (see 43 4 SrFe12O19 platelets outer lines of the spectrum in fgure 4.5b). According to this and due to the diÿculty for providing a unique ftting for the spectrum measured in platelets at 26 K, a qualitative analysis of the spectrum was done. When trying to ft this spectrum various trends were observed: (i) there is a strong tendency to obtain as the most intense sextet the one having hyperfne parameters close to tetrahedral coordination; depending on the parameters fxed, the intensity of the sextet corresponding to site 2a increases noticeably but its isomer shift goes to very low values, again compatible with a coordination lower than the octahedral one which is the one expected (it is known that the hyperfne parameters of sites 4f1 and 2a are strongly correlated [129]) and (ii) the area of the sextet corresponding to the octahedral site 12k has a tendency to decrease; in some of the ftting models tried the area goes down to a half of the expected value. The large increase of the contribution corresponding to “tetrahedral/lower than octahedral coordination sites” at the 26 K spectrum of the platelets might be understood on the basis of their recoil free fraction. Since, as the ILEEMS data have suggested, these sites are preferentially located at the surface of the platelets, it could be assumed that their recoil free fraction at room temperature is low [133] and that it increases dramatically at low temperature. Taken together the results seem to indicate that the platelets contain a large amount of “tetrahedral/lower than octahedral coordination sites” which are mainly located at the surface. It must be taken into account that given the shape of the platelets, whose lateral dimensions are several orders of magnitude larger than their thickness, the amount of sites unsaturated in oxygen which are located at the surface has to be larger than the number of these sites in the bulk. It would follow that the broadening observed in the XRD data, and at some extent in the RT Mössbauer data, would refect then the various confgurations arising from the distribution of iron ions which cannot complete their octahedral coordination and that can show either tetrahedralor penta-oxygen coordination. As the low temperature data have suggested, this situation would imply, consequently, a reduction in the number of well-defned 12k sites existing in the platelets. This should be refected in the magnetic moment measured in these particles (see below). 44 4.3 Morphological, structural and compositional characterization Transmission (%) 10 5 0 5 10 Velocity (mm/s) 295 K 26 K a) b) 97 98 99 100 98.4 99.0 99.6 100.0 Transmission (%) 10 5 0 5 10 Velocity (mm/s) 295 K 295 K 26 K c) d) e) 100.0 100.8 100.6 Effect (%) 99.6 99.2 99.25 99.5 99.75 100.0 100.0 Transmission (%) Transmission (%) Figure 4.5: a) and b) Mössbauer spectrum of the SFO commercial powder recorded in transmission mode at 295 K and at 26K, respectively. c), d) and e) Mössbauer spectrum of the SFO platelets in electron detection mode at 295 K and transmission mode at 295 K and at 26 K, respectively. The fve sextets in each spectrum correspond to the sites where the iron cations are positioned. The color for each sextet was selected according to the SFO structure image depicted in the SrFe12O19 structure section. 45 4 SrFe12O19 platelets Site Charge (e) MM (µB) 12k 6.37 4.07 4f1 6.41 -3.96 4f2 6.29 -3.93 2a 6.32 4.08 2b 6.46 3.99 TOTAL 220.00 20.00 Table 4.3: Bader charge and magnetic moment of each Fe site and total values per formula unit including also the oxygen and Sr ions. the double peak at 532 eV is explained by the hybridization of 2p antibonding oxygen states with iron 3d states. The separation between the two peaks that compose it is related to the splitting between the t2g and eg orbitals in the iron cations due to the crystal feld [154]. However, as is the case for maghemite and magnetite [153], the presence of three di˙erent iron environments (tetrahedral, octahedral, and trigonal bipyramidal) smears out the clear double peak detected in hematite. The peaks at higher energies (536-576 eV) refected transitions into oxygen p-states hybridized with extended 4p and 4s iron states. Their particular origin has been assigned by means of multiple scattering cluster calculations in Ref. [152], to intrashell multiple scattering (peak at 545 eV) and single scattering between the absorber and consecutive oxygen shells (peaks at 563 and 550 eV). In fgure 4.10 the XAS spectrum was compared with the density of unoccupied states projected (PDOS) on the O 2p state directly calculated by DFT (dark line). Most of these peaks are correctly reproduced both in shape and energy location within the expected accuracy of the DFT formalism. Therefore, a reasonable overall correlation was attained. Furthermore, comparison versus the DOS projected on the Sr atoms and the Fe -s, -p and -d states revealed that the infuence of the former on the oxygen p-states is negligible compared with the latter and that, particularly at higher energies, there is certain correlation between the O-p and Fe-p PDOS. The study of the XAS spectra by DFT calculations was only carried out for the oxygen at the K edge since the excitations are calculated from the 1s core level being much simpler computationally than for the excitations of the principal quantum number n = 2 (L edge). The frst-principles calculations at the L edge must take into account more electronic transitions as well as the peak-splittings emerging from the presence of a strong spin-orbit coupling in the 2p orbitals. 52 4.4 Magnetic characterization Figure 4.10: Blue line: X-ray absorption spectrum at the O K absorption edge, recorded at 2 K and an applied magnetic feld of 6 T from the SFO platelets and averaged for both light helicities. Rest of lines: Calculated density of states projected on the Sr atoms, Fe-d,p and s and the O-p states (see legend). The PDOS curves for each species have been vertically shifted and rescaled for visual inspection. The measurements of X-ray spectroscopy shown above (fgure 4.7) correspond to the signal associated to a macroscopic quantity of powder, as it was seen in fgures 4.1 and 4.2. In order to determine the XAS-XMCD images and spectra on a single platelet, we used photoemission electron (PEEM) microscopy [74]. For this analysis, the powder was dispersed in ethanol, diluted several times, sonicated, and then deposited on a Si(100) wafer covered with its native oxide. Once introduced the sample into the chamber, we searched the surface for platelets lying on the plane. The XAS average image (fgure 4.11b) was taken at maximum energy of the XAS at the Fe L3 edge spectrum (fgure 4.11a) while the dichroism image (fgure 4.11c) was acquired at the most intense energy peak of the dichroic signal of the XMCD spectrum (fgure 4.11a). The XAS spectrum resembled those shown in fgure 4.7 and the XMCD spectrum is somewhat noisier than those determined above since it is obtained from a sub-micron area at room temperature. Nonetheless, the features 53 4 SrFe12O19 platelets resemble what is seen in the high-quality dichroism spectra acquired with a macroscopic amount of powder. The averaged XAS image shows platelets that appear partially stacked on top of each other. In addition, there is a detectable shadow together to some particles edges due to the platelets height prevents the incident X-ray beam. So, knowing the incident X-ray beam angle and the shadow size, the platelet thickness can be determined. The thickness estimated for these platelets was 18 nm. The XMCD image shows several domains (black and white regions) on the platelet. It is well-known that the asymmetry contrast is proportional to the local magnetization along the X-ray beam direction [155]. Hence, this platelet presents magnetic domains. 700 710 720 730 740 Photon Energy (eV) X-ray absorption (arb. units) XAS at Fe L2,3 edge XMCD at Fe L2,3 edge a) b) c) 500 nm Figure 4.11: a) XAS and XMCD spectra acquired at the Fe L2,3-edge on the platelet by PEEM. b) X-ray absorption image collected at the Fe L3 maximum and c) dichroism image extracted from the most intense peak in the XMCD signal as a result of the di˙erence between both X-ray helicities. Proceeding with the characterization of the vector magnetization pattern, we measured XMCD images with three non-coplanar orientations [156] of the X-ray beam relative to the sample (fgure 4.12). For the three images, the same magnetic domains are observed on the same areas in the platelet. So, it can be concluded that the magnetization orientation in the platelet is mostly out-of-plane. The result was expected since this material presents a large anisotropy and the crystallographic c-axis is perpendicular to the platelet, and thus also the magnetic easy axis. 54 4.4 Magnetic characterization -55º 0º 55º -55º 0º 55º M M M Figure 4.12: Top: XMCD images acquired at di˙erent angles respect to the X-ray beam. Botton: Magnetization orientation in the sample (out-of-plane) at each angle. From the combination of this XMCD three images, the pixel-by-pixel magnetization vector can be determined, as shown with a color map in fgure 4.13a. The platelets in the image are multidomain, with their magnetization vector mostly in and out-of-plane (black and white areas) with 180◦ domain walls between them. Using the 3D rendering software "Muview" the magnetization was depicted with arrows that pointed in the local magnetization direction (fgure 4.13b). The green and pink colors correspond to domain walls (DWs) between black and white areas and also, to a lesser extent, to the noise in the determination of the magnetization vector. It is important to consider that the polar angle of incidence of the X-ray beam is fxed at 74◦ with respect to the normal to the sample, where 90◦ indicates the in-plane direction. Therefore, the out-of-plane magnetization sensitivity is reduced. Additionally, the domain walls observed were quite sharp. In fact, their width is likely p smaller than the experimental resolution. Their expected width is π A/K = 13 nm, where A is the exchange sti˙ness and K the uniaxial anisotropy constant [157]. A cut across a domain wall is shown in fgure 4.13e, confrming this prediction. The pixel width in the images is 8.5 nm, and the overall experimental lateral resolution is around 20 nm [74]. To better understand the observed magnetic domain distributions in the platelet, micromagnetic simulations have been carried out with the Mumax software [86]. For this purpose, we have used the experimental magnetization map as the initial confguration of a micromagnetic simulation for an object with a shape similar to the one experimentally determined. The voxel size in the simulation was 4.23 nm, to reproduce accurately the domain walls. The material constants employed for the saturation magnetization, exchange sti˙ness and magnetocrystalline hexagonal anisotropy were Ms = 3.8 × 105 Am−1 , As = 6 × 10−12Jm−1 and Ku = 3.6 × 105 Jm−3 , respectively [91, 158, 108]. When the energy is minimized, the initial confguration gives rise to the magnetization pattern shown in fgure 4.13c. The domains closely resemble the experimental ones, with a similar curvature of the domain walls. A cut across the domain walls gives the magnetization profle in fgure 4.13e. Another simulation was performed using instead the maximum magnetization measured for the platelets by VSM. The resultant simulation is presented in fgure 4.13d. An important observation is that the magnetic domain distribution relaxes to a di˙erent one (compared to the experimental image shown in Fig. 4.13a) if the measured Ms-value is used. Specif55 4 SrFe12O19 platelets 50 100 150 200 Assymetry (arb. units) Distance (nm) Wall ~8 nm 90º 0º 270º 180º 45º 315º 135º 225º a) b) e) c) d) Figure 4.13: a) Reconstructed magnetic state of the SrFe12O19 platelet pixel by pixel. b) 3D projection of the magnetization vector in the platelet represented by arrows. c) Relaxed simulation of the initial confguration using the reported bulk magnetization saturation of SrFe12O19. d) Relaxed simulation of the initial confguration using a reduced magnetization saturation measured experimentally on the SrFe12O19 platelets by VSM. e) comparison between domains wall width of the experiment and simulations. Note that the color palette in the right corner represents the spin direction in the magnetic domains. cally, the domain walls have less curvature and reproduce worse the experimental results. Probably, the main reason for the change in the magnetic domain distribution at the latter simulation could be due to that the VSM magnetization measured at 1.5 T underestimates the saturation magnetization. In summary, in this chapter, a full characterization of strontium hexaferrite platelets grown by hydrothermal synthesis has been carried out. For this purpose, several experimental techniques have been used. For the morphological, compositional and structural characterization, microscopies (TEM and TXM), di˙raction (XRD) and spectroscopies (XPS and Mössbauer) have been applied. The magnetic properties have been determined using VSM and X-ray absorption techniques and have been understood according to the cationic distribution discussion done in the Mössbauer spectroscopy section. Furthermore, calculations (DFT and multiplet approach) and micromagnetic simulations have been supported the results obtained. 56 4.5 Conclusions 4.5 Conclusions •Strontium hexaferrite platelets have been synthesized by the hydrothermal method. The particles have a hexagonal shape with lateral sizes of micrometres and tens of nanometers thickness and are found forming aggregates. •In addition to confrming the presence of Fe3+ species and the existence of the pure SFO phase observed by XPS and XRD, Mössbauer spectroscopy provided information on the cationic distribution. Platelets show an increase of iron cations in tetrahedral sites with respect to octahedral sites in the region near the surface. This fact plays an essential role in the understanding of the magnetic moments values. •XAS and XMCD spectra of the L2,3 iron absorption edges and oxygen K edge both for the platelets and for the reference commercial material have been presented. In order to understand the origin of each feature in the iron L edge spectra, multiplet calculations were carried out, taking into account the crystal feld symmetry of each iron cation. It was determined that the contributions from Fe in octahedral and tetrahedral sites turned out to be responsible for the observed features showing the importance of performing the XAS measurement with high resolution. •The sum rules applied to the XMCD spectra gave estimates for the net magnetic moments of Fe. The lower net magnetic moment of Fe in the platelets grown by the hydrothermal method compared to the value of the commercial SFO was attributed to the increase of the iron in tetrahedral positions in the former where the spin moments are aligned antiferromagnetically to the net magnetization. The O K edge spectrum was reproduced by the unoccupied density of states calculated by DFT by Jorge Cerda et al. •XMCD-PEEM imaged the magnetic domains in platelets showing a magnetization mostly perpendicular to the platelet plane. Micromagnetics simulations using the bulk parameters for SFO reproduced the distribution of the magnetic domains. 57 5 Magnetic interactions in magnetic single domain platelet with cobalt overlayer 5.1 Introduction The purpose of this chapter is to understand the magnetic coupling between a hard magnetic layer and a soft magnetic layer. Specifcally, the experimental system corresponds to a magnetic single domain SFO platelet with a metallic cobalt layer on top. As explained in Chapter 1, this combination of soft-hard layers is a resource much studied and discussed during the last decades for the development of permanent magnets [159, 160]. By rigidly coupling both layers, an enhancement in its magnetic properties can be achieved. The magnetically soft material brings high magnetization while the hard material provides high coercivity. Nonetheless, several studies [161, 162, 163] reported a decrease in the coercive feld in a robust rigid coupling. This decline is promoted by the propagation of unpinned domain walls generated at the soft phase, which facilitates the reversal magnetization [22]. Therefore, this chapter, with the analysis of the platelet/metal system by means of X-ray absorption techniques and micromagnetism simulations, addresses the question about the magnetic interaction conditions necessary to improve the magnetic properties of a bilayer system. In particular, we will research to what extent soft magnetization direction might align with that of hard by pure magnetodipolar interaction and/or in low exchange-coupling regime, which is the interaction that induces the domain wall propagation mechanism. The results presented in this chapter have been published in: •G. D. Soria, C. Granados-Miralles, A. Mandziak, P. Jenus, M. Saura-Múzquiz, M. Christensen, M. Foerster, L. Aballe, J. F. Fernández, J.d.l. Figuera, and A. Quesada, “Uncorrelated magnetic domains in decoupled SrFe12O19/Co hard/soft bilayers", Journal of Physics D: Applied Physics, vol. 54, no. 5, p. 054003, 2020. 59 5 Magnetic interactions in magnetic single domain platelet with cobalt overlayer 5.2 Cobalt Cobalt is a transition metal that crystallizes at room temperature in the hexagonal closepacked structure (hcp). Its lattice parameters are a = 2.5 Å and c/a = 1.632 Å. Since its magnetization easy axis is the c-axis, presenting a strong uniaxial anisotropy (4.1 × 105 Jm−3) [164, 165, 166]. However, in thin flm form, the packing can be modifed by changing the particle size and the growth temperature. Above 725 K, the cobalt crystal structure is the face-centered cubic (fcc) one [167]. We selected this material for the coupling study because its high value of saturation magnetization (1.4 × 106 Am−1) [168]. This property suggests that in a rigid or partial exchange-coupling composed of cobalt metal as a soft magnetic material and a hard ferrite, an enhancement in the energy product of the system might be found [169, 170]. In contrast, for instance, such improvement would not be so obvious using a soft phase such as nickel whose Ms is 4.8 × 105 Am−1 [168] (similar than to SFO ferrite). 5.3 XAS-XMCD characterization from a single domain platelet In the previous chapter, we studied SFO platelets which presented a crystallographic structure with the c-axis oriented normal to the platelet plane. Thus, taking into account that the magnetocrystalline anisotropy is along the c-axis, the net magnetization was found to be perpendicular to the platelet-plane. These results are in agreement with other studies of strontium hexaferrite platelets reported in the literature [13, 91]. That particular platelet showed several magnetic domains. For this study, a SFO platelet was selected, displaying only a single magnetic domain in order to avoid the possible e˙ects of domain walls and internal demagnetizing felds associated with multidomain structures. The characterization of this platelet and the subsequent magnetic coupling experiment with a cobalt layer was performed at the ALBA synchrotron by means of X-ray absorption microscopy. The platelets that were used for this study were synthesized by the same hydrothermal method explained previously. As a result, the obtained particles showed similar shape and dimensions to those previously synthesized. For their characterization, the platelets were dispersed in ethanol and sonicated. Subsequently, a drop of the dispersion was deposited onto a Si (100) substrate without removing the native Si oxide and left to dry. The sample was then introduced in the ultra-high vacuum (UHV) chamber of a PEEM microscope and degassed at 100 ◦C. Since photoemission electron microscopy allows the characterization of a single platelet by means of the collection of images in real-time, as it has already been done in chapter 4, an isolated platelet was selected. Figures 5.1a and 5.1b present the Fe L3 edge X-ray absorption and XMCD spectra measured on a single strontium hexaferrite platelet which are characteristic of this phase (see the previous chapter). As discussed, the peaks arise from the Fe3+ cations in di˙erent chemical environments in the SFO structure [102, 171]. The images shown in the right panel of fgure 5.1 correspond to the image of the micron size platelet measured by PEEM 60 5.3 XAS-XMCD characterization from a single domain platelet at the energy of maximum intensity of the absorption spectrum (a), the XMCD image obtained at an energy of 707.0 eV (b) and, the XMCD image acquired at an energy of 708.0 eV (c). From these pictures, it is interesting to comment on the change in the intensity (gray level) on the platelet between the di˙erent dichroism images. The XMCD contrast indicates the magnetization component along the X-ray direction. The frst observation of the XMCD image is that the gray level in each image is the same through the platelet indicating the platelet has a single magnetic domain. The reversal of intensity, black in fgure 5.1d and white in fgure 5.1e, can be understood looking at the XMCD spectra (5.1b): the XMCD intensity is reversed at the two energies, 707 eV and 708 eV. This result is in agreement with the prediction that established that the SFO single domain threshold size is approximately 1 µm [91, 109]. 714712710708706704702700 XMCD XAS X-ray absorption (arb. units) Photon energy (eV) a) c) d) e) b) 708 eV 707 eV Figure 5.1: a) Fe L3 edge XAS spectra of a SFO platelet before deposition of the cobalt layer. b) XMCD spectra associated to Fe L3 edge XAS spectra of a SFO platelet before deposition of the cobalt layer. c) XAS-PEEM image recorded at the energy of the maximum in the Fe L3 edge XAS spectrum. d) and e) corresponding XMCD-PEEM images obtained at the energies indicated by the arrows. 61 5 Magnetic interactions in magnetic single domain platelet with cobalt overlayer 5.4 Growth and characterization of cobalt overlayer Once the SFO platelet was analysed, a metallic cobalt overlayer was deposited on top of it by molecular beam epitaxy. The growth was carried out in the UHV PEEM chamber, where the XAS spectra and images are acquired. An electron bombardment doser was used to deposit cobalt from a cobalt metallic rod. A 2 kV di˙erence voltage between the tungsten flament of the doser and the rod was used, and an emission current of 14 mA (corresponding to a heating power of 28 W) provided one cobalt layer for an evaporation time of 4 minutes. The evaporator calibration was carried out previously to the deposition on a ruthenium single crystal. The growth of cobalt was performed at room temperature. The total monolayers evaporated were 7, giving a thickness close to 1.2 nm [172]. The Co L2,3 edge X-ray absorption and dichroism spectra are presented in fgure 5.2a. The XAS spectrum of the cobalt layer is typical of metallic cobalt [70]. This indicates that cobalt has not been intercalated among the surface atoms of the strontium hexaferrite platelet or, if so, it has done to a minimal extent. Otherwise, the oxidized cobalt formed at the interface should be detected in the absorption spectrum giving distinctive XAS, and XMCD spectra [173]. Since atomic di˙usion at room temperature is low, the formation of a metallic cobalt overlayer is not an unexpected result [174, 175]. Based on a comparable work by Kang et al. [26] on Fe/SFO bilayers and considering the growth was carried out at RT, one might expect the soft cobalt layer to be polycrystalline, although it has not been experimentally checked. In addition, the Co L2,3 edge XMCD spectrum indicates a magnetic signal in the cobalt layer. The Fe L2,3 edge XAS-XMCD spectra after cobalt evaporation is shown in fgure 5.2b. It is similar to the iron spectra acquired before cobalt deposition. XMCD XAS 770 780 790 800 810 820 830 Photon energy (eV) X-ray absorption (arb. units) XMCD XAS 700 710 720 730 740 Photon energy (eV) X-ray absorption (arb. units) a) b) FeCo Figure 5.2: a) XAS and XMCD spectra at Co L2,3 edge after cobalt deposition. b) XAS and XMCD spectra at Fe L2,3 edge after cobalt evaporation. 62 5.8 Conclusions However, although in this chapter, the objective of the magnetic coupling between both phases has not been achieved, a very promising result has been obtained. And this is that the micromagnetic simulations suggest that at low exchange couplings, the system’s magnetic properties could be substantially improved, i.e., the hard-soft layers allow aligning the soft layer with the hard phase without facilitating massive domain wall propagation. This development would avoid a loss of coercivity as it was observed in previous works for fully rigid coupled systems [183, 163, 184, 185]. 5.8 Conclusions A bilayer system composed of a single magnetic domain SFO platelet and a cobalt metal layer has been studied by means of X-ray absorption techniques at the ALBA synchrotron in order to study the magnetic coupling between both layers. Cobalt metal layer 1.2 nm thick has been deposited by MBE on strontium hexaferrite platelet at room temperature. Dichroism images at the Co and the Fe L-edges at di˙erent azimuthal angles have determined that the magnetization is in the plane for the soft layer and perpendicular to the sample plane for the platelet. The lack of correlation between the magnetic domains of both phases indicated the absence of exchange coupling between them. Theoretically, this hypothesis was supported by a consideration of the energy contributions involved in the system. The shape anisotropy of the soft layer overcomes the external feld created by the SFO layer. In addition, micromagnetic simulations were carried out that corroborate the absence of the interlayer exchange-coupling as it was seen in the experiment. Furthermore, these simulations provided a possible path for a substantial improvement in the energy product, (BH)max, for hard-soft bilayer systems. It was shown that at low couplings between both layers (κ < 0.25), it is possible to maximize the magnetization without a high loss of the coercitivity of the system. For this exchange coupling range, the magnetization of both layers is aligned while, at the same time, the domain wall propagation activated in more rigid coupling conditions is avoided. 69 6 SrFe12O19 thin flms 6.1 Introduction In the previous chapter, we have studied and demonstrated that in the SFO/soft confguration, the shape anisotropy from the soft layer overcomes the out-of-plane magnetization from SFO layer. In addition, we have proved that there is no alignment between both without the occurrence of interlayer exchange-coupling. In this chapter, we present results on a SFO/Co bilayer system but with the peculiarities that the SFO layer is continuous (thin flm) and presents magnetization within the plane. With this approach, we want to avoid the problem of shape anisotropy from the soft layer. Further, we seek the occurrence of some kind of magnetic coupling in this di˙erent bilayer system (again in the absence of exchange-coupling). The main growth techniques used to prepare strontium hexaferrite flms are pulsed laser deposition [186], chemical solution deposition [187], metal–organic chemical vapor deposition [188], spin coating sol–gel process [189] and RF magnetron sputtering [42]. The latter (and used in this study) can produce SFO flms with magnetic easy axes distributed randomly in plane, or aligned perpendicularly, depending on the sputtering process [190]. This change in the easy axis orientation is attributed to the variation in the c-axis growth direction’s orientation relative to the flm plane. This c-axis orientation can depend on di˙erent deposition parameters such as: substrate temperature, sputtering power, oxygen pressure, annealing conditions [191, 192, 42, 193]. Also, the flm thickness and the growth in an epitaxial intermediate layer are factors involved in the prefered c-axis direction [194, 195]. Here, the infuence of several factors will be discussed in-depth, such as annealing treatment, sputtering power and thickness. The structure of this chapter follows the next points: 1. Growth of SFO thin flms by RF magnetron sputtering and the annealing e˙ect. 2. Magnetization orientation study of the thin flms as a function of the growth parameters. 3. Understanding of the coupling with a Co overlayer in in-plane conditions. Part of the results presented in this chapter has been published in: •G. D. Soria, J. F. Marco, A. Mandziak, S. Sánchez-Cortés, M. Sánchez-Arenillas, J. E. Prieto, J. Dávalos, M. Foerster, L. Aballe, J. López-Sánchez, J. C. GuzmánMínguez, C. Granados-Miralles, J. d. l. Figuera, and A. Quesada, “Infuence of the growth conditions on the magnetism of SrFe12O19 thin flms and the behavior of 71 6 SrFe12O19 thin flms Co/SrFe12O19 bilayers", Journal of Physics D: Applied Physics, vol. 53, p. 344002, 2020. 6.2 SFO target For the growth of the strontium hexaferrite thin flms, we have used radiofrequency magnetron sputtering. As explained in the frst chapter, a plasma is generated, which etches on the substrate, and the material is a target of the compound to be deposited, in our case, SrFe12O19. The target was sintered from commercial SrFe12O19 powders. Initially, the commercial powders were sieved. These were compacted using a 7% by weight solution of 20% paraloid (resin-adhesive) and 80% acetone. Then they were pressed by an automatic pressing machine at 150 kg/cm2 . Finally, they were put in an oven whose temperature was ramped at 5◦C/min from RT to 1250 ◦C and kept at that temperature for three hours. The organic part disappeared at 600 ◦C. The resulting target dimensions were 5.1 cm in diameter and 2.5 mm thickness. In order to determine if the sintered sputter target presented a pure strontium hexaferrite phase and, in consequence, was a suitable target, transmission Mössbauer and Raman spectra were acquired at room temperature (fgure 6.1). The Mössbauer spectrum was ftted to fve sextets. Each sextet corresponds to Fe3+ in a di˙erent crystallographic site[143]. Table 6.1 collects the hyperfne parameters used for the ft. All parameters are characteristic of strontium hexaferrite [131]. Thus, the Mössbauer data indicates that the target is composed only by SFO. The Raman spectrum shows the expected vibration bands for strontium hexaferrite [196, 197, 198, 199]. Specifcally, the Raman modes at 409, 469, 618, 688, and 729 cm−1 map vibration modes arising from di˙erent Fe3+ sites. The peak observed at 409 cm−1 is assigned to the octahedral site (dominated by 12k and 2a), the peak at 469 cm−1 is a mixture of octahedral sites (12k and 2a), 620 cm −1 corresponds to the octahedral sites (4f2), and the peaks at 688 and 734 cm−1 are due to the bipyramidal sites (2b) and tetrahedral sites (4f1) vibration modes, respectively. Hence, again it is confrmed that the target is entirely SFO. 200 400 600 800 1000 Raman Intensity (a. u) Wavenumber (cm-1) SrFe12O19 (1250ºC) 10 5 0 5 10 Velocity (mm/s) 99.70 99.85 100.00 Transmission (%) a) b) Figure 6.1: a) Mössbauer spectrum and b) Raman spectrum recorded at RT from the SFO target. 72 6.3 Annealing e˙ect in SFO thin flms formation Site δ (± 0.03 mms−1) 2ε (± 0.05 mms−1) H (± 0.05 T) Γ (± 0.05 mms−1) Area (%) 12k 4f1 4f2 2a 2b 0.35 0.26 0.39 0.29 0.27 0.38 0.18 0.30 0.02 2.28 40.9 48.8 51.4 50.7 40.6 0.36 0.34 0.32 0.34 0.34 48 18 19 9 6 Table 6.1: 57Fe Mössbauer parameters obtained from the ft of the spectra shown in Figure 6.1a. The symbols δ , 2ε, H, Γ correspond to isomer shift, quadrupole shift, hyperfne magnetic feld and linewidth, respectively. 6.3 Annealing e˙ect in SFO thin flms formation Once the purity of the phase in the target is confrmed, SFO thin flm deposition can be carried out. It has been reported that the growth of the crystalline phase of SFO requires two steps: deposition and post-annealing treatment at high temperature (800◦C - 900◦C) [158, 192, 42]. Several works have demonstrated the need for the annealing step on the sample either in-situ or ex-situ of the deposition chamber [191, 200, 201]. Such studies pointed out, by X-ray di˙raction analysis, that the as-grown sample is amorphous. However, only a few studies discuss the character of the as-grown flm and how it is a˙ected by the annealing step. Snyder et al. [202, 203, 204] reported the local structural anisotropy in the as-grown flm before annealing for barium ferrite thin flms (a compound which is isostructural to SFO) by EXAFS. The data suggested networks of iron atoms surrounded by their oxygen nearest neighbors, with barium atoms ftting into in between spaces. Apparently, this determines the fnal crystalline texture (and the resulting magnetic anisotropy), which forms upon annealing. According to the phase diagram of SrO-Fe2O3 (fgure 6.2), the formation of SFO appears at the composition of 10 % weight of strontium oxide together with iron oxide at elevated temperature [205, 100]. One might expect that the thin flm without annealing step would be composed of the species mentioned (SrO, Fe2O3). Figure 6.2: The binary phase diagram SrO-Fe2O3 reprinted from reference [205]. 73 6 SrFe12O19 thin flms To understand the infuence of annealing on the formation of the strontium hexaferrite structure in our thin flms, two flms were grown in identical conditions using the parameters collected in Table 6.2. Subsequently, one of them was annealed at 850◦C in air for three hours. Substrate Si (100) Temperature 25◦C Target distance 60 mm Base pressure 1x10−6mbar Time pre-sputter 15 min Working pressure 7× 10−3 mbar Ar pressure 1 mbar Ar/O2 ratio 2% Power 260 W Time deposition 5 hours Table 6.2: Parameters used in the deposition process rf-magnetron sputtering. 6.3.1 Compositional and structural characterization The composition and thickness of the di˙erent flms were determined by Rutherford backscattering spectroscopy (RBS) in a 5 MV tandem accelerator using 4He+ at 3.045 MeV, fgure 6.3. This analysis was carried out in the Centre for Micro Analysis of Materials (CMAM). The RBS technique is based on elastic collision between atoms. The ion beam (He+) hits the sample, and depending on the atom it collides with, the backscattered ions will have di˙erent energies. Hence, RBS provides elemental chemical identifcation since these energies are specifc of the scattering atoms and, furthermore, the number of ions reaching the detector at these specifc energies is proportional to the concentration of scattering atoms in the flm examined, allowing for quantitative characterization. A energy of the incident He+ beam of 3.045 MeV was chosen as it is the appropriate both to fnger all the di˙erent atoms in this particular flm as well as to enhance the oxygen cross section (resonant backscattering). The depth distribution and quantifcation of the various elements were determined with the SIMNRA simulation software package [206]. The ion fuence for the Rutherford backscattering and resonant backscattering experiments was set at 10 µC. A silicon barrier detector, at a scattering angle of 170.0 ◦ , collected the backscattering ions while a three-axis goniometer was employed to control the crystal position. First of all, a very similar composition was found for both samples. Both the non-annealed sample as well as the post-annealed flm present the same elements in the spectra: O, Si, Fe, Sr and Ba (fgure 6.3). These atoms come from the SFO flm as well as from the substrate. The small amount of barium arises from a small Ba impurity in the commercial powder. The ft is done using the SIMRA program resulted in good agreement with the experimental spectra. The concentrations of the di˙erent elements are not too di˙erent between both samples. The atomic Fe/O ratio expected for pure SFO is 0.63. In the case of the non-annealed flm, the Fe/O ratio is 0.66, while for the annealed flm is 0.63. Thus, the missing O atoms during deposition are recovered after the annealing treatment. The flms thicknesses have also been determined. The concentration obtained for the annealed sample is 1.53 × 1019 atoms/cm2 . Assuming that this flm is strontium hexaferrite and using a density of 5.14 g/cm3 as corresponds to pure SFO [18], the calculated thickness is 1.62 µm. For the non-annealed sample, the concentration of the atoms is 1.75 x 1019 atoms/cm2 what results, using the same density value, in a thickness of 1.86 µm. The di˙erence in thickness (13%) between both flms probably arises from the systematic error of the growth method. The results above are collected in table 6.3. 74 6.3 Annealing e˙ect in SFO thin flms formation Counts As-grown film Simulated O Si Fe Sr Ba Channel 9008007006005004003002001000 9.000 8.000 7.000 6.000 5.000 4.000 3.000 2.000 1.000 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 Energy [keV] a) b) Annealed film Simulated O Si Fe Sr Ba Channel 1.000900800700600500400300200100 Counts 10.000 9.000 8.000 7.000 6.000 5.000 4.000 3.000 2.000 1.000 0 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 Energy [keV] As-grown film Annealed film Figure 6.3: RBS experimental spectra (red circles) of SFO: a) as-grown flm and b) annealed flm. The simulationed spectrum obtained by the SIMRA code for each sample is defned by blue solid line. The signals corresponding to the diferent elements are depicted in the legend of each spectra. Samples Fe/O ratio Concentration (at/cm2) Thickness (µm) As-grown flm 0.66 1.75 × 1019 1.86 Annealed flm 0.63 1.53 × 1019 1.62 Table 6.3: Parameters calculated for the flms from RBS spectra. The crystal structure and texture of the flms was analyzed by X-Ray Di˙raction (XRD). Figure 6.4 presents the XRD patterns recorded from each sample. The XRD pattern of a commercial SFO powder has been included for comparison. The annealed flm (black pattern) presents all the di˙raction peaks that correspond to a crystalline strontium hexaferrite structure, with the exception of the Si peaks arising from the Si substrate. The higher intensity peaks appear at 30.4◦ , 35.2◦ , 37.1◦ , 54.0◦ and 63.2◦ which match with the (110), (201), (203), (300), (220) Miller indices, respectively of SFO [112, 207]. Additionally, the most prominent peak is the one at 30.4◦ . This suggests a preferential growth of the structure with the c-axis of the SFO parallel to the flm plane. The crystallite size was determined by the Scherrer formula [128], and the lattice parameters by the distance formula between adjacent planes in the set (hkl) for hexagonal crystal structure [48]. The value of particle size is 55 nm and the lattice parameters are a = 5.89 Å and c = 23.49 Å. The unit cell values are in good agreement with the literature [91, 207]. The non-annealed sample shows a few broad di˙raction peaks indicating a low crystallinity of the flm. As it was mentioned previously, some authors found the as-grown samples to be amorphous. Our as-grown flm shows broad di˙raction peaks which do not correspond to SFO. In fact, the di˙raction peaks at 30.1◦ and 66.5◦ , could be associated to magnetite (Fe3O4) or maghemite (γ-Fe2O3) [208, 209, 210, 211, 212]. XRD does not allow to distinguish well between Fe3O4 and γ-Fe2O3, therefore we resort to Mössbauer spectroscopy to achieve a more confdent identifcation (see section 6.3.2). The crystallite size of this flm determined from the XRD data is 12 nm. 75 6 SrFe12O19 thin flms 30 40 50 60 2 (º) Intensity (arb.units) As grown film Annealed film Reference powder (SrFe12O19) θ * ** * * ** * * * * Si Si Si Figure 6.4: XRD di˙raction patterns from the SFO flms: as-grown (blue color) and postannealed (black color), together with the reference powder (bottom one). Raman spectroscopy was then used for the chemical characterization of the flms (fgure 6.5). A reference SFO powder pattern was used for identifcation purposes. The annealed sample shows the bands characteristic of strontium hexaferrite [213, 196, 197, 198, 199]. Such Raman modes were explained in a previous section as arising from the iron cations in di˙erent crystallographic sites of SFO. The as-grown sample shows a vibration mode with a broad peak at 690-700 cm−1 . According to the literature, the broad and most intense peak for maghemite is around 703 cm−1 , which could be assigned to the A1g vibration mode (O-Fe-O bridge) [214, 215, 216]. This correlates well with the XRD data in that the as-grown sample seems to have a maghemite-like structure. 200 400 600 800 Wavenumber (cm-1) Raman Intensity (a.u) Reference powder (SrFe12O19) Annealed film As-grown film Figure 6.5: Raman spectra of SrFe12O19 recorded from the flms and the reference powder. 76 6.3 Annealing e˙ect in SFO thin flms formation 6.3.2 Magnetic characterization From a Mössbauer spectrum, information about the chemical, structural and magnetic state of a particular iron cation can be obtained. In fgure 6.6a, the transmission Mössbauer spectra recorded from the as-grown SFO flm at 298 K (RT) and 35 K are presented. In the spectrum recorded at RT, a paramagnetic doublet is observed instead of the usual fve magnetic components expected for SFO [131]. This doublet presents an isomer shift (δ = 0.34 mm/s) which is characteristic of Fe3+ [217]. The collapse of the magnetic interactions at room temperature is compatible with the poorly crystalline/superparamagnetic character of the as-deposited flm and probably arises from the small size of the flm grains. This assumption is in good agreement with the results observed by Raman and XRD. When the non-annealed sample was measured at 35 K, two magnetic contributions were observed, which can be associated to Fe3+ cations in octahedral and tetrahedral environments. The hyperfne parameters determined from the spectra of the non-annealed sample are collected in Table 6.4. The low temperature spectrum recorded from the non-annealed sample does not correspond, then, with that expected for SFO and resembles clearly that shown by many oxides having an spinel-related structure. In this particular case, the 35 K spectrum, and the corresponding hyperfne parameters, are very similar to those reported for 5 nm maghemite nanoparticles measured at comparable temperatures (40 K) [218]. Both the RT and 35 K Mössbauer spectra endorse the XRD and Raman results and confrm that the iron phase contained in the as-grown flm is superparamagnetic maghemite. Opposite to this, the RT and 35 K spectra recorded from the annealed flm (fgure 6.6b) are fully characteristic of SFO [131], except by the small Fe3+ paramagnetic doublet (3%) observed at RT, which possibly arises from a very minor fraction of SFO/maghemite particles of nanometer dimensions. -10 -5 0 5 10 Velocity (mm/s) 99.6 99.8 100.0 99.8 99.9 100.0 Transmission (%)Transmission (%) RT 35 K 99.2 99.6 100.0 Transmission (%) -10 -5 0 5 10 Velocity (mm/s) 99.7 99.85 100.0 Transmission (%) 35 K RT a) As-grown film b) Annealed film Figure 6.6: Transmision Mössbauer spectra recorded from both flms at RT and 35 K. The asgrown spectrum at 35 K was ftted by 2 sextets corresponding to tetrahedral sites (red) and octahedral sites (yellow). The annealed spectra was ftted by fve sextets in each spectrum corresponding to the sites occupied by the iron cations in the SFO structure. 77 6 SrFe12O19 thin flms The intensity ratio of the absorption lines of the sextets (3:x:1:1:x:3) can be used to determine the average orientation of the magnetization in each sample. The spectrum of the as-grown sample (6.6a) at 35 K is best-ftted using a value of x = 2, indicating a random orientation of the magnetization. For the annealed sample x = 3.5 correspondings to an average magnetization mostly within the plane (15◦). T Site δ 2ε HΓ Area (K) (± 0.03 mms−1) (± 0.05 mms−1) (± 0.05 T) (± 0.05 mms−1) (%) 298 Fe3+ 0.34 0.80 - 0.40 100 35 Fe3+ T d 0.39 -0.03 44.2 0.40 28 Fe3+ Oh 0.46 0.01 49.3 0.40 72 Table 6.4: 57Fe Mössbauer parameters obtained from the ft of the spectra shown in fgure 6.6a. The symbols δ , 2ε, H, Γ correspond to isomer shift, quadrupole shift, hyperfne magnetic feld and linewidth, respectively. Magnetization curves were measured from both samples under a maximum applied external feld of 1.25 T. Figure 6.7 shows the RT hysteresis loops for the as-grown sample and for the annealed-sample, a and b, respectively. In the case of the sample without annealing treatment, no easy magnetic axis nor hysteresis cycles are observed. The noise of the hysteresis loops increases with the applied magnetic feld due to the coupling of other signals resulting from the VSM setup. The shape of the curves can be interpreted in terms of a paramagnetic behavior of the flm as already reported by Mössbauer spectroscopy at room temperature (fgure 6.6), however, given the low quality of the data, it can not be discerned whether the coercive feld is zero or not. The saturation magnetization calculated for this sample is approximately 20 emu/g. For the annealed flm, a clear easy axis is appreciated within the sample plane. These results agree with a crystalline preference of the c-axis oriented parallel to the plane of the flm as indicated by XRD, fgure 6.4 since the magnetization is along that axis. Then data also confrm the in-plane magnetization orientation determined by Mössbauer spectroscopy at RT and 35 K (fgure 6.6b). However, within the plane of the flm, there is not preferential magnetic orientation (red, green and black curves). A possible explanation of this might be that the sample is polycrystalline, and each grain points its magnetization in a di˙erent in-plane direction. The saturation magnetization and the coercive feld determined with the in-plane confgurations were 63.1 emu/g and 0.49 T, respectively. These values are within the range of magnetization values reported for this compound [136, 91]. When the hysteresis cycle was recorded perpendicular to the sample plane, a di˙erent saturation magnetization value was found. This evidence results from the need for a higher applied feld to reach the threshold value of magnetization seen previously in the in-plane measurements. We conclude from the characterization of these thick flms by di˙erent techniques that the annealing treatment is crucial for obtaining strontium hexaferrite. The results suggest a nanocrystalline maghemite phase for the as-grown flm deposited by sputtering and probably with the presence of amorphous strontium oxide. The latter compound was presumed according to the phase diagram for SFO, and due to no crystalline phase was evidenced in the XRD data, we suggested an amorphous state for SrO. 78 6.4 Thin flms -1 0 1 Magnetic Field (T) -100 -50 0 50 100 M (emu/g) Out-of-plane In-plane 90 º In-plane 45 º In-plane 0 º -0.2 0 0.2 Magnetic Field (T) -80 -40 0 40 80 M (emu/g) In-plane Out-of-plane a) b) Figure 6.7: Room-temperature hysteresis loops recorded from as grown SFO flm and postafter annealing treatment. The blue curve was recorded with a magnetic feld applied perpendicular to the sample. Red, green and black curves were recorded with a magnetic feld applied parallel to the sample. 6.4 Thin flms Knowing the importance of the annealing step for growing good quality SFO thin flms, several strontium hexaferrite samples were deposited to achieve one of the goals of this work: create a thin flm with a magnetization easy axis parallel to the sample plane. For this purpose, the samples were grown by RF magnetron sputtering using the parameters specifed in the previous section 6.2. The only factor which was varied in the growth method of each thin flm was the sputtering power (140 W, 180 W, 220 W and 260 W). The thin flms were deposited for 30 minutes sputtering time. All samples were then annealed at 850 ◦C for three hours. To determine their thickness, the step between the substrate and the flm was measured by means of a proflometer. As it can be seen in the inset of fgure 6.8, the amount of sample deposited in a certain deposition time increases linearly with the sputtering power. 200 400 600 800 1000 Distance ( m) 0 100 200 300 400 Thickness (nm) 140 W - 160nm 180 W - 240nm 220 W - 310nm 260 W - 360nm 100 150 200 250 300 Sputtering power (W) 4 6 8 10 12 14 Thickness/Time (nm/min) m Figure 6.8: SFO flms thicknesses measured by a proflometer. Inset: thickness-time ratio as a function of the sputtering power (sputtering time, 30 min). 79 6 SrFe12O19 thin flms other hand, sample B show an increased density of columns (60 to 95 nm tall), Fig.6.14b. The RMS roughness in this flm is 8 nm. 2 µm 129 nm 0 40 60 80 100 a) b) 2 µm 95 nm 0 20 30 40 50 60 70 80 Figure 6.14: a) and b) surfaces of the samples A and B, respetively. The vertical sizes of the pictures are 5 ×5µm. Taking into account these AFM images with those from the previous section (images 6.11b and 6.11d), we have suggested how the protrusions distribution varies in function to sputtering power and thickness. According to Sui´s works [221, 222] this grain arrangement can be related to the c-axis orientation in the structure. At low values of sputtering power, typically, the grains grow in a more pronounced and isolated columnar shape. When the flm thickness is small, isolated tall grains are observed. As the thickness of the flm increases, the grains tend to change their orientation within the plane. At high sputtering power values, a more homogeneous columnar growth, although of lower height, was found. For this case, at small thicknesses, we suspect that part of the grains tends to grow perpendicular to the plane while another part parallels it. The flm thickness increase causes the majority of the grains to be oriented in the sample plane. Assuming that grain preferential orientation determines, to some extent, the change of the average c-axis orientation of SFO flms, the thickness and sputtering power are parameters involved in such change. The magnetic behaviour in this new set of samples was also studied by Mössbauer spectroscopy. Figures 6.15a and 6.15b show the CEMS spectra recorded from Sample A and Sample B, respectively, which are both characteristics of strontium hexaferrite [131]. The area ratio of the spectral lines obtained from the ft of the spectrum recorded from Sample A was 3:3.5:1:1:3.5:3. This implies that the sample shows a magnetization practically in the flm plane (magnetization/surface angle of 15 ◦). This is remarkable comparing this result with that shown in fgure 6.15c, which corresponds to a flm produced with the same sputtering power but having a much smaller thickness, a signifcant di˙erence is observed in the averaged orientation of the magnetization. However, correlating the sample A spectrum with that from the flm with the same thickness (360 nm) but di˙erent sputtering power (fgure 6.15d), a similar average magnetization orientation is detected (18 ◦). In the case of Sample B (fgure 6.15b), an area ratio of 3:2.2:1:1:2.2:3 was obtained, which corresponds to a magnetization having a signifcant out-of-plane magnetization component (average 33◦). In comparison with a flm grown with the same sputtering power (260 W) but thicker, we can observe the di˙erence in the average magnetization orientation between both (fgure 6.15d). A di˙erence in the magnetization direction is also seen when comparing sample B spectrum with the flm grown at lower sputtering power with the identical 86 6.5 Infuence of thickness and power sputtering on the magnetic behaviour of thin flms thickness (fgure 6.15c). So, despite the low sputtering power used, a thicker flm makes the magnetization practically in-plane (Sample A). In contrast, a larger sputtering power but a smaller thickness (Sample B) gives place to an intermediate situation in magnetic behaviour, implying that the thinner flm contains a signifcant out-of-plane magnetization component. Again, considering that the net magnetization is along the c-axis, Mössbauer spectra present the same behaviour as the di˙raction data, fgure 6.13. a) 15º -10 -5 0 5 10 Velocity (mm/s) 100 102 104 Effect (%) b) 33º -10 -5 0 5 10 100 101 102 103 Velocity (mm/s) Effect (%) c) -10 -5 0 5 10 Velocity (mm/s) 100 102 104 Effect (%) d) -10 -5 0 5 10 Velocity (mm/s) Effect (%) 55º 100 102 104 106 Sample A (140 W – 360 nm) Film grown at 140 W (160 nm) Sample B (260 W – 160 nm) Film grown at 260 W (360 nm) 18º Figure 6.15: Mössbauer spectra for SFO flms: a) Sample A, b) Sample B, c) flm grown at 140 W (160 nm) and d) flm grown at 260 W (360 nm). Therefore, the present results indicate that the sputtering power used during deposition is not, by itself, the main factor that determines the magnetization orientation and that the thickness and consequently the morphology of the sample are also crucial. The magnetization orientation in thin flms depends on the interplay among the shape anisotropy of the grains, the shape anisotropy of the overall deposited flm and the magnetocrystalline anisotropy of the SFO flms, the latter term being most probably the dominant one. In any case, within the context of these experiments in which we are using sputtering powers between 140 W and 260 W and thicknesses between 160 nm and 360 nm, the results indicate that: i At low sputtering powers, the thickness is the determining parameter since the magnetization orientation can be tuned by an appropriate choice of the flm thickness. ii At high thicknesses, the sputtering power does not infuence the magnetization due to this is oriented in the sample plane for both cases. iii At low thicknesses, the sputtering power a˙ects moderate the magnetization direction of the sample. 87 6 SrFe12O19 thin flms 6.6 Interaction between strontium hexaferrite thin flm with cobalt overlayer In this section, we will consider the sample deposited at 260 W and having a thickness of 360 nm, which has its magnetization oriented within the flm plane. As mentioned previously, one of the stated objectives of this research is to obtain a SrFe12O19 sample with magnetization in the plane to study its magnetic coupling with a magnetically soft layer specifcally with a thin cobalt layer deposited on top. This coupling experiment was decided in order to avoid the competition between the shape anisotropy arise cobalt layer and magnetocrystalline anisotropy from SFO platelet observed in the previous chapter. Therefore, the sample grown at 260 W is a suitable candidate for such experiment. Hysteresis loops were measured by VSM to confrm the magnetic orientation determined from the CEMS spectrum, fgure 6.16. The black curve was measured by applying a magnetic feld parallel to the sample plane, while the red curve was recorded by applying a magnetic feld perpendicular to the sample plane. For the black curve, the measured coercive feld is 0.42 T, and the saturation magnetization is achieved for an applied feld of 1.8 T. From the red curve, a smaller coercive feld (0.37 T) is measured while the saturation magnetization is reached at a slightly larger applied feld (2.3 T). Figure 6.16 also shows a larger remanence magnetization in the black curve than in the red one. These observations confrmed that the flm has its magnetic easy axis, mainly in the sample plane. -4 -2 0 2 4 Magnetic Field (T) -1 0 1 M / M s In plane Out-of-plane r Mr/Ms = 52% Hc = 0.42 T Mr/Ms = 36% Hc = 0.37 T Figure 6.16: Room-temperature hysteresis loops recorded from a 360 nm thick SFO flm deposited at 260 W after annealing. The black curve was recorded with the magnetic feld applied within the plane. Red curve recorded with a magnetic feld applied perpendicular to the sample. Then, a cobalt layer two nm thick was evaporated using MBE in the UHV chamber of CIRCE beamline ALBA Synchrotron (Barcelona). Firstly, the characterization of the bilayer system was carried out by X-ray absorption analysis. XAS and XMCD spectra and images were measured at Fe L2,3 edge on the strontium hexaferrite thin flm before and after deposition of the cobalt layer (fgure 6.17a). The upper left image from fgure 6.17a corresponds to XMCD image acquired at the Fe L2,3 edge before depositing the cobalt overlayer and the upper right image once the cobalt was 88 6.6 Interaction between strontium hexaferrite thin flm with cobalt overlayer grown. Both images show a magnetic contrast as expected from XMCD spectra signal (fgure 6.17a lower panel). It should be noted that XAS spectra recorded at Fe L2,3 before and after cobalt growth are slightly di˙erent. An additional feature is observed at a lower photon energy (706.9 eV) in the spectra of the Co-covered SFO flms. In fgure 6.17b XAS and XMCD spectra and XMCD image acquired at the Co L2,3 edge are shown. XAS spectrum is quite similar to that shown by metallic cobalt [172]. However, the spectrum shows minor peaks at 776.0 eV and 778.5 eV, which are compatible with the presence of some Co2+ arising from the oxidation of a fraction of metallic cobalt during deposition. This evidence, together with the XAS spectrum obtained in the SFO flm once cobalt is deposited, suggests the following: the cobalt evaporated on the strontium hexaferrite flm interacts with oxygen atoms from the SFO flm producing oxidation of part of cobalt atoms and consequently a reduction of some Fe3+ of the SFO to Fe2+ at the interphase. Note also the two peaks appearing beyond the Co L3 (784.0 eV) and L2 edges (797.5 eV). They correspond to a residual barium contamination [172, 223, 224] stemming from the SFO target used in the magnetron sputtering system. 770 780 790 800 810 820 X-ray absorption (arb.units) Photon Energy (eV) XAS XMCD a) b) 700 710 720 730 740 Photon Energy (eV) X-ray absorption (arb.units) XAS Before XAS After XMCD Before XMCD After Figure 6.17: a) Upper: X-ray magnetic circular dichroism images at Fe L3 edge before(left) and after(right) cobalt deposition. Lower: XAS and XMCD spectra at Fe L2,3 edge before and after cobalt exposition. b) Top: X-ray magnetic circular dichroism images at Co L3. Botton: XAS and XMCD spectra at Co L2,3 edge. In order to fnd whether the addition of a cobalt overlayer infuences at some extent the average magnetization orientation and the magnetic properties of the SFO flms, Mössbauer spectroscopy (ICEMS) and hysteresis loop data were recorded from the Co/SFO bilayer. The ICEMS spectrum recorded at room temperature is found in fgure 6.18a. As well as for the other strontium hexaferrite samples studied, the spectrum was ftted to the fve sextets (fgure 6.12). Similar spectral areas were obtained for the di˙erent components to 89 6 SrFe12O19 thin flms those of the sample without the cobalt coating. The area ratio of the sextet lines is 3: 3.4: 1 in all cases, which corresponds to an average magnetization orientation virtually within the plane (16◦). This magnetization orientation is comparable to that registered for the SFO sample without cobalt. The recorded hysteresis loop of this bilayer, fgure 6.18b, also suggests a magnetization lying preferably in the sample plane. Furthermore, the values of the di˙erent magnetic flm properties (remanent magnetization/saturation magnetization ratio, Mr/Ms, and the coercive feld, Hc) are comparable with those acquired from the single strontium hexaferrite for both in-plane and out-of-plane hysteresis loops (fgure 6.16). In the in-plane data, the coercive feld is 0.42 T, and the magnetization is saturated under an applied feld of 1.8 T, while in the out-of-plane confguration, the coercive feld is 0.33 T, and a larger magnetic feld is required to magnetize the sample completely (2.2 T). Both results indicate no signifcant change in magnetization with the deposition of the metal layer. This fact is not surprising given the small thickness of the cobalt layer (2 nm) compared to the SFO thin flm (360 nm). 10 5 0 5 10 Velocity (mm/s) 100.0 101.5 103.0 104.5 Effect (%) 16º -4 -2 0 2 4 Magnetic Field (T) -1 0 1 Mr/Ms In plane Out-of-plane Mr/Ms = 55% Hc = 0.42 T Mr/Ms = 31% Hc = 0.33 T a) b) Figure 6.18: a) Room temperature ICEMS spectrum recorded from the Co/SFO bilayer. The arrow in the upper right side represents the average angle of the magnetization direction respect to the sample surface. b) Room temperature hysteresis loops recorded the from Co/SFO bilayer system. The blue curve was recorded with a magnetic feld applied within the sample surface plane, while the yellow curve was recorded with a magnetic feld applied perpendicular to the sample. To elucidate the nature of the coupling between the soft magnetic layer (cobalt) and the hard magnetic thin flm (SFO), a vector magnetization map was measured at the Fe-L3 X-ray absorption edge before cobalt deposition (fgure 6.19a) and at the Fe-L3 (fgure 6.19b) and Co-L3 X-ray absorption edges (fgure 6.19c) after cobalt deposition. The vector magnetization is proportional to the magnetization in each layer, and it can be obtained by combining three XMCD images along three non-coplanar directions with azimuthal angles of 0◦ , 60◦ and 120◦ , and a polar angle of 16◦ [156]. There was no signifcant change in the SFO domains upon cobalt deposition: fgure 6.19f shows a comparison between the Fe L3 X-ray absorption edge of SFO fgure 6.19b shown in f as a color pattern and fgure 6.19a shown in f as the contour. The most prominent result is that the SFO flm shows a uniaxial anisotropy large enough as to show a magnetization vector preferentially aligned in the directions (160◦ and 340◦) within the flm plane, fgure 6.19d. No correlation is apparent between the cobalt domains and the SFO ones (fgure 6.19c). Hence, the magnetic domains of the hexaferrite layer are not imposed on the cobalt layer. This strongly suggests a lack of exchange-coupling between the hard and soft layers. Nonetheless, it has been determined that the easy axis of the cobalt layer is the same as the in-plane easy axis of the hexaferrite layer (fgure 6.19e). This is evidenced by the same colors (green and red) in fgures 6.19b 90 6.6 Interaction between strontium hexaferrite thin flm with cobalt overlayer and 6.19c, with the magnetic domains of cobalt pointing either parallel or antiparallel sense to the underlying hard domains. Thus, the point to be debated in this context is: in the absence of interchange coupling between both layers, what mechanism intervened so that an alignment of the easy axes occurs. First, let us consider the hexaferrite structure in the [110] direction, which is the flm growth direction. Strontium hexaferrite is a ferrimagnet, with iron cations in some crystallographic sites pointing along the net magnetization direction and others pointing in the opposite direction. Along (110), there are planes within the unit cell with di˙erent populations of each iron cation. Thus the net surface magnetization is opposite along di˙erent planes. If some grains of the hexaferrite flm present such di˙erent terminations, a dipolar magnetic coupling with the cobalt layer should give rise to grains where the magnetization of the cobalt, coupled to the net surface magnetization of the hexaferrite, has opposite directions between di˙erent grains. a) b) c) Fe L3 before cobalt deposition Fe L3 after cobalt deposition Co L3 0° 45° 90° 135° 180° 225° 270° 315° d) 90º 0º 270º 180º 45º 315º 135º 225º e) 0° 45° 90° 135° 180° 225° 270° 315° f) g) after and before Iron and cobalt layers 3 3 3 Fe L3 cobalt deposition Figure 6.19: Vector magnetization maps for: the Fe L3 X-ray absorption edge before a), after b) cobalt deposition and c) the Co L3 X-ray absorption edge. d) and e) polar plots representing the magnetization distribution in the surface plane extracted from a) and c) images, respectively. f) Overlay images at Fe L3 X-ray absorption edge before and after cobalt deposition. g) Overlay images at Fe L3 and Co L3 X-ray absorption edges after cobalt deposition. Note that the color palette in the upper corner represents the spin direction in the magnetic domains. 91 6 SrFe12O19 thin flms Figure 6.20 shows a scheme of the situation discussed above. The iron cations corresponding to the di˙erent chemical environments have been represented in blue or yellow spheres according to the direction of their spins. Assuming di˙erent terminations in the surface grains, the cobalt deposited on top should have the same magnetic orientation as the termination where it is found due to the magnetostatic interaction. Thus, for the lowest terrace (to the right), the iron cations (blue) present their spin oriented in one direction, so the upper cobalt should also be oriented in the same direction. For the middle terrace, the iron cations (yellow) have their spin in the opposite direction, and consequently, the coupling promotes such magnetization orientation in the cobalt layer. However, this scenario is considered unlikely for several reasons. On the one hand, the flm needs to present a substantial number of grains with di˙erent terminations, although the crystal termination is often determined by the lowest surface energy. On the other hand, this situation would require a correlation between the domain walls in the cobalt overlayer and the hexaferrite layer, even if the magnetization were coupled either ferromagnetically or antiferromagnetically (depending on the underlying termination). No such correlation is observed in fgure 6.19g. Co Fe Fe O Sr Figure 6.20: Strontium hexaferrite structure with di˙erent terminations. The elements are depicted in the legend. The arrows on top of Co atoms represent the magnetization direction in the soft layer. The explanation we propose is that both directions (but not sense) of the magnetization in the two layers are coupled structurally. This could happen in two ways. On one side, the strain imposed at the interface by the ferrite layer could favor a specifc growth direction of the cobalt layer that leads to the alignment of the easy axes. On the other, the epitaxial relationship alone could explain this alignment as well. Given the very low thickness of the cobalt layer, our data, especially XRD, are not able to confrm the epitaxial relationship. Thus, it was suggested that the coupling between the two layers is structural instead of magnetic. This result is not entirely surprising, as, for instance, growth of cobalt on W(110) always produces some uniaxial anisotropy of the cobalt layer [225, 226] and this does not require di˙erent surface terminations. A consequence of this interpretation, already suggested by the lack of correlation between domain patterns in the two layers, is that in the absence of exchange-coupling, dipolar interactions alone do not lead to the alignment of the spins of the soft layer with the magnetization of the hard one. In order to further understand the magnetic behavior observed in the bilayer PEEM images, micromagnetic simulations have been performed [86] in a simplifed system. Such system 92 6.6 Interaction between strontium hexaferrite thin flm with cobalt overlayer consists of a strontium hexaferrite slab having a well-defned in-plane magnetization easy axis (100) covered by a layer of cobalt on top. The cobalt overlayer has its easy axis of magnetization oriented along the same direction (100) as the easy-axis of the hexaferrite layers. The thickness of the two layers was set to that shown by the experimental sample studied in PEEM (SFO thin flm 360 nm thick with a 2 nm cobalt overlayer). The following set of magnetic parameters was thus used as input in the simulations: exchange sti˙ness of hard and soft phases: As(SF O) = 6 × 10−12Jm−1 and As(Co)=1.5 × 10−11Jm−1 from ref.[91, 179]; saturation magnetization Ms(SF O)=3.8×105 Am−1 and Ms(Co) = 1.4×106 Am−1 from ref. [108, 168] and magnetocrystalline uniaxial anisotropy Ku(SF O)=3.6×105 Jm−3 and Ku(Co)=4.1 × 105 Jm−3 from ref. [158, 227, 166]. A random multi-domain structure for the cobalt layer and a single domain confguration for the hexaferrite layer was used. Then, the confguration was relaxed. The frst simulation was carried out, considering no exchange coupling between the two layers. It was observed that the magnetic domains in the soft and the hard magnetic layers are indeed not correlated (fgure 6.21a). The hard magnetic layer (SFO) shows only one magnetic domain (color red in the fgure) that corresponds to the orientation of the magnetization at 0◦ , while in the soft magnetic layer (cobalt), there are two magnetic domains (blue and red) that represent the spin in the same directions but in the opposite sense, 180◦ and 0◦ , respectively. The second simulation was performed, incorporating a 25 % interlayer coupling. In this case, the domains in the cobalt and ferrite layers are totally aligned (fgure 6.21b). For both layers, a single magnetic domain (red color) can be seen. For the third case, no magnetocrystalline anisotropy direction in the cobalt layer was set and the simulation running in the absence of exchange coupling with the SFO layer. This micromagnetic simulation showed that the soft layer presents magnetic domains within the plane in all directions irrespectively of the orientation of the magnetic domain of the hard ferrite layer (fgure 6.21c) if there is no exchange coupling between both layers. Specifcally, the strontium hexaferrite layer has a single magnetic domain (red color), while the cobalt layer exhibits di˙erent magnetic domains representing the distinct magnetizations (rainbow). The simulations thus support that the dipolar coupling is unable to impose the magnetization domain pattern of the hard layer onto the soft layer in the thickness range investigated in the absence of a correlation between the easy axis of both layers. (a) (b) (c) Figure 6.21: Micromagnetic simulations of bilayer SFO/Co: a) without exchange coupling, b) with 25 % exchange coupling and c) cobalt without magnetocrystalline anisotropy and without exchange coupling. Note that the color palette in the upper corner represents the spin direction in the magnetic domains. 93 6 SrFe12O19 thin flms 6.7 Conclusions First, we have determined that the annealing step after growing SFO results crucial for obtaining a genuine SFO flm having the right composition and crystal structure. To this end, two samples with and without annealing treatment were characterized by di˙erent spectroscopic and microscopic techniques. The results for the sample as-grown indicate that it is composed of nanosized maghemite. Strontium is presumed to be present forming part of SrO amorphous oxide. Subsequently, SFO thin flms were deposited at several sputtering powers with constant deposition times (and thus di˙erent thickness) in order to adjust the net magnetization orientation. XRD data have shown that the flms are textured and that their structural orientation changes with the sputtering power and thickness. Mössbauer spectroscopy was used to determine the average magnetic easy axis in each thin flm. Taken together, the Mössbauer and XRD data suggest that the magnetization of the SFO flms is oriented along the c-axis growth direction. SFO flms grown at the highest power used (260 W) showed preferential in-plane magnetization. Trying to understand the role of thickness and sputtering power in the magnetization easy axis orientation from the samples studied, two new thin flms were grown. AFM images, x-ray di˙ractions and Mössbauer spectra for the new samples pointed out that both factors were involved in the c-axis direction of SFO flms and then the magnetization orientation of them. To obtain a hard-soft bilayer system, a cobalt layer was deposited on an SFO flm grown at a power of 260 W (360 nm thickness) by molecular beam epitaxy. The characterization of the bilayer was performed using synchrotron PEEM. The XAS spectra at the Co and Fe-L2,3 edges revealed the presence of a small amount of Fe2+ and Co2+ likely due to the interaction of the deposited cobalt with oxygen atoms from the SFO surface. Acquiring XMCD images at Co and Fe-L2,3 edges, we obtain vector magnetization maps showing the magnetic domains associated with each layer. The polar graphs calculated from 3D magnetization images indicated the same uniaxial easy axis for both the SFO flm and cobalt layer. However, the magnetic domains in the cobalt overlayer are not correlated with the magnetic domains in the SFO surface. This result suggests a lack of exchangecoupling between the layers. We suggest the coincidence in the direction of the uniaxial easy axis arises from a structural coupling. This is further supported by micromagnetic simulations, which confrmed that dipolar interactions alone do not lead to an alignment of the soft spins with the hard layer magnetization. 94 7 CoFe2O4 ultra-thin flms 7.1 Introduction Cobalt ferrite, CoFe2O4 (CFO), has attracted much interest since it shows a variety of electronic and magnetic properties such as the magnetoelectric e˙ect, the magnetooptical e˙ect, and others, resulting in its prospective application for data storage and switching devices, actuator and transducers, hyperthermia applications and magnetic feld [15, 228, 229, 17, 230, 231, 232, 233]. CoFe2O4 leads to the applications mentioned above mainly due to its high magnetocrystalline anisotropy constant, which is larger by over an order of magnitude than other spinel ferrites (K1 in the range of 2-4×105 J/m3), its large magnetostriction constant among all iron magnetic oxides, its high Curie temperature and its large saturation magnetization for a ferrite [234, 235, 15]. Much of the research has centered upon the magnetic properties of polycrystalline and monocrystalline bulk or thin flm materials [236, 237, 238, 239]. Specifcally, in this chapter, we will focus on CFO flms. CFO flms have been grown by many di˙erent methods, such as by the sol-gel process [240, 241], by dual ion beam sputtering [242, 243], by pulsed laser deposition [244, 245, 246], by magnetron sputtering [247, 248, 244], and by molecular beam epitaxy, to cite a few. The latter has been employed by depositing cobalt and iron in atomic oxygen [249, 250] or molecular oxygen [251, 252], using post-oxidation steps of metal layers [253], depositing cobalt on magnetite [254] or even annealing oxide layers [255, 256]. In the present chapter, CFO thin flms have been deposited by oxygen assisted molecular beam epitaxy on Pt(111) at 523 K and subsequently annealed up to 723 K. Many of the studies about cobalt ferrite thin flms have been carried out using oxides or insulator as substrates. Unlike for other spinel ferrites, only a few works, such as that carried out by Santis et al. [253], report the growth of CFO on a metallic substrate, in their case on Ag (100). Thus, the use of a metallic substrate (platinum) for the growth of thin layers is something relative unexplored and important to consider as it opens the door for spintronics applications. Also, the growth and annealing temperatures were chosen in order to prevent the formation of multiphasic flms, island growth, or dewetting phenomena [251, 252]. The dependence of the magnetic properties with the growth conditions and subsequent annealing treatments has been already reported [257]. As we will see in the following section, cobalt ferrite displays a spinel structure in which the cation distribution depends on the growth method and particular growth conditions. Therefore, the main purpose of this work is to understand the structural and magnetic properties of CFO samples with di˙erent thicknesses (5 nm and 20 nm) grown by MBE on platinum metallic substrates and determine the e˙ect on these properties of di˙erent 95 7 CoFe2O4 ultra-thin flms shown in the Auger spectra which could be readily accommodated in such a FexCo1−xO termination. We hence suggest that our conditions are such as to promote a rocksalt termination of the flm. The surface morphology of the as-grown and post-annealed treatment at (673 K and 773 K) ferrite flms were observed by scanning tunneling microscopy. A direct real-space image of a surface is achieved by moving a tiny metal tip across the sample surface and recording the electron tunnel current between tip and sample as a function of position [52]. Figure 7.5a shows an STM image recorded from the as-grown flm. The image shows the presence of aggregates/particles with a size around 15–20 nm, with surface root mean square (RMS) roughness of 0.3 nm. Figure 7.5b, which corresponds to the previous sample after annealing up to 773 K in UHV, shows a flm with somewhat larger clusters of 20–25 nm and slightly larger 0.4 nm RMS roughness. These results indicate that with increasing temperature, the particle size and roughness on the surface of the flms increases. A previous study by Lee et al [241] had already addressed the dependence of surface roughness on annealing temperatures. This work pointed out that the surface roughness increases linearly with increasing annealing temperature. Furthermore, Oujja et al. [245] reported increasing particle size with increasing temperature in good agreement with our results. 50 nm 50 nm a) b) Figure 7.5: STM images of the 20 nm CFO thin flm: a) as grown and b) annealed to 773 K. 7.4.2 Magnetic characterization To determine the change in the magnetic behavior experimented in the 20 nm CFO thin flm after the di˙erent treatments, Mössbauer spectroscopy has been used. However, the interpretation of the Mössbauer spectra is complicated as the data can be ftted in various reasonable ways. In fact, we will show how the Mössbauer spectrum recorded from the as-grown sample at room temperature can be ftted using di˙erent models, presented in fgure 7.6. The spectra were ftted using the Recoil program with Lorentzian multiple analysis (a) and Voigt-based ftting (VBF) (b)2 . The second ft model was proposed by 2The Lorentzian multiple analysis is taking into account the analysis with Gaussian singlets, doublets or sextets according to paramagnetic components with or without quadrupole splitting and magnetic components with magnetic hyperfne felds. The Voigt-based ftting allows the use of a distribution of hyperfne parameters. This corresponds to the Gaussian sum for representing a quadrupole splitting distribution in paramagnetic sites or magnetic hyperfne distribution in magnetic sites. 102 7.4 20 nm thin flm our collaborators from "Jerzy Haber Institute of Catalysis and Surface Chemistry" for the discussion of the research results. The values of the hyperfne parameters for the various components obtained from each ft are collected in table 7.2. Fit a, fgure 7.6a, shows strong paramagnetic signals (one singlet and two doublets) in its central part and a magnetic component with broader lines. The paramagnetic singlet corresponds to residual metallic iron contamination present in the platinum substrate. Such contribution will be constant in the spectra studied hereinafter. The hyperfne values of the doublets point out to Fe3+ in distorted octahedral coordination and Fe2+ also in octahedral environment. The magnetic contribution exhibits hyperfne parameters characteristic of Fe3+ . In ft b, fgure 7.6b, a doublet and two contributions associated with magnetic components are obtained. On the one hand, the doublet corresponds to Fe3+ taking into account a quadrupole splitting distribution. On the other hand, the high-contribution magnetic component presents a very low hyperfne magnetic feld distribution to be able to adjust the asymmetric peak the central part while another magnetic component adjusts the broad magnetic pattern. Both magnetic components present an isomer shift value compatible with Fe2.5+ . Thus, like the previous ftting model, this ft shows Fe3+ and a certain concentration of iron in 2+ oxidation state. Despite the di˙erences in the two fts, it can be deduced from both fts that the spectrum presents a paramagnetic part together with magnetic contributions, caused probably by a lack of structural cationic order/small size particle distribution. This is supported by the STM image for the as-grown thin flm, where very small diameter particles (sizes between 10 and 15 nm) are observed. To understand the discussion raised in the Mössbauer spectra, we will briefy explain what superparamagnetism consists of. The particle size, as well as the temperature, are factors that can modify the magnetic behavior of the system. A ferromagnetic compound that must be magnetically oriented at a certain temperature, due to the small volume of the particles, exhibits a paramagnetic state (superparamagnetism). However, when it is measured at lower temperatures, it does present ferromagnetic behavior. The threshold temperature in which both states coexist is called the blocking temperature (TB). Above TB , thermal excitations cause continuous changes in the orientation of the magnetization for individual particles, and therefore, there is a distribution of magnetic orientations. Below TB , the magnetic moments are frozen. In the Mössbauer spectra, this change in magnetic behavior for small particles is refected by a transition from a doublet (superparamagnetic) to a sextet (ferromagnetic/antiferromagnetic) with the temperature decrease. In addition, the particle size distribution causes line broadening since the moment alignment for all particles does not occur at the same temperature. One of the frst works about superparamagnetic e˙ect in nanoparticles was carried out by Schuele et al [276] where CoFe2O4 and NiFe2O4 ultrafne particles (30 to 200 Å) were studied. In the literature, we found many publications on this topic [62, 277]. In the case of thin flms, di˙erent authors have pointed out this superparamagnetism behaviour associated with the small size of the grains in their samples. J.G. S. Duque [240] reported this phenomenon for CFO flms with particle sizes between 10-20 nm. Also, López et al. [278] commented on the infuence of the grain size with an average size of 103 7 CoFe2O4 ultra-thin flms 10-40 nm in their CoFe2O4 flms. In addition, Yanagihara´s group [279] working on 13 nm thick cobalt ferrite flms on α-Al2O3 (0001) reported the occurrence of a broad magnetic component which was interpreted as a result of a thermally fuctuating magnetic order near the critical temperature or the blocking temperature associated to a superparamagnetic character of the flms. In any case, the type of ftting model chosen for the as-grown spectrum has been (a) because the second one might contribute to a more complicated interpretation associated with magnetite formation and possible electron hopping processes. -12 -8 -4 0 4 8 12 Velocity (mm/s) Effect (%) 10 0.0 104.0 106.0 102.0 -12 -8 -4 0 4 8 12 Velocity (mm/s) b) a) Figure 7.6: Mössbauer spectrum from as-grown sample measured at RT resolved for two di˙erent models ft: a) Lorentzian multiple analysis and b) Voigt-based ftting. Spectrum Site δ (± 0.03 mms−1) 2ε (± 0.05 mms−1) H (± 0.05 T) Area (%) As grown a) As grown b) Fe0 Fe2+ Fe3+ Fe3+ Fe3+ Fe2.5+ Fe2.5+ 0.33 1.25 0.41 0.43 0.35 0.56 0.53 - 0.75 0.80 0.02 0.75 (δ 0.414) 0 0 - - - 25.1 - 9.3 (δ 7.0) 42.8 (δ 6.5) 11 14 58 17 41.2 43 15.8 Table 7.2: 57Fe Mössbauer parameters obtained from both fts of the as-grown sample spectrum measured at RT from fgure 7.6. The symbols δ, 2ε, H, and δ correspond to isomer shift, quadrupole shift, hyperfne magnetic feld, and distribution of the hyperfne parameter, respectively. The isomer shift values are quoted relative to α-Fe at room temperature. Figure 7.7 shows the Mossbauer spectra measured at RT and 125 K after the sample has been heated to 673 K and 773 K in UHV. In the spectrum measured at RT (fgure 7.7a), a notable di˙erence is observed as compared with the spectrum of the as-grown spectrum. The annealing treatment causes a rearrangement and short distance di˙usion of nearby atoms promoting the crystallization as well an increase of the aggregates sizes in the thin flm. This is refected in the spectrum as defned magnetic components, although some magnetic relaxation is still observed in the sample responsible for broadening the lines. For this reason, it is necessary to ft the spectrum using, in addition to discrete components, a hyperfne magnetic feld distribution. Therefore, for this type of ftting, code NORMOS [280] is used. In this spectrum, two discrete sextets corresponding to Fe3+ in tetrahedral sites and Fe3+ in octahedral sites can be seen as well as broad magnetic component ftted by a hyperfne 104 7.4 20 nm thin flm magnetic feld distribution of Fe3+ cations. Furthermore, the spectrum also presents the paramagnetic components already previously observed in the as-grown sample from ft (a): a singlet that refers to Pt metallic and the two doublets that correspond to Fe3+ and Fe2+ . It should be noted that the Fe3+ doublet is less intense here in comparison with that of the as-grown sample. This can be understood by the increase in particle size upon annealing. In addition, the presence of a Fe2+ contribution supports the possible explanation of rock salt terminations suggested by the LEED pattern. Discrete sextets can be associated with the expected sites for oxides with a spinel-like structure [60]. However, the cobalt ferrite presents a ratio between iron in tetrahedral and in octahedral sites Fe3+/Fe3+ equal to 1, while for this spectrum, it was obtained a A B value of 3. It is well-known that the cation distribution of cobalt ferrite cannot be precisely determined from its RT Mössbauer spectrum. In general, the tetra/octa site ratio is usually overestimated from a such type of measurement. Several publications from our group have discussed this issue in detail [246, 243, 264, 281]. This is mainly due to the strong overlap of the sextets corresponding to Fe3+ in both sites, the result being very much dependent on the constraints imposed to the linewidths of both sextets during ftting. In general, the tetrahedral sextet tends to be broader, and this appears to be related with the occurrence of supertransferred magnetic felds in the spinel structure [282]. Due to the supertransfer mechanism, a signifcant percentage of the Fe3+ at the octahedral sites experience hyperfne magnetic felds, which can be very similar and even smaller than the average hyperfne feld felt by the tetrahedral Fe3+ cations. This broadens the tetrahedral sextet, and, thus, it results in an area that is larger than that expected. The spectrum of the sample measured at 125 K (fgure 7.7b) was ftted using NORMOS program too. This spectrum shows less superparamagnetic relaxation because the sextets appear to be much better resolved and the contribution of the hyperfne magnetic felds distribution is reduced. However, the measurement temperature should be lowered further to completely remove superparamagnetic e˙ects. The singlet associated with the substrate is still observed, but the paramagnetic doublets have dissapeared. If we compare this spectrum with respect to the spectrum measured at RT, the central part do not allow the doublets to be included in the ft. Likewise, the values of the hyperfne parameters for the sextets are compatible with those expected for cobalt ferrite [246]. The obtained Fe3+/Fe3+ ratio (1.6) is still much higher A B than that expected for a canonical cobalt ferrite. Certainly, although recording data at low temperatures usually helps in determining the cation distribution, in the present case there still exists a signifcant contribution of the hyperfne magnetic feld distribution at 125 K which complicates such determination. However, the isomer shift which characterizes this distribution has character markedly octahedral which indicates that most of the cations contributing. In addition, it should be noted that the intensity of the sextet lines in the spectrum reveals the average magnetization orientation in the sample. Considering that the area ratio of the magnetic lines is 3:3.5:1:1:3.5:3, the average magnetization direction of the CFO thin flm is practically within the sample plane (14◦). The values of the hyperfne parameters used in the fts of spectra measured at RT and at 125 K are found in the table 7.3. 105 7 CoFe2O4 ultra-thin flms 10 0.0 100.5 101.0 101.5 Effect (%) -12 -8 -4 0 4 8 12 Velocity (mm/s) 100.0 100.5 101.0 101.5 Effect (%) 102.0 102.5 102.0 a) RT b) 125 K 0 2 4 6 8 Probability (%) a.1) 10 20 30 40 50 0 1 2 3 4 5 Probability (%) b.1) Magnetic hyperfine field (T) Figure 7.7: Left: Mössbauer spectra obtained for the 20 nm cobalt ferrite thin flm after postannealed treatment: a) measured at RT and b) measured at 125 K. Right: Hyperfne magnetic feld distribution used to ft of the broad magnetic component in the spectra on the left. Spectrum Site δ (± 0.03 mms−1) 2ε (± 0.05 mms−1) H (± 0.05 T) Area (%) Annealed, RT Annealed, 125 K Fe0 Fe3+ B Fe3+ A Fe3+ Fe2+ Fe3+ Fe0 Fe3+ B Fe3+ A Fe3+ 0.23 0.38 0.30 0.31 0.87 0.35 0.30 0.51 0.41 0.47 - -0.09 -0.01 0.93 0.78 -0.04 - -0.04 -0.03 0.02 - 49.3 46.4 - - 44.0 (HAV G - 37.1)a - 52.5 49.2 43.0 (HAV G - 31.2)a 5 12 37 6 6 34 2 29 47 22 a In the case of the distribution component, H corresponds to the maximum of the distribution while HAV G refers to the average feld of distribution. Table 7.3: 57Fe Mössbauer parameters obtained from the ft of the spectra shown in fgure 7.7 measured at room temperature and 125 K. The symbols δ , 2ε, H, HAV G correspond to isomer shift, quadrupole shift, hyperfne magnetic feld and average magnetic feld, respectively. The isomer shift values are quoted relative to α-Fe at room temperature. 106 7.5 5 nm thin flm 7.5 5 nm thin flm 7.5.1 Compositional, structural and morphological characterization The AES spectra recorded from the as-grown 5 nm thick CFO flm as well as from the annealed flms both in vacuum and in oxygen are shown in fgure 7.8. Again, the O KLL, Fe LMM and Co LMM lines are observed. However, unlike for the 20 nm sample that showed an increase in the cobalt to iron ratio near the surface region after the heating treatments, the Auger peaks intensities of the three elements for all the treatments are very similar or practically the same suggesting that a surface cobalt enrichment does not occur here. 500 600 700 800 Electron Energy (eV) -4 -2 0 2 4 Intensity (arb. units) As grown Annealed to 673 K Annealed to 773 K Annealed in O2 to 773 K 750 760 770 780 0 -0.4 0.4 Figure 7.8: Auger spectra of the 5 nm flm for each process performed. The spectra were normalized to the intensity of the Fe peak at 598 eV. The LEED di˙raction patterns from the as-grown and annealed flms are shown in fgure 7.9. The di˙raction pattern of the substrate, Pt(111), is shown frst (fgure 7.9a). As for the thicker flm, all the patterns are hexagonal with the same symmetry and orientation as the substrate (1×1 di˙raction pattern). The LEED pattern for the as-grown sample presents a slightly smaller spacing compared to the Pt pattern, implicating a higher distance between lines of atoms in-plane and more di˙use spots, suggesting the occurrence of structural disorder in the thin flm, fgure 7.9b. As the sample is annealed, the spots become sharper, indicating a better crystallinity of the flm (fgure 7.9c,d). This feature is more evident in the annealing step with oxygen (fgure 7.9e). In addition, the spots show atoms planes separation that approaches those of the substrate. Considering that the distance between planes of atoms is proportional to the distance between atoms within the plane, a graph has been made showing the in-plane lattice parameter of the 5 nm thick flm after each treatment, taking as a reference the Pt in-plane lattice parameter (a = 0.28 nm), fgure 7.9f. After annealing in oxygen, the lattice parameter for the sample is practically the same as the platinum lattice parameter. Regarding the observation of a 107 7 CoFe2O4 ultra-thin flms 1×1 LEED di˙raction pattern, instead of expected 2×2, we suggest the same explanation put forward for the thicker sample. 300 400 500 600 700 800 Temperature (K) 0.27 0.28 0.29 0.30 0.31 Lattice parameter (nm) Pt(111) As grown Annealed to 673 K Annealed to 773 K Annealed in O2 to 773 K f) a) b) c) d) e) Figure 7.9: Top panel: LEED di˙raction patterns for: a) platinum substrate, b) As grown 5 nm flm, b) annealed in UHV to 673 K, d) annealed in UHV to 773 K e) annealed in O2 to 773 K. Botton panel: f) lattice parameter from LEED for each processing step. STM images from the as-grown flm as well as from the flm annealed at 773 K both in vacuum and in oxygen, are shown in fgure 7.10. They show aggregates of particles 10-15 nm in size with an rms roughness of 0.3 nm, 15-18 nm and a rms roughness value of 0.4 nm and 20-25 nm with an rms roughness of 0.5 nm, respectively. As we have seen for the 20 nm flm, annealing in an ultra-high vacuum and in an oxygen atmosphere favors the increase in the size of the aggregates. a) b) c) 50 nm 50 nm 50 nm Figure 7.10: STM images for CFO thin flm 5 nm thick: a) as-grown, b) annealed to 773 K in vacuum, and c) annealed to 773 K in oxygen. 108 7.5 5 nm thin flm 7.5.2 Magnetic characterization Figure 7.11 shows the Mössbauer data recorded from the as-grown sample at RT, the annealed flm in vacuum both at RT and 115 K and the flm annealed in oxygen also both at RT and 125 K. These spectra were ftted using the NORMOS code. Table 7.4 collects the values of the corresponding hyperfne parameters. The spectrum for the as-grown sample was ftted to a singlet arising from Fe0 , a Fe3+ doublet and a broad magnetic component ftted with a hyperfne magnetic feld distribution. This spectrum di˙ers from the one of the 20 nm thin flm mainly in its magnetic component. In this case, it is much more signifcant (81 %) so it seems that this sample is more magnetically ordered than the previous one. This is remarkable since one might expect that the increase in thickness would confer the sample magnetic properties closer to those observed in the bulk. Moreover, the average grain size shows an similar value from both flms, fgure 7.5a and 7.10a. Several studies on magnetite, a spinel-type iron oxide, reported similarly shaped Mössbauer spectra for low thickness thin flms. These have attributed their spectrum shape to possible e˙ects produced by antiphase domain boundaries, resulting in frustration of the interdomain interactions [283, 284]. However, this explanation is unlikely since these works indicate the total elimination of the superparamagnetic component upon increasing the thickness of the sample, unlike what was obtained in the present investigation. From our point of view, as explained in the previous sample, the spectrum is compatible with a distribution of aggregate sizes with a range of sizes that includes those grains small enough to be in a paramagnetic state at room temperature (superparamagnetic behaviour), as well as some grains that exhibit magnetic state. Nonetheless, the grains’ atomic structuring might be the causative factor of this di˙erent magnetic behaviour between both as-grown flms. I.e., we assumed that both samples present a poor crystallinity; however, the increase magnetic contribution shown in the thinner flm appears to point out a higher atomic order compared with 20 nm flm. This fact might be explained as a consequence of the deposition rates used for each sample. Deposition rates for the 5 nm flm have been slower than for the thicker flm in a factor of almost two. Considering that the temperature growth was relatively low (523 K), slower deposition rates favour an arrangement of the atoms more orderly way in the structure. So, we suggest that the deposition rate di˙erence between both samples has led to the atomic order variation, and thus, the change in the magnetic behaviour. As in the case of the thicker flm, the annealing treatments induce signifcant changes in the nature of the deposited flm. The RT spectra recorded from the annealed flm both in vacuum and in the presence of oxygen are very similar and they are also similar to the RT spectrum recorded from the annealed 20 nm CFO flm: they show much better defned magnetic components and signifcantly less intense paramagnetic contributions. Although, a priori, the as-grown samples for both thicknesses start from di˙erent crystalline states within the grains, the increase in temperature up to 773 K produces the atomic reorganization to a spinel-like structure [241]. Therefore, spectra have been all ftted using the same procedure. Thus, the same considerations mentioned in the case of the annealed 20 nm-thick flm are of application here. From the results, compiled in table 7.4 for the RT Mössbauer data of the annealing in UHV and oxygen atmosphere, we cannot detect a signifcant change in the structural and magnetic properties with the thickness of the flms. 109 7 CoFe2O4 ultra-thin flms At LT the spectra show much narrower sextet lines although some magnetic relaxation is still present, hence the need of including a low-intensity hyperfne magnetic distribution. Contrarily to the LT spectrum of the thick flm, the 125 K spectra of these flms continues to show a Fe2+ contribution. The appearance of Fe2+ cations can be ftted again. Important information to consider when comparing the spectra at LT of the 5 nm thick sample with that of the 20 nm flm is the Fe3+/Fe3+ area ratio. For the thicker flm, A B we observed a ratio of 1.6, while for the thinner flm they are 1.9 and 1.1 for those flms annealed in vacuum and oxygen, respectively. It seems that annealing in oxygen favors the formation of a inverse cobalt ferrite structure [243, 240]. Certainly, these ratios are overestimated since for all the spectra we have a certain magnetic relaxation. To eliminate this contribution, the temperatures should be lowered even further. However, these temperatures cannot be reached with the present experimental setup. Further, for the 5 nm-thick flm the area of the sextet lines follow the ratio 3:3.1:1:1:3.1:3 which corresponds to an average angle of the magnetization to the sample surface of 20◦ . Thus, in the present flm, the magnetization is slightly more out-of-plane than in the 20 nm-thick flm. Hence, the thickness increase promotes the orientation of the magnetization within the flm plane, in line with the results of Khodaei et al [235]. In any case, the results indicate that the flm thickness does not play a crucial role in the characteristics of the flm, and that the annealing treatment helps to achieve a cation distribution close to that expected for cobalt ferrite. Additionally, the thinner flm has shown an improvement of the cationic order after annealing in oxygen. 110 7.5 5 nm thin flm 100.0 100.5 101.0 101.5 Effect (%) 10 0.0 100.5 101.0 101.5 Effect (%) a) RT 10 0.0 100.5 101.0 101.5 -12 -8 -4 0 4 8 12 Velocity (mm/s) 100.0 100.5 100.1 101.5 Effect (%)Effect (%) -12 -8 -4 0 4 8 12 Velocity (mm/s) 100.0 100.5 101.0 101.5 Effect (%) -12 -8 -4 0 4 8 12 Velocity (mm/s) d) RT e) 125 K c) 115 K b) RT Annealed to 673 K / 773 K Annealed in O2 to 773 K As grown 0 10 20 30 40 50 60 Magnetic hyperfine field (T) 0 1 2 3 4 5 6 7 Probability (%) a.1) 0 1 2 3 4 5 6 7 Probability (%) 10 20 30 40 50 0 1 2 3 4 5 6 Probability (%) b.1) Magnetic hyperfine field (T) c.1) 0 1 2 3 4 5 6 7 Probability (%) 10 20 30 40 50 Magnetic hyperfine field (T) 0 1 2 3 4 5 6 7 Probability (%) d.1) e.1) Figure 7.11: Left: Mössbauer spectra obtained for the 5 nm cobalt ferrite thin flm for the di˙erent stages: a) as-grown measured at RT, b) and c) annealed to 673 K and 773 K measured at RT and 115 K, respectively, d) and e) annealed in oxygen atmosphere to 773 K measured at RT and 125 K, respectively. Right: Hyperfne magnetic feld distribution used in the ft of the broad magnetic component from the spectrum on the left. 111 8.3 Micromagnetic simulations of CoFe2O4/FeCo system 8.3 Micromagnetic simulations of CoFe2O4/FeCo system Micromagnetic simulations have been carried out to support the experimental results. Different micromagnetic parameters have been changed in order to study the spring magnet response predicted by the theory: (i) the exchange sti˙ness of the soft layer (As), (ii) exchange-coupling between the soft and the hard layer (κ)and (iii) saturation magnetization of the soft layer (Ms). The simulation cell size was set to 4 x 4 x 2 nm3 for both layers and slabs with the inplane size 800 x 800 nm3 with several replicas were employed to mimic periodic boundary conditions avoiding isolated system behavior. The thicknesses of CoFe2O4 and FeCo layers have been chosen to be 76 nm and 4 nm, respectively, i.e. the same as the experimental bilayer with the thinnest soft layer that showed a spring magnet e˙ect. The exchange sti˙- ness and anisotropy values were taken from the literature [269, 294], while the saturation magnetization was obtained from experimental measurements of single-layer flms of the soft and hard layers. Thus, the following starting point magnetic parameters were used in simulations for hard (H, i.e. CoFe2O4) and soft (S, i.e. FeCo) layers: exchange sti˙ness AH = 1.10 × 10−11 J/m, AS = 1.7 × 10−11 J/m, saturation magnetizations MH = 3.41 × 105 A/m and MS = 1.85 × 106 A/m, magnetocrystalline constants KH = 5.10 × 105 J/m3 and KS = 0.472 × 103 J/m3 . 8.3.1 Exchange sti˙ness of soft layer Figure 8.3 presents the simulated magnetization curves and their derivative for four di˙erent exchange sti˙ness values at an applied magnetic feld of up to 2 T in [110] direction in these simulations. A single-domain initial confguration for both layers is defned. Moreover, taking into account that the experimental data indicate a good crystalline order of the interface, an interface exchange coupling of 75% has been assumed. First, the hysteresis loop with an exchange sti˙ness value of AS = 1.7 x 10−11 J/m for the soft layer, corresponds to the red curve in fgure 8.3a. In this case, the hysteresis cycle shows a rigid coupled system, in agreement with the predictions of classic spring magnet theory [30, 294]. In fact, performing the derivate dM/dH from this curve, a soft layer reversal at low negative felds is not observed (see fgure 8.3b same color pattern). In order to force the spring magnet behaviour seen in the experiment, we have estimated the value of the required exchange sti˙ness according to the classic spring theory equation 8.1 using experimental parameters obtained (saturation magnetization, MS = 1.85 × 106 A/m, and nucleation feld, 0.04 T). A value of AS = 3.1 ×10−12 J/m was determined. For this soft material exchange sti˙ness, we observe in fgure 8.3a the typical "jump" characteristic from the spring magnet regime. This feature is related to the reversal of the soft layer, which occurs at a lower magnetic feld than for the hard layer. This can be better observed in the dM/dH plot where a clear maximum is appreciated (fgure 8.3b). Hysteresis cycles were calculated with exchange sti˙ness values intermediate between those two. In fgure 8.3a, we see that for both AS = 8.5 × 10−12 J/m and AS =7× 10−12 J/m it is diÿcult to defne whether there is a spring behaviour. However, looking at the derivatives, a slight maximum is observed for AS = 8.5 × 10−12 J/m while for the other 119 8 Magnetic interactions in CoFe2O4/FeCo bilayer thin flms -0.5 -0.4 -0.3 -0.2 -0.1 Magnetic Field (T) dM/dH A = 3.1 x 10-12 A = 7.0 x 10 -12 A = 8.5 x 10-12 A = 1.7 x 10 -11 FeCo FeCo FeCo FeCo J/m J/m J/m J/m a) b) -2 -1 0 1 2 Magnetic Field (T) -1 0 1 M (A/m) AFeCo= 1.7 x 10-11 J/m AFeCo= 8.5 x 10-12 J/m AFeCo= 7.0 x 10-12 J/m AFeCo= 3.1 x 10-12 J/m Exchange Coupling 75% Figure 8.3: (a) Simulated dM/dH curve for four di˙erent exchange sti˙ness values for a 76-nm CoFe2O4/ 4-nm FeCo bilayer. (b) Simulated magnetization curves corresponding to the highest and lowest sti˙ness values. The arrow indicates the curve maximum which indicates the spring magnet behavior. value, the maximum has practically disappeared (curves in colors cyan and orange from fgure 8.3b, respectively). Therefore, from the micromagnetism simulations performed in this section, it is confrmed that the system behaves as a rigid/spring magnet above/below AS = 8 × 10−12 J/m. We note, however, that such value of exchange sti˙ness is much too low compared with reported values for FeCo (typically 6x larger, AS = 1.7 × 10−11 J/m) [294]. 8.3.2 Interlayer exchange coupling Another way to induce a partial decoupling of the two layers is by reducing the exchange coupling between the soft and hard layer by the κ parameter. κ is the factor that rescales the exchange coupling value between both layers, ranging from κ = 1 (perfect coupling) to κ = 0 (complete exchange decoupling). In fgure 8.4, hysteresis loops of the CoFe2O4/FeCo bilayer at an applied magnetic feld of up to 2 T in the [110] direction are plotted for two values κ (0.05 and 1) of the interlayer exchange coupling, keeping the other magnetic parameters constant at the values indicated at the beginning of the section. A singledomain initial confguration for both layers was defned in these simulations. It was found that in order to observe the exchange-spring behaviour in the bilayers, the value of κ has to be as low as 0.05. This exchange coupling value is unrealistic since the experimental observations pointed out a coherent and sharp interface with a few separate misft dislocations. Thus, as for the exchange sti˙ness parameter discussion, the magnetic spring behaviour can only be reproduced by forcing the material parameters to unrealistic values. 120 8.3 Micromagnetic simulations of CoFe2O4/FeCo system -2 -1 0 1 2 Magnetic Field (T) -1 -0.5 0 0.5 1 M\Ms Exchange Coupling 5% Exchange Coupling 100% Figure 8.4: Simulated hysteresis loops of the CoFe2O4/FeCo bilayer for two di˙erent κ values (0.05 and 1.0) 8.3.3 Saturation magnetization of soft layer As it has already been mentioned in the "experimental background" section, a study performed by Laborato et al. [293] used magnetite as the soft magnetic layer (Fe3O4) instead of FeCo. In such work, the spring magnet thickness limits of the classic theory were fulflled. The main di˙erence with our study comes from the fact that the saturation magnetization of magnetite is considerably lower than that of iron-cobalt alloy. In this part, the infuence of the saturation magnetization of the soft is studied, keeping the rest of the parameters constant (beginning of the section). For this, two micromagnetism simulations were performed using the magnetization value of magnetite (Ms = 4.8 × 106 A/m) and the magnetizing value of the cobalt-iron alloy (Ms = 1.85 × 106 A/m). For both, the exchange sti˙ness was set to AS = 3.1 × 10−12 . This is the value calculated to reproduce the spring magnet behavior in the CFO/FeCo bilayer system (see fgure 8.3). Further, a single-domain initial confguration for both layers was defned in these simulations. Figure 8.5 shows the magnetization curves of the CoFe2O4/FeCo bilayer at an applied magnetic feld of up to 2 T in [110] direction for the two saturation magnetization values. Using the Ms for FeCo leads to the exchange-spring reversal, while the Ms of Fe3O4 yields a ’rigid’ behavior. Although, the fact that CFO and Fe3O4 have very similar values makes it incredibly diÿcult to discern rigid from spring behavior only from magnetization curves. This observation proves the important infuence of the Ms of the soft phase and hints at reasons for the discrepancy with Lavorato et al. results. In addition, it supports the idea that considering the exchange length of the hard phase as the main parameter determining critical thickness is an oversimplifcation. 121 8 Magnetic interactions in CoFe2O4/FeCo bilayer thin flms -2 -1 0 1 2 Magnetic Field (T) -1 -0,5 0 0,5 1 M/Ms Ms=4.8 X 105A/m Ms=1.85 x 106A/m Exchange Coupling 100% Figure 8.5: Simulated hysteresis loops of the CoFe2O4/FeCo bilayer for two di˙erent magnetization values of FeCo 8.3.4 Simulating multiand single-domain confgurations In this section, we have attempted a slightly more realistic model of the hard layer. Instead of a monocrystalline hard layer, a multi-domain state was obtained by defning grains with an average in-plane size of 100 nm in the CoFe2O4 layer using the Voronoi tessellation. The exchange coupling between these grains was set to zero, i.e., the grains are exchangedecoupled from one another. It is worth noting that exchange-coupling does occur between these grains/domains in reality, with magnetic domain walls pinned at antiphase boundaries (APB)1 . Both rotational processes within each domain/grain and depinning of magnetic domain walls from APB are expected to contribute to the magnetization process in the CFO layer alone. Thus, it is important to keep in mind that the micromagnetic model presented here still constitutes a simplifcation compared to the real experimental system. A single continuous soft layer is covering the CFO grains. The exchange coupling set between the CFO multi-domain state from cobalt ferrite and iron-cobalt single domain state was κ = 0.75 to approach the experimental conditions. Further, the micromagnetic simulation was performed using the same parameters from both phases showed at the beginning except the soft exchange sti˙ness constant, which was set AS = 3.1 × 10−12 J/m since spite of to be an unrealistic value, we want to simulate for the present study the spring magnet behavior observed in the experimental part. This is not reproduced with literature AS value. 1APB separates two domains of the same ordered phase. It arises from symmetry breaking that occurs during ordering processes, which might begin at di˙erent locations in a disordered lattice. The antiphase boundary is formed when two such regions come into contact and there a mismatch in the composition across the interface [297]. 122 8.3 Micromagnetic simulations of CoFe2O4/FeCo system Figure 8.6 presents the magnetization change of the bilayer system with an applied magnetic feld of up to 3 T and depicts the bilayer magnetization state at di˙erent magnetic felds. At remanence (0 T) the soft layer is found coupled with the CoFe2O4 domains. When the applied magnetic feld increases, the magnetization of the top of the FeCo layer di˙ers from the bottom side, which is in contact with the hard CoFe2O4 multidomain state. At a magnetic feld of 0.38 T the average magnetization of the soft layer changes direction to that of the applied feld, showing domain walls (separation between the orange and yellow regions), and it is not until 1.5 T that the soft flm is fully oriented. In contrast, the domain walls formed in the CFO disappear at around 2.2 T. Therefore, in this simulation, a propagation of the domain wall in the soft phase is observed, which is the factor that promotes the magnetization orientation in the bilayer. 00.5 11.5 22.5 3 Magnetic Field (T) 0 0.5 1 M\M s Multi-domain CoFe2O4 / Single-domain FeCo As = 3.1 x 10-12 J/m Figure 8.6: Simulated magnetization curve for the CoFe2O4/FeCo bilayer starting from a multidomain initial confguration. Inset shows magnetization states during the magnetization process. 123 8 Magnetic interactions in CoFe2O4/FeCo bilayer thin flms 8.4 Conclusions Micromagnetic models that follow the accepted spring theory force us to use unrealistic sti˙ness and exchange coupling conditions in order to reproduce the experimental observations. In addition, saturation magnetization confrms that domain wall propagation plays an important role in the soft layer’s demagnetization. Indeed, iron-cobalt layer saturation magnetization is an infuencing factor in the bilayer magnetic behaviour, demonstrating that the hard phase exchange length is not the only parameter involved. The micromagnetic simulation using a multi-domain confguration of the hard layer and a value of exchange sti˙ness of AS = 3.1 × 10−12 J/m, which is the value calculated from experiment results, reproduced the bilayer behaviour. Virgin magnetization curve and images from the bilayer magnetic state evolution showed a propagation of domains in the soft refecting an increase of the magnetization at a feld of 0.38 T. Reversal of the magnetically hard layer is produced at higher magnetic felds. Therefore, micromagnetic simulations confrmed the spring magnet-like behaviour due to domain wall propagation observed in the CoFe2O4/FeCo bilayer system. 124 9 General conclusions In this thesis, the magnetic and structural properties of hard ferrites, strontium hexaferrite and cobalt ferrite have been investigated by di˙erent characterization techniques. These materials are of high interest due to their widespread use as permanent magnets, recording media and components in di˙erent electronic devices. The interaction of these ferrites with a magnetically soft layer has brought forward the understanding of the magnetic coupling regime experimented in this type of system as well as elucidated the interaction conditions needed to enable the improvement of magnetic properties, making these materials more competitive. The most important conclusions of this research are presented below. 1. SrFe12O19 platelets Strontium hexaferrite platelets grown by hydrothermal synthesis have been studied through microscopies, di˙raction and spectroscopic techniques. These platelets showed a lateral size of microns and a thickness of tens of nanometers. The results pointed out an increase of the iron cations in tetrahedral sites with respect to the octahedral ones in the region near the platelet surface. This fact resulted in a lower net magnetic moment for the platelets than that strontium ferrite commercial powder since the iron in tetrahedral environments in the former presents spin moments aligned antiferromagnetically to the net magnetization. The platelets have displayed magnetic domains, which represented a magnetization perpendicular to the platelet plane. This is in good agreement with the c-axis direction determined due to the magnetization easy axis lies along it, which is also perpendicular to the platelet plane. 2. SFO platelets with cobalt layer With the aim to improve the SFO platelets magnetization, a layer of a magnetically soft material (cobalt) was deposited on top of them by MBE at room temperature. Specifcally, a platelet with a single magnetic domain and its magnetic interaction with the cobalt overlayer was characterized by X-ray absorption techniques. These revealed a magnetization orientation in the sample plane for the soft layer and out-ofplane magnetization for the platelet. The lack of correlation between the magnetic domains of both phases indicated the absence of exchange coupling. Such noncorrelation was interpreted as a consequence of the metallic layer shape anisotropy dominating over the external perpendicular feld produced by the hexaferrite layer. 125 9 General conclusions Micromagnetic simulations corroborated the experimental result and predicted that a low exchange-coupling strength between both layers should lead to an eÿcient alignment of the soft moments with the hard phase. This outcome gave rise to a new strategy of developing hard/soft systems since domain wall propagation is avoided. 3. SFO thin flms and their interaction with cobalt layer As previously observed for the platelet/metal system, coupling between the two phases to improve the hard/soft system magnetic properties is not easy to achieve. In this chapter, an attempt was made to promote the magnetostatic interaction between both layers by growing a strontium hexaferrite thin flm with the magnetization direction within the plane to prevent the competition observed in the previous chapter and thus facilitate alignment with the magnetically soft layer. SrFe12O19 flms were grown by RF magnetron sputtering. The annealing step in air was required to obtain SFO. Indeed, the non-annealed SFO flm presented a nanosized maghemite phase and likely a SrO amorphous oxide. The net magnetization orientation for SFO thin flms was adjusted in the sample plane, depositing at several sputtering powers with constant deposition times (and thus di˙erent thickness). These samples’ characterization by spectroscopic and di˙raction methods has demonstrated that the flms are textured, and their c-axis orientation, and consequently their magnetization easy axis, changes with the sputtering power and the thickness. SFO flms 360 nm thick grown at the highest power used (260 W) showed preferential in-plane magnetization. On this thin flm, the cobalt layer was grown by MBE to study the magnetic interaction with a soft magnetic layer. The bilayer analysis performed using XAS, and XMCD techniques showed the oxidation of some cobalt atoms in the interface with the SFO and the magnetic domains corresponding to each layer. The magnetization vector alignment was along the directions at 160◦ and 340◦ in the flm plane for both layers. However, the magnetic domains in the two layers again were not correlated, implicating a lack of exchange-coupling between them. We suggested that the coincidence in the direction of the uniaxial easy axis is caused by structural coupling. 4. CoFe2O4 thin flms In this chapter, we have studied the properties of the cobalt ferrite ultrathin flms (5 nm and 20 nm thick) grown by oxygen assisted molecular beam epitaxy on a Pt(111) single crystal with the particularity that the characterization was carried out in-situ under UHV conditions. As-grown thin flms presented a superparamagnetic/poorly crystalline state composed of a distribution of Fe3+ containing-aggregates of di˙erent sizes in the nanometer scale and a minor Fe2+ phase. The unexpected 1×1 LEED pattern for CFO suggested a FexCo1−xO termination of the flms. The thinner flm presented higher magnetic ordering than the thicker sample, which was explained due to the lower deposition rate employed, allowing a cation distribution similar to canonical CoFe2O4. 126 Annealing in vacuum promoted an increase in particles size, resulting in the development of magnetic ordering at RT, although the annealing treatment has not entirely removed the superparamagnetic contribution. The iron in tetrahedral and octahedral sites ratios revealed cobalt ferrite formation after the annealing process achieving the thinner flm a cation distribution close to that expected for a CFO inverse spinel after annealing in oxygen. Thus, the O-MBE process provided cobalt ferrite thin flms using intermediate temperatures (up to 773 K), preventing the formation of di˙erent phases, island features, or dewetting phenomena. 5. CFO with iron-cobalt layer An experimental system composed of cobalt ferrite (76 nm) and iron-cobalt layer (4 nm) layers showed a spring magnet behaviour in conditions where a rigid coupling was expected. So, the purpose of this section was to understand this interaction through micromagnetic simulations. These simulations reproduced such scenario using unrealistic values of exchange sti˙ness, magnetization saturation and exchange coupling. Considering that the simulations followed the accepted spring theory that does not include the domain wall propagation mechanism, the results evidenced the latter mechanism’s importance in the hard/soft bilayer systems’ spring magnet regime. 127 10 Conclusiones generales En esta tesis se han investigado las propiedades magnéticas y estructurales de las ferritas duras, hexaferrita de estroncio y ferrita de cobalto, mediante diferentes técnicas de caracterización. Estos materiales son de gran interés debido a su amplio uso como imanes permanentes, soportes de grabación y componentes en diferentes dispositivos electrónicos. La interacción de estas ferritas con una capa magnéticamente blanda ha permitido avanzar en la comprensión del régimen de acoplamiento magnético experimentado en este tipo de sistemas, así como dilucidar las condiciones de interacción necesarias para permitir la mejora de las propiedades magnéticas, produciendo que estos materiales sean más competitivos. A continuación se presentan las conclusiones más importantes de esta investigación. 1. Plaquetas de SrFe12O19 Las plaquetas de hexaferrita de estroncio crecidas por síntesis hidrotermal se han estudiado mediante técnicas de microscopía, difracción y espectroscopia. Estas plaquetas mostraron un tamaño lateral de micras y un espesor de decenas de nanómetros. Los resultados señalaron un aumento de los cationes de hierro en sitios tetraédricos con respecto a los sitios octaédricos en la región cercana a la superfcie de la plaqueta. Este hecho dio lugar a un momento magnético neto menor para las plaquetas que el del polvo comercial de ferrita de estroncio, ya que el hierro en entornos tetraédricos en las primeras presenta momentos de espín alineados antiferromagnéticamente a la imanación neta. Las plaquetas han exhibido dominios magnéticos que presentaban una imanación perpendicular al plano de la plaqueta. Esta evidencia está en concordancia con la dirección del eje c obtenida debido a que el eje fácil de imanación se encuentra a lo largo del mismo, siendo también perpendicular al plano de la plaqueta. 2. Plaquetas de SFO con una capa de cobalto Con el objetivo de mejorar la imanación de las plaquetas de SFO, se depositó sobre ellas una capa de un material magnéticamente blando (cobalto) mediante epitaxia de haces moleculares a temperatura ambiente. En concreto, se caracterizó una plaqueta con un único dominio magnético y su interacción magnética con la capa de cobalto mediante técnicas de absorción de rayos X. Éstas revelaron una orientación de la imanación en el plano de la muestra para la capa blanda y una imanación fuera del plano para la plaqueta. La falta de correlación entre los dominios magnéticos de ambas fases indicó la ausencia de acoplamiento de 129 B List of publications 1. G. D. Soria, K. Freindl, J. E. Prieto, A. Quesada, J. de la Figuera, N. Spiridis, J. Korecki, and J. F. Marco, “Growth and characterization of ultrathin cobalt ferrite flms on Pt(111)", in preparation. 2. G. D. Soria, C. Granados-Miralles, A. Mandziak, P. Jenus, M. Saura-Múzquiz, M. Christensen, M. Foerster, L. Aballe, J. F. Fernández, J.d.l. Figuera, and A. Quesada, “Uncorrelated magnetic domains in decoupled SrFe12O19/Co hard/soft bilayers", Journal of Physics D: Applied Physics, vol. 54, no. 5, p. 054003, 2020. 3. A. Mandziak, G. D. Soria, J. E. Prieto, M. Foerster, J. d. l. Figuera, and L. Aballe, “Di˙erent spin axis orientation and large antiferromagnetic domains in Fe-doped NiO/Ru(0001) epitaxial flms", Nanoscale, vol. 12, pp. 21225–21233, 2020. 4. G. D. Soria, J. F. Marco, A. Mandziak, S. Sánchez-Cortés, M. Sánchez-Arenillas, J. E. Prieto, J. Dávalos, M. Foerster, L. Aballe, J. López-Sánchez, J. C. GuzmánMínguez, C. Granados-Miralles, J. d. l. Figuera, and A. Quesada, “Infuence of the growth conditions on the magnetism of SrFe12O19 thin flms and the behavior of Co/SrFe12O19 bilayers", Journal of Physics D: Applied Physics, vol. 53, p. 344002, 2020. 5. S. Ruiz-Gómez, A. Mandziak, J. E. Prieto, M. Aristu, E. M. Trapero, G. D. Soria, A. Quesada, M. Foerster, L. Aballe, and J. de la Figuera, “A real-time XAS PEEM study of the growth of cobalt iron oxide on Ru(0001)", The Journal of Chemical Physics, vol. 152, p. 074704, 2020. 6. A. Mandziak, G. D. Soria, J. E. Prieto, P. Prieto, C. Granados-Miralles, A. Quesada, M. Foerster, L. Aballe, and J. de la Figuera, “Tuning the Néel temperature in an antiferromagnet: the case of NixCo1−xO microstructures", Scientifc Reports, vol. 9, p. 13584, 2019. 7. G. D. Soria, P. Jenus, J. F. Marco, A. Mandziak, M. Sanchez-Arenillas, F. Moutinho, J. E. Prieto, P. Prieto, J. Cerdá, C. Tejera-Centeno, S. Gallego, M. Foerster, L. Aballe, M. Valvidares, H. B. Vasili, E. Pereiro, A. Quesada, and J. de la Figuera, “Strontium hexaferrite platelets: a comprehensive soft X-ray absorption and Mössbauer spectroscopy study", Scientifc Reports, vol. 9, no. 1, p. 11777, 2019. 8. A. Quesada, G. D. Soria, L. Pascual, A. M. Aragón, P. Marín, C. Granados-Miralles, M. Foerster, L. Aballe, J. E. Prieto, J. de la Figuera, J. F. Fernández, and P. Prieto, “Exchange-spring behavior below the exchange length in hard-soft bilayers in multidomain confgurations", Physical Review B, vol. 98, no. 21, p. 214435, 2018. 137 B List of publications 9. A. Mandziak, J. de la Figuera, S. Ruiz-Gómez, G. D. Soria, L. Pérez, P. Prieto, A. Quesada, M. Foerster, and L. Aballe, “Structure and magnetism of ultrathin nickeliron oxides grown on Ru(0001) by high-temperature oxygen-assisted molecular beam epitaxy", Scientifc Reports, vol. 8, p. 17980, 2018. 10. G. D. Soria, J. R. Espinosa, J. Ramirez, C. Valeriani, C. Vega, and E. Sanz, “A simulation study of homogeneous ice nucleation in supercooled salty water", The Journal of Chemical Physics, vol. 148, p. 222811, 2018. 11. J. R. Espinosa, G. D. Soria, J. Ramirez, C. Valeriani, C. Vega, and E. Sanz, “Role of Salt, Pressure, and Water Activity on Homogeneous Ice Nucleation", The Journal of Physical Chemistry Letters, vol. 8, pp. 4486–4491, 2017. 138 Bibliography [1] A. Basak, Permanent-magnet DC linear motors. Clarendon Press, 1996. [2] O. Gutfeisch, M. A. Willard, E. Brück, C. H. Chen, S. G. Sankar, and J. P. Liu, “Magnetic materials and devices for the 21st century: stronger, lighter, and more energy eÿcient,” Advanced Materials, vol. 23, no. 7, pp. 821–842, 2011. [3] S. Sugimoto, “Current status and recent topics of rare-earth permanent magnets,” Journal of Physics D: Applied Physics, vol. 44, no. 6, p. 064001, 2011. [4] J. M. D. Coey, “Permanent magnetism,” Solid State Communications, vol. 102, no. 2, pp. 101–105, 1997. [5] J. Coey, “Permanent magnets: Plugging the gap,” Scripta Materialia, vol. 67, no. 6, pp. 524–529, 2012. [6] B. D. Cullity and C. D. Graham, “Hard magnetic materials,” in Introduction to Magnetic Materials, pp. 477–504, 2008. [7] K. H. J. Buschow and F. R. de Boer, “Permanent magnets,” in Physics of Magnetism and Magnetic Materials, pp. 105–129, 2003. [8] P. Jenus, A. Ucakar, S. Repse, C. Sangregorio, M. Petrecca, M. Albino, R. Cabassi, C. de Julian Fernandez, and B. Belec, “Magnetic performance of SrFe12O19-Zn0.2 Fe2.8O4 hybrid magnets prepared by Spark Plasma Sintering,” Journal of Physics D: Applied Physics, 2021. [9] O. Gutfeisch, “Controlling the properties of high energy density permanent magnetic materials by di˙erent processing routes,” Journal of Physics D: Applied Physics, vol. 33, no. 17, pp. R157–R172, 2000. [10] N. Haque, A. Hughes, S. Lim, and C. Vernon, “Rare earth elements: Overview of mining, mineralogy, uses, sustainability and environmental impact,” Resources, vol. 3, no. 4, pp. 614–635, 2014. [11] V. Balaram, “Rare earth elements: A review of applications, occurrence, exploration, analysis, recycling, and environmental impact,” Geoscience Frontiers, vol. 10, no. 4, pp. 1285–1303, 2019. [12] S. H. Ali, “Social and environmental impact of the rare earth industries,” Resources, vol. 3, no. 1, pp. 123–134, 2014. 139 Bibliography [13] J. Smit and H. Wijn, “Ferrites, Philips technical library,” Eindhoven, The Netherlands, vol. 278, 1959. [14] J. Went, G. Rathenau, G. E.W, and G. Van Oosterhout, “Ferroxdure, a class of new permanent magnet materials,” Philips Technical Review, vol. 13, no. 7, p. 194–208, 1952. [15] V. A. M. Brabers, “Chapter 3 Progress in spinel ferrite research,” in Handbook of Magnetic Materials, vol. 8, pp. 189–324, 1995. [16] J. L. Snoek, New developments in ferromagnetic materials: with introductory chapters on the statics and the dynamics of ferromagnetism. No. 19, Elsevier Publishing Company, 1947. [17] A. Goldman, Modern Ferrite Technology. Springer, 2 ed., 2006. [18] F. Kools, “Hard magnetic ferrites,” in Concise Encyclopedia of Advanced Ceramic Materials, pp. 200–206, 1991. [19] E. E. Fullerton, J. S. Jiang, M. Grimsditch, C. H. Sowers, and S. D. Bader, “Exchange-spring behavior in epitaxial hard/soft magnetic bilayers,” Physical Review B, vol. 58, no. 18, pp. 12193–12200, 1998. [20] G. E. Fish, “Soft magnetic materials,” Proceedings of the IEEE, vol. 78, no. 6, pp. 947– 972, 1990. [21] L. H. Lewis and F. Jiménez-Villacorta, “Perspectives on Permanent Magnetic Materials for Energy Conversion and Power Generation,” Metallurgical and Materials Transactions A, vol. 44, no. 1, pp. 2–20, 2013. [22] A. Quesada, C. Granados-Miralles, A. López-Ortega, S. Erokhin, E. Lottini, J. Pedrosa, A. Bollero, A. M. Aragón, F. Rubio-Marcos, M. Stingaciu, G. Bertoni, C. d. J. Fernández, C. Sangregorio, J. F. Fernández, D. Berkov, and M. Christensen, “Energy product enhancement in imperfectly exchange-coupled nanocomposite magnets,” Advanced Electronic Materials, vol. 2, no. 4, p. 1500365, 2016. [23] G. C. Hadjipanayis, “Nanophase hard magnets,” Journal of Magnetism and Magnetic Materials, vol. 200, no. 1, pp. 373–391, 1999. [24] A. López-Ortega, M. Estrader, G. Salazar-Alvarez, A. G. Roca, and J. Nogués, “Applications of exchange coupled bi-magnetic hard/soft and soft/hard magnetic core/shell nanoparticles,” Physics Reports, vol. 553, pp. 1–32, 2015. [25] O. Masala, D. Ho˙man, N. Sundaram, K. Page, T. Pro˙en, G. Lawes, and R. Seshadri, “Preparation of magnetic spinel ferrite core/shell nanoparticles: Soft ferrites on hard ferrites and vice versa,” Solid State Sciences, vol. 8, no. 9, pp. 1015–1022, 2006. 140 Bibliography [26] Y.-M. Kang, J.-B. Park, Y. J. Kang, and K. Ahn, “Magnetic properties of M-type Sr hexaferrite/Fe bilayers,” Thin Solid Films, vol. 594, pp. 40–44, Nov. 2015. [27] S. M. Parhofer, J. Wecker, C. Kuhrt, G. Gieres, and L. Schultz, “Remanence enhancement due to exchange coupling in multilayers of hardand softmagnetic phases,” IEEE Transactions on Magnetics, vol. 32, no. 5, pp. 4437–4439, 1996. [28] C. d. J. Fernandez, C. Sangregorio, J. d. l. Figuera, B. Belec, D. Makovec, and A. Quesada, “Topical review: Progress and prospects of hard hexaferrites for permanent magnet applications,” Journal of Physics D: Applied Physics, 2020. [29] S. Sabet, A. Moradabadi, S. Gorji, M. Yi, Q. Gong, M. H. Fawey, E. Hildebrandt, D. Wang, H. Zhang, B.-X. Xu, C. Kübel, and L. Al˙, “Impact of interface structure on magnetic exchange coupling in MnBi/FexCo1−x bilayers,” Physical Review B, vol. 98, no. 17, p. 174440, 2018. [30] E. E. Fullerton, J. S. Jiang, and S. D. Bader, “Hard/soft magnetic heterostructures: model exchange-spring magnets,” Journal of Magnetism and Magnetic Materials, vol. 200, no. 1, pp. 392–404, 1999. [31] G. Asti, M. Ghidini, R. Pellicelli, C. Pernechele, M. Solzi, F. Albertini, F. Casoli, S. Fabbrici, and L. Pareti, “Magnetic phase diagram and demagnetization processes in perpendicular exchange-spring multilayers,” Physical Review B, vol. 73, no. 9, p. 094406, 2006. [32] A. J. Zambano, H. Oguchi, I. Takeuchi, Y. Choi, J. S. Jiang, J. P. Liu, S. E. Lofand, D. Josell, and L. A. Bendersky, “Dependence of exchange coupling interaction on micromagnetic constants in hard/soft magnetic bilayer systems,” Physical Review B, vol. 75, no. 14, p. 144429, 2007. [33] J. S. Jiang, J. E. Pearson, Z. Y. Liu, B. Kabius, S. Trasobares, D. J. Miller, S. D. Bader, D. R. Lee, D. Haskel, G. Srajer, and J. P. Liu, “A new approach for improving exchange-spring magnets,” Journal of Applied Physics, vol. 97, p. 10K311, May 2005. [34] J. S. Jiang and S. D. Bader, “Magnetic reversal in thin flm exchange-spring magnets,” Scripta Materialia, vol. 47, no. 8, pp. 563–568, 2002. [35] G. Asti, M. Solzi, M. Ghidini, and F. M. Neri, “Micromagnetic analysis of exchangecoupled hard-soft planar nanocomposites,” Physical Review B, vol. 69, no. 17, p. 174401, 2004. [36] D. M. Mattox, “Chapter 1 - Introduction,” in Handbook of Physical Vapor Deposition (PVD) Processing, pp. 1–24, William Andrew Publishing, second edition ed., 2010. [37] P. J. Kelly and R. D. Arnell, “Magnetron sputtering: a review of recent developments and applications,” Vacuum, vol. 56, no. 3, pp. 159–172, 2000. [38] S. Swann, “Magnetron sputtering,” Physics in Technology, vol. 19, no. 2, pp. 67–75, 1988. 141 Bibliography [39] K. V. Madhuri, “10 - Thermal protection coatings of metal oxide powders,” in Metal Oxide Powder Technologies, Metal Oxides, pp. 209–231, Elsevier, 2020. [40] D. A. Glocker, S. Shah, and W. Westwood, Handbook of Thin Film Process Technology. No. 1-2, Institute of Physics Publication, 1995. [41] J. M. D. Coey, M. Venkatesan, and H. Xu, “Introduction to Magnetic Oxides,” in Functional Metal Oxides, pp. 1–49, John Wiley & Sons, Ltd, 2013. [42] B. R. Acharya, S. Prasad, N. Venkataramani, S. N. Shringi, and R. Krishnan, “The e˙ect of deposition and annealing conditions on textured growth of sputter-deposited strontium ferrite flms on di˙erent substrates,” Journal of Applied Physics, vol. 79, no. 1, pp. 478–484, 1996. [43] A. Y. Cho and J. R. Arthur, “Molecular beam epitaxy,” Progress in Solid State Chemistry, vol. 10, pp. 157–191, 1975. [44] M. A. Herman and H. Sitter, Molecular Beam Epitaxy: Fundamentals and Current Status. Springer Series in Materials Science, 2 ed., 1996. [45] J. R. Arthur, “Molecular beam epitaxy,” Surface Science, vol. 500, no. 1, pp. 189–217, 2002. [46] B. Fultz and J. M. Howe, Transmission electron microscopy and di˙ractometry of materials. Springer Science & Business Media, 2012. [47] F. Adams and C. Barbante, “Chapter 6 - X-Ray Imaging,” in Comprehensive Analytical Chemistry, vol. 69 of Chemical Imaging Analysis, pp. 213–268, Elsevier, 2015. [48] B. D. Cullity, Elements of x-ray di˙raction. Addison-Wesley Publishing Company, Inc., 1978. [49] J. F. Watts, “X-ray photoelectron spectroscopy,” Vacuum, vol. 45, no. 6, pp. 653–671, 1994. [50] B. D. Cullity and C. D. Graham, “Experimental methods,” in Introduction to Magnetic Materials, pp. 23–86, John Wiley & Sons, Ltd, 2008. [51] S. Foner, “Versatile and sensitive vibrating-sample magnetometer,” Review of Scientifc Instruments, vol. 30, no. 7, pp. 548–557, 1959. [52] H. Lüth, Surfaces and interfaces of solid materials. Springer Study Edition, 3 ed., 1995. [53] H. H. Telle, A. G. Uren , a, and R. J. Donovan, Laser chemistry: spectroscopy, dynamics and applications. John Wiley & Sons, 2007. [54] H. J. Butler, L. Ashton, B. Bird, G. Cinque, K. Curtis, J. Dorney, K. Esmonde-White, N. J. Fullwood, B. Gardner, P. L. Martin-Hirsch, M. J. Walsh, M. R. McAinsh, 142 Bibliography N. Stone, and F. L. Martin, “Using Raman spectroscopy to characterize biological materials,” Nature Protocols, vol. 11, no. 4, pp. 664–687, 2016. [55] B. Voigtlaender, Scanning probe microscopy: Atomic force microscopy and Scanning tunneling microscopy. Nanoscience and technology, Springer-Verlag, 2015. [56] V. D. Lyles, W. K. Serem, J.-J. Yu, and J. C. Garno, “Surface Characterization Using Atomic Force Microscopy (AFM) in Liquid Environments,” in Surface Science Techniques, pp. 599–620, Springer, 2013. [57] L. E. Davis, Handbook of Auger electron spectroscopy: a reference book of standard data for identifcation and interpretation of Auger electron spectroscopy data. PerkinEmer Corp., 1979. [58] J. B. Pendry, “Low-Energy Electron Di˙raction,” in Interaction of Atoms and Molecules with Solid Surfaces, Physics of Solids and Liquids, pp. 201–211, 1990. [59] G. Binnig and H. Rohrer, “Scanning tunneling microscopy,” Surface Science, vol. 126, no. 1, pp. 236–244, 1983. [60] N. N. Greenwood and T. C. Gibb, Mössbauer Spectroscopy. Dordrecht: Springer Netherlands, 1971. [61] W. Keune, “Application of Mössbauer spectroscopy in magnetism,” Hyperfne Interactions, vol. 204, no. 1, pp. 13–45, 2012. [62] G. J. Long, ed., Mössbauer Spectroscopy Applied to Inorganic Chemistry. Modern Inorganic Chemistry, Springer US, 1984. [63] Y. Yoshida and G. Langouche, eds., Mössbauer Spectroscopy: Tutorial Book. Berlin Heidelberg: Springer-Verlag, 2013. [64] R. L. Mössbauer, “Kernresonanzfuoreszenz von Gammastrahlung in Ir191,” Zeitschrift für Physik, vol. 151, no. 2, pp. 124–143, 1958. [65] J. R. Gancedo, M. Gracia, and J. F. Marco, “Practical aspects of Mössbauer spectroscopy instrumentation,” Hyperfne Interactions, vol. 83, no. 1, pp. 71–78, 1994. [66] F. Salvat and J. Parellada, “Theory of conversion electron Mössbauer spectroscopy (CEMS),” Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, vol. 1, no. 1, pp. 70–84, 1984. [67] J. R. Gancedo, M. Gracia, and J. F. Marco, “Cems methodology,” Hyperfne Interactions, vol. 66, no. 1, pp. 83–93, 1991. [68] J. F. Marco, J. R. Gancedo, M. Monti, and J. d. l. Figuera, “Mössbauer spectroscopy and surface analysis,” in Mössbauer Spectroscopy, pp. 455–469, 2013. [69] N. Spiridis, K. Freindl, J. Wojas, N. Kwiatek, E. Madej, D. Wilgocka-‚l¦zak, 143 Bibliography P. Dró»d», T. ‚l¦zak, and J. Korecki, “Superstructures on epitaxial Fe3O4(111) flms: Biphase formation versus the degree of reduction,” Journal of Physical Chemistry C, vol. 123, no. 7, pp. 4204–4216, 2019. [70] F. d. Groot and A. Kotani, Core Level Spectroscopy of Solids. CRC Press, 2008. [71] G. van der Laan and A. I. Figueroa, “X-ray magnetic circular dichroism—A versatile tool to study magnetism,” Coordination Chemistry Reviews, vol. 277-278, pp. 95–129, 2014. [72] C. C. Chusuei and D. W. Goodman, “X-Ray Photoelectron Spectroscopy,” in Encyclopedia of Physical Science and Technology (Third Edition), pp. 921–938, Academic Press, 2003. [73] F. Jiménez-Villacorta, C. Prieto, Y. Huttel, N. D. Telling, and G. van der Laan, “X-ray magnetic circular dichroism study of the blocking process in nanostructured iron-iron oxide core-shell systems,” Physical Review B, vol. 84, no. 17, p. 172404, 2011. [74] L. Aballe, M. Foerster, E. Pellegrin, J. Nicolas, and S. Ferrer, “The alba spectroscopic leem-peem experimental station: layout and performance,” Journal of Synchrotron Radiation, vol. 22, no. 3, pp. 745–752, 2015. [75] J. Stöhr, “Exploring the microscopic origin of magnetic anisotropies with X-ray magnetic circular dichroism (XMCD) spectroscopy,” Journal of Magnetism and Magnetic Materials, vol. 200, no. 1, pp. 470–497, 1999. [76] T. Funk, A. Deb, S. J. George, H. Wang, and S. P. Cramer, “X-ray magnetic circular dichroism—a high energy probe of magnetic properties,” Coordination Chemistry Reviews, vol. 249, no. 1, pp. 3–30, 2005. [77] E. Bauer, Surface Microscopy with Low Energy Electrons. Springer, 2014. [78] J. Stöhr and H. C. Siegmann, Magnetism: From Fundamentals to Nanoscale Dynamics. Springer Series in Solid-State Sciences, 2006. [79] P. Fischer, “Magnetic Imaging with Polarized Soft X-rays,” in Magnetism and Synchrotron Radiation: Towards the Fourth Generation Light Sources, Springer Proceedings in Physics, pp. 155–183, 2013. [80] W. Kuch and C. M. Schneider, “Magnetic dichroism in valence band photoemission,” Reports on Progress in Physics, vol. 64, no. 2, p. 147, 2001. [81] A. Barla, J. Nicolás, D. Cocco, S. M. Valvidares, J. Herrero-Martín, P. Gargiani, J. Moldes, C. Ruget, E. Pellegrin, and S. Ferrer, “Design and performance of boreas, the beamline for resonant x-ray absorption and scattering experiments at the alba synchrotron light source,” Journal of Synchrotron Radiation, vol. 23, no. 6, pp. 1507– 1517, 2016. 144 Bibliography [82] C. T. Chen, Y. U. Idzerda, H.-J. Lin, N. V. Smith, G. Meigs, E. Chaban, G. H. Ho, E. Pellegrin, and F. Sette, “Experimental Confrmation of the X-Ray Magnetic Circular Dichroism Sum Rules for Iron and Cobalt,” Physical Review Letters, vol. 75, no. 1, pp. 152–155, 1995. [83] S. Günther, B. Kaulich, L. Gregoratti, and M. Kiskinova, “Photoelectron microscopy and applications in surface and materials science,” Progress in Surface Science, vol. 70, no. 4, pp. 187–260, 2002. [84] T. Schmidt, S. Heun, J. Slezak, J. Diaz, K. C. Prince, G. Lilienkamp, and E. Bauer, “SPELEEM: Combining LEEM and Spectroscopic Imaging,” Surface Review and Letters, vol. 05, no. 06, pp. 1287–1296, 1998. [85] G. Bracco and B. Holst, eds., Surface Science Techniques. Springer Series in Surface Sciences, 2013. [86] A. Vansteenkiste, J. Leliaert, M. Dvornik, M. Helsen, F. Garcia-Sanchez, and B. Van Waeyenberge, “The design and verifcation of MuMax3,” AIP Advances, vol. 4, p. 107133, 2014. [87] T. L. Gilbert, “A lagrangian formulation of the gyromagnetic equation of the magnetization feld,” Physical Review D, vol. 100, p. 1243, 1955. [88] L. D. Buda-Prejbeanu, Modelling of spintronic devices: from basic operation mechanisms toward optimization. Habilitation à diriger des recherches, Communauté Université Grenoble Alpes, 2017. [89] V. Adelsköld, “X-ray studies on magneto-plumbite, PbO. 6Fe2O3, and other substances resembling “beta-alumina”, Na2O. 11Al2O3,” Arkiv for Kemi Mineralogioch Geologi, Series A-12, vol. 29, pp. 1–9, 1938. [90] H. Kojima, “Chapter 5 Fundamental properties of hexagonal ferrites with magnetoplumbite structure,” in Handbook of Ferromagnetic Materials, vol. 3, pp. 305–391, Elsevier, 1982. [91] R. Pullar, Hexagonal Ferrites: A Review of the Synthesis, Properties and Applications of Hexaferrite Ceramics, vol. 57. 2012. [92] J. Kreisel, H. Vincent, F. Tasset, M. Paté, and J. P. Ganne, “An investigation of the magnetic anisotropy change in BaFe12−2xTixCoxO19 single crystals,” Journal of Magnetism and Magnetic Materials, vol. 224, no. 1, pp. 17–29, 2001. [93] K. Momma and F. Izumi, “VESTA 3 for three-dimensional visualization of crystal, volumetric and morphology data,” Journal of Applied Crystallography, vol. 44, no. 6, pp. 1272–1276, 2011. [94] W. Heisenberg, “Mehrkörperproblem und Resonanz in der Quantenmechanik,” Zeitschrift für Physik, vol. 38, no. 6, pp. 411–426, 1926. 145 Bibliography [95] P. A. M. Dirac and R. H. Fowler, “On the theory of quantum mechanics,” Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character, vol. 112, no. 762, pp. 661–677, 1926. [96] E. Gorter, Saturation magnetization and crystal chemistry of ferrimagnetic oxides. PhD thesis, 1954. [97] C. Tejera-Centeno, S. Gallego, and J. I. Cerdá, “An ab initio study of the magnetic properties of strontium hexaferrite,” Scientifc Reports, vol. 11, no. 1, p. 1964, 2021. [98] A. Grill and F. Haberey, “E˙ect of diamagnetic substitutions in BaFe12O19 on the magnetic properties,” Applied physics, vol. 3, no. 2, pp. 131–134, 1974. [99] J. M. D. Coey, Magnetism and Magnetic Materials. Cambridge: Cambridge University Press, 2010. [100] F. X. N. M. Kools, Science and technology of ferrite magnets : modelling of coercivity and grain growth inhibition. PhD thesis, 2003. [101] G. Bertotti, “Chapter 5 - Exchange and Anisotropy,” in Hysteresis in Magnetism, pp. 129–162, Academic Press, 1998. [102] F. Kools and P. J. v. d. Valk, “Ferrites,” in Kirk-Othmer Encyclopedia of Chemical Technology, 2004. [103] F. Hellman in Encyclopedia of Materials: Science and Technology, pp. 4761–4767, Elsevier, 2001. [104] E. F. Bertaut, A. Deschamps, R. Pauthenet, and S. Pickart, “Substitution dans les hexaferrites de l’ion Fe3+ par Al3+ , Ga3+ , Cr3+,” Journal de Physique et le Radium, vol. 20, no. 2-3, pp. 404–408, 1959. [105] L. Jahn and H. G. Müller, “The Coercivity of Hard Ferrite Single Crystals,” physica status solidi (b), vol. 35, no. 2, pp. 723–730, 1969. [106] C. Zener, “Classical Theory of the Temperature Dependence of Magnetic Anisotropy Energy,” Physical Review, vol. 96, no. 5, pp. 1335–1337, 1954. [107] G. Herzer, “Grain size dependence of coercivity and permeability in nanocrystalline ferromagnets,” IEEE Transactions on Magnetics, vol. 26, no. 5, pp. 1397–1402, 1990. [108] A. Verma, O. Pandey, and P. Sharma, “Strontium ferrite permanent magnet-An overview,” Indian Journal of Engineering Materials Sciences, vol. 7, no. 5-6, 2000. [109] M. Sugimoto, “The Past, Present, and Future of Ferrites,” Journal of the American Ceramic Society, vol. 82, no. 2, pp. 269–280, 1999. [110] D. Y. Chen, Y. Y. Meng, D. C. Zeng, Z. W. Liu, H. Y. Yu, and X. C. Zhong, “CTAB146