Tailoring the Magnetic Properties of Amorphous TbCo Nano Films

UPTEC Q 18001

Examensarbete 30 hp Mars 2018

Tailoring the Magnetic Properties of Amorphous TbCo Nano Films

Viktor Djurberg

Teknisk- naturvetenskaplig fakultet UTH-enheten

Besöksadress:

Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0

Postadress:

Box 536 751 21 Uppsala

Telefon:

018 – 471 30 03

Telefax:

018 – 471 30 00

Hemsida:

http://www.teknat.uu.se/student

Abstract

Tailoring the Magnetic Properties of Amorphous TbCo Nano Films

Viktor Djurberg

The possibility to change magnetic anisotropy of amorphous TbCo films from out-of-plane to in-plane has been investigated. The effects of TbCo film's thickness and composition on the magnetic anisotropy were investigated together with the effects of growing the TbCo films on a SmCo seed layer. This was studied by sputtering TbCo films of composition Tb_xCo_(100-x) x=16,18,20,22 and 24, with thickness ranging between 2-20 nm, with and without the presence of a 20 nm Sm_15Co_85 seed layer. All films were grown in a 130 mT magnetic in- plain field to imprint an in-plane anisotropy. The structure and

composition of the films were examined with Rutherford backscattering spectrometry, X-ray reflectivity, and Grazing incidence X-ray

diffraction. The magnetic properties of the films were studied with magneto-optic Kerr effect measurement, vibrating sample magnetometer, Kerr microscopy and magnetic force microscopy. The magneto-optic Kerr effect measurement showed that it was possible to change TbCo film's preferred magnetization direction from out-of-plane to in-plane by reducing the film thickness. The SmCo layer made it easier for the TbCo films to change preferred magnetization direction from out-of- plane to in-plane.

ISSN: 1401-5773, UPTEC Q18 001 Examinator: Åsa Kassman Rudolphi Ämnesgranskare: Petra Jönsson Handledare: Gabriella Andersson

Skräddarsy de magnetiska egenskaperna hos ultra tunna TbCo skikt

Utvecklingen och upptäckter av nya magnetiska material har gått hand i hand med utveckling- en av ny teknologi. Ett exemplet är att upptäckten av naturligt magnetiska stenar möjliggjorde utvecklingen av kompasser. Utvecklingen av magnetiska material har kommit långt sedan dess och idag är kraven väldigt höga på nästa generation magnetiska material som ska kunna möta de nya teknologiska utmaningarna som vi står inför. Idag behövs material med specifika kombinationer av magnetiska egenskaper för att fungera till morgondagens högteknologiska tillämpningar, såsom nästa generations magnetiska hårddiskar och sensorer.

Ett sätt att tillverka material med specifika magnetiska egenskaper är att utgå från ett ma- terial vars magnetiska egenskaper går att ändra på. Examensarbetes mål vara att utforska olika sätt att påverka magnetiska egenskaper i amorft Terbium-Kobolt (TbCo). TbCo valdes eftersom det har visatsig vara relativt enkelt att påverka de magnetiska egenskaperna i mate- rialet. Anledningen till detta är att materialet är amorft och ferrimagnetiskt.

Att ett material är amorft betyder att atomerna i materialet sitter samman på ett sätt som gör att de saknar ordning över längre avstånd men de kan fortfarande ha ordning på kortare avstånd. Figuren 1 visar att skilnaden mellan ett amorft och ett kristalint material. Eftersom ett amorft material saknar regelbundenhet går det att göra små ändringar i hur atomerna för- håller sig till varandra genom att, till exempel, tillverka ett magnetiska amorfa materialet i ett magnetfält. Förändring i atom positionerna kommer påverka elektronerna i materialet vilket i sin tur påverkar de magnetiska egenskaperna.

Figur 1: En beskriving för hur atomerna har för ordning i ett kristallint material (a) och i ett amorft material (b).

Ett ferrimagnetisk material består av flera olika atomslag med olika magnetiska moment som linjerar upp sig åt olika håll, se figur 2. Ett magnetiskt moment berättar hur starkt magne- tisk atomen är samt i vilken riktning magnetisering är riktad, vilket beskris med en pil i bilden.

Genom att förändra sammansättningen går det att ändra förhållandet mellan de motriktade

magnetisk momenten och därigenom ändra de magnetiska egenskaperna hos materialet. Vid

en viss sammansättning kommer den magnetiska momenten åt båda hållen vara lika stor, vil-

ket leder till ett provet blir omagnetiskt. Denna sammansättning kallas kompensationspunk-

ten.

Figur 2: Ett ferrimagnetiskt material under kompensationspunkten (A) och ovanför (B). Pi- larna beskriver atomernas magnetiska moment. Den stora pilen beskriver det totala magnetiska momentet hos materialet.

I detta arbetet studerades den magnetiska anisotropin hos TbCo. Att ett magnetiskt materi- al är anisotropt betyder att det är enklare att magnetisera materialet i vissa riktningar. Ett tunt skikt av amorft TbCo har den anisotropin som gör att materialet är enklare att magnetisera ut ur skiktet. Det som undersöktes var om det går att ändra så att det är enklare att magnetisera TbCo längs med skiktet istället för ut ur skiktet och hur detta, bland annat, är beroened av tjockleken och sammansättning av filmen.

För att undersöka anisotropin hos tunna skick av amorft TbCo så tillverkades det över 50 prover. Alla prover tillverkades med en metod som kallas sputtring. Medans proverna tillvär- kades utsattes de för ett magetfält. Detta gjordes för att påverka atomerna att lägga sig på ett sådant sätt att de vill magnetiseras i planet. Proverna skilde sig gällande tjocklek, sammansätt- ning och om TbCo var växt ovan på amorft SmCo. Provernas tjocklekar var mellan 2-20 nm och de sammansättningar som undersöktes var 16, 18, 20, 22 och 24 atom% Tb i filmerna vilket täckte in båda sidorna av kompensationspunkten.

Flera experimentella metoder användes för att undersöka de magnetiska och fysiska egen- skaperna på proven. Två röntgen baserade metoder som kallas för röntgenreflektivitet och grazing-incidence-röntgendiffraktion användes för att undersöka om proverna var amorfa och deras tjocklekar. Rutherford backscattering spektrometry, som är en metod baserad på en jon- stråle, användes för att bestämma sammansättningarna. Olika metoder baserade på magneto- optiska Kerreffekten, som beskriver hur ljus roterar när det reflekteras från en mangetisk yta, användes för att undersöka de magnetiska egenskaperna på proven.

Slutsattsen är att det går att påverka de magnetiska egenskaperna så att TbCo hellre vill

vara magnetiserad i skiktet istället för ut ur skiktet. För proverna där TbCo var växt på SmCo

hände detta vid en tjocklek på runt 8 nm och för enbart TbCo så började detta hända vid

ungefär 2 nm.

Contents

1 Introduction 1

2 Theory 3

2.1 Introduction to magnetic materials . . . . 3

2.2 Materials . . . . 6

2.3 Analyzing methods . . . . 9

3 Sample Fabrication and Structural Analysis 12

3.1 Sputtering of the TbCo films . . . 12

3.2 The Methodology for Creating the Samples . . . 14

3.3 Measurement set-ups . . . 15

3.4 Results and Discussion of the Structure and Composition . . . 17

4 Magnetic Properties of TbCo Films 23

4.1 Measurement set-ups . . . 23

4.2 Result and Discussion of the Magnetic Properties . . . 27

5 Magnetic Domain Structure 43

5.1 Measurement set-up . . . 43

5.2 Result and Discussion of the Magnetic Domain Structure . . . 46

6 Conclusions 51

6.1 Future work . . . 51

A 57

A.1 MOKE . . . 57

A.2 MFM-images . . . 90

A.3 Kerr microscope images . . . 91

A.4 Codes . . . 93

Chapter 1

Introduction

The first magnetic material discovered was a ferrimagnetic iron ore called Lodestone. This material was later used to create the first compass which revolutionized the long distance sea- faring, since it made it possible to navigat during cloudy days. Today, new discoveries in magnetic materials could lead to improvements in existing technology or enable new technol- ogy. However, the next generation magnetic materials needs to have very specific combination of magnetic properties to be used in e.g. high density hard drives, sensors, and electrical com- ponents. One way to meet this technological challenge, of engineering these new materials, is to find different ways of tailoring the magnetic properties of materials.

The aim of this project was to investigate different ways of tailoring the magnetic proper- ties of materials. Amorphous TbCo was chosen as the material to be investigated and mag- netic anisotropy was chosen as the magnetic property to be tailored, to demarcate the project.

Amorphous TbCo was chosen because of two reasons; Firstly, amorphous TbCo has a large out-of-plane anisotropy which is a sought after property for many applications. Secondly, amorphous TbCos magnetic properties can be change through changes in compensation and growth conditions [1, 2, 3]. For example, one way to change the magnetic properties is by growing the TbCo in a magnetic field [4].

Four different ways of affecting the anisotropy were investigated. The first one was grow- ing the samples in a in-plane magnetic field. The method has been shown to imprint an anisotropy along the magnetic field direction in e.g. amorphous SmCo films [4, 5]. The second way was to grow the TbCo on top of a layer of amorphous SmCo. The idea was that these layer would interact and that the anisotropy of the SmCo would affect the anisotropy of the TbCo. Lastly, the anisotropy dependence on the thickness and the compensation of the films were also investigated.

Amorphous TbCo has been suggested for a wide range of different applications because

of its attractive combination of properties. One of these areas is in new technologies for im-

proving the reading and writing speeds of magnetic bits used in magneto-optical recording

technologies [6]. Magneto-optical recording store bits as magnetic domains with reversed mag-

netization but, compared to traditional hard drives, uses light to read and write the bits. One

subgroup of magneto-optical recording is "all optical switching" (AOS) where a polarized laser

pulse is used to switch the domain magnetization. This method could revolutionize today’s

hard drives with faster reading and writing. AOS gained attention after it was discovered

that it was possible to switch domains in Gd

Fe

Co

using just a 40 femtosecond circular

polarized laser pulse [7]. It has also been demonstrated that low remanent magnetization is

important for AOS [8] which is a reason why it is important to be able to tailor the magnetic

properties of the magnetic material.

Figure 1.1: A Schmatic image of how an all optical switching (AOS) and heat-assisted magnetic switching devise work.

Another promising application, for denser magnetic storage, is heat-assisted magnetic switch- ing. It pushes the superparamagnetic limit down by using a material with high out of plane anisotropy but during switching momentary heating it up with a laser, thus decreasing the co- ercivity [9]. TbCo is a promising material for this application, because of its large out of plane anisotropy and its heat-sensitive coercivity [10].

There are other applications for TbCo beside using it as a medium for magnetic storage. Ex- amples ranging from MEMS material [11], as a material in reader heads [12], magnetic tunnel junctions [13, 14] and magnetic sensors [15], just to mention a few. In many of these applica- tions, TbCo is not used alone but together with other materials in layer combinations, so called heterostructure. The different layers can augment each other properties by, for example, creat- ing exchange bias between the layer or in other ways changing the magnetic properties of the layers.

One last application where amorphous TbCo can be used is in spintropics devices. Spin-

tronics refers to devices that use spin current to change magnetic domains. There are a mul-

titude of interesting applications such as magnetic hard-drives and magnetic random access

memory (using spin currents to change the magnetisation). Using materials that have a strong

out-of-plane magnetic anisotropy gives better thermal stability [16] and lower currents for

spin-transfer switching [17].

Chapter 2

Theory

2.1 Introduction to magnetic materials

This section is a general introduction to magnetic materials and how to interpret magnetic measurements.

2.1.1 General Magnetic Theory

One way to think about a magnetic material is as a collection of tiny bar magnets. Each atom in the material is one of these bar magnets and the strength of these bar magnets is defined by the magnetic moment ~ m. The atomic magnetic moment can be defined through the energy (E

) that the atomic magnetic moment has if it is aligned along the applied magnetic field ( _H). ~

E

= −~ m · ~ H (2.1)

However, it is hard to measure one atomic magnetic moment in a material, so the total magnetic moment Σ ~ m of a sample is usually measured. The total magnetic moment is normal- ized with the volume (V) of the sample to get a material constant, giving the magnetization (M).

M = ^Σ ~ m

V (2.2)

2.1.2 Magnetic anisotropy

Magnetic anisotropy denominates the direction dependencies of the magnetic properties. The magnetic anisotropy is usually described with an easy-axis, along which it is energetically fa- vorable for the magnetic moment to align. Hence, the magnetic moment will spontaneously align along the easy-axis if it is not affected by an external magnetic field. However, there might be more than one easy-axis in a material so all the magnetic moments in the entire ma- terial might not align with each other.

There are different fundamental reasons for the magnetic anisotropy. For example, the atomic structure of the material or from the shape of the magnetic object. The magnetic anisotropy from the structure can approximately be described as an interaction between the crystal electric field and the electrical orbits (connected to the electrons’ angular moment).

Thus, changing the structure of the material in some way will usually affect the magnetic

anisotropy. The magnetic anisotropy from the shape of the magnetic sample arises from that

the demagnetizing field is not equal in all directions. For example, thin films usually has an

in-plane magnetic anisotropy because of the energy associated with demagnetizing field.

2.1.3 Magnetic Domain Structure

A magnetic domain is a part of the material where all the atomic moments aligned in a way that decreases the energy. Between these magnetic domain are regions where the magnetic moment does not line up with either domain. Separating the different domain are region, called domains walls, where the moment gradually change one direction to the other. An example of magnetic material where different magnetic is shown in figure 2.1.

Figure 2.1: The magnetic domian strucutre of a thin film of TbCo, the dark and light area are domains with revers magnetization. The magnetic field is out-of-plane.

2.1.4 Magnetic Hysteresis

How the magnetic moment of a magnetic material respond to an applied magnetic field is not trivial. It is energetically favorable for the magnetic moment to align along the magnetic field.

However, there are cases in which the moment does not align with the magnetic field because of the energy cost associated with the magnetic moment deviating from the easy-axis in the material. This is why the magnetic moment will be keep is magnetic domain structure if not a large enough magnetic field is applied. This is the reason behind that the magnetization does not only depends on the applied field but also on the previous arrangement of the magnetic moment, which is called magnetic hysteresis. How magnetic hysteresis is seen in magnetic measurements is explained below.

Different shapes of the hysteresis curves, in which the magnetization is plotted against the

applied field, are shown in figure 2.2. The example will be a single domain magnetic particle

with one easy-axis, for simplicity. If the applied magnetic field is along the easy-axis for a single domain magnetic particle, its moment will not change directly when the applied field change direction, because of the anisotropy. The size of the applied field that is needed to change the direction of the moment is a material parameter called the coercive field (H

). Fig- ure 2.2 b shows a hysteresis loop of a single domain magnetic particle measured along the easy-axis. Lastly, for a single domain magnetic particle with an easy-axis which is perpendic- ular to the applied field will not have a sharp transition from one direction to another. Instead, the moment will gradually align itself along the applied field until it parallel to the field. The hysteresis loop will have the form is shown in figure 2.2 c.

(a) Hysteresis loop of a mag- netic isotropic material.

(b) Hysteresis loop of a mag- netic anisotropic material along easy-axis.

(c) Hysteresis loop of a mag- netic anisotropic material perpendicular to easy-axis

Figure 2.2: Three examples of hysteresis loops for single domain magnetic particle with only

one easy-axis at zero kelvin.

From a magnetic hysteresis curve can important magnetic material parameters be extracted.

The mentioned coercive field (H

) which is the strength of the magnetic field for where the spin

reversal happens. The largest magnetization is called saturation magnetization (M

) and can

be related back to the size of the magnetic moments in the sample. Lastly, the remanent mag-

netization (M

) is the magnetization of the sample at zero field after it has previously been

magnetized. The quantities are shown in figure 2.3

Figure 2.3: The magnetic hysteresis curve from a fictive magnetic sample. In the curve is the co- ercive field (H

), saturation magnetization (M

) and remanent magnetization (M

) shown.

2.2 Materials

2.2.1 Ferrimagnetic and compensation point

A ferrimagnetic material has two or more different kinds of atoms which have opposing mag- netic moments. In a ferrimagnetic material, the coupling between the moments is the same as in an antiferromagnet, but the magnetic moments does not cancel out each other leading to a net magnetic moment. However, by tuning either the temperature or composition of atoms in the material is it possible to get sum of the opposing moments to the same size and through it cancel out the net magnetic moment. This is called the compensation temperature and the compensation composition respectively. Close to this point the coercive field (H

) will increase drastically [18].

An example of how the opposing moments change with the composition of the ferrimag- netic material TbCo is shown in figure 2.4. The net moment of the Tb will be greater than the net moment of Co above the compensation point and therefore it align with the magnetic field.

The Co moment, that is antiferromagneticly coupled to the Tb, will thus align in the opposite direction.

2.2.2 Amorphous materials as functional material

First in 1960, Gubanov theoretically proved that magnetic ordering in amorphous materials

was possible. Before that, it was believed that magnetic ordering would only be possible in

crystalline materials [19]. The interest in amorphous materials has grown, during the last few

years, for many reasons. Firstly, the structure of amorphous magnetic materials have many

Figure 2.4: How the net moment of Tb and Co behaves close to the compensation point.

advantages. They are almost free from defects such as atomic steps and grain boundaries which leads to a decrease the domain wall pinning[20, 21, 22]. Secondly, lattice matching be- comes a non-issue for heterostructure, which gives a great range of material combinations[23].

Lastly, the composition of a amorphous material can vary greatly without changing the lattice constant or "structure", making tuning the materials magnetic properties through changes in composition easy.

2.2.3 Imprinted in-plane anisotropy in amorphous materials

Amorphous materials have no long range ordering, however, it has been shown that it can still exist a short or medium range ordering in amorphous materials [24]. By applying a magnetic field during the growth of the material, the short to medium range ordering can change in such a way that it induces a magnetic anisotropy [4, 25]. The origin of the induced magnetic anisotropy is still not exactly known, however, it can be both the result of both macroscopic and microscopic effects. Examples of macroscopic effects are stress induced magnetostrictive effects and heterogeneous voids [25]. Microscopic effects that induce magnetic anisotropy in- cludes alignment of different atomic clusters and effects on the direction of bonding between atoms [26, 27].

The medium range ordering make it possible to induce anisotropy by growing the film in a magnetic field which has been shown, for example, in SmCo systems [25]. However, the exact type of ordering that gives rise to the anisotropy is still unknown. A group at Uppsala University has shown that imprinted anisotropy does not only depends on magnetostrictive effects [25]. That is why a further investigates of the local configuration is needed to answer these question.

2.2.4 Exchange bias in ferrimagnetic heterostructure

Exchange bias comes from an interaction between two magnetic layers. This interaction make it more suitable for one layer to be magnetized in a certain direction. The effect was first observed by Meiklejohn and Bean in Co/CoO particles over 60 years ago [28]. Co is a fer- romagnet (FM) and CoO is an antiferromagnet (AF), which makes the exchange coupling of FM/AF type. However, exchanges bias also occurs in FM/FM and Ferrimagnet/Ferrimagnet (FI/FI) systems [29].

A strong perpendicular exchange bias in FI/FI-system for transition metals (TM) and rare

earth metals (RE) has been seen in the TbFeCo system [30]. The suggested mechanism for

the exchange bias is that a "hard" (hard to change direction of magnetic moment) FI-layer is pinning the "soft" (easy to change direction of magnetic moment) FI layer, see figure 2.5.

The interaction at the interface is the same as inside the material, i.e. the RE-RE and TM-TM is coupled FM and RE-TM AF. This can have several advantages compared to FM/AF. One example is that FM/AF system can be very sensitive to surface roughness because interlayer defected/interlayer roughness will result in compensated spins, a limitation that FI/FI do not have [30]. This uncompensated spins could result in a higher exchange bias compared to FM/AF.

Figure 2.5: An RE-TM FI/FI-system with exchange bias. The under layer is hard FI layer and is pinning the softer top layer.

2.2.5 Amorphous TbCo and SmCo

Amorphous TbCo is a material with out-of-plane magnetic anisotropy. V. G. Harris showed that the out-of-plane magnetic anisotropy, in the related system TbFe, is dependant on a slight structural change between out-of-plane- and in-plane-direction [27]. Since a materials electron structure, and through it magnetic properties, is highly depend on the atomic structure, can the magnetic properties be changed by changing the structure. A change in structure can be achieved by altering the annealing and manufacturing parameters or thickness [31, 32, 33].

That is why the Tb

Co

composition point at RT for can vary, but it is around 20-23 % Tb [4, 34, 35].

SmCo

Crystalline SmCo is well known for its technical use because of its large coercivity, fairly large

magnetic moment and a high Curie temperature. Amorphous SmCo has less moment and co-

ercivity than its crystalline cousin but has the advantage that its magnetic properties are highly

tunable. There is also the possibility of imprinting a large anisotropy in amorphous SmCo by growing the material in a magnetic field [5]. Amorphous SmCo of composition Sm

Co

has 1 mm magnetic domains[36]. The composition used within this thesis was Sm

Co

which is well below compensation point at RT. Hence, the total magnetic moment of Co is greater compared with the total moment of Sm [5].

Figure 2.6: Thin films of TbCo and SmCo sputtered on two diffrent pieces of silicon waffers.

2.3 Analyzing methods

2.3.1 Grazing incidence X-ray diffraction

Grazing incidence X-ray diffraction (GIXRD) is a surface sensitive X-ray diffraction method commonly used to study thin films. As the name suggests the angle of the incidence X-ray is fixed to an value shortly above the critical angle. This makes sure that the X-rays just travel through the surface and not the bulk. Since the angle of the ingoing X-rays is fixed, only the de- tector moves. This means that the q-vector is not perpendicular to the films surface at all times, which it is for X-ray diffraction. The q-vector describes the moment transfer of the sample to the light when it gets reflected from the sample. The direction of the q-vector is correlated to the analyzed direction in the sample. Hence, the direction of the q-vector is important for structural anisotropic materials. However, the GIXRD spectrum will be the same as the X-ray diffraction spectrum for structural isotropic materials [37].

In amorphous material, no long order exists between atoms so no systematic destructive or constructive interferes can occur. Therefore, the intensity will be weak and it will depend on the radial distribution of atoms.

2.3.2 X-ray reflectivity

X-ray reflectivity (XRR) is a method which is used to determine layer thickness and layer

roughness/inter-layer mixing of thin films. X-ray reflectivity is a typical X-ray diffraction

measurement, where the intensity of the reflected X-ray is measured as a function of the re-

flected angle θ, but only for low angles [37]. The angles are chosen close above the angle of

total reflection because the reflectivity approximately decreases with 1/q

which is shown with

equation 2.3.

q = 4π sin θ/λ (2.3) q is moment transfer of the X-rays„ where λ is the wavelength of the X-rays [38].

2.3.3 Rutherford backscattering spectrometry

Rutherford backscattering spectrometry is commonly used when a non-destructive elemental depth analysis is needed. The sample gets probed by high energy ions, usually helium, and the backscattered ions intensity and energy is recorded. Then loss of energy is correlated to the ions travel-distance in the material and to the weight of the nucleus it was reflected from. The ions are assumed to be scattered elasticity.

2.3.4 Vibrating Sample Magnetometer

A vibrating sample magnetometer (VSM) is a measurement system used to measure the mag- netic moment of a sample as a function of an applied magnetic field. The method works by vibrating the magnetic sample close to two copper coils. The movement of the magnetic sam- ple induces a voltage in the coils, which can be described through Lenz’s law. The voltage can then be related back to the magnetic moment of the sample.

2.3.5 Measurements Techniques based on Magneto-optic Kerr effect

The interaction between light and a magnetic material will give rise to magneto-optic effects.

These effects come from an optical anisotropy in the material that originates from the magnetic moments in the material.

In 1876, Kerr showed that polarized light changes rotation after it gets reflected from a magnetic surface, an effect later named the Kerr effect [39]. Furthermore, the rotation is to a first order approximation proportional to the magnetic moment of the sample, which makes it possible to use this effect in different surface sensitive magnetic measurements. However, the rotation depends, in a non-linear way, on the composition of the sample, the temperature, and the wavelength of the polarized light, which makes the effect hard to use to find the absolute moments [40].

Magneto-optic Kerr effect measurement

Magneto-optic Kerr effect measurement (MOKE) is a standard method for analyzing the mag- netic properties of surfaces. It is used because of its surface sensitivity (around 20 nm), local probing nature and the simplicity. A simple MOKE set up measures how much the polarisa- tion of the light rotates after it gets reflected from a magnetic surface. The magnitude of the rotation is to first order approximation proportional to the magnetization of the surface [41].

Polar-MOKE (P-MOKE) is a MOKE-measurement which is sensitive to the out-of-plane

magnetization of the sample surface. This is achieved by having the incident light be paral-

lel to the norm of the surface. Longitudinal-MOKE (L-MOKE) is a method where the surface

is probed with s-polarized light where the samples magnetization is parallel to the reflective

surface and perpendicular to the scatter plane (see figure ??). However, if the samples magne-

tization has a out-of-plane component with that also be measured [41]. The disadvantage with

MOKE is that the magnetization only can be measured in relative terms compared for example

with vibrating sample magnetometer where the total magnetization can be measured.

Figure 2.7: The definition of the scatter plane, s- and p-polarized light.

Kerr Microscope

The Kerr microscope uses the magneto-optic Kerr effect to show the domain structure on the surface of a magnetic sample. The Kerr microscope works much in the same way as a light microscope with the added feature that intensity also depends on magnetic domain structure.

This is achieved by using polarized incident light and having a second polarizer for the re- flected light. By polarising the light a second time, it is possible to see the effects of the Kerr rotation and through it get information on the magnetization of the surface. One thing to note is that the samples need to be flat and uninform so the change of light intensity only will depend on the Kerr rotation and not on the surface structure.

2.3.6 Magnetic force microscopy

Magnetic force microscopy (MFM) is a method used to analyze the domain structure of a mag- netic surface. The method is based on having a sharp magnetic tip that runs over the surface.

The induces a force F is described by:

F =

(~ m ∇) ~ H (2.4)

Where µ

is the permeability, H is the stray field from the surface, m magnetic moment of the tip and F induces a force.

To measure very small magnetic interaction a magnetic tip is vibrates, at close to its res-

onant frequency, over the surface, in a so called tapping mode. The resonant frequency will

change depending on the magnetic force on the tip. The change in resonant frequency can be

related to a changes in the phase and amplitude which is measured.

Chapter 3

Sample Fabrication and Structural Analysis

To investigate if it was possible to change the out-of-plane magnetic moment of TbCo films to in-plane, 49 samples were grown. The samples varied in composition, thickness, and exis- tence of a 20 nm Sm

Co

underlayer. The films composition were analyzed with Rutherford backscattering spectrometry (RBS) and their structures were examined with X-ray reflectivity (XRR), and grazing incidence X-ray diffraction (GIXRD). A more detailed explanation of how the experiment was carried out, together with a short explanation of the analyzing software, follows in this section. The experiment was carried out by the author if nothing else is men- tioned. The magnetic measurements are described in Chapter 4.

3.1 Sputtering of the TbCo films

The films were grown in an ultra-high vacuum sputtering system Sleipnir, shown in figure 3.1,

which have four magnetrons connected to direct current power supplies. The four targets

used, one of each magnetron, were Tb, Sm, Co, and Al

Zr

. During the growth 99.9999 %

pure Argon was used as a sputtering gas at a pressure of 2.00 mTorr.

Figure 3.1: The sputtering system used for the growth of the samples. A) The load lock there the sample is load into. Its function is to be a intermediater between the outside and the high vaccum inside the main chamber. B) The main chamber, where the sample is sputtered. C) The electronics that is needed to control Sleipnir.

The films were grown on (100)Si substrates having a natural oxide layer. The composition

and thickness of the films were controlled by choosing the sputtering times and power for the

different magnetrons. The power for the Tb and Sm magnetrons was kept constant and the

power of the Co magnetron was varied in order to change the composition. AlZr was both

used as a seed (to facilitate amorphous growth) and a cap layer (to prevent oxidation of the

top layer). A schematic image of the structure of the films is shown in the figure 3.2.

Figure 3.2: A schematic image over the samples made. On the left hand side is the hetrostruc- tures and on the right hand side is the single films.

All films were grown in a 130 mT in-plane magnetic field, to facilitate an in-plane magnetic anisotropy. The magnetic field came from two NdBFe magnets that were mounted on the sample holder. The sample holder rotated during the growth to achieve an even composition and layer thickness.

3.2 The Methodology for Creating the Samples

The thickness and composition of the TbCo samples were not planed before all samples were grown. Instead, the samples thickness and composition were decided in parallel with the growth of the samples. This methodology was chosen because it was unknown in what in- terval, if it was at all possible, the TbCo moment switch from out-of-plane to in-plane. The compositions were chosen to cover both sides of the compensation point. The starting point for each composition was to grow 5 nm and 20 nm thick films. The out-of-plane hysteresis loop was measured with P-MOKE on the existing films parallel with the growth of films. The results were used to find the thickness interval where the out-of-plane anisotropy disappeared.

A new sample in the middle of that interval was then grown. The described process was re-

peated until the thickness interval was sufficiently decreased. This process was done for both

single films of Tb

Co

and for Tb

Co

on Sm

Co

(heterostructure). Table 3.1 shows

which films were created for both single films and heterostructure.

Table 3.1: The thickness (T) and composition (C) of the samples that were grown. Table a) shows parameters for the Tb

Co

-Sm

Co

heterostructure and table b) for the single films of Tb

Co

. The thickness listed in nm and the composition in atomic

% of Tb.

a) Heterostructure

C [%] 16 18 20 22 24 T [nm]

5 x x x x x

7 x x

8 x x x x x

9 x x x

10 x x x x x

12 x

15 x

20 x x x x x

b) Single Films

C [%] 16 18 20 22 24 T [nm]

2 x x x x x

3 x x x

5 x x x x x

10 x x x

20 x x

3.3 Measurement set-ups

The different measurements methods and set-ups for analyzing the structure and composition are described below. If nothing else is written was the measurement preformed by the author.

3.3.1 Grazing incidence X-ray diffraction

Grazing incidence X-ray diffraction (GIXRD) was measured on a few films to investigate the film’s structures, namely wherever the films were crystalline or amorphous. The films were measured with Cu K

X-rays at the Philips X-ray diffractometer, which is shown in figure 3.3.

The system had a 1/8 mm slit for the source, a 0.2 mm slit in front of the detector and a nickel filter. The incident angle α was set to the angle of the first interferes peak in the XRR data, which corresponds to the greatest layer distance that reflected X-rays. α were around 0.67

degrees for the films measured.

The peaks in the spectrum can be described by Bragg’s law:

2d sin (

) =

(3.1)

2θ is the angle between the incoming and outgoing X-rays, d is the distance between the atoms electron density and λ is the wavelength of X-rays. If the atoms are ordered in a crystalline structure, systematic destructive or constructive interferes can occur, and a few peaks are vis- ible. An amorphous material lacks the order for destructive or constructive interferences to occur, instead peaks deepened on the a radial distribution function describing the distance between the atoms.

Seven samples were investigated, chosen to cover both high and low composition of single films and heterostructure. The thickness of the single films were 20 nm and the compositions were Tb

Co

, Tb

Co

, and Sm

Co

. The thickness of the TbCo layer in the heterostructure was also 20 nm and the composition were Tb

Co

, Tb

Co

and Tb

Co

. An untreated substrate ((100)Si) was also measured as a reference.

3.3.2 X-ray reflectivity

Roughness and thickness of all the different layers that made up the samples were analysed

with X-ray reflectivity. The measurements was performed on all samples with D8 Discovery

Figure 3.3: The measuring system used for measured GIXRD togheter with the definition of the angle 2θ and α. A) X-ray source, B) sample holder and C) detector.

using Cu K

(1.54 Å) X-rays with 0.2 mm slits in front of the X-ray source and the detector.

The measurement was performed for 2θ between 0.2

to 8

degrees with a resolution of 0.04

degrees, and the time per point was 1 s. Approximately half of the measurements were per- formed by the author, and the residuary half was measured by the co-supervisor.

GenX: XRR Analyses Software

GenX is a program analyzing X-ray or neutron reflectivity data [42]. It starts from a model which has parameters for each layer in the sample. Example of parameters are layer thickness, layer density, and roughness. The parameters then change by the program to fit to the XRR data by using a differential evolution algorithm which is used to avoid ending up in a local minimum. An example of an model sample is shown in the figure 3.4.

Figure 3.4: The GenX model which was used for 5 nm of Tb

Co

/ SmCo.

3.3.3 Rutherford Backscattering Spectrometry

Rutherford backscattering spectrometry (RBS) was measured to calculate the compositions of the calibration samples and to adjust from any drift in composition. The RBS measurements were not performed by the author. The measurement was performed with 4He at 2000 keV and a scattering angle of 170 degrees.

The data was analyzed with the program SimNRA. From the program was the intensity for elements calculated, by integration of the element specific peaks. The intensity of an element peak, I, of a film is proportional to the relative concentration of that element in the film, x, and the differential cross section. The differential cross is proportional to the square of the atom number of the element, Z

[43]. This make it possible to calculate the relative concentration of element 1 and 2 from there intensities:

x

Z

x

Z

= ^I

I

(3.2)

3.4 Results and Discussion of the Structure and Composition

3.4.1 X-ray reflectivity

The thickness and roughness were calculated from the X-ray reflectivity data using GenX. In Figure 3.5 is an example of reflectivity data together with the fit from Genx shown.

Figure 3.5: XRR data from 20 nm Tb

Co

(blue dots) together with the fit from GenX (red line). The fit is good.

Table 3.2 displays the modelled thickness and roughness/interlayer mixing for all samples.

Table 3.2: The calculated thickness (t) and layer roughness /inter layer mixing(σ) form XRR data. All data are in Å

order Sample t

t

14 20 nm Tb

Co

on SmCo 203 200 12 20 9 8 15

16 5 nm Tb

Co

/ SmCo 197 48 9 18 12 5 13

18 20 nm Tb

Co

- 196 - 13 6.6 4 8

19 5nmTb

Co

- 48 - 8 5 1.2 8.6

20 20 nm Tb

Co

- 198 0 10 4 10 8.5

21 5 nmTb

Co

- 49 - 9 5 2 8.8

22 20 nm Tb

Co

/ SmCo 201 204 9 16 7.5 1 9.3

23 5 nm Tb

Co

/SmCo 202 47 16 12 5 3.5 13

24 10 nm Tb

Co

/ SmCo 193 103 12 14 7.8 5 7.6

25 10 nm Tb

Co

- 99 - 8 5 1 8.9

26 2 nm Tb

Co

- 20 - 9 7 3 12.1

27 5 nm Tb

Co

- 49 - 7.5 4 3 11

28 10 nm Tb

Co

/ SmCo 191 103 9 13 7.4 7 12

30 8 nm Tb

Co

/ SmCo 199 73 15 13 7 5 8

31 5 nm Tb

Co

/ SmCo 190 51 14 9 4.5 8 9

32 2 nm Tb

Co

- 16.6 - 9 3.7 3.6 10

33 8 nm Tb

Co

/ SmCo 195 79 10 12 6.6 4.4 8

35 10 nm Tb

Co

/ SmCo 188 103 11 11 4.2 0 9.4

36 2 nm Tb

Co

- 34 - 7 3 2.1 14

37 12 nm Tb

Co

/ SmCo 190 121 9 11 4 4.8 9.5

38 5 nm Tb

Co

/ SmCo 191 50 15 9 4.2 7.9 9

39 5 nm Tb

Co

- 45 - 6 4.5 8.8 8.9

40 2 nm Tb

Co

- 24 - 7 5 3.1 20.5

41 20 nmTb

Co

/ SmCo 198 201 29 21 7.6 5.7 11.9

42 3 nm Tb

Co

- 33 - 7 3.8 0.7 1

43 20 nmTb

Co

/ SmCo 205 199 14 19 8.7 8.9 13

44 7 nm Tb

Co

/ SmCo 189 71 11 10 4.5 3.8 10

46 9 nm Tb

Co

/ SmCo 193 88 8 13 8.2 5.3 8.1

47 9 nm Tb

Co

/ SmCo 193 85 9 11 4.5 2.3 9.4

48 5 nm Tb

Co

/ SmCo 187 52 13 10 4 7.9 9.4

49 5 nm Tb

Co

- 47 - 8 4.3 4.2 8.8

50 10 nm Tb

Co

/ SmCo 190 98 9 11 6.3 2.4 9.8

53 15 nm Tb

Co

/ SmCo 189 150 11 13 6.4 2.5 9.3

54 2 nm Tb

Co

- 16 - 9 3.2 3.4 9.7

55 10 nm Tb

Co

/ SmCo 191 102 9 13 6.5 5 8

56 8 nm Tb

Co

/ SmCo 191 74 13 10 5 4.9 7.3

57 20 nm Tb

Co

/ SmCo 195 199 24 17 9.5 7 12.4

58 8 nm Tb

Co

/ SmCo 188 104 11 11 4.2 0 9.4

61 7 nm Tb

Co

/ SmCo 190 67 14 14 0 2.8 10.7

62 8 nm Tb

Co

/ SmCo 193 74 8 11 7 7.5 7.8

63 3 nm Tb

Co

- 26 - 7 10 2.7 7.6

64 3 nm Tb

Co

- 25 - 10 0 1.5 18.9

65 10 nmTb

Co

- 93 - 9 8.1 3.1 8.8

66 9 nm Tb

Co

/ SmCo 193 87 5 8 5.5 5 13.5

67 10 nm Tb

Co

- 95 - 9 0 4.1 7.6

68 20 nm SmCo 195 - 12 0 - 13 3

Table 3.2 shows that the samples had a high layer roughness/interlayer mixing. The rough- ness/interlayer mixing probably comes from an uneven oxide layer that had grown on top of the Si wafer. Baking the Si wafer before growth would both help with creating a smoother oxide layer and get rid of any water which was absorbed into the surface.

The thickness of the SmCo layer was 194(5) Å which was slightly thinner than expected.

The thickness is frequently thicker for the early samples and thinner for the last samples, but the difference is not statistically significant. The thickness is obtained by a fitting procedure using the software GenX. The fitting could give an incorrect thickness if the model is wrong or if the XRR-measurement was performed incorrectly. However, the same model was used for all GenX fits, making it an unlikely source of the drift in film thickness.

3.4.2 Grazing incidence X-ray diffraction

Grazing incidence X-ray diffraction (GIXRD) measurements are shown for the single films and heterostructure in figure 3.6 and 3.7. The incident angle for all measurement was around 0.4

degrees.

Figure 3.6: The GIXRD measurement of 20 nm Tb

Co

, Tb

Co

and Sm

Co

. The small

peak around 54

degree is caused by the substrate. The spectrum are displaced

from each other in y-direction with steps of 1 [a.u.]

Figure 3.7: The GIXRD measurements of 20 nm Tb

Co

, Tb

Co

, and Tb

Co

on Sm

Co

. The small peak around 54

degree is caused by the substrate. The spec- trum are displaced from each other in y-direction with steps of 1 [a.u.]

Figure 3.8 depicts the GIXRD measurement of the (100)Si substrate. The substrate had a peak around 54

degree corresponding to SiO

(210) plane.

Figure 3.8: The GIXRD measurement of (100)Si substrate with an natural oxide layer

The results from the grazing incidence X-ray diffraction showed that the sample was amor- phous because of the lack of crystalline peaks from the film and a broad peak around 37

de- gree. The peak around 37

degree indicates that the X-rays penetrated the sample and the width of it shows that it comes from an amorphous structure. However, a secondary peak detected around 54

degrees for some samples. The peak correspond to the SiO

(210) peak from the substrate, see figure 3.8. This is caused by a too high incident angle. That only one sharp peak was visible in the substrate spectrum suggest that the substrates was well oriented.

However, the signal to noise ration was low, hence, other peaks could have gone unnoticed.

3.4.3 Rutherford Backscattering Spectrometry

An example of a Rutherford backscattering spectrometry (RBS) spectrum where all the peaks are explained for the calibration sample of Tb

Co

can be seen in the figure 3.9.

Figure 3.9: The RBS spectrum for the calibration sample of Tb

Co

with labeling of all the peaks.

The calculated compositions from the RBS are shown in table 3.3. The compositions are calculated from the spectrum using equation 3.2.

Table 3.3: The calculated compositions form the RBS measurements

Fabrication order Expected sample composition Real composition of RE from RBS [atomic %]

8 Sm

Co

16.5(3)

68 Sm

Co

17.9(3)

13 Tb

Co

26.4(6)

45 Tb

Co

24.4(2)

17 Tb

Co

21.7(4)

12 Tb

Co

19.4(4)

29 Tb

Co

17.4(3)

38 Tb

Co

19.1(5)

65 Tb

Co

18.7(5)

The result from the RBS measurement in table 3.3 shows that the difference in the nominal

and real compositions of the films was 1.2 atomic % on average. The first samples had compo-

sitions with around 2 atomic % more RE compared to the nominal values. The samples done

later has compositions that are much closer to the nominal composition, except for sample 245

had 1.6 atomic % less Tb than predicted. There is usually a slight increase in deposition rate

with time because the target get deeper "race tracks". However, the results point towards a de-

crease in deposition rate for the Co target. That it is the deposition rate of the Co that decrease

rather than the Tb and Sm target that increase comes from that the current and voltage of the

Co-target started changing around sample 230. However, the results could also come from a

high fluctuation in composition or from high uncertainty in the measurement which were not

included in the calculated standard deviation. Different theory’s for the source of the error was investigated by trying to correct the data based on the theory. All attempt were unsuccessful.

Hence, for the rest of the report the nominal composition will be used.

Chapter 4

Magnetic Properties of TbCo Films

The magnetic properties were analyzed with both longitudinal and polar magneto-optic Kerr effect measurement (L-MOKE and P-MOKE), Vibrating Sample Magnetometer (VSM) and the magnetic domain structure was analyzed with Magnetic force microscopy (MFM), and Kerr microscopy. The measurement set-ups and the results are described in this chapter.

4.1 Measurement set-ups

4.1.1 Polar Magneto-Optic Kerr Effect Measurement

The out-of-plane hysteresis loops were measured on all samples with P-MOKE. About half of the measurements were performed by the author, and the residuary half was measured by the co-supervisor. This division was made because the measurement was performed alongside the growth of the sample. From the measurements was the coercivity obtained together with the shape of the curves. The hysteresis curve shape is related to the switching behavior of the films.

The P-MOKE setup was designed in a way that the laser hit the sample surface parallel

to the samples normal vector, see figure 4.1. The set-up is purely sensitive to the out-of-plane

magnetic moment, because of the incident angle of the laser. About one third of the measure-

ments were performed with a 660 nm red laser and the residuary was measured with a 787 nm

infra-red laser. A few samples were measured with both lasers to make sure that there was no

difference between the results from the two lasers.

Figure 4.1: The measurment setup for PMOKE. A) laser, B) optics for the laser C) electro mag- net and sample holder

4.1.2 Longitudinal Magneto-Optic Kerr Effect Measurement

The L-MOKE measurement was used to measure the in-plane hysteresis loops, both in the di- rection of the growth field and perpendicular to it, for all samples.

The setup, shown in figure 4.2, consisted of a 660 nm laser with an incident angle of 45 de-

grees to the sample surface. The 45-degree incident angle gave both in-plane and out-of-plane

components in the hysteresis loop. One hysteresis loop consisted of on average 5-10 measure-

ments which each took 100 seconds. The hysteresis loop was normalized so the maximum

intensity was at one and minimum intensity at minus one. So the intensity in the figures can

not be related back to size of the Kerr rotation, and should not be compared between different

figures.

Figure 4.2: The measurment setup for LMOKE. A)laser, B) first polarizer C) lenses to focus the light, D) sample holder and magnet, E) secound polarizer, F) detector

Calculation of Coercivity

The coercive field is usually calculated as the interception of the hysteresis curve and the x- axis. However, the normalization made on the data, occasional drifts in the data and noise made this method unreliable. An alternative way of calculating coercivity was used based on finding where the large change of magnetic moment happens, which should correspond to the moment flip in the material, by looking at the derivative. This method is in some sense more unreliable then just looking at the interception. It is possible to write the matlab code using the derivative even with noise in the data. All results were checked afterwards to make sure that the right coercivity was found. The method that was used is described below.

The switch from one moment direction to the other usually happened over 5 measurement

points. To find the middle of this change, the derivative of M(H) was calculated and the two

points with the highest derivative was assumed to correspond to the coercivity. To not get the

wrong point, the data was first pre-smoothed and a numerical derivative was calculated with

a 5 points method. Both these methods made it easier to find the point in the data where the

switch actually happened. The smoothing algorithm used a local regression to a first-degree

polynomial using weighted squares.

4.1.3 Vibrating Sample Magnetometer

VSM measurement was done on Lakeshore (model 7404 with resolution of 0.1 µemu (in the mode used)) to measure the magnetic moments of each single film composition. The thickest measurable film of all single film compositions was chosen. This was done to minimize the interface effects on the magnetic moment and to get a high signal-to-noise ratio.

The VSM was calibrated with a multipoint calibration. After the calibration, the sample was attached to the sample holder with duplex postesque tape. To center the sample between the four measuring coils, the sample was first magnetized with the build-in magnets. The rest of the centering was done under zero applied field. This was done to minimize the diamagnetic contribution from the substrate and the sample holder. However, the centering was still dif- ficult because of the films low magnetic moment, and it was not possible to find the halfway distance between the coils by measuring the magnetic moment from the sample. Thus, the halfway distance was determined with a ruler and kept constant for all samples.

The samples were measured for the range 1.8 T to - 1.8 T in steps of 0.04 T. Each point was measured for 10 seconds in point by point mode. The background was also measured, for each sample, by replacing the sample with a substrate of similar size and measuring again with the same centering. The background curve was subtracted from the sample curve to get the mo- ment of the thin film. An example of the background curve, the measured curve, and end result is shown in figure 4.3. The backgound also had a ferromagnetic contribution, probably from the magnets iron core or form some form of dirt on the sample holder.

The volume of the films was needed to calculate the magnetization of the films. The sample was weighted and the area of the film was calculated using the thickness and density of the substrate. This method assumed that the substrate edge was straight which was not always the case. However, the error of the area was still small in comparison to the uncertainty of the magnetic moment in the thinnest films. Other ways of calculated the area of the sample are based on taking an image of the surface and calculating the pixels.

(a) Raw VSM measurements.

(b) The final data, after the subtracting the back- ground from the figure

Figure 4.3: a) The VSM measurements of 20nm Sm

Co

and the background. b) The VSM

measurement after the background curve has been subtracted and the moment nor-

malized with the weight.

4.2 Result and Discussion of the Magnetic Properties

4.2.1 Magneto-Optic Kerr Effect Measurement

The results form the Magneto-Optic Kerr Effect (MOKE) measurements are summarized be- low. All the hysteresis loops from the MOKE measurements are shown in the appendix.

Shape of hysteresis loops composition dependences

Figure 4.4 is examples of how the hysteresis curve can look, both below and above the compen- sation point. Below the compensation point, the hysteresis has a background with a positive slope, and above the compensation point does the background have a negative slope. Note that all MOKE data have been flipped (if necessary) so the moment flip for positive magnetic field give increase in intensity.

(a) Below compensation point. (b) Above compensation point.

Figure 4.4: An example of a P-MOKE hysteresis curves below the compensation point (figure a) and above the compensation point (figure b). a) is a measurement of 10 nm Tb

Co

on Sm

Co

and b) is a measurement of 10 nm Tb

Co

on Sm

Co

The out-of-plane hysteresis loop for the heterostructure changed shape depending on which side of the compensations point the composition is. One possible explanation is that these changes can be attributed to a change in Kerr rotation of the Tb

Co

layer. If the Kerr rota- tion of Tb

Co

has different sign depending on which side of the compensations point the Tb

Co

composition is, would explain the curves shapes. Figure 4.5 shows how the differ- ence in Kerr rotations of the Tb

Co

layer together with the unchanged Kerr rotation of the Sm

Co

film gives the different shapes of the hysteresis loops.

A possible explanation for the change in Kerr rotation is that the Kerr rotation is dominated

by one atom type. It is not necessary domination caused by its magnetic moment but because

of its electron structure is more Kerr rotation active. Hence, the direction of dominating atom

type’s moment will decide the direction of the Kerr rotation and not the magnetization of the

sample. Because TbCo is a ferrimagnetic material, which atom type’s moment that is aligned

with the magnetization will flip when crossing the compensation-point, thus, flipping the sign

of the Kerr rotation. That all figures shows this behaviour indicates that the laser penetrate

both magnetic films for all thickness.

Figure 4.5: A sketch over the possible explaination of the magneto-optical response from the Tb

Co

and the Sm

Co

layer.

Shape of hysteresis loops thickness dependences

There were three different shapes of the out-of-plane hysteresis loops for Tb

Co

on 20 nm

Sm

Co

. The thick films had rhombic loops, films close to where the out-of-plane anisotropy

dispersed, shown in table 4.4, had a small horizontal step and below that thickness the out-of-

plane anisotropy disappeared. An example of all three hysteresis-loops shapes are shown in

figure 4.6.

Figure 4.6: P-MOKE hysteresis curves of 5, 10 and 20 nm Tb

Co

on 20 nm Sm

Co

. The hysteresis loops are all normalized to ± 1.

The shape of the out-of-plane hysteresis loop for the heterostructures depend on the thick- ness of the Tb

Co

layer. The shape of the hysteresis loop suggest that the anisotropy of TbCo goes from being out-of-plane, for thick films, to in-plane for thin films. There where three forms of the hysteresis loops for 5, 10 and 20 nm Tb

Co

on Sm

Co

(figure 4.6).

For thick layer of Tb

Co

the moment of the two films behave independently from each other, except that the moment of the two films might be tilted towards each other. In the tran- sition thickness, where the Tb

Co

moment goes from being out-of-plane to in-plane, a step shown up at for low out-of-plane moments in the hysteresis loop. A suggestion for the direc- tion of the films magnetic moments is shown in figure 4.7. For even thinner films, the moment of Tb

Co

film followed the Sm

Co

magnetic moment.

The transition from out-of-plane to in-plane direction of the Tb

Co

layer comes from that it is more energetically favourable for the Tb

Co

moment to align along the Sm

Co

moment. One possible explanation for this is that the out-of-plane anisotropy of the Tb

Co

decreases with the thickness. This might come from thin that films has less material to build

up the structure that give its out-of-plane anisotropy to begin with. Another, possible reason

is that the relative size of the Sm

Co

magnetization compare with the Tb

Co

magnetiza-

tion increases with thinner films and this make it possible for the Sm

Co

pull the Tb

Co

moments in-plane.

Figure 4.7: An possible explanetion to the appearance of a step in P-MOKE data for the Tb

Co

on Sm

Co

.

Figure 4.8 shows hysteresis loops measured with L-MOKE in the direction of the growth field and perpendicular to it for the thickness 5 and 12 nm of Tb

Co

on 20 nm Sm

Co

. 5 nm is below the critical thickness, the thickness where the SmCo and TbCo behave as one film, and 12 nm is above.

(a) The L-moke hysteresis loop of 12 nm Tb₂₄Co₇₆ on 20 nm Sm₁₅Co₈₅

(b) The L-moke hysteresis loop of 5 nm Tb₂₄Co₇₆on 20 nm Sm₁₅Co₈₅

Figure 4.8: The L-MOKE hysteresis loops measured both perpendicular and in the direction of

the growth field. Films shown are 5 and 12 nm of Tb

Co

on 20 nm Sm

Co

from that L-MOKE measures both the in-plane and out-of-plane component but mostly the out-of-plane. The SmCo magnetic moment is assumed to be in-plane and the TbCo moment slightly tilted out of plane. The more the TbCo moment is tilted in-plane by the applied mag- netics field the smaller is the out-of-plane component will be which leads to a smaller kerr rotation. This is seen as the increase of intensity in the plot. For the figure 4.8 (b) it is the two layers switching together which give it the square shape. Figure 4.9 shows how the moment of the two layers are assumed to be directed.

(a) Above critical thickness. (b) Below critical thickness.

Figure 4.9: The assumed configuration of the magnetic moment for the TbCo and SmCo which gives the L-MOKE hysteresis curves for films above (a) and below (b) the critical thickness for the magnetic moment of the TbCo layer to stay in-plane.

Figure 4.10 shows the L-moke measurement for 2, 3 and 20 nm of Tb

Co

. The hystere-

sis loops from the L-moke measurement has both out-of-plane and in-plane components. The

mixing of out-of-plane and in-plane components makes the analysis more difficult. However,

figure 4.10 and table 4.3 shows that for thin enough films of Tb

Co

it is possible to have

in-plane anisotropy. As shown was the in-plane anisotropy was also observed for the het-

erostructure, however, it was because the Sm

Co

layer effected the Tb

Co

layer. The

mechanism for the single films must therefore be different. All single film that had in-plane

anisotropy also had out-of-plane anisotropy. One possible explanation for this is that films are

so thin and the surface so uneven that it might be clusters of different compositions that are

responsible for the in-plane components.