A range of different setups and techniques are used to characterize the samples. To study the magnetic properties of the sample, the Magneto-optical Kerr Effect is used. The Magneto-optical Kerr Effect is utilized by the laser MOKE setup and the optical Kerr microscope and will be dis-cussed in the next section. They are mainly used to quantify the OOP magnetic properties. To quantify the IP magnetic properties, for which higher magnetic fields are necessary, the SQUID-VSM is used and explained in section 4.3.2. All current related experiments on the Hall crosses make use of the anomalous Hall effect (AHE), which is a measure for the OOP component of the magnetization. This effect is explained together with a brief overview of the setup in section 4.3.3.
4.3.1 Magneto-optical Kerr Effect
The Magneto-Optical Kerr Effect is widely used to characterize the magnetic properties of a sam-ple by looking at the reflection of light shine on the samsam-ple surface [43]. It can be used very locally, in the form of the laser MOKE, or with a wide field-of-view by using Kerr microscopy.
Sample
x x
y y
Figure 4.5: Principle of the Kerr effect. Linearly polarized light, consisting of right and left handed circularly polarized light, is incident on the sample. The two circular modes will be absorbed and reflected differently, depending on the magnetization direction of the sample. This will result in rotation and ellipticity changes in the reflected light.
Linearly polarized light can be seen as a superposition of right-handed and left-handed circularly polarized light. If this light is incident on a magnetic surface, the two circular modes will be absorbed and reflected differently depending on the magnetization direction of the sample. This will result in a change of rotation and ellipticity of the reflected light. The effect is schematically shown in figure 4.5. The origin of this effect can be found in the permittivity of a material that varies with the magnetic orientation. The permittivity effects the speed of light in the material.
Hence, the speed of light varies depending on the magnetization direction and causes fluctuations in the phase of the polarized incident light, resulting in a change of rotation and ellipticity of the reflected light.
The Kerr effect can be used in three different configurations, corresponding to the direction of the magnetization. The polar configuration is used for out-of-plane magnetization, longitudinal for magnetization along the plane of incidence and transverse for magnetization perpendicular
to the plane of incidence. They are shown in figure 4.6. A more fundamental description of the Kerr effect, involving the spin-orbit coupling in the material, can be found in [44].
Polar Longitudinal Transverse
a. b. c.
Figure 4.6: Three different magnetization modes in which a MOKE can be used. In each case the polarization of the light changes after reflecting on the sample surface.
Laser MOKE
The laser MOKE setup uses a 632.8 nm HeNe laser with a spot size of ≈ 300µ m. The laser light is sent through a linear polarizer and photo-elastic modulator (PEM) onto the sample. The PEM slightly changes the optical path for one polarization component at a certain frequency. After reflecting on the surface of the sample, the light travels through a analyzer before being collected by a photo diode. Using the PEM as a reference, a lock-in technique is used to detect the change in rotation or ellipticty of the light to measure a change in magnetization. The magnetization can be changed by sweeping a magnetic field between -370 mT and +370 mT. The setup can be used in either polar or longitudinal configuration. A step motor is used to scan wedge samples and determine the hysteresis curves at each position.
It should be noted that the Faraday effect, caused by transmission of the light through an optical element of the MOKE setup, can also influence the polarization. This effect depends linearly on the applied magnetic field and can therefore easily be subtracted from the measurement. All results shown in chapter 5 are corrected for linear background.
Optical Kerr microscope
Another way of utilizing the Kerr effect is by using a Kerr microscope. Instead of a laser, a halogen lamp is used in combination with a conventional microscope. The light is again polarized before incident on the sample. After reflection it is analyzed and captured by either a camera for digital imaging or directly by the naked eye. A sample image captured by the Kerr microscope is shown in figure 4.7. Lighter and darker regions indicate differences in the magnetization. The main advantage of a Kerr microscope over the laser MOKE setup is the wide field-of-view compared to the single spot laser. However, sensitivity and contrast are generally much lower in this case. The resolution for the Kerr microscope is limited to the optical diffraction limit. Magnetic domains down to a width of approximately 500 nm can be visualized, which is in the same order of magnitude as the typical size of the micro-structured samples.
In this project, an Evico Kerr microscope with a built-in CCD (charge-coupled device) camera is used to visualize the magnetic sample. The contrast in images can be enhanced dramatically software adjustment. Several different modules can be connected to the microscope to make it
a. b.
10 μm
Figure 4.7: Kerr microscope images of a Hall cross. The difference between the two figures in intensity of the horizontal bar of the Hall cross in figure is the result of a different magnetization direction, which changes the polarization of the reflected light.
a very versatile setup. A three-dimensional magnet around the microscope is used to apply a magnetic field in any direction of up to 60 mT to the sample.
4.3.2 SQUID-VSM
The SQUID-VSM is a combination of two different setups. A SQUID (super-conducting quantum interference device) and a VSM (vibrating sample magnetometer). The VSM is based on Faraday’s law:
Vind= N∆ΦB
∆t , (4.1)
which states that a voltage Vindis induced by a changing magnetic flux ∆Φ∆tB through a coil with N windings. The sample is vibrated between two pick-up coils. By using a lock-in technique with the vibration frequency as a reference, the magnetic moment of a sample can be detected quantitatively by the pick-up coils. As it only detects the change in magnetic flux by the movement of the sample, any external magnetic field is filtered out. As the magnetic layer studied are very thin, the absolute magnetic moment is very small. Using a SQUID technique, extremely sensitive detection of the magnetic moment of a sample is possible. The SQUID is based on superconducting coils containing Josephson junctions that can measure individual flux quanta [45]. An external magnetic field of up to 7 T can be applied in the VSM-SQUID to measure hysteresis curves of the sample. An oven module can be used for heating the sample, as already discussed in section 4.2.
4.3.3 Anomalous Hall effect
A substantial amount of the experiments in this thesis are done on the micro-sized Hall crosses which have been introduced in section 4.1.4. Optical techniques lacked the resolution to
success-fully determine the magnetization in such small structures. As an alternative way to measure the magnetization, the anomalous Hall effect is used.
The reader may be familiar with the ordinary Hall effect (OHE, or simply: Hall effect) that causes a voltage difference VHacross a conductor transverse to an electric current if a magnetic field Hz is applied perpendicular to it, see figure 4.8a [46]. The AHE is an extra contribution to the Hall voltage that occurs in ferromagnetic materials [47]. It causes a voltage difference across an F-layer transverse to an electric current and perpendicular magnetization Mz, shown in figure 4.8b. The AHE can be much stronger than the OHE in ferromagnetic materials. The total extra resistivity ρH of a material due to the OHE and AHE is given by:
ρH= cOHEµ0Hz+ cAHEµ0Mz, (4.2) with cOHEand cAHEthe Hall coefficients. In the experiment, there is no external Hzwhen measur-ing the VH. Therefore, a change in VHis directly related to a change in Mzand only consists of the contribution from the AHE. In the results, instead of the Hall voltage, the change in resistance is usually presented which is equal to RAHE= VIH
x, with Ix the probe current. As the Hall coefficient cAHE is unknown, only a relative change of the magnetization can be measured by monitoring VH.
Figure 4.8: Schematic overview of (a) the ordinary Hall effect and (b) the anoma-lous Hall effect . Both generate a voltage transverse to the applied current direction.
The origin of the AHE is strongly related to the spin Hall effect introduced in chapter 3 as it also depends on spin-dependent scattering processes and intrinsic contributions originating from the Berry-phase. For a detailed description of the AHE, see [48].
4.3.4 Anomalous Hall effect setup
This section discusses the AHE setup that is used for all current related experiments. The AHE setup is basically the Kerr microscope setup with the sample placed in a chip carrier, which allows for the connection of the Hall cross to other equipment. Figure 4.9 schematically shows the connections for the Hall cross. Via a pulse generator connected to one bar of the Hall cross, a current pulse can be sent through the sample. By measuring the voltage drop over a resistance placed in series with the current line, the current pulse magnitude can be measured. The other bar of the Hall cross is used to detect the anomalous Hall voltage by connecting it to a Volt meter. A small probe current of 10−4 A is used during the AHE measurement. The voltage can be converted to the anomalous Hall resistance by dividing through the probe current magnitude.
It is related to the magnetization of solely the center of the Hall cross through which the current flows (the red region in the figure).
v v
R
AHEI
pulseI
pulseR
extFigure 4.9: Schematic overview of the anomalous Hall effect setup. The voltage is monitored over one of the bars of the Hall cross to determine the anomalous Hall resistance. A pulse generator is connected to the other bar of the Hall cross in series with a resistance. The voltage over the resistance is monitored to determine the current magnitude of the pulse. A three-dimensional magnet is used to apply a field in any direction (not shown in the figure). The whole setup is remotely controlled by LabVIEW software to conduct the desired measurements.
The Hall cross is surrounded by a three-dimensional magnet capable of applying a field in any direction of up to 60 mT. An Agilent 33250A pulse generator can apply (a sequence of) pulses as short as 10 ns with a maximum amplitude of 10 V as well as a bias voltage which is used as the probe current to measure VH. The AHE voltage is monitored by an Agilent 34410A Volt meter. The voltage over the external resistance (to measure the pulse current) is measured by an analog-to-digital converter with a sample rate of 200 kHz. This limits the pulse duration to about 50 µs to be able to detect the magnitude of it.
4.3.5 Measurement sequence
All equipment is actuated via a custom-made LabVIEW program. The program is very flexible and designed to measure the AHE after various applied current or field sweeps. A basic measurement sequence it depicted here:
1. Initialize all equipment
2. Make a list of all desired measurement points (a current magnitude or field magnitude sweep)
3. Set bias voltage (for probe current to measure VH) 4. Set desired field magnitude and/or pulse magnitude 5. Apply pulse
6. (Together with 5) Measure current pulse magnitude 7. Measure resulting VH (with probe current)
8. Repeat 4-7 for desired currents/fields 9. Plot & save data points
10. Close all equipment
At each point, the passed time, magnetic fields, pulse voltage, pulse current and AHE voltage are saved to a file. Other basic parameters used are a bias voltage of 0.1 V, pulse duration of 50 µs, pulse rise time of 500 ns. If any field is applied, a 0.5 s saturation delay is built-in after setting or removing the applied field.
This measurement scheme and parameters are used in all measurements that are shown in chap-ter 6, unless stated otherwise.
Creating an orthogonal exchange bias
This chapter describes the results on creating an orthogonal exchange bias in a F/AF system. In such a system, the spin orientation of the ferromagnetic layer is orthogonal to the antiferromag-netic spin configuration. Following an introduction to material choices and sample properties, a successful demonstration of an orthogonal exchange biased sample is presented in section 5.2.
A study on the influence of the various layer thicknesses on the sample properties is given in section 5.3. To get an better understanding of the behavior of this orthogonal exchange bias, the stability and magnitude of the exchange bias are investigated in section 5.4 and 5.5.
5.1 Desired sample properties
The three main properties that the sample created in this chapter should exhibit are: (i) An OOP ferromagnetic layer, (ii) orthogonally coupled to an IP anti-ferromagnetic layer and (iii) a have layer for spin Hall current injection. All three are critical to make the sample suitable for the switching experiments in chapter 6. In this section, the material choices for the various layers are discussed, related to the three main properties. Figure 5.1 shows the resulting multilayer stack.
(i) OOP ferromagnetic layer
To create an OOP ferromagnetic layer, a Pt/Co bilayer is used, grown on top of a silicon substrate with a thin Ta buffer layer. A very thin layer of Co is known to have a large OOP magnetic anisotropy at room temperature when combined with Pt, due to its strong surface anisotropy caused by the spin-orbit coupling effects in the Pt. Other possibilities for the material of the F-layer have also been considered, but either it consisted of a multilayer structure which leaves more layers to optimize ([Co/Ni]x, for example) or it did not result in OOP magnetization in the sample (CoFeB did not result in any OOP magnetization in the experiments). The thin Ta buffer layer was found to enhance the growth properties and interfaces of the subsequent layers.
(ii) IP anti-ferromagnetic layer
IrMn with an atomic ratio of 1:5 is used for the AF-layer. It is grown directly on top of the Pt/Co bilayer. IrMn has large exchange bias in combination with Co [29]. Moreover, this widely studied antiferromagnet has a blocking temperature of around 450 K and a Neèl temperature of 693 K,
51
Capping
Figure 5.1: Schematic overview of the different layers in the sputter deposited sam-ples. The relative thicknesses are not to scale. In some samples, a dusting layer of Pt is added to increase the PMA and stability of the sample. The spins indicate the de-sired orthogonal configuration of the exchange bias. The y-direction and x-direction correspond to OOP and IP, respectively.
resulting in stable exchange bias at room temperature. Both IP exchange biased system and OOP exchange biased systems are frequently reported in literature.
To prevent oxidation of the AF-layer, a thin capping layer of either Ta or Pt is grown on top of the sample. For samples which are studied under the optical Kerr microscope, Ta is used. It oxidizes and thus becomes transparent, making is very suitable for optical study. TaOx is also an insulator, so it reduces current shunting through the layer. For samples in which optical or electrical properties are not important, a Pt capping layer is used. Pt is a noble metal so there is no risk of under or over-oxidizing.
In some samples a small Pt dusting layer is inserted between the Co and the IrMn layer. This is done to enhance the anisotropy in the Co, as this will add an extra surface anisotropy con-tribution. The dusting layer also influences the exchange coupling, as the interface is altered.
Therefore, one might expect a decrease in exchange bias. However, the addition of a Pt spacer layer is said to induce a better collinear alignment of the Co spins out of the plane, which can even induce an increase in exchange bias for very thin (<0.2 A) Pt layers [49]. Thicker spacers can reduce the exchange bias, presumably by weakening interfacial exchange bonds. Other dust-ing layers such as Ta have been tried, but resulted in immediate disappearance of the exchange bias.
(iii) Spin Hall current injection
The Pt layer that is necessary for the OOP anisotropy in the Co is also very suitable as spin Hall current generator. It has a relatively high spin Hall angle of 0.07 and short spin diffusion length of 1.4 nm [32]. Other materials with a high spin Hall angle such as Ta have been tried, but did not result in OOP anisotropy in the Co layer.
5.1.1 Typical hysteresis curves for as-deposited and OOP field-cooled samples A range of samples are fabricated using the materials depicted in the previous section. This section shows typical hysteresis curves of the stack for two samples with different layers:
Sample 1: Ta(3)/Pt(3)/Co(1.35)/IrMn(6)/Pt(2).
Sample 2: Ta(1)/Pt(3)/Co(0.7)/Pt(0.3)/IrMn(6)/Ta(1.5).
c.
c.
d.
e.
d. e.
b.
a.
H
extH
extOOP MOKE, as dep IP VSM, as dep
Ma g n e ti za ti o n (a .u .)
Figure 5.2: a: Easy-axis hysteresis curve for the as-deposited sample without dust-ing layer, measured by usdust-ing the polar MOKE setup. The field sweep direction is OOP, as indicated in the figure. A double loop is observed, explained by assuming domain formation after growth in the AF layer. Figures c-e correspond to different applied fields indicated in figure a and show the partial rotation of the F layer. Both loops show the same coercivity of 38 mT and an exchange bias of ±78 mT. b: Hard-axis hysteresis loop measured by the VSM-SQUID. No exchange bias or coercivity is visible along the IP field sweep direction.
Sample 1 is referred to as the sample without Pt dusting layer, sample 2 is referred to as the sample with Pt dusting layer. An in-depth layer thickness dependence study of these samples is given in section 5.3.
After material deposition, the samples are in their ‘as-deposited’ state. IP and OOP hysteresis curves of the sample without dusting layer are shown in figure 5.2ab. Clear hysteresis is seen in the OOP loop, a result of OOP anisotropy in the sample. However, instead of two magnetization levels (with the Co magnetization pointing up or down), three levels are visible in the form of a double hysteresis loop. This remarkable result can be explained by domain formation of the Co during growth. During the Co deposition, the Co forms random OOP domains in either
up or down direction to reduce its demagnetization field. Subsequently, the IrMn is deposited.
Spins couple to the Co and the random domain state is transferred to the IrMn layer. During a field sweep, the orientation of the IrMn remains fixed and impose a directional OOP anisotropy onto the Co. The orientation of the Co and IrMn during a field sweep is schematically shown in figure 5.2cde, corresponding to different parts of the hysteresis loop indicated in figure 5.2a. The domain formation of the Co can also be visualized in a Kerr microscope image, shown in figure 5.3. The image is taken just after deposition without any field applied. Domains of micrometers in size are visible, which are typical for such systems [28]. It confirms the given explanation for the double hysteresis loop. A domain structure is also visible in the hysteresis curves as they show gradual switching between states, most noticeable in the region between -80 mT and -120 mT. This indicates that different domains are switching at different applied fields.
From the double loop, a coercivity of µ0HC=38 mT, very large OOP exchange bias of µ0HEB=78 mT and a remanence of ~100% can be extracted.
5 μm
Figure 5.3: Kerr microscope image of the domain structure in the as deposited sam-ple. At zero applied field, the visible Co grains correspond to the AF grain structure
Figure 5.3: Kerr microscope image of the domain structure in the as deposited sam-ple. At zero applied field, the visible Co grains correspond to the AF grain structure