Schematic illustration of PPMS - 鎳鈀多層膜中的自旋霍爾效應與磁鄰近效應

The sample is first put on the Horizontal & Vertical Sample Rotators, and the rotators is inserted in the sample chamber. The sample chamber can be pumped to 15 torr to avoid thermal drift effect. The sample chamber is put in the vacuum flask which is equipped with liquid He to lower the temperature. Next to the sample, there is a thermal circuit to raise the temperature. The thermal circuit is controlled by the PID circuit.

Figure 3.7 the Horizontal & Vertical Sample Rotators (from PPMS Horizontal rotator manual)

The magnet current can control the field induced by the superconducting magnet coil.

The current flow into the superconducting magnet coil is controlled by a well-designed persistent switch. As Figure 3.8 shows, the left one is superconducting magnet coil, the middle is the heater, and the right one is power supply.

Figure 3.8 Schematic illustration of persistent switch

Chapter 4

In 2011, Oepon’s group constructed Pt/Co/Pt trilayer structures and discovered that when the current passed through the sample, which was parallel to the interface, the resistivity had an abnormal result when the field was rotate in the plane perpendicular to the current direction. That is the resistivity of field perpendicular to the plane (Perpendicular MR, PMR) was larger than the one with field parallel the plane (Transverse MR). This result defied the Geometric Size Effect (GSE) in the single ferromagnetic thin film. As Figure 4.1 shows, they changed the thickness of Co in the Pt/Co/Pt structure and discovered that the MR ratio increased with the increasing Co thickness when the Co thickness was thinner than 7nm. But when the Co thickness was larger than 7nm, the MR ratio decreased with the trend of 1/tco. They considered this particular magnetoresistance phenomenon to be unrelated to the inner part of Co layer.

Therefore, this phenomenon was due to the Pt/Co interface contribution. They called this phenomenon anisotropic interface magnetoresistance (AIMR). [10]

Figure 4.1 MR ratio versus the thickness of Co [10].

In the last decade, there have seen growing interest on research in spintronics such as Spin Hall Effect (SHE)[11][12], spin Seeback Effect (SSE)[14] and spin pumping Effect [13]. Take Spin Hall Effect for example, SHE can convert a charge current in a nonmagnetic metal, which has strong spin-orbit coupling (SOC), into a pure spin current in the transverse direction. Although this spin current cannot easily be detected by usual electrical measurement method, it can be detected by some pure spin detectors.

When passes through materials with strong SOC, the spin current converts into charge current resulting in charge accumulation in the transverse direction and a voltage drop, which can be detected easily. Pt is the most popular pure spin detector due to its large SOC.

Pt/YIG (yttrium iron garnet, Y3Fe5O12) bilayers have been widely investigated due to the essential roles mentioned above. In 2012, C. L. Chien’s group measured the magnetoresistance in the Pt/YIG structure. They found that Pt/YIG exhibited a new type of MR with unique characteristics that were very different from those of other known MR phenomena, as shown in Figure 4.2. Pt/YIG shows a pronounced MR (Longitudinal MR equal to Perpendicular MR, both were larger than Transverse MR).

As we know, Pt is a normal metal with no ferromagnetic characteristics, and YIG is an insulator. Therefore Pt and YIG should not show MR phenomenon individually.

Because Pt is a material which is close to the Stoner Criterion, it will acquire induced magnetization by YIG layer and causes the MR effect.

Pt is a well-known material with strong spin orbital coupling (SOC) and has usually been used as a spin current detector. In the emerging spintronic research, the relation between magnetic proximity effect (MPE) and pure spin current detector must be examined carefully.

Figure 4.2 The scan direction of the sample and the Hybrid MR [?].

E. Saitoh’s group from Japan proposed a different aspect to explain the abnormal magnetoresistance in the Pt/YIG structure. They published Spin Hall magnetoresistance theory in Physic Review Letter in 2013. In their theory, the charge current pass through the Pt layer will be converted into a pure spin current in the transverse direction. With spin-angular-momentum exchange between magnetization M in YIG and conduction-electron spin polarization in Pt, Spin-flip scattering is activated when the spin direction of spin current and M are not collinear. Some part of the spin current is then absorbed by the magnetization as a spin-transfer torque even at an interface to a magnetic insulator and the spin current reflection is suppressed. This absorption is maximized if M is perpendicular to the spin direction and zero if parallel. When the absorption is zero, the reflected spin current will be converted into charge current due to the inverse Spin Hall effect (ISHE), which enhances the conductivity. Conductivity enhancement due to SHE and ISHE is expected to be maximized (minimized) when M is in transverse (longitudinal and perpendicular) direction. The Pt film resistance is therefore affected by the magnetization direction in YIG, giving rise to the spin Hall magnetoresistance (SMR)[15].

Figure 4.3 Schematic for SMR [15].

To further investigate the physical origin of hybrid magnetoresistance, C. L. Chien’s group did two measurements on the altered YIG surface and use SiO2

(7% Fe) to replace the YIG in 2014[26]. The altered YIG surface called the YIG

BB greatly reduces the spin-mixing conductance at the Pt/YIG interface, thus, blocking the spin current transmission. In

SiO

(7% Fe) sample, the concentration of the Fe is too low to be ferromagnetic, but it still contains Fe composition so that the SiO

(7% Fe) still can simulate the MPE but without spin current.

Interesting results shown in Figure 4.4, Pt(3nm)/SiO2(7% Fe) exhibits very similar MR behavior to that of Pt/YIGBB at all fields. However, the Pt(3nm)/SiO2(7% Fe) and Pt/YIGBB show different results to that of conventional Pt/YIG at small field.

The Pt(3nm)/SiO2(7% Fe) and Pt/YIGBB show the same angular dependence as that of Pt(3nm)/YIG. All of them exhibit (cos 𝜃)² angular dependence in the ϕxy scan and θyz

scan with saturation field. Hence, they claim that in the Pt/YIG sample, the SMR influences the results at small field while the MPE dominates at high field.

Figure 4.4 the MR and AHE results of Pt(3nm)/SiO2(7% Fe) and Pt/YIGBB. [26]

Au is a material used as a spin current detector with a relatively long spin diffusion length. Besides, Au is far from the stoner criteria so that it couldn’t have magnetization induced by the YIG. Therefore, if the Au is inserted into the Pt/YIG, the MPE will diminish but the spin Hall magnetoresistance should survive. C. L. Chien’s group did the experiment as mention above. Interesting results were obtained as shown in Figure 4.5.

Figure 4.5 Field dependence of the (a) Hall resistance RH at different temperatures, (b) R∥ and RT for Pt3 nm/Au2 nm/YIG. Insert of (a) shows the temperature dependence of the anomalous Hall resistance RAHE for Pt3 nm/Au2 nm/YIG. Field dependence of the (c) Hall resistance RH at different temperatures, (d) R∥ and RT for Pt3 nm/Au6 nm/YIG. [26]

In the Pt(3nm)/Au(2nm)/YIG sample, the AHE results still show the negative AHE

signals which means that the MPE still exists in this sample. The AHE signal is

increasing with decreasing temperature which agrees with the MPE model. Although

the AHE signals are smaller than that of Pt(3nm)/YIG, the inserted Au layer is still too

thin to eliminate the MPE. As a result, the MR ratio of Pt(3nm)/Au(2nm)/YIG increases

with increasing field similar to the Pt(3nm)/YIG results. In the Pt(3nm)/Au(6nm)/YIG,

the MPE is obstructed, shows only ordinary Hall effect signals but without AHE. The

MR ratio of Pt(3nm)/Au(6nm)/YIG decreases with increasing field. This is the intrinsic

spin current related MR without the MPE.

Chapter 5

Sample parameters

The samples were deposited by sputter deposition, the sample geometry is like a key which is defined by the e-beam lithography. The samples were measured by the standard four probe measurement. We can measure the Hall voltage by using the electrode VH and Vb at transverse direction.

Figure 5.1.1 sample geometry

5.1 Resistance measurements in multilayers sample with rotating angle

We fix the magnetic field direction and change the sample direction by using the PPMS Horizontal & Vertical Sample Rotators. The field we applied was 3T which is larger than the saturation field. The rotation is from field parallel to the sample plane to field perpendicular to plane. When the field is parallel to plane, it can be divided into two direction, current parallel to field (longitudinal) and current perpendicular to field (transverse). Hence, the rotating direction can be divided into three parts, the field is rotating from transverse to perpendicular (TP rotation), the field is rotating from longitudinal to perpendicular (LP rotation), and the field is rotating from longitudinal to transverse (LT rotation) directions, as shown in the Figures below.

(a)

(b)

Figure 5.1.2 Schematic illustration of field rotation

In MR measurement, we applied the field at three directions, longitudinal, transverse and perpendicular. To exclude the thermal disturbance effect we measured the resistance at the temperature 10K.

In the Section 5.1, we fix the Pd layer thickness at 4.8nm, and change the thickness of the Ni layers. The number of bilayer is 20 in my samples. We measure the angular dependence of resistance at 10K and 300K respectively. The results are shown in Figure 5.1.3.

t=5nm t=5nm

t=1.6nm t=1.6nm

Figure 5.1.3 Angular dependence measurement of saturation resistance in the multilayers sample with fixed Pd thickness and changing Ni thickness.

To elaborate the figures above more clearly, we take the 10K measurement data and plot in polar axes. The 0 degree and 180 degree represent the field parallel and antiparallel to the current direction (Longitudinal), respectively, and the 90 degree and 270 degree represent the field perpendicular to the plane.

Figure 5.1.5 The figure of polar axes from Figure 5.1.4

As shown in Figure 5.1.5, there are interesting results. When the Ni thickness decreases from 12nm to 1.6 nm, the angular dependence of saturation resistance shows totally different behavior. When Ni thickness is 12.5nm, the AMR effect dominates because most part of the sample is Ni. Hence, the results only show the Ni contribution, the Pd contribution is too small to be observed. But when Ni thickness decreases to 5nm which is approximately equal to the Pd thickness, the resistance behavior at 10K shows bi-axial anisotropy where the maximum ρ occur when θ=45 ° between the perpendicular and longitudinal direction.

Interesting results as shown in Figure 5.1.4, the 300K measurement of the Ni 5nm sample shows only AMR behavior. When the thickness of Ni decreases to only 1.6nm, we can observe an abnormal phenomenon, that is, the resistivity with field in perpendicular direction is larger than the field in longitudinal direction. This abnormal magnetoresistance has yet been discovered. What’s worthy to talk about is that this phenomenon can be observed whether the temperature is 10K or 300K. However the phenomenon in 10K measurement is more notable. Besides, because of the different behavior in 10K and 300K of the tni=5nm sample, we do the measurement with variable temperature.

Figure 5.1.6 Resistance measurement with variable temperature in LP rotation of the sample Pd4.8nm/(Ni5nm/Pd4.8nm)20/Si.

As shown in Figure5.1.5, we measure the angular dependence of resistance with variable Ni thickness, when the Ni thickness is thick enough compared to the Pd thickness, the results show conventional AMR as we expected. With decreasing Ni thickness, the abnormal magnetoresistance phenomenon appears gradually. The MR behavior of the tNi=5nm sample show bi-axial anisotropy. These results can prove that the AMR and our new MR exist concurrently. Whether the AMR or our new MR dominates depends on the thickness of Ni. Besides, as shown in Figure 5.1.6, we can observe the new MR phenomenon increases with decreasing temperature while the AMR is almost the same with decreasing temperature. In the measurement of rotating angle, some data show the difference of the resistance due to the thermal drift.

5.2 Measurements with variable Pd thickness

We can observe abnormal magnetoresistance in the tNi=1.6nm sample whether the temperature is 10K or 300K except the measurement at 300k of the Pd2.4nm/(Ni1.6nm/Pd2.4nm)20/Si . Hence, we focused on the tNi=1.6nm sample with variable Pd thickness.

Figure 5.2.1 Resistance measurement in LP rotation and the MR results of the Pd2.4nm/(Ni1.6nm/Pd2.4nm)20/Si.

Figure 5.2.2Resistance measurement in LP rotation and the MR results at 10K and 300K of the Pd3.6nm/(Ni1.6nm/Pd3.6nm)20/Si.

Figure 5.2.3 Resistance measurement in LP rotation and the MR results at 10K and 300K of the Pd4.8nm/(Ni1.6nm/Pd4.8nm)20/Si.

Figure 5.2.4 Resistance measurement in LP rotation and the MR results at 10K and 300K of the Pd6nm/(Ni1.6nm/Pd6nm)20/Si.

Figure 5.2.5 Resistance measurement in LP rotation and the MR results at 10K and 300K of the Pd7.2nm/(Ni1.6nm/Pd7.2nm)20/Si.

Figure 5.2.6 Resistance measurement in LP rotation and the MR results at 10K of the Pd8.4nm/(Ni1.6nm/Pd8.4nm)20/Si.

Figure 5.2.7 ∆ρ/ρ versus the thickness of Pd , the black line P-L means that the perpendicular resistance (PMR) minus longitudinal resistance (LMR) and then divided by the perpendicular resistance (PMR), same method were used in red line and black line.

As shown in Figures 5.2.1, to Figure 5.2.3, we measure the MR at 300K with variable Pd thickness. The results show that the resistance decreases with increasing magnetic field. This result is due to the decreasing of the spin disorder scattering with increasing field and some thermal disturbance. For MR measurement at 10K, as shown in Figures 5.2.1 to 5.2.6, the resistance is almost constant when the field is large enough. Hence we focus on the 10K measurement. In the data shown above, we can observe the abnormal magnetoresistance in all sample whether the temperature is 10K or 300K except the measurement of the Pd2.4nm/(Ni1.6nm/Pd2.4nm)20/Si sample at 300K. This transition of the MR results between 10K and 300K of the Pd2.4nm/(Ni1.6nm/Pd2.4nm)20/Si sample may due to the Pd have the induced magnetization and the Curie temperature of this induced magnetization for 2.4 nm Pd is between 10 and 300 K.

In this series of sample, we can see the new MR phenomenon increases with decreasing temperature. I consider this phenomenon is due to the increase of the spin diffusion length with decreasing temperature. Detailed discussion will be given in Section 5.6. Besides, we can observe the giant magnetoresistance (GMR) occur when the Pd thickness is thick enough. But whether the GMR occur, the new MR phenomenon still exists.

In Figure 5.2.7, we show that the new MR phenomenon is prominent when the Pd thickness is 6 to 8.4 nm. Hence, we consider this new MR is due to not only the interlayer contribution but also the bulk contribution from Pd.

5.3 Resistance measurement with varying temperature and rotating field

Figure 5.3.1 Resistance measurement in LP rotation with variable temperature of the Pd2.4nm/(Ni1.6nm/Pd2.4nm)20/Si sample.

Figure 5.3.2 Resistance measurement in LP rotation with variable temperature of the Pd3.6nm/(Ni1.6nm/Pd3.6nm)20/Si sample.

Figure 5.3.3 Resistance measurement in TP rotation with variable temperature of the Pd3.6nm/(Ni1.6nm/Pd3.6nm)20/Si sample.

Figure 5.3.4 The ∆ρ/ρ of the LP and TP rotation versus temperature. Red line means that perpendicular resistance (PMR) minus longitudinal resistance (LMR) and then divided by the perpendicular resistance (PMR). Black means that perpendicular resistance (PMR) minus transverse resistance (TMR) and then divided by the

perpendicular resistance(PMR).

Figure 5.3.5 Color coded ∆ρ/ρ in LP rotation of Pd2.4nm/ (𝑁𝑖1.6𝑛𝑚/

𝑃𝑑2.4𝑛𝑚)₂₀/𝑆𝑖 sample.

圖 5.3.6 Color coded ∆ρ/ρin LP rotation of Pd3.6nm/(𝑁𝑖1.6𝑛𝑚/𝑃𝑑3.6𝑛𝑚₎₂₀/𝑆𝑖 sample.

5.4 Canting angle measurement

Figure 5.4.1 Schematic illustration of canting angle determination.[27]

The canting angle means that when there is no external field acting on the magnetization, the magnetization will lie on the energy minimum state which has an angle to the perpendicular axis, as shown in Figure 5.4.1. When the external field increases, the magnetization will be aligned from the easy axis to the applied field direction. If the external field is high enough to maintain a single-domain state but not sufficient to completely align the magnetization in field direction, the magnetization will lie at the angle between canting angle and the applied field direction. As we know, the resistance of the ferromagnetic materials depends on the magnetization direction, hence, there is a method that allows a very accurate and unambiguous determination of the canting angle via magnetoresistance measurements indirectly.

Figure 5.4.2 Magnetic anisotropy versus angle [27].

H. P. Oepen’s group presented in the Journal of Applied Physics in 2013[27] a method that allow us to measure the canting angle via electrical measurement.

The free-energy density for a system in an external field can be approximated by the anisotropy constants,K1,eff and K2 and the Zeeman energy. The equation can be described as below

From Figure 5.4.2, when the field strength along the normal is reduced, the magnetization cannot overcome the high energy barrier at 0 degree and 90 degree, hence, the field aligned cone is populated. In order to determine the canting angle of the strongly canted sample, magnetotransport measurements are proposed. If the applied field is strong enough to saturate the magnetization, the magnetization will be aligned to the field direction, if not, with varying field magnitudes, reducing the field strength from the saturation, the field aligned cone angle will move toward the canting

angle gradually. Therefore, in the magnetotransport measurements, the resistance will be changed by the field strength. If we measure the resistance with rotating field angle, there will be some difference between the resistance with saturation field and unsaturation field. One then take the difference between two of them and plot ∆R versus field angle. If at some angle, ∆R is zero, it means that this angle is the canting angle. That is because if the applied field is at the canting angle, the magnetization will be aligned to the field direction whether the field is strong enough to reach saturation or not. See Figure 5.4.3.

Figure 5.4.3 ∆R versus field angle for different field changes [27]

Figure 5.4.4 the 10K resistance measurement with field rotating in LP plane of the Pd6nm/(Ni1.6nm/Pd6nm)20/Si.

Figure 5.4.5 ∆R versus field angle of the Pd6nm/(Ni1.6nm/Pd6nm)20/Si (𝜃_𝐶 = 73).

Figure 5.4.6 ∆R versus field angle of the Pd8.4nm/(Ni1.6nm/Pd8.4nm)20/Si (field in LP plane).

Figure 5.4.7 ∆R versus field angle of the Pd8.4nm/(Ni1.6nm/Pd8.4nm)20/Si (field in TP plane).

As shown in Figures 5.4.5, 5.4.6, and 5.4.7, we can find that the canting angle is about 70~75 degree to the plane normal. These results can be caused by four reasons.

First, the Ni thickness is thin enough so that the easy axis can lie at the angle between in plane and out of plane. In our sample, no perpendicular magnetic anisotropy was found. Secondly, because our thin film structure is deposited by sputtering, in which depositing time is short, hence, the thin film structure may form the corrugated plane instead of smooth plane. Thirdly, in the Pd/Ni multilayers structure, the Pd is the material near the stoner criteria so that the Pd layer may have magnetization which is induced by the Ni layer. If the magnetization direction of the Pd layer is not collinear with that of the Ni layer, it may cause the canting angle. Lastly, the Pd layer is the material which have strong spin orbital interaction, when the adjacent layer is ferromagnetic material, the Dzyaloshinskii-Moriya interaction (DMI) effect may occur. Hence, form the corrugated magnetization. The canting angle results are all the same in all the samples, even in the samples after annealing.

5.5 Anomalous Hall effect measurement

We measure the anomalous Hall effect with the applied field rotated in the LP and TP plane. In our measurement, the current flow is defined as the X direction, the applied field is defined as the Z direction, and we measure the Hall voltage in the Y direction.

The voltage signal is proportional to the Z component of the magnetization, as the equation below.

𝜌_𝐻 = 𝑅_𝐻𝐵 + 𝑅_𝑆4𝜋𝑀

, where B is the applied field, M is the magnetization，𝑅_𝐻、𝑅_𝑆 are Hall coefficient and anomalous Hall coefficient, respectively. In ferromagnetic materials, the 𝑅_𝐻 is small enough to be ignored. Hence, 𝜌_𝐻 ≈ 𝑅_𝑆4𝜋𝑀 ∝ M.

We measure the anomalous Hall effect with the applied field rotate in the LP plane of the. If the field is rotated in the LP plane, the Hall voltage should proportional to the Z component of the magnetization which is the perpendicular projection of the magnetization. Thus, this value should correspond to the Sine function. However, the experiment results are out of our expectation, as shown in Figures 5.5.1 to 5.5.4.

In the Pd2.4nm/(Ni1.6nm/Pd2.4nm)20/Si and Pd3.6nm/(Ni1.6nm/Pd3.6nm)20/Si samples,when the applied field is in the longitudinal direction, the voltage signal is nonzero. This nonzero voltage signal is more notable at low temperature. We consider that there is a Z component of the magnetization, when the applied field is in the

在文檔中鎳鈀多層膜中的自旋霍爾效應與磁鄰近效應 (頁 36-92)