5. Results and Discussion

5.1 Magneto-optical properties

In order to gain understanding about the intrinsic magneto-optical properties of TbFeCo/Pd, the multilayer structures were examined using a wide wavelengths magneto-optical Kerr spectroscope apparatus[3.4]. When plotting _K and  versus wavelength it is possible to investigate if the intrinsic properties of the current materials are suitable for future short wavelength magneto-optical recording. In Fig 5.1 and 5.2 the polar Kerr rotation angle, _K, is plotted for t_TbFeCo=30Å and t_TbFeCo=50Å respectively.
 
 

Fig 5.1 Polar Kerr rotation, _K, of TbFeCo 30Å/Pd tÅ

Fig 5.2 Polar Kerr rotation, _K, of TbFeCo 50Å/Pd tÅ

As seen in the figures (Fig 5.1-2), the absolute value of the Kerr rotation is not significantly enhanced in the major part of the spectra. Some small enhancement was detected at the ultraviolet photon energies beyond 4 eV. The peak position of the _K enhancement is at about 4.5 eV. The optimal structure for maximum enhancements of the polar Kerr rotation at shorter wavelengths are t_Pd about 12 Å for both t_TbFeCo=30Å and t_TbFeCo=50Å. When increasing the Pd layer thickness even further the enhancement becomes saturated. According to this result one can estimate the range from the interface where Pd atoms are likely being polarized. If saturation is assumed to occur at t_Pd=12Å this range will be about 6Å, half the Pd thickness.

When comparing the obtained data with previous work done on TbFeCo/Pt and TbFeCo/NdCo (Fig 5.3), it is clear that the enhancement of the polar Kerr rotation angle is much smaller in the TbFeCo/Pd multilayers. The peak observed at around 5 eV for TbFeCo/Pt multilayers is almost not detectable when changing element from Pt to Pd. A comparison of the magnitude of _K at 4.5 eV shows that the enhancement of the polar Kerr rotation is about four times larger in the TbFeCo/Pt multilayer than in the TbFeCo/Pd multilayer. Even NdCo has got a significantly larger enhancement in that part of the spectra.
 
 

Fig 5.3 Polar Kerr rotation of TbFeCo and TbFeCo/Pd,
TbFeCo/Pt and TbFeCo/NdCo

The behavior of the polar Kerr ellipticity in TbFeCo/Pd shows little improvements compared with the single layer TbFeCo. In fact for the major part of the wavelength spectra the magnitude of the ellipticity decreases (Fig 5.4 and 5.5). Pd saturation thickness seems to occur at very thin Pd layers. In the case of t_TbFeCo= 30Å the best structure seems to be at a Pd thickness of only 3Å. This seems also to be the case when t_TbFeCo= 50Å.
 
 

Fig 5.4 Polar Kerr ellipticity of TbFeCo 30Å/Pd tÅ

Fig 5.5 Polar Kerr ellipticity of TbFeCo 50Å/Pd tÅ

Comparison of the ellipticity spectra of TbFeCo/Pd with TbFeCo/Pt and TbFeCo/NdCo shows that the property seems quite similar in the range of 1.4-4.5 eV. Beyond approximately 4.5 eV, peaks are arising in TbFeCo/NdCo and especially in the TbFeCo/Pt structures. This peak has its maximum at about 5.5 eV in both TbFeCo/Pt and TbFeCo/NdCo. The peak is not to be found in the TbFeCo/Pd structure.
 
 

Fig 5.6 Polar Kerr ellipticity of TbFeCo and TbFeCo/Pd,
TbFeCo/Pt and TbFeCo/NdCo

5.2 Magnetic properties
 

5.2.1 Coercivity versus t_Pd
In order to examine the magnetic properties, the coercivity at room temperature was measured for all samples using polar Kerr looper in fields up to 15 kOe. The results of these measurements are shown in figure 5.7. It is clearly shown that with increasing Pd layer thickness the coercivity increases and then decreases when the Pd thickness increases even more. A change in the polarity was observed at already t_Pd=1Å in the case of t_TbFeCo=30Å and at t_Pd=3Å in the case of t_TbFeCo=50Å.
 
 

Fig 5.7 Coercivity versus t_Pd

This result indicates that the multilayer structures? composition have changed from RE rich to TM rich and the compensation composition is at about 0.75Å and 2Å Pd thickness for 30Å and 50Å TbFeCo thickness respectively. This behavior suggests that Pd act as a TM-element in this specific structure. A schematic description of the subnetworks in TbFeCo/Pd is given in Fig 5.8.
 
 

Fig 5.8 Schematic description of subnetworks in TbFeCo/Pd

The samples with the structure TbFeCo 50Å/Pd 1Å and TbFeCo 50Å/ Pd 2Å both have too high coercivity, more than 15 kOe, to be determined at room temperature.
 
 

5.2.2 Hysteresis loops
Figure 5.9 display polar Kerr hysteresis loops of TbFeCo/Pd at various Pd thicknesses. As shown in the figure, the TbFeCo/Pd multilayers exhibit high squareness of the loops for a wide range of Pd layer thicknesses. When t_TbFeCo=50 Å a square loop is obtained even at t_Pd = 24Å. In the case of t_TbFeCo=30 Å square loops are feasible until the Pd layer exceeds 6 Å. Notice that the TbFeCo 50Å/Pd 24Å hysteresis curve is given in a different scale to present a clearer view of the squareness of the loop.
 
 

Fig 5.9 Hysteresis loops

5.2.3 Temperature dependence of the Coercivity
The temperature dependence of the coercivity was measured using polar Kerr looper in the temperature range from ambient to 200C (Fig 5.10 and Fig 5.11).  It is clearly shown that the compensation temperatures, T_comp, are shifted towards lower temperatures when increasing the Pd thickness in the multilayer structure.
 
 

Fig 5.10 Coercivity versus temperature (C), t_TbFeCo= 30 Å

Fig 5.11 Coercivity versus temperature (C), t_TbFeCo= 50 Å

In the figure 5.12 the estimated compensation temperatures are plotted versus the Pd thickness.
 
 

Fig 5.12 Compensation temperature (C) versus t_Pd

5.2.4 Saturation Magnetization
To obtain information about the saturation magnetization the samples were measured using VSM, in fields up to 15 kOe. Figure 5.13 visualizes the relationship between M_s and t_Pd for three different TbFeCo layer thicknesses, t_TbFeCo= 10Å, 30Å and 50Å. The total sample volumes were taken into account when deriving M_s. When t_Pd increases M_s decreases and reaches a minimum in the vicinity of the compensation composition. When increasing t_Pd further, M_s increases up to a point when t_Pd becomes large compared with the bilayar thickness, ?. The rapid decrease of the saturation magnetization in the t_TbFeCo=10Å series when t_Pd increases from 12Å up to 24Å can be explained by the fact that the Pd volume fraction of the TbFeCo 10Å/Pd 24Å sample is about 70%.
 
 

Fig 5.13 M_s versus t_Pd, total sample volume

The increase in M_s with t_Pd was clearly found for all sample series. This behavior could be due to polarization of the Pd atoms. An attempt to model the structure was made in figure 5.14.
 
 

Fig 5.14 Structure of multilayers with alloyed region

The model proposes a structure containing three different phases. In the center of the TbFeCo layers there are assumed to be uninfluenced TbFeCo regions like as well as in the center of the Pd layers uninfluenced Pd regions. The uninfluenced layers are assumed to act as single TbFeCo and Pd layers respectively, i.e. the Pd layers are non-magnetic. At the interfaces an interfacial alloyed region with polarized Pd are presumed to appear. The alloyed region is supposed to be twice as thick as t_Pd up to a certain maximum alloyed region thickness t_max. When plotting M_s against the sum of all the interfacial alloyed region thicknesses, an almost linear result is observed when assuming a maximum interfacial alloyed region thickness, tmax, of 10Å(Fig 5.15).
 
 

 Fig 5.15 M_s versus total interfacial alloy thickness

Structural analysis with high angle x-ray diffraction, -2 mode using Co-K, indicates a possible crystalline structure in the Pd layers when t_Pd exceeds 10Å(Fig 5.16).
 
 

The fcc Pd(111) peak in the single layer TbFeCo film is due to the 30Å Pd over layer. An increase in intensity is clearly observed when the Pd multilayer thickness increases to a thickness of more than 10Å.

When examining the shape of the M-H loops, a step was detected in the single TbFeCo film. This step was significantly larger than in the multilayers and could not be explained as background impurities nor could polar Kerr looper detect the step. This indicated a possible in-plane magnetized layer at the substrate interface. Films with different thicknesses were examined with polar Kerr looper and it was found that films thicker than 150Å performed square loops. Thinner films showed in-plane behavior.
 
 

Fig 5.17 Polar Kerr loops of thin TbFeCo layers and
M-H loop of TbFeCo single layer

To avoid this property, a buffer layer of Pd was grown on the substrate before the deposition. It was found that 10Å buffer layer was sufficient to avoid any in-plane magnetization.
 
 

Fig 5.18 Kerr loop and M-H loop with 10Å Pd buffer layer

5.2.5 Magnetic Anisotropy
The uniaxial magnetic anisotropy constant, K_u, was measured using a torque magnetometer with both rotational hysteresis loss, W_r, and L2 Fourier analysis techniques[3.3]. The methods showed reasonable agreement at compositions not to close to the compensation point. In the vicinity of the compensation composition, however, the conventional Fourier analysis is not valid to determining Ku, since the maximum field strength of 15 kOe used in the measurement is still insufficient to saturate the magnetization. In this case distorted torque curves was detected (Fig 5.19).
 
 

 Fig 5.19 Distorted torque curve

The uniaxial anisotropy of RE-TM is reported to increase at compositions close to compensation point[16]. The data gained from Fourier analysis demonstrates a drop in K_u near the compensation point. This drop is only apparent, i.e. it does not correspond to a decrease of the anisotropy energy required to bring the subnetwork magnetization to in-plane direction. This feature originates from the canting between the RE and TM subnetwork magnetization. Even a small canting can cause large discrepancy in K_u[16]. Therefore the data obtained by rotational hysteresis loss measurements was assumed to be more accurate nearby the compensation point, though they showed higher values of K_u. Even though K_u increases very rapidly with the Pd thickness, the K_u values at thin Pd layers are believed to be even higher than the measured data.

Figure 5.20 shows the Pd thickness dependence of K_u for t_TbFeCo=30Å and 50Å. As seen in the figure K_u increases with t_Pd in both cases and becomes significantly larger at t_Pd=3 Å. When t_Pd exceeds 6Å a difference in the behavior between the t_TbFeCo=30Å and t_TbFeCo=50Å series was detected. The samples with t_TbFeCo=30Å showed a decrease in K_u while the t_TbFeCo=50Å samples remained at a constant level of approximately 2.5*10⁶erg/cc. At t_Pd =12Å the t_TbFeCo=30Å series ceased to decrease and became constant to t_Pd and with a K_u about 7*10⁵erg/cc.
 
 

Fig 5.20 Perpendicular magnetic anisotropy versus t_Pd

The behavior of TbFeCo 50Å/Pd tÅ is in reasonable agreement with the corresponding TbFeCo 50Å/Pt tÅ reported by Y. Itoh et al[17].
 

5.2.6 Investigations of the Coercivity mechanism in TbFeCo/Pd  multilayers

To investigate the coercivity mechanism in TbFeCo/Pd multilayers, measurements were made using an alternating gradient field magnetometer with applied fields up to 20 kOe[3.6].

Figure 5.21 presents a set of minor hysteresis loops of TbFeCo 30Å/Pd 6Å. From this example it is clear that wall pinning is a great contributor to the coercivity in TbFeCo/Pd multilayers. The hysteresis loops were measured at room temperature and they exhibit high squareness.
 
 

Fig 5.21 Hysteresis loops of Pd 6Å/TbFeCo 30Å

The observed coercivity changes with the maximum applied field, which is plotted in the figure 5.22 below. The pinning field was found to be about 480 Oe and the actual coercivity around 550 Oe. These observations, also, leads to a conclusion that the coercivity mechanism in TbFeCo/Pd is due to wall pinning.
 
 

Fig 5.22 Apparent Hc against applied field for TbFeCo30Å/Pd 6Å

A way to gain understanding about the coercivity mechanism is to use the magnetization relaxation phenomenon under the influence of the magnetic field. This analysis has been demonstrated in magnetic multilayers such as Co/Pt [18]. The relaxation of magnetization may be written as

where ? is the relaxation time, M₀ the magnetization at t=0 and M? is the magnetization after infinite time (equal to either Ms or -Ms). When derivating the equation above (5.1) with respect of time, the dependence of the magnetization is gained.

?(H) is the following expression where V_B is the Barkhausen volume and H_p the wall pinning field.

From the time dependence of the magnetization expressed by the equation above, it is possible to derive the coercivity as a function of the sweeping speed of the applied magnetic field.

According to this equation the coercivity is linearly dependent on ln(dH/dt) and from this relation the Barkhausen volume,  V_B, can be estimated.
 
 

Fig 5.23 Barkhausen volume estimations

In order to evaluate the Barkhausen volume, measurements of the coercivity as a function of sweeping speed of the applied magnetic field for the TbFeCo/Pd multilayers were made. The sweeping speed was varied from 50 Oe/s to 1000 Oe/s and the measurements were performed at ambient temperature. The relation between Hc and ln(dH/dt) is plotted in the figure 5.23 and all measured samples exhibits linear relationships. The Barkhausen volume is easily estimated from the slope using equation (5.4) above.
 
 

Fig 5.24 Barkhausen volume and diameter versus t_Pd

When calculated the Barkhausen volume, an estimation of the Barkhausen diameter, d_B, was made assuming the Bakhausen volume stretches all across the sample and has a cylindrical shape. The results are shown in Fig 5.24 (right axis).
 
 
 

5.3 Magneto-optical Recording Performance

The recording performance of the fabricated discs was tested using a disc tester, Nakamichi OMS-2000, at different reading and erasing powers but at a constant linear velocity of 5 m/s. The pulse width, the duty, was set to 30% and the frequency 1 MHz. In ordinary drives the bias power are approximately the same as the writing power, but using the present disc tester the bias power can be reduced significantly. The bias power was set to be 0.1mW.
 

5.3.1 Recording
In order to determine the appropriate recording powers for the media. The C/N ratio dependence of recording power was measured for different mark lengths (Fig 5.25a-5.25c). When comparing the results no great difference in C/N ratio could be detected between the different discs.
 
 

Fig 5.25a C/N versus recording power

Fig 5.25b C/N ratio versus recording power

Fig 5.25c C/N versus recording power

The Carrier to Noise, C/N, or Signal to Noise ratio S/N, was measured at different recording fields. For positive values of the field the ideal signal is zero, but in reality this is not so very often the case. This phenomenon is due to the antenna effect. When heating a bit to the Curie temperature the recorded bits in the vicinity will affect the bit with their stray field. This may cause an involuntary recorded bit.
 
 

Fig 5.26 Recording field versus C/N ratio

5.3.2 Reading Power Dependence
There are several different sources of noise in magneto-optical readout signals. They are electric noise, shot noise in the detector, media noise and laser noise. These noise sources' amplitude is at almost the same level, but different noise mechanisms are dominant at different reading power intervals. The amplifier noise is the largest contributor at low powers, at slightly higher powers the carrier to noise (C/N) ratio are mostly determined by the shot noise.  When increasing the power even more media and laser noise becomes the dominant factors. The servo system when using Nakamichi OMS-2000 needs about 0.5 mW reading power. In this case that servo power is enough to make the media noise the dominant noise contributor. As seen in figure 5.27 the C/N ratio is almost independent of the reading power. The marks, bits, were recorded with two times the minimum recording power.
 
 

Fig 5.27 C/N ratio versus reading power

5.3.3 Erasing
When deciding the erasing power one has to be careful not to set a too high power. If the power is set to high this will lead to an annealing effect that will decrease the perpendicular magnetic anisotropy and if the power is set high enough the anisotropy may become in-plane. To avoid this phenomenon the erasing power of TbFeCo should not be higher than three times the minimum recording power. An erasing power below three times the minimum recording power will allow 10⁵-10⁶ erasing cycles. Lowering the erasing power even more to below 2.5 times the minimum recording power one will easily obtain more than 10⁶ erasing cycles. While a higher erasing field like four times the minimum recording field will only allow 100 or even 10 erasing cycles and the magnetic properties will change after only one erasing. Following this, the erasing power is a very critical property to determine. In figure 5.28 the erase ratios versus the erasing power are plotted for different reading powers.
 
 

Fig 5.28 Erase ratio (dB) versus erasing power (mW)
for different writing powers

5.3.4 MFM and AFM investigations
The written marks were observed with MFM[3.8] and the surface structure by AFM[3.8]. Damaged parts of the discs can easily be seen in the AFM pictures (Fig 5.29). These damaged areas correspond to the heated laser spots when writing the marks. It can clearly be seen that the damaged areas correspond to written marks in the MFM picture. It is thereby assumed that the laser power was too high in the case of the TbFeCo and TbFeCo 30Å/Pd 3Å discs. The TbFeCo 50Å/Pd 6Å disc on the contrary managed this heating much better. Maybe the thicker Pd layers were acting as heat sinks. A way to improve this structure could be to ad an aluminum reflecting layer. This layer would act as a heat sink and transport the heat away from the SiN layer. This would possibly decrease the damages to the film.
 
 

Fig 5.29 MFM and AFM pictures

Other information gained by the pictures (Fig 5.29) is that the tracks were not fully erased before the writing process began. Some remaining background magnetization can easily be detected at numerous places in the MFM pictures.

When examining the MFM pictures carefully one can observe a reversed magnetized area in the center of the written bits. This phenomenon may be due to the stray field of the neighboring bits. During writing, the bit was heated above the Curie
 
 

Fig 5.30 Section Analysis of TbFeCo 30Å/Pd 3Å

temperature and the cooling of center area was maybe a little bit slow and thereby the stray field of the neighboring bits may reverse the magnetization direction of the center of the bit.

When decreasing the laser power, the damage of the films can be avoided. As seen in the pictures below (Fig 5.31).
 
 

Fig 5.31 AFM and MFM pictures