Samples characterization

Figure 2.8: Experimental setup for characterization (a) and zoom on sample position (b).

2.3.1 Samples characterization

As a figure of merit to evaluate their performances, the resistance of different MR sensors was evaluated as a function of an external magnetic field provided by a Helmholtz coil. Specifically, the voltage change was measured being related to the device magnetoresistance when a constant current is supplied to the leads. Initially, a constant (DC) current was employed but then, in order to improve the signal-to-noise ratio, an AC driving current was used in combination with a lock-in amplifier read-out (useful especially in the case of the small voltages resulting from PHE-based structures with magnitude of hundreds of 𝜇𝑉).

The experimental setup employed for the characterization of GMR and TMR based structures is shown in Figure 2.8 (a): the GMW coils can provide a uniform magnetic field between its poles until several hundreds of 𝑚𝑇 and in the bottom part is possible to see the sensing circuits realized for the measurements connected with excitation and readout instrumentation. The system is driven by a LabView home-made software, with which we also performed the calibration of the current flowing into the coils by measuring the generated magnetic field through a GMW teslameter located at the samples’ position on system startup and at regular timings, to be sure that eventual changes in temperature due to Joule’s effect didn’t affect the characterization of different samples.

(a) (b)

42 Figure 2.9: Electrical scheme of experimental setup used for GMR and TMR characterization. The sensing circuit consist in an AC current generator feeding the series of MR sensor with 1 𝑘Ω resistor.

A SR830 Lock-in amplifier measures the voltage drop at the structure’s terminals using as reference the frequency signal at the far end of the resistor.

More in detail, with reference to Figure 2.9, the sensing circuit was arranged with the series of a MR structure (GMR or TMR) and a constant resistance 𝑅 = 1𝑘𝛺, while the current was supplied by a Keithley 6221 AC current generator set with a signal at 𝑓 = 1.00387 𝑘𝐻z and amplitude 𝐴 = 1 𝑚𝐴. The frequency was selected with the purpose to reduce noise from external electromagnetic sources (including their harmonics), while the amplitude came from a compromise in term of signal and reduction of structures degradation and warm up due to the Joule effect. The voltage drop at the MR structure was then measured at different external magnetic field values using a lock-in amplifier (Stanford Research SR830) and the voltage drop at R as its reference signal. The samples were preliminary bonded on a home-made carrier realized on a printed circuit board (PCB) during the beginning of this PhD work.

Figure 2.11 shows a typical response obtained from a GMR sensor. Specifically, in the range larger range [−20 𝑚𝑇; 20 𝑚𝑇] investigated in (a), it’s possible to notice the free layer fast rotation near zero field, where the pinned layer magnetization remains fixed pointing to the positive x axis, while, on the contrary, the latter starts to rotate for more negative field than the former, obtaining a decrease in resistance (proportional to the voltage drop) with a lower drop with respect to B magnitude. It’s possible to assert these considerations on the dynamic change of the magnetic configuration since it’s well known from theory that the GMR effect depends

43 upon the angle between the ferromagnetic layers spaced by a non-magnetic conductor and considering their different pinning strength in this structure, the experimental data are in good agreement with theory. Figure 2.10 (b) shows a typical characterization obtained in the magnetic field range [−10 𝑚𝑇; 10 𝑚𝑇], so below the coercive field of the pinned layer, obtaining an excursion depending only upon the effect of the external field on the free layer.

The left axis scale in both figures represents the voltage drop measured, while the right one indexes the value of magnetoresistance, defined as 𝑀𝑅(𝐵) = ^{( )} . According to the magneto-resistive ratio definition, expressed in Equation 1.1, we calculated for these GMR structures the average value to be 4.1%. Furthermore, another key parameter for a sensor is the sensitivity, namely the derivative of the signal respect to the stimulus. This quantity is a constant only in case of a linear dependence between them, but in general it can depend also on the value of the independent variable, so it can follow a complex behavior and it could be useful to define a sensing range where sensitivity can be considered approximately constant and near its maximum value. In our case, we numerically extrapolated this quantity from the experimental data and considered its estimate as the average value calculated in the range of free layer sharp rotation, obtaining a value of 6 %/𝑚𝑇.

Figure 2.10: Example of response curves obtained by GMR sensors in range [−20 𝑚𝑇; 20 𝑚𝑇], obtaining almost the complete switch of all magnetic layers for negative fields with respect to the setup positive direction (a) and in a smaller one [−10 𝑚𝑇; 10 𝑚𝑇] where there is the complete rotation of the free layer only with the pinned one remaining fixed, pointing along positive x axis

during the field sweep .

(a) (b)

44 TMR’s structures characterization is similar to GMR’s and it was performed with the same experimental setup. Since also in this case the magnetoresistance depends upon the angle between the two ferromagnetic electrodes, the signal obtained has similar shape, as shown in Figure 2.11 but with some substantial differences.

Figure 2.11: Example of response curves obtained by TMR sensors in range [−17.5 𝑚𝑇; 17.5 𝑚𝑇], where there succeeds only the switching of the free layer, since the pinned one has much higher

coercive field.

First of all, in the range investigated, [−17.5 𝑚𝑇; 17.5 𝑚𝑇] , there is only the rotation of the

“almost” free layer, which is weakly pinned in orthogonal direction with respect to the other electrode at zero field, since the pinned layer coercive field should be more than twice the maximum magnitude of the field applied considering the results we obtained with MOKE measurement during the first part of fabrication phase with a thinner synthetic antiferromagnet for 0.9 𝑛𝑚 of 𝑅𝑢 spacer between 𝐶𝑜𝐹𝑒 and 𝐶𝑜𝐹𝑒𝐵 layer with same thicknesses.

Furthermore, it’s possible to notice that the GMR change of resistance is sharper than TMR’s, although the latter exhibit a slightly larger variation, since we obtained an average value of 103% for the magnetoresistive ratio for this series of sensors, meaning the high resistance status has about twice the magnitude of the low resistance one. Regarding the sensitivity, calculated with the same approach of the previous case, we obtained a value of 11.8 %/𝑚𝑇, so about twice the value of our Giant Magnetoresistance-based sensors, originated from the two characteristics exposed above. Furthermore, it’s possible to notice slightly smaller hysteresis

45 for TMR’s than GMR’s, probably due to the influence of shape anisotropy on the free layer, since in some region of the structure it let a system have an easy axis parallel to the pinned layer magnetization, while in others it points in perpendicular direction, considering the zig-zag layout employed.

For PHE based structures, the experimental setup employed was similar with respect to the one described above for GMR’s and TMR’s sensors. The main differences are reported in the electrical scheme shown in Figure 2.12, where it’s possible to notice in the bottom-left part the different magnitude of the series resistance used for getting the frequency of the reference signal, which value of 100 Ω is close to ohmic resistance of the sensors in the flowing current direction, so the one which determines the charge applied between the power supply outputs.

Figure 2.12: Electrical scheme of experimental setup used for PHE sensors characterization. The sensing circuit consist in an AC current generator feeding the series of MR sensor with 100 Ω resistor. A SR830 Lock-in amplifier measures the voltage drop at perpendicular structure’s terminals

using as reference the frequency signal at the far end of the resistor.

Figure 2.13 (a) reported a sketch of the structures, patterned as crosses to let the current flow through the easy axis direction 𝑥 and to measure the voltage drop in perpendicular direction 𝑦 by two terminals connected to the lock-in amplifier as shown in Figure 2.12. It’s known from theory (Section 1.1.3) the dependence of the planar Hall voltage on the angle 𝜙 between magnetization and current density, specifically it results 𝑉 ∝ sin(2𝜙), so by applying an external magnetic field along 𝑦 it’s possible to exert a torque on 𝑀 which causes its rotation,

46 obtaining 𝜙 to change in the range − ; . The shape of 𝜙(𝐵) this kind of system is stated in Equation 1.20, from which it’s possible to derive the dependence of the planar Hall voltage upon the applied field.

A typical response obtained by the characterization of this structures in the range [−8.5 𝑚𝑇; 8.5 𝑚𝑇] is reported in Figure 2.13 (b), where left axis shows the voltage drop magnitude. Specifically, its shape near zero field is approximately linear, as expected from Equation 1.18, while moving to more positive (negative) values it decreases (increases), until reaching a stationary point for 𝜙 = − where it starts increasing (decreasing) until reaching a saturation value, characterized by the angle 𝜙 = − . Such odd behavior is fully explained by the dependence upon sin(2𝜙) , furthermore it’s not useful to investigate a larger field range since there are no more magnetic phenomena possible of interest for our purpose in these structures.

Figure 2.13: Sketch of PHE based sensor layout with the indication of the measurement parameters (a) and typical response obtained in the investigated range [−8.5 𝑚𝑇; 8.5 𝑚𝑇], enough to obtain the

signal saturation near its edges (b).

The right axis of Figure 2.13 (b) represents the percentage change in planar Hall resistance. In this case, we used the definition of PHR stated in Section 1.1.3, but we normalized this parameter to obtain the relative change as

𝑃𝐻𝑅(𝐵) =𝑉 − 𝑉

Δ𝑉 , (2.3)

(a) (b)

47 to take into account that the angle 𝜙 follows the opposite sign of the external magnetic field and the very small order of magnitude of the related impedance: since 𝑃𝐻𝑅 ∈ [24 𝑚Ω; 88 𝑚Ω], a small change in the length of the external electrodes employed for the reading can affect the measurement.

Equation 2.3 leads to a magnetoresistive ratio of 100% by construction and a sensitivity of 47 %/𝑚𝑇 near zero, so in the linear region. The last parameter is usually defined in terms of microvolt change in the planar Hall voltage per one Oersted (1 𝑂𝑒 = 0.1 𝑚𝑇) of field in literature, so we also calculated the sensitivity with this unity of measurement to be 8𝜇𝑉/𝑂𝑒.

A consistent number of sensors for any series has been characterized for performing a statistical analysis of signal change (Vs magnetic field), sensitivity and typical impedance magnitude and Table 2.1 summarizes the evaluated parameters. It’s possible to notice the biggest sensitivity belonging for the PHE-based sensors among the devices under test, which is a key parameter for the magnetic particles’ sensing, considering that they reach the saturation magnetization only for very high external fields of the order of Tesla, so they will show only a small fraction of them to be detected for fields obtained in miniaturized devices.

GMR PHE TMR unconventional characterization performed during my last secondment period in Bielefeld, with a different setup configuration. An AC current flowing through a PHE structure in the 𝑟 direction (cylindrical coordinates) generates in the outside a magnetic field with the same frequency directed along 𝜙 according to Ampere’s law. This field can magnetize some particles