Ferroelectric Characterisation

3.4 Ferroelectric Characterisation

3.4.1 P-V Measurements

MFM structure

Polarisation switching can be observed in a ferroelectric hysteresis loop when measured by using a Sawyer-Tower circuit, Sawyer and Tower [1930]. The polarisation is measured by the change in current when applying an AC voltage, at a small frequency, that is varied from 0 V to voltages greater than the coercive field of the ferroelectric being measured.

Figure 3.13 shows a typical ferroelectric hysteresis curve for a PZT thin film deposited with the same multiple target magnetron sputtering technique used in this thesis. The remanent polarisation, P_r, is the polarisation at which there is no applied electric field and the coercive field, E_c, is the coercive field in which there is zero net polarisation. For the ideal ferroelectric these values are expected to be symmetric, +P_r=-(-P_r) and +E_c=-(-E_c).

However, there are many factors that do not allow for this symmetry such as film thickness, charged defects, mechanical stresses, and deposition conditions, Damjanovic [1998]. For example in thin films this hysteresis loop is often tilted due to the depolarisation field, Tagantsev et al. [1995]. The PZT film shown in figure 3.13 is not symmetric and has an internal bias field, E_b.

Figure 3.13: Ferroelectric hysteresis loop of 300 nm PZT(45:55) (111) oriented, where the curve is started with a) a positive DC bias and b) a negative DC bias, Hiboux [2002].

Jaffe et al. [1971] summarise very well the fact that ferroelectric hysteresis loops can often be misinterpreted due to assuming that the layers resistance is linear and that its capacitance is non linear. The P-V curve can prove that a layer is ferroelectric: by showing that it is possible to switch the dipole moment of the structure and have it retain this net polarisation. However, it is much better to compliment this measurement with another measurement technique such as observing the Curie temperature, the high dielectric constant and domain structure with PFM. Determining if a layer is ferroelectric or not would be easy if the layer was an insulator due to the fact that all charges are capacitive. However, when the layer is not an insulator it is necessary to determine if the layer possesses non-linear or linear capacitive and dielectric behavior.

48 CHAPTER 3. MEASUREMENT TECHNIQUES

MFS structure

The ferroelectric hysteresis curve is instrumental for observing the ferroelectric properties of a ferroelectric sandwiched between two electrodes. However, this is not the case for the samples used in this thesis where it is possible to think of the configuration as metal -ferrroelectric - semiconductor - metal. With this configuration it is no longer so straight-forward to observe the ferroelectric switching behavior as the semiconductor impacts this measurement. This is due to the fact that it becomes more difficult to associate what cur-rent is a result of the polarisation switching in the ferroelectric layer. If the semiconductor is highly doped and the channel is directly in contact with the ferroelectric layer maybe it would be possible to measure the change in current flow and thus get a ferroelectric hysteresis loop.

Figure 3.14: Ferroelectric hysteresis loop of 130 nm PZT(53:47) on a n-doped GaN with an electron density of approximately 1x1017 electrons/cm³, Li et al. [1999].

Li et al. [1999] were able to measure a ferroelectric hysteresis loop on a PZT/GaN struc-ture. 130 nm of PZT(53:47) which was deposited by pulsed laser deposition, PLD, onto a thick n-doped GaN layer with a final annealing step at 600^◦C. The n-doped GaN layer had an electron density of approximately 1x1017 electrons/cm³ that was directly in contact with the PZT which allowed for making a polarisation hysteresis curve, shown in figure 3.14. However, Jaffe et al. [1971] give a thorough description of when it is necessary to be careful in interpreting the P-V hysteresis curve. Since, if a lossy linear capacitor, with linear dielectric and conduction behavior, is measured it will give the same ellipsoidal curve shown in figure 3.14.

3.4.2 C-V Measurements

MFM structure

Capacitance-Voltage, C-V, curves can also be represented as the relative dielectric per-mittivity, ₃₃, vs the DC electric field E_dc. The C-V curve of a typical PZT thin film deposited with the same magnetron sputtering technique as used in this thesis is shown in figure 3.15, Hiboux [2002]. The butterfly shape that this C-V curve makes is charac-teristic of ferroelectric films and shows that it is possible to separate intrinsic lattice and extrinsic domain contributions. The maximum capacitance occurs at the coercive field.

3.4. FERROELECTRIC CHARACTERISATION 49

Figure 3.15: Relative permittivity vs gate electric field characteristic of a PZT thin film, Hiboux [2002].

These experiments will be useful in determining if the layer is ferroelectric, whether it switches and partly retains the polarisation when switched. Capacitance-Voltage, C-V, Measurements were performed using a Hewlett Packard 4284A precision LCR meter with an AC voltage frequency range of 20 Hz to 1 MHz. The external DC bias was also applied through this LCR meter. The AC voltage was smaller than the coercive field and the additional DC poling voltage was switching the ferroelectric.

This measurement technique is complementary to the ferroelectric polarisation hysteresis loop measurement, since it is not always possible to apply equation 3.18 to the ferroelectric hysteresis curve in order to obtain the C-V curve, Damjanovic [1998]. This is due to the fact that the dielectric constant is frequency dependent and the P-V curve is usually done at a much smaller frequency than the C-V curve. Also problems in domain switching occur when making C-V measurements due to applying the large DC field that can permanently block the ferroelectric domains, where the AC field supplied by the lock amplifier is not capable to sense them. Only if the sample being measured is a ideal single crystal, will it be possible to apply equation 3.18 and observe equivalent results in the C-V and P-V measurement techniques.

_f_o = ∂P_f

∂E (3.18)

MFS structure

Unlike the ferroelectric hysteresis loops it is possible to measure the C-V curves of fer-roelectrics deposited onto semiconductors that are not highly conducting. Shen et al.

[2002] pulsed laser deposited, PLD, 400 nm PZT(53:47) onto 75 nm Al_0.22Ga_0.78N/ 3 nm Al0.22Ga0.78N/ GaN heterostructure. The C-V curve of this structure while using an alu-minium top electrode is shown in figure 3.16. Note that it is not similar to the C-V curve of the ferroelectric layer sandwiched between two metal electrodes as in figure 3.15, but exhibits counter clockwise behavior. The reason for the change in the capacitance is due to the modulation of the electron sheet concentration in the 2DEG with the gate voltage, explained by the simple equation 2.1.

The C-V measurement technique is thus a useful method that allows for both observing if

50 CHAPTER 3. MEASUREMENT TECHNIQUES

Figure 3.16: Counterclockwise capacitance vs gate voltage curve in a PZT(53:47)/ Al0.22Ga0.78N/GaN heterostructure exhibiting ferroelectric behavior, Shen et al. [2002].

the ferroelectric layer is indeed so and if the polarisation induced when the ferroelectric/se-miconductor is poled modulates the electrons in the channel or 2DEG. The depletion of electrons is observed by comparing the capacitance at the largest positive applied electric field to the largest negative applied field. If there is a difference in this capacitance it is assumed that there is a depletion of electrons in the 2DEG. The ferroelectricity of the film should be visible from C-V curves, as in figure 3.16 where the curve is counterclockwise due to a retention of the capacitance, when the capacitance is either increased or decreased after obtaining the maximum negative and positive electric field, respectively, see Sze [1985]. If this curve was clockwise this would be an indication of no ferroelectricity in the film due to the observation of absolutely no retention. If a depletion in the channel occurs for a positive DC bias then what was stated above will be reversed: clockwise curves will indicate retention and ferroelectricity, whereas counterclockwise curves indicate no retention and ferroelectricity. C-V curves that do not show a complete cycle of the DC poling voltage do not contain enough information to determine if the thin film is ferroelectric or not, but only indicates if it is possible to modulate the channel with the application of a DC bias. The voltage gap at the average capacitance value in the C-V curve can be considered the memory window of the ferroelectric device, which gives an indication to the retention of the ferroelectric device. This memory window is often dependent on factors as the thickness of the ferroelectric film, largest electric field applied and frequency used for the C-V measurement.