CSD PZT Gate

devices, as described in section 1.3.2. Probably the most important obstacle in achieving this is the optimisation of the necessary processing steps and material compatibility.

The PZT samples were predominantly deposited and processed as Hall bar structures where the active area was poled with the conductive cantilever from the PFM. Whereas, the testing of P(VDF/TrFE) was done on capacitor structures, after which more infor-mation was obtained from top electroded Hall bar structures.

7.2 CSD PZT Gate

The GaN sample used for the development of the chemical solution deposited, CSD, PZT had a different structure than that used for the other experiments. From top to bot-tom the structure was as follows: 8 nm of GaN n-doped, 12 nm AlGaN n-doped with 15% Al, 1.5 − 3 µm GaN n-doped on a sapphire substrate. The most important dif-ference with this heterostructure and the one used throughout the thesis, is the low aluminium concentration in the upper AlGaN layer. The transport properties of its 2DEG are noted in table 7.2. At 300 K its 2DEG had transport properties of n_s = 4x10¹²− 3x1013 electrons/cm⁻² and µ = 280 − 1000^cm2/Vs. At 77 K its 2DEG had transport properties of n_s = 1.6x10¹²− 2.7x1013 electrons/cm⁻² and µ = 9000 − 16000^cm2/Vs.

Table 7.1: Mobility and electron sheet concentration of the 2DEG in the GaN/AlGaN/GaN structure used for the experiments with CSD PZT.

ns µ

[electrons/cm²] [cm²/Vs]

298 K 4x10¹²− 3x10¹³ 280-1000 77 K 1.6x10¹²− 2.7x10¹³ 9000-16000

A 400 nm thick polycrystalline PZT film was deposited by the CSD technique onto this GaN/AlGaN/GaN heterostructure with a 2DEG located 20 nm below the ferroelectric layer. The annealing process done to crystallise the PZT layer was with an RTA at 700^◦C for 30 s in air. This annealing process was optimised in order to obtain the best crystalline quality of PZT, without severely degrading the transport properties of the 2DEG. See section 4.1.2 and 5.2 for the details regarding this technique and the optimisation of this ferroelectric layer on the AlGaN heterostructure.

Mesas, of Hall bar structure, with a depth of 500 nm, figure 7.3, were defined by ECR-RIE.

Electrodes of Ti/Al/Ti/Au, 30 nm/100 nm/30 nm/30 nm respectively, were deposited by electron beam evaporation as bottom electrodes to the 2DEG.

The PZT film was poled using the SFM domain writing technique, described in section 3.3.4, with a DC voltage of ±40 V applied to the SFM conductive cantilever tip, figure 7.2. No top electrode was used for this experiment and the created polarisation pattern was controlled by piezoresponse force microscopy, figure 3.10.

130 CHAPTER 7. FERROELECTRIC GATE OPERATION

Figure 7.3: The PZT/AlGaN/GaN Hall bar structure without top electrode. Showing the 50x50 µm² active area that was poled with ±40 V to the conductive cantilever tip using a SFM tech-nique.

7.2.1 Transport Measurements

In order to evaluate the effect of the polarised ferroelectric film on the 2DEG the pref-erential polarisation was induced in the area of 50x50 µm² and then the 2DEG was characterised by the Hall effect. Firstly, measurements were done without poling in order to define the initial electron concentration. The concentration was found to be 5x1012 electrons/cm² and remained virtually unchanged within the temperature range from 77 K to 298 K. Then the studied area was poled with −40 V, corresponding to the deple-tion effect of electrons in the 2DEG and the measurements were repeated. As the last step, the polarity on the same area was inverted being poled by +40 V and the third series of transport measurements were performed. Figure 7.4 shows that the electron concentration changes approximately by a factor of two as the sign of polarisation in PZT switches. These values are summarised in table 7.2. At 298 K and 77 K it was possible to modulate the electron sheet concentration by a factor of two when poling with ±40 V.

Figure 7.4: Effect of spontaneous polarisation in the PZT film on the electron concentration in the 2DEG, when poling with ±40 V applied to the SFM conductive cantilever tip.

The fact that the sheet concentration measured in the ”as deposited” PZT state did not

7.2. CSD PZT GATE 131

Table 7.2: Electron sheet concentration of the 2DEG in the CSD PZT/GaN/AlGaN/GaN structure when being modulated with a DC bias of ±40 V.

n_s [electrons/cm²]

As deposited After −40 V After +40 V

298 K 5x10¹² 2.8x10¹² 6x10¹²

223 K 5x10¹² 3.3x10¹² 5.8x10¹²

173 K 5x10¹² 3.7x10¹² 5.8x10¹²

123 K 5x10¹² 3.7x10¹² 5.8x10¹²

77 K 5x10¹² 3.0x10¹² 5.8x10¹²

change with temperature, at 5x1012 electrons/cm² was as expected. This is due to the fact that the electron sheet concentration in a AlGaN heterostructure is controlled by the total polarisation in each layer which is fairly stable with temperature, to ∆n_s≈5%. Larger modulations of the sheet concentration are normally due to parallel conduction at the substrate interface. Therefore, if a stronger depletion effect can be observed at lower temperatures it is more informative to study the electron mobility or sheet resistance.

The measurements of the electron mobility was temperature dependent, however it was virtually independent of the poled polarisation, measured as 1200^cm2/Vs at 298 K and 4200^cm2/Vsat 77 K. The fact that the electron mobility varies with temperature confirms the fact that if a stronger modulation/depletion effect is expected at low temperatures then it is of interest to study the mobility instead of the sheet concentration. The reason why µ was not impacted when poling the ferroelectric gate is of concern and needs to be further investigated.

Although the results presented here are not strong, they are a positive indication to further optimise device fabrication in order to observe a stronger effect of depolarisation due to the change of the spontaneous polarisation in the ferroelectric layer. More extensive measurements need to be done, especially in determining the dependence of polarisation in the PZT layer and the mobility in the 2DEG with temperature. It could be that many of the problems occurring here are due to the poor quality of PZT deposited. This poor quality is due to the sensitivity of both the crystalline quality of PZT and preservation of transport properties of the 2DEG with the high temperature annealing process. It is thus of interest to investigate other PZT deposition processes that are done at lower temperatures, without annealing steps, and different ferroelectric layers. Also of interest is to use an AlGaN heterostructure with a higher concentration of aluminium in an attempt to limit inter-diffusion occurring at the PZT/AlGaN interface and obtain a higher quality PZT layer.

7.2.2 PZT/MgO/AlGaN

Before switching to a different deposition process of the ferroelectric layer an MgO buffer layer was deposited by pulsed laser deposition onto the standard Cree AlGaN/GaN het-erostructure used throughout this thesis. The most important reason for implementing this buffer layer was to inhibit the diffusion occurring at the PZT/AlGaN interface during the rapid thermal annealing process, preserving the 2DEGs transport properties. Also this layer could help preserve the original ferroelectric properites of the PZT layer

de-132 CHAPTER 7. FERROELECTRIC GATE OPERATION

posited by CSD, thus allowing for the observation of a stronger depletion effect in the 2DEG than that presented above, in section 5.2.

The MgO layer deposited was 10 nm ± 5 nm by PLD, pulse laser deposition, from a MgO target onto the AlGaN/GaN heterostructure. Onto which a PZT(40:60) layer of 300 nm was deposited by the CSD technique and annealed at 700^◦C for 30 s. There was no degradation of the transport properties of the 2DEG after the deposition of MgO and the high temperature annealing process for the CSD PZT. No seeding layer was used since the direction of orientation of MgO is (111), so it was assumed that this will help grow the PZT in the preferential direction of (111). However, this was not the case and the PZT layer had random orientation without any signs of pyrochlore phases. Circular gold electrodes were deposited onto a non etched structure for simple ferroelectric and depletion measurements.

C-V measurements

C-V measurement were done at 100 kHz with an AC modulation voltage of 0.01 V_AC and a DC poling volatage of ±5 V_DC, applied for 2 s. Both C-V measurements and PFM measurements concluded that there was no ferroelectric switching of the PZT film in the MFS structure, see figure 7.5. This could possibly be due to large screening effects produced by the MgO buffer layer, that might be too thick 15 nm, impinging the 2DEG of being affected by the charge associated with the ferroelectric polarisation, or the poor quality of the CSD PZT. With the PLD system used it is not possible to control the thickness of the layer grown to a better accuracy than 10 nm ± 5 nm. Therefore, it is not possible to further optimise the deposition conditions in order to grow a thinner MgO layer. For the use of a buffer layer it is important to have an ”ultra-thin” buffer layer in order to not completely screen the polarisation, if the MgO layer is 15 nm there is a risk that it is too thick.

Figure 7.5: Normalised C-V curve of 300 nm CSD PZT/10 nm MgO/AlGaN heterostructure.

Unfortunately, there was only one trial done for the deposition of CSD PZT on a MgO buffer layer. Possible optimisation of the multitude of processing steps and conditions could lead to a better insight of the mechanisms behind the non-switching of the ferro-electric layer deposited onto MgO/AlGaN.