Summary

PZT was deposited onto the AlGaN/GaN heterostructure and then tested for ferroelec-tricity. The use of the additional MgO buffer layer was also investigated. The samples that successfully underwent testing in this chapter that will be used for future experiments in chapter 7 are:

• CSD 400 nm PZT gate

• CSD 300 nm PZT gate with 10±5 nm MgO buffer layer

• Sputtered 130 nm PZT gate

• Sputtered 195 nm PZT gate

5.6.1 TiO

Seeding Layer for (111)PZT

The deposition of a 2 nm TiO₂ seeding layer by magnetron sputtering favored the (111) growth of PZT on AlGaN. Without this nucleation layer random oriented PZT was grown.

5.6.2 CSD PZT

Depositing PZT by CSD methods proved highly destructive to the 2DEG due to the rapid thermal annealing process in oxygen. The best annealing process was for a time of 30 s at 700^◦C, allowing for the preservation of most of the 2DEGs properties. However, this PZT(40:60) layer was not of optimal quality, since the complete elimination of all

5.6. SUMMARY 107

pyrochlore phases was not possible. The PZT that is in the pyrochlore phase will act as a ”dead” layer, that is it will not have the switching and retention of the spontaneous polarisation that the perovskite phase possesses.

5.6.3 PLD MgO Buffer Layer

An attempt to minimise the diffusion was done by the addition a MgO buffer layer. The deposition of MgO by pulsed laser deposition, PLD, onto AlGaN was successful since they have a small lattice mismatch. There was little degradation to the 2DEG, from conductance measurements, after the CSD deposition of randomly oriented PZT(40:60).

5.6.4 Sputtered PZT

Due the volatility of the CSD deposition process the PZT(40:60) deposition process was changed to multiple target RF magnetron sputtering. After the optimisation of this method it was possible to deposit 100% (111) textured PZT with absolutely no unwanted pyrochlore phases. This process still degraded the 2DEG but only increased the sheet resistance by a factor of three, decreased the electron sheet concentration by a factor two and decreased the mobility by a factor of two thirds.

5.6.5 Outlook

After these measurements it would have been interesting to do more depositions of sput-tered PZT on AlGaN/GaN: minimising the temperature and pre-heating time of the deposition, minimising the diffusion at the interface, preserving the transport properties of the 2DEG, trying different compositions and thicker PZT films. Also of interest would have been to see how these parameters directly influence the modulation of the transport properties of the 2DEG. Especially of interest is the observation of the correlation of the PZT thickness, the Zr:Ti composition in the PZT to the modulation of the transport properties. However, it was extremely unfortunate that after the first series of successful experiments the magnetron sputtering machine had a series of mechanical and electronic failures which did not allow for further depositions of PZT.

Is PZT really the most optimal ferroelectric to be used as the gate on the AlGaN het-erostructures? The combination of the two materials allow for the use in applications that have demands for high temperature and high frequency. However, for commercial use there is probably no use for lead based electronics due to environmental reasons. So for the sole reason of attempting to research towards a commercial ready device it would be more interesting to study organic ferroelectrics such as the P(VDF/TrFE) ferroelectric polymer as will be done in the following chapter 6.

The integration of PZT with GaN for the use in RF micro-electro-mechanical systems, MEMS, for bandpass filters is also being currently investigated, Dey et al. [2004]. In this configuration the switching of the spontaneous polarisation in the PZT layer is not being used but its capability of coupling with the RF signal while maintaining a stable large polarisation.

Chapter 6

P(VDF/TrFE) Deposition and Characterisation

The ferroelectric layer of poly(vinylidene fluoride/trifluoroethylene), P(VDF/TrFE), was also tried as the gate ferroelectric on the AlGaN/GaN heterostructure. It has multiple interesting characteristics, one of which being a smaller spontaneous polarisation, than PZT, of 8.5^µC/cm², which should be still sufficient to modulate the 2DEG. Another ad-vantage is its small dielectric constant, P(VDF/TrFE)=13, so that when it is sandwiched with the AlGaN, Al0.3Ga0.7N=10.29, Ambacher et al. [2000], a significant amount of the voltage drops across the ferroelectric layer and not the semiconductor layer, and the de-polarisation field is minimised. Lastly, and possibly most importantly P(VDF/TrFE) has a low crystallisation temperature, of 130^◦C that can better preserve the great transport properties of the 2DEG, than the higher temperature processing of PZT.

This chapter will describe in detail the deposition of the co-polymer ferroelectric layer, poly(vin-ylidene fluoride/trifluoroethylene), P(VDF/TrFE), onto the AlGaN/GaN het-erostructure and its characterisation. Also described will be the optimisation of the depo-sition process for a HfO₂ buffer layer to limit charge injection between the P(VDF/TrFE) and AlGaN layers. Listed below are the different ferroelectric, and buffer layers whose depositions were investigated in this chapter.

• Spin casted P(VDF/TrFE)(70:30)

• Sputtered HfO₂ buffer layer

• Sputtered HfO2/Hf buffer bi-layer

After the ferroelectric and/or insulator layer was deposited various tests were done in order to characterise the ferroelectric thin film and determine if the 2DEG retained its great transport properties. The experimental investigations of the P(VDF/TrFE) and HfO₂ layers that were done in this chapter are listed below. When these results were positive it was then possible to further proceed with capacitance-voltage measurements and the fabrication of Hall bar and van der Pauw structures.

• XRD spectra

• AFM topography