Bi-Layered HfO 2 /Hf Buffer Layer

Assuming that it was the initial oxygen flow that degraded the AlGaN layer and its 2DEG, it was of interest to try the two step deposition process presented by Lu et al. [2006] done on silicon. That is deposit approximately 3 nm of pure Hf and then a few nanometers of HfO₂, in an attempt to minimise the degradation of the 2DEG.

124 CHAPTER 6. P(VDF/TRFE) DEPOSITION AND CHARACTERISATION

To determine the deposition rate of the hafnium layer, a layer of pure hafnium was RF sputtered at 300^◦C with a constant flow of 33 sccm argon for 60 min. The measured thickness deposited with an alpha step profilometer was 201 nm giving a deposition rate of 3.35^nm/min. The XRD diffraction pattern of this sample was representative of hexagonal (001) hafnium, only one peak was present at 32.7^◦which is similar enough to the database file summarised in appendix C. The capability to deposit such a high quality hafnium layer can not be surprising as there is only a small in-plane lattice mismatch between it and AlGaN, that of hafnium being 3.32 ˚A and that of GaN being approximately 3.186 ˚A, see appendix C.

A trial deposition was done in order to observe this possibility with a layer of hafnium deposited by RF sputtering for 60 s for a thickness of 3.35 nm onto the AlGaN layer at 300^◦C with a constant flow of 33 sccm of argon. After which oxygen was quickly introduced for 2 s at 33 sccm and reduced to the deposition flow rate of 4 sccm for the HfO₂ layer. The deposition was done for 180 s (approximately 2 nm) at 300^◦C, 500 W and an argon flow rate of 28 sccm. The benefits of this method was evident as there was no increase in the two point probe resistance measurements using this double layer deposition process, making it the ideal buffer layer for AlGaN based devices.

Figure 6.16: TEM image of HfO2/Hf/AlGaN/GaN interfaces. Where at constant pressure, 4.4x10⁻³Torr, and temperature, 300^◦C, hafnium was deposited by RF sputtering for 60 s and HfO2was deposited by RF sputtering for 180 s with a total oxygen flow of 4 sccm.

Making a linear extrapolation to estimate the thickness of the hafnium layers deposited was not accurate as can be seen in the TEM image in figure 6.16. From the TEM image it is possible to measure the hafnium layer as approximately 8 nm and the HfO₂ layer as 9 nm. The hafnium layer alone was too thick and will most likely act as a metallic layer and screen all of the polarisation from the 2DEG. The HfO₂ layer was also too thick to act as an efficient buffer layer. The hafnium layer is being deposited to limit the diffusion of oxygen into the AlGaN heterostructure, but should ideally be consumed with oxygen during the HfO₂ deposition process so that there is absolutely no metallic layer in the buffer layer. It is thought possible that in figure 6.16 that the arrow is pointing to a layer of hafnium that did not get at all affected by oxygen and is left to screen the polarisation from the 2DEG. The HRTEM images confirm this, figure , in that it is possible to use the layer of hafnium still existing, as it was not consumed with oxygen. It is possible to observe a layer of approximately 6 nm of hafnium that remains to screen the 2DEG from the ferroelectric polarisation. This is noted since the hafnium layer has a higher

6.4. SUMMARY 125

density than HfO₂, which is observed from the darker contrast in the HRTEM image.

It is possible to observe in figure 6.17b that the layers deposited are of high crystalline qualitty, however it is possible that the HfO₂ deposited on Hf is not the same as when it is deposited directly on AlGaN. It would thus be of interest to deposit a thicker layer and observe its XRD diffraction spectrum. The experiments involved in optimising this buffer layer is not exhausted and it is of interest to minimise both these layers while preserving the transport properties of the 2DEG.

(a) (b)

Figure 6.17: Two HRTEM images of the same HfO₂/Hf layer grown on AlGaN/GaN heterostructure.

Where at constant pressure, 4.4x10⁻³Torr, and temperature, 300^◦C, hafnium was de-posited by RF sputtering for 60 s and HfO₂was deposited by RF sputtering for 180 s with a total oxygen flow of 4 sccm.

Another two layered attempt was done with a deposition time of approximately 30 s to deposit a thinner layer hafnium, 4 nm. After which an oxygen flow of 4 sccm was introduced and the argon flow reduced to 28 sccm for the deposition of approximately 2 nm(180 s) of HfO₂. To control the 2DEG two point probe resistance measurements were done, with the Ti/Al/Ti/Au electrodes annealed in nitrogen, the resistance changed from 0.56 kΩ to 0.7 kΩ showing a small and unsubstantial degradation. The important thing to note here is that this sample had no visible metallic look due to the deposition of a thinner hafnium layer, while the above sample with 8 nm of hafnium showed a metallic appearance.

6.4 Summary

P(VDF/TrFE)(70:30) was deposited onto the AlGaN/GaN heterostructure and then tested for ferroelectricity. The use of an additional HfO₂ buffer layer was also investi-gated to reduce charge leakage. The samples that successfully underwent testing in this chapter and will be used for depletion experiments in chapter 7 are:

• Spin casted 250 nm P(VDF/TrFE) gate

• Spin casted 250 nm P(VDF/TrFE) gate with a 5 nm HfO₂ buffer layer

126 CHAPTER 6. P(VDF/TRFE) DEPOSITION AND CHARACTERISATION

• Spin casted 250 nm P(VDF/TrFE) gate with a 9 nm HfO₂/8 nm Hf buffer layer

• Spin casted 250 nm P(VDF/TrFE) gate with a 9 nm HfO₂/4 nm Hf buffer layer