Twist's Review #1 - Implementation of AlScN-based NCFET

Paper : Negative capacitance field-effect transistors based on ferroelectric AlScN and 2D MoS2

This week I read this paper on a relatively newfound ferroelectric material, AlScN (Aluminium Scandium Nitride), which was integrated into a NCFET device with MoS2 as its channel. AlScN’s ferroelectricity was only found in 2019, so unlike previous materials like HfO2, ZrO2, HZO, perovskite-based materials, and others, there is not a “best” method to implement AlScN (not that a perfect method for previous materials have been found either; although it has been more extensively researched).

Now the NCFET itself, or the Negative Capacitance Field Effect Transistor, is device that is aimed to sort of defeat a limit of conventional transistors, which is essentially to minimize the amount of voltage that needs to be applied to the transistor for it to turn on measured by a parameter called the Subthreshold Swing (SS) which is limited at 60 mV/dec (the lower the better). This kind of device is enabled by ferroelectric materials as they exhibit negative capacitance in a certain phase, that differs between each materials.

In summary this paper fabricated the NCFET in two ways, one through mechanical exfoliation (kind of a fancy way of saying carefully scraping a material; also used to make graphere) of MoS2 and also through Chemical Vapor Deposition (CVD) as it is more suitable for large area applications. In the end different interlayers/dielectrics were used with different thicknesses in the gate stack to investigate the effects and find the best between HfOx, AlOx and also h-BN (hexagonal boron nitride). Through mechanical exfoliation, it was found that 0.5 nm of HfOx showed the best performing device, reaching an SS of about 30.7 mV/decade and on the other hand h-BN showed the worst with 316 mV/dec. Another important parameter to think about is the hyteresis effect within these devices, and although it was small on the HfOx device it was much smaller on the AlOx device; a trade-off.

Through CVD, 6 devices with only AlScN (no interlayer/dielectric) were tested and all were at or below the SS limit, with the best being 36 mV/dec. Testing with AlOx and HfOx found that the device exhibited nearly-free hysteresis characteristics but only reached an SS of 90 mV/dec, while the latter had hysteresis but reached an SS of 51 mV/dec. Furthermore another experiment of varying the channel length was done to investigate its effects on the device’s SS, which showed lower channel length being more advantageous.

[1] S. Song et al., “Negative capacitance field-effect transistors based on ferroelectric AlScN and 2D MoS2,” Applied Physics Letters, vol. 123, no. 18, Oct. 2023, doi: Negative capacitance field-effect transistors based on ferroelectric AlScN and 2D MoS2 | Applied Physics Letters | AIP Publishing.

Personally I thought this paper was interesting, but I never really read up much about AlScN so I wonder what would happen if we use different methods of depositions for either the AlScN or MoS2 (deposition method for AlScN was not mentioned in the paper) as there have been studies with HfO2-based ferroelectric devices that shows ALD deposited HfO2 would perform better than with other methods.