Reaney- Characterisation of Ferroelectric Thin Films For NvFRAMs Flashcards
What was one of the biggest issues in holding back the development of NvFRAMs?
The yield of bits during deposition and their reliability during operation
Why is control of the microstructure more important in thin films than bulk ceramics?
Any pore/second phase particle will act as an electric field concentrator and potentially cause an electric short within its vicinity. This results in individual bits of the memory being unusable. In bulk ceramics, small amount of unwanted phases or porosity aren’t much of an issue if the overall microstructure is dense with the correct phases present
What are film properties often dominated by?
The contact between the electrode and the ceramic since these interfaces are a significant fraction of the total volume unlike in bulk ceramics where they can be ignored
Main problem with sol-gel spin coated thin films
They are notoriously prone to the formation of pores which is disastrous for the hysteresis properties over time
Evolution of microstructure of PZT sol-gel film as a function of temperature
Up to 300C the film is amorphous. At 400C reduction occurs due to the combustion of C in the organic molecules and Pb crystals form in the centre of the film. At 500C the Pb particles oxidise to PbO and the intermediate pyrochlore phase forms. At 650C pyrochlore phase transforms to perovskite but the phase separation of the Pb/PbO from the rest of the film results in an inhomogeneous distribution of the cations inhibiting the transformation.
When do inhomogeneity problems occur for sol-gel?
Often if the as-deposited layer is too thick and the local PO2 (pressure of O2?) is dominated by the presence of C. Means thin layers (50-100nm) are usually deposited followed by crystallisation.
General transformation for sol-gel thin film
Amorphous to pyrochlore to perovskite
When is pyrochlore retained?
I’d the control of stoichiometry is not accurate as crystallisation proceeds, pyrochlore is retained at the surface and eventually becomes trapped at the film/top electrode interface
How is the top electrode usually deposited?
Usually sputtered which is a high energy process
Problem with sputtering electrodes on
It damages the surface of the film during deposition creating an amorphous layer in between the ceramic and electrode
How to improve the microstructure of the damaged interface after sputtering top electrode
Post deposition anneal in O2 at about 500C
Is a thin film made using MOCVD susceptible to the problems caused by sputtering the top electrode?
Yes even though it produces a high crystalline quality and the films perform well
Orientation of sputtered films
Often don’t show any strong orientation normal to the substrate and often the crystals are randomly oriented throughout the film. This causes problems with the ferroelectric switching behaviour. Therefore a need to control the orientation during film growth
Common way of controlling orientation during film growth from sputtering (and PZT example)
Film surface coated with a seed compound to control the orientation of the bulk of the film. For PZT most common method to deposit PbTiO3 prior to deposition of PZT. Perovskite structured PbTiO3 has a natural growth habit plane of {001} which occurs irrespective of the surface upon which it is deposited. Then perovskite PZT nucleates and grows on a surface with the same structure but oriented so it adopts the {001} habit plane of the underlying PbTiO3.
The 3 main parameters of a thin film intended for a NvFRAM
Switching characteristics (hysteresis behaviour)
Fatigue properties
Leakage behaviour
Impact of thickness of thin films on switching characteristics
Thinner films have more tilted hysteresis loops and lower remanent polarisation. As films get thicker the loops resemble normal bulk ceramics becoming more square with high remanent polarisation. This is due to the increasing influence of the interfaces as the film gets thinner. A non-ferroelectric dead layer is believed to exist in the vicinity of the interface and increasingly contributes to the overall properties as the thickness is reduced
What is the degree of fatigue as a function of the number of cycles controlled by?
In general the choice of electrodes. Pt electrodes give the worst performance whereas oxide electrodes show good resistance to fatigue of the remanent polarisation. One example arrangement is
Ir/IrO2/PZT/IrO2/Ir
What is leakage current related to?
Or current density related to the hysteresis loop
How does leakage current occur?
As the ferroelectric thin film becomes polarised, first there is reorientation of the dipoles within the ferroelectric domains which oppose the direction of the applied field. Then there is a displacement of space charge not associated with the ferroelectric dipoles. This manifests itself as a slow increase in polarisation from the remanent (Pr) to the saturated (Ps). If the ferroelectric dipoles contribution is extracted from the total displaced charge the leakage current density is what remains.
Problem with leakage current density
This non-dipolar charge is able to redistribute itself once the field is removed since it is not structurally fixed in place like the ferroelectric dipoles. Therefore it can compensate for the stored dipoles charge and slowly over time cause the remanent polarisation to decrease.
Current density relative to hysteresis loop
There is a peak at low positive voltage and comes back down quickly before slowly curving up again at higher voltages. Stops at the voltage for the Ps of the hysteresis loop
