BackgroundIn the electronics and optics industries there are widespread applications of materials that have piezoelectric and ferroelectric prop-erties. Piezoelectric materials are crystalline materials, often ceramics, that develop an electrical charge when subjected to mechan-ical stresses; ferroelectric materials are a subgroup of these that demonstrate spontaneous electrical polarisation, the direction ofwhich can be altered by an external electrical field. The value of ferroelectric ceramics is clearly illustrated by their use in the semi-conductor industry.Semiconductor memories such as dynamic random access memories (DRAMs) and static random access memories (SRAMs) cur-rently dominate the market. However, the main problem with these memories is that they are "volatile" - in other words if the powerfails, the stored information is lost. Ferroelectric random access memories (FRAMs), on the other hand, are non-volatile and have theadded benefits of greater radiation hardness and higher speed. FRAMs made from ferroelectric thin films store data by altering (with an external electrical field) the magnitude and direction of elec-trical polarization. Their non-volatility is because the polarization remains in the same state after the voltage is removed, and their radi-ation hardness allows devices containing these memories to be used in harsh environments, such as outer space.One of the problems with ferroelectric memories is their tendency to lose the ability to store data after a certain number of read/writecycles. This phenomenon is called fatigue. At the moment the fatigue resistance of FRAMs is not sufficient for them to replace com-pletely semiconductor memories, but this can be improved with further optimisation of both composition and microstructure.Lead zirconate titanate (PZT) is a ceramic material that exhibits excellent ferroelectric properties. PZT thin films are commonly usedto make FRAM devices and the prospects are very bright for such devices to replace semiconductor memories in the near future.This application note focuses on the microstructure of a commercial PZT thick film deposited on a single crystal silicon wafer. Thesample has been extensively thermally cycled, and automated EBSD analyses are used to characterise the microstructural proper-ties of two contrasting grain populations. The resulting data may have important implications for the application of this PZT in FRAMdevices.
VIII Contrasting grain typesin a ceramic thick film
Q: What are the growth mechanisms of different grain populationsin a PZT ceramic thick film?
Experimental Set-UpSample Preparation Raw materials with 200 Å carbon coatingSEM type W- filament EBSD System HKL CHANNEL 5Acc. V. 30 kVProbe Current ~10 nA
Total grid dimensions 200 x 100 139 x 104Grid spacing 3 µm 1 µmNumber of points 20,000 14,456Scanning speed 9 patterns/second 8 patterns/secondNoise filtering level Low Low
ResultsFigure 1: Secondary electron image
This secondary electron micrograph shows the generalmicrostructure of the PZT thin film. Large grains, often with reg-ular forms (e.g. star shaped), are isolated in a matrix of muchsmaller grains. These shapes may be explained by interpene-trative twinning, resulting in regular twin domains.
Figure 2: EBSPs from within a single large grain
The CHANNEL 5 acquisition software, Flamenco, has been used to view diffraction patterns collected from the individual domains inone of the large star-shaped PZT grains in figure 1. It is clear that the EBSPs 1 and 3 are identical, as are the EBSPs 2 and 4. Indexingthese EBSPs as tetragonal (pseudo-cubic) PZT confirms that the orientation pairs are twin related, with a 60° rotation about a <111>axis.
Small orientation maps were collected of individual large grains or clusters of large grains. A number of these are shown in figure 3,along with the EBSP quality maps. In all images the grain boundaries (>10°) are marked in black, with subgrains (2-10°) in grey. Twinboundaries (60° about <111>) are marked in red.It is clear that all the large grains are twinned. The twinning characteristics vary from grain to grain, but typically they have 2-fold or3-fold structures.
Figure 4: Fine-grained matrix characteristics
The microstructure of the matrix was characterised using automated EBSD mapping. These images show the results of a typicalEBSD analysis.
Figure 4 (a): EBSP quality map, with grain boundaries (>10°) shown in black and twin boundaries in red. The scale bar = 50 µm.Figure 4 (b): Orientation map of the same area, with colours corresponding to the Euler angles. The wide range of colours indicatesthat there is almost no texture in this sample. Once again, grain boundaries (black) and twin boundaries (red) are marked.
ConclusionsSeveral PZT thick film specimens have been studied using both manual and automated EBSD. Two contrasting grain types have beenidentified on the surface of the silicon single crystal wafer:
• A matrix population of small grains (mean < 8 µm), with no texture and very few twin boundaries. • Isolated large grains, typically 100-250 µm diameter, all characterised by distinctive twinning. These twins typically form two-fold orthree-fold domains.
It is likely that some form of abnormal grain growth has occurred in order to produce these large twinned grains. The EBSD data showthat they do not have any preferred orientation with respect to the thick film: it is therefore possible that such grains have grown atisolated nucleation sites, although further EBSD analyses are required to confirm this interpretation. In summary, EBSD analyses have identified contrasting microstructural characteristics of two grain populations in these samples: thiscould have a damaging effect on the electrical properties of the PZT film.
Acknowledgements
The National Institute of Standards and Technology, USA, is thanked for providing the sample and for allowing presentation of these data.
