Artificial Superlattices Promise Better Memories
Alexander E. Braun, Senior Editor -- Semiconductor International, 5/15/2008 8:03:00 AM
Transition metal oxides have, for years, been the subject of research because of the great variety of extremely useful properties that they exhibit. They can be dielectrics, ferroelectrics, piezoelectrics, magnets or superconductors.
Professor Matthew Dawber is currently leading research in this field at Stony Brook University (Stony Brook, N.Y.). He began his investigations leading to superlattices and their characteristics four years ago, together with researchers from the University of Liège in Belgium and the University of Geneva in Switzerland.
“The group in Europe had a long history of making high-quality epitaxial thin films, so from there to doing superlattices of ferroelectric materials only took a small step,” Dawber said. The thinking behind the effort was that because they were producing multiple stacks of thin films, it should be possible to make superlattices because the process would be analogous to creating film stacks. “This would provide a far more delicate control of the boundary condition than what is usually obtained with thin films,” he said.
Atomic scale structure of a 1/1 PbTiO3/TiO3 superlattice; strontium atoms are blue, titanium atoms green and oxygen atoms red. The associated electron cloud is yellow. The rotations of oxygen atoms in consecutive layers result from the structure’s artificial layering, and are a feature of ferroelectric behavior. (Source. University of Liège)
Because of experience working with TiO3 and success in the production of thin films, research was started on a PbTiO3/SrTiO3 system because the group had worked with TiO3, and stacking it with SrRiO3 would be fairly easy because of the two materials’ good lattice match. PbTiO3 and SrTiO3 are two familiar and well-characterized oxides, with the first presenting a ferroelectric structural instability and the second a non-polar structural instability.
Initially, Dawber’s group started with that system because it is easy to grow and provides a good model system for studies of its interior. It was also possible to tune the ferroelectric properties by adjusting both the strain (to enhance the polarization) and composition. “We found that it is possible to do quite effectively things like changing the polarization on the dielectric constant or transition temperature, but when we attained all this, we discovered that something quite new was taking place at the interfaces,” Dawber said. What happens at the interfaces of these superlattice structures is the coupling between different kinds of distortions in the structure; this, in turn, gives rise to improper ferroelectricity.
Improper ferroelectricity is a kind of ferroelectricity that rarely occurs in natural materials, and when it does, its effects are too minute to be useful. This is what made the results so interesting because it opens a way of introducing ferroelectricity into materials that normally might not be ferroelectric. “This is a new route to the production of ferroelectric materials using those that are normally excluded because they have, for example, magnetic properties that are normally antagonistic to ferroelectric properties,” Dawber said. Presently, there is considerable interest in trying to do multiferric materials, and now that the researchers have the ability to build up these kinds of materials atomic layer by atomic layer, it becomes possible to produce new ones with unheard of properties.
This work is of great interest to the semiconductor sector because it should assist in the quest after better dielectrics, which is why, to increase memory density, for example, some device makers are already using HfO2. Dawber’s work offers a direct pathway to that because the superlattice he and his team have produced has a dielectric constant of ~600, quite higher than other existing materials. While naturally occurring ferroelectric materials inherently exhibit a high dielectric constant, it is usually temperature-dependent. Because it is an improper ferroelectric, the superlattice’s high dielectric constant is not temperature-dependent. The application of such a material to semiconductor devices is obvious because device temperature is a constant concern to designers.
“Ours is a ferroelectric system — you’re not looking for things like low resistivity. But if you’re after an interface to obtain high mobility, there are several other materials that can be used,” Dawber said, adding that the kind of interface quality obtained would make it easier to couple together the different properties of materials and come up with a different kind of coupling, something that is very discrete because of a very well-defined interface. “When you have high-quality interfaces, you’re better able to define a region over which whatever you are looking to manipulate in your device is taking place,” he said. “This whole area of oxide interfaces has become very exciting this last year because we’re finally starting to discover things — different effects — that are unique to the various interfaces.” The research group plans to study several new interfaces between oxides that have both rotational distortions and some kind of magnetic ordering. The expectation here is to come up with a strongly coupled multiferric material.
In terms of devices, particularly memory chips, there is a long-standing problem in that while they are fairly easy to write to, they can be quite difficult to read. As a counterpoint, for magnetic devices, the opposite holds true: They are difficult to write to but easy to read. Dawber thinks that it should be possible to create a material that will couple those two qualities, leading to a major breakthrough in memory. This is only one of many approaches looking toward new multiferric materials — introducing that kind of ferroelectricity into materials that are already magnetically ordered. The result would be non-volatile memories that are easy to read, easy to write, very fast and not power hungry. Other possibilities, because these are all piezoelectric materials, are expected to open up for future MEMS applications.
semiconductor.net
The article is recent but superlattice research has been going on for years all across the semiconductor industry. L'Intel alone has blown tens of millions (at least) on it in connection with polymer based FeRAM since @the late 90's. What has changed now is the level of understanding of the ferroelectric mechanism which has really spiked in the last year, e.g.,
pda.physorg.com
Certainly it may only be serendipity... but it is certainly an "interesting" coincidence that L'Intel's move to spin off Phase Change development should occur just when the characterization of ferroelectric materials takes a critical step forward.
Of course if I were to bet on it... and I certainly have... I would bet that last year's "oxide interface" discoveries will/have lead to the elimination of any major impediments to further scaling of F-RAM below the 90nm node... and that, considering the incredibly bad performance of Phase Change as a memory IP, L'Intel has yet again made a massively poor decision on the future course of a non-cpu technology.
Then again it remains to be seen whether we... the industry... or civilization for that matter... will survive to see these wagers pay off. I should think the results would be obvious to all by 2010... barring planetary catastrophe of some kind.
All things considered... you might want to place a small side bet on Pestilence. There are no sure things on the pessimist's side of the ledger anymore than there are on the optimist's side. But I see no reason to expect humanity will be permitted to run amok with impunity... when sweet, innocent, little lemmings aren't allowed to get away with it. <Hoo><ack><ack><pHooie>
0|0 |
