Navigation and service

"Tantalizing concept "

Jülich scientists study new computer memories

[11. April 2006]

Future computer memories should be able to do everything: save large amounts of data, switch quickly and store data even without the supply of electric current. How this can all be integrated in one ceramic system, Professor Kristof Szot and his team of scientists from Research Centre Jülich demonstrated in a recent edition of the renowned specialist journal Nature Materials: he reduces the memory modules to the size of only a few atoms. With the tip of a scanning probe microscope, he can manipulate their conductance by many orders of magnitude and in this way quickly switch back and forth between two states. In the accompanying commentary in the same journal, Angus Kingon from North Carolina State University says: "This tantalizing demonstration of the concept lays an exciting foundation for a potential new memory technology."

Reference: K. Szot et al., Nature Materials 5, 312-320 (2006)

Also read the contribution about "phase-change memories":

Further up-to-date information on Research Centre Jülich’s exciting field of information research can be found at:

Background information on the recent publication in Nature Materials:

In the seventies of the last century, it was already observed that certain transition metal oxides show a switchable resistance effect after a so-called formation, for example thermal pre-treatment in reducing atmospheres. In the late nineties, a group under the supervision of George Bednorz from the IBM research laboratory in Zurich, who was awarded the Nobel Prize jointly with his colleague Alex Müller for the discovery of high-temperature superconductivity in oxides, suggested the effect as a potential basis for future computer memories. However, up to now, it was not clear how the resistance switching takes place and what structure properties it is associated with in the oxidic material.

In recent years, Professor Kristof Szot from the Institute of Solid State Research at Research Centre Jülich has developed a special version of so-called scanning probe microscopy that makes it possible to adsorb extremely small currents with an extraordinarily high spatial resolution on atomically smooth surfaces. He then applied his method to switching oxides, such as strontium titanate as suggested by Bednorz, to track down the physical and chemical mechanism of the effect. As reported in the April edition of the renowned specialist journal Nature Materials, he was able to demonstrate that after the formation of a single crystal (and in the same way of thin films) made of strontium titanate, the conductance does not occur homogeneously on the whole surface, but is limited to extended lattice defects, especially so-called dislocations in the crystal lattice of the oxide. The conductance between the exit of a dislocation on the surface and a defect-free siteonly one nanometre (equal to one 30,000th of the diameter of a hair) away can vary by many orders of magnitude. With the help of a positive voltage which he applied to the scanning probe tip, he was also able to switch off the high conductance of this dislocation exit again. Negative and positive voltages above a limit of approx. 2 V could optionally switch the conductance of the dislocation on and off. Obviously, this was exactly the same effect other researchers observed with large covering electrodes as resistance switches.

It seems that during the formation of the oxide, at first a tiny amount of oxygen extends along the dislocations in the crystal. In this way, the conductance of the material increases. As Gustav Bihlmayer and Kristof Szot demonstrated in the article by theoretical calculations on the basis of quantum-mechanical simulations, the conductance will indeed remain strongly localized – as was observed experimentally. The switching can obviously be seen as an electrochemical effect on a nanometre scale (like a nano battery effect) in which some oxygen ions are displaced and the access of the scanning probe tip to the conductive interior of the dislocation is interrupted and established again by local oxidation or reduction of the dislocation near the surface. On account of the very short distances, the ions are shifted very rapidly under the effect of electric power.

The possibility of using this effect in future computer memories becomes clear by taking account of the limitations of present memories. So-called dynamic random access memories, DRAMs, are volatile information memories; their information is lost when the operating voltage is turned off. In addition, it is hardly possible to reduce the size of the small capacitor which, as a charge-coupled memory, is the heart of a DRAM cell. Today’s other standard storage technology, the so-called flash memory in digital cameras and MP3 players, is in fact non-volatile memories, but the write cycle takes ten thousand times longer than in DRAMs. The resistive switching of the dislocations in transition metal oxides may combine the advantages of both memories: it is a non-volatile effect, and first assessments show that switching can take place very quickly. In addition, the small dimensions of the dislocations show the great potential that is in principle possible with regardto a future increase in package density. If scientists should succeed in pushing forward to the limits of the effects found, a further increase in density by a factor of 1000 is possible compared with today’s DRAM and flash generations.

In this way, a vision of the inventor of the transistor, William Shockley, could come true. In the eighties, he thought about whether dislocations in crystals could one day be used as electronically active structural elements and thus achieving a miniaturization still unimaginable at that time.

However, before the effect found can be used in practical storage elements, researchers have to solve a large number of very difficult tasks. The dislocations are statistically distributed in single crystals and grown thin films. However, they have to be arranged in a regular manner and positioned precisely for use in computer memories. To solve this central problem, Jülich researchers are now working on pre-structuring the wafers by means of appropriate nanometre-scale nuclei and, in this way, if possible predetermining the dislocations where they should grow. Other challenges involve the search for methods for the series production of nanometre-scale bottom and top electrodes which permit long-time and stable switching, as well as for the production of conductor connections on this small scale.

The future will tell to what extent it will be possible to solve these tasks and, in this way, push the door open to a completely new generation of computer memories.

Press contact:

Kosta Schinarakis
science journalist
Research Centre Jülich
52425 Jülich, Germany
Tel. +49 (0)2461 61- 4771, Fax +49 (0)2461 61-4666