How you can never fail to be impressed by engineering creativity

The results of this impetus are all around us: consumer electronics products of all kinds, automotive electronics, transportation and more. Electronics is the enabling technology for the modern world. Yet Moore's Law is now encountering the laws of physics. Each move to the next process technology node becomes more difficult as layers get thinner, there are fewer electrons and many more process steps. Memories are one problem area. While dram is scaling reasonably well, flash memory is facing some challenges as process technologies head towards the 20nm node and beyond. The reason? It's all about storing charge. At these incredibly small dimensions, storing charge becomes harder and harder. Electrons leak and the effects of the quantum world become more apparent. Some years ago, scientists started to think that, instead of storing charge, memories could be created using an electron's spin. In theory, instead of needing hundreds of electrons to represent a logic 1 or 0, that value could be represented by the direction of a single electron's spin. While that might seem a simple concept, converting it into reality is a significant challenge: how do you impart the correct spin?; how long you can control the spin for?; how do you read it? And another problem: can you do it repeatably at room temperature? Recent work by IBM Zurich, in conjunction with Swiss technology centre ETH, is beginning to answer some of these questions. The research has shown that spin can be maintained for 1.1ns – the cycle time for a 1GHz processor. However, the work is being undertaken at 40K. There will be a limit to how far conventional cmos processes can be scaled; there will be an apparent brick wall. But you never fail to be amazed by the creativity of researchers and engineers looking to break through the wall.