Su Suntan Tiny Capacitors May Overcome Physical Limits of Hard Drives
Suntan Technology Company Limited
---All kinds of Capacitors
Storage—there is never enough of it. I still remember when I thought my 700MB hard drive was huge... until I tried to copy an entire CD onto it for faster access. After that, I spent a period stuck choosing music to stick on my three GB hard drive. Two weeks ago, I ditched six months' worth of simulation data because my 320GB hard drive was full. One TB of new drive later, and I'm wondering how soon it will be before I start feeling the squeeze again. Maybe never, if some of the latest research coming out of Korea and Germany bears fruit.
One of the cool things about hard drive technology is how it has actually kept pace with computer needs. The basic mechanism for hard drive storage, however, does have some fundamental limitations, which manufacturers will have to deal with fairly soon. Bits are currently stored in the orientation of tiny magnets, called ferromagnetic domains, on a hard drive platter. The smaller the domain, the easier it is for that orientation to be scrambled by temperature or stray electromagnetic fields. At a certain size, thermal photons (e.g., heat energy from the surrounding case or the underlying disk) have enough energy to flip a domain's orientation. Manufacturers will have to keep their domain sizes significantly bigger than that threshold size to ensure data integrity, which puts a ceiling on storage density, one we're rapidly approaching.
An alternative is to use ferroelectric domains. Unlike ferromagnetic domains, ferroelectric domains have a natural electric field with an orientation that can be used to represent data. Until recently, these haven't looked that attractive because they have pretty much the same limitations that ferromagnetic domains have, but they lack the cool read-out tricks. Ferroelectric materials, however, do have one big advantage over ferromagnetic materials: they can be used to make really good capacitors. This is exactly what the latest research, published in Nature Nanotechnology, is about.
The authors created a very dense array of nanocapacitors by a combination of masks and pulsed laser deposition. Essentially, an aluminum material was used to create a honeycomb-like structure on a platinum substrate. The voids in the honeycomb were then filled with ferroelectric material that had been blasted off a ceramic by laser pulses. Then, in the same way your mom removes cake from a cake tin, the aluminum mask was peeled off to leave lumps of ferroelectric materials. These were then coated by another layer of platinum, creating thousands of 60nm sized capacitors with a density of about 176 billion capacitors per square inch.
Of course, these capacitors aren't much use if they are too leaky to store any charge, so the researchers placed a needle on top of some of them to measure their electrical properties. They found that the capacitor could hold its charge for more than three days and that the array of capacitors had quite similar properties—they all behaved identically within about 10 percent.
As a side note, and possibly more remarkably, the researchers were able to demonstrate that something called substrate clamping exists. These materials, in addition to being ferroelectric, are also piezoelectric—they expand and contract in response to an applied voltage. However, thin films of piezoelectric materials generally do not exhibit the same percentage expansion for a given electric field as do thicker materials. This is because the layer consists of many tiny crystals that are at different orientations to each other. However, at the boundaries between crystals and at the boundary between the substrate and the layer, the spacing between the atoms must match. This causes a great deal of strain in the layer, which acts to prevent its expansion when a voltage is applied. The tiny capacitors also contain more than a single crystal, which are more closely aligned to each other so that they naturally match at the boundaries. This leaves them free to expand when a voltage is applied.
Given the important role of nanopositioning systems in much of modern research, I can see that last point becoming a significant focus of their research. However, the title and the conclusion don't quite reflect the contents of the paper. One capacitor stores one bit, therefore 176 Gb/inch2 is not all that close to one Tb/inch2. On the upside, I can imagine that read/write systems may not be too difficult to develop, since the feature size is within the resolution of current lithographic technology. This means that the paper could be a step in the direction of truly high density solid-state storage.
