An IBM-Zurich team including Nobel laureates has extended the Atomic Force Microscopy (AFM) concept to data storage. IBM group manager and probe storage team member Michel Despont explains.
While today’s magnetic data storage techniques are reaching some impressive levels, at some point in the not too distant future they will hit the physical limits of the technology. IBM has recognized that new approaches will be needed.
Ten years ago, the IBM Zurich Research Lab team, including researchers who collaborated on the Nobel-prize winning scanning tunneling microscope (STM), were looking for a way to extend related scanning probe techniques— such as atomic force microscopy, or AFM— beyond the scientific lab into the commercial domain. AFM, now the most widely used probe technique, does not use lenses or bouncing electron beams. Rather, it drags a tiny cantilever over a surface, measuring the dips and rises with atomic precision. A computer processes the cantilever deflection data into a graphical image.
However, the earliest attempts at AFM sometimes left indentations on the surface. And is so often the case, what was a bug became a feature: why not use the indentations as “1’s” and no-indentations as “0’s”—just like the old days of data punch cards?
But AFM is slow, and data storage applications require high data rates. The team reasoned that speed could be achieved by arranging arrays of thousands of cantilevers operated in parallel, each with a heatable tip that could create indentations in a polymer-coated table.
Recently, we have fabricated arrays of 4096 cantilevers in a 40.96mm² array on a 100mm² chip. We also demonstrated first working system prototypes that are able to store data at densities greater than 0.5 Tb/in². In single cantilever experiments we even achieved storage densities well beyond 1Tb/in²—more than 200 times higher than those of DVDs. With the potential for building structures on the molecular scale, this is just the beginning. The probe storage technology is optimized for minimal power consumption, ideally suiting it for mobile devices like digital cameras and cell phones.
The core components of the probe storage system are:
• a two-dimensional array of silicon probes (cantilevers) with heatable tips and
integrated sensors, and
• a micro-mechanical scanner that moves the storage medium relative to the array.
A sophisticated design positions the probes with nanoscale accuracy over the storage medium and ensures the system is not significantly affected by external vibrations or impact. For the device to perform its reading, writing and erasing functions, the cantilever tips are briefly heated and brought into contact with the storage medium, which is a thin film of a custom-designed, cross-linked polymer coated on a silicon substrate that is moved in the x- and y-directions.
The cantilevers used in the array are of a three-terminal design, with separate heaters for reading and writing, and a capacitive platform for electrostatic actuation of the cantilevers in the z-direction. The cantilevers are approximately 70 µm long and 0.5 µm thick, with a 500–700 nm long tip integrated directly above the write heater.
The apex of each tip has a radius of about ten nanometers, allowing data to be written at extremely high densities (greater than 1 Tb/in²). In addition to the cantilevers, the array chip also carries eight sensors that are used to provide x/y positioning information for closed-loop operation of the micro-scanner.
The cantilevers and thermal-positioning sensors are fabricated in the top, monocrystalline layer of SOI wafers, which will later be transferred and interconnected to a CMOS wafer containing the cantilevers’ driving electronics. The sort of SOI wafer that Soitec produces gives us precise control of the silicon membrane thickness in which the cantilevers are fabricated, which is crucial for the homogeneity of their mechanical characteristics. The insulating BOX of the SOI wafer is used as an etch stop during the dissolution of the “handle” part of the SOI stack, a process step performed before the cantilever transfer.
Ultimately, we believe that such very largescale integrated (VLSI) micro-nanomechanics will also generate as yet unthought of VLSIMEMS applications and opportunities.