Eliminating Probe Interactions, Other Metrology Issues

Michael Garner, manager of the External Materials Research Group, Technology Strategy, at Intel Corp. (Santa Clara, Calif.), gave a presentation at the 2007 International Conference on Frontiers of Characterization and Metrology for Nanoelectronics in which he outlined a few of the hurdles metrology faces in encountering the nanotech era.¹

An area he has concentrated on is our industry's focus on alternate-state devices that use technologies such as spin for storage and computing purposes. "Currently, we can detect lower concentrations of charge than we can spin," Garner said. "Right now, we can inject spin into a semiconductor, but there aren't many ways of detecting spin concentration in the material. Thus, often times, there's a need for some sort of a device fabricated to actually determine the effective injection of spin-polarized electrons into the semiconductor. We're looking at ways to transport and manipulate spin in very small devices." Garner added that we must be able to detect spin in device structures, as well characterize the concentration of spin-polarized electrons, and map it diagnostically to determine where spins are being lost or transported.

Other issues deal with characterizing behavior in embedded interfaces. This has been a problem with molecular devices for many years, making it difficult to exactly determine what takes place. Some experiments are headed in the right direction. "If we look at nanoscale devices," Garner said, "we don't want to destroy the embedded interface, but it's getting challenging to characterize the transport of charge across the interface, as well as characterizing the interface structure. Now, with the addition of properties such as spin, it's even more difficult to characterize the relationship of the structure of the interface to the propagation or spin interaction at those interfaces. With very small features and new materials, we no longer look at a planar interface, but at nanostructured interfaces."

A major issue is that measurements are affected by probe fields. At small scales, the probe interacts with the sample and can perturb the local state of the material. This problem will worsen, and there is great need for models that will help decouple these perturbations — the changes that the probe introduces to the materials — and either find ways to reduce the effect or extract the interaction from the probe sample. "As you get down to things like quantum dots — nanometer scale structures — this requirement becomes increasingly acute."

Another looming issue is sample preparation. Preparing samples at the nanoscale level using something like ion milling, for example, can considerably change a device structure. Therefore, the preparation and probe sample interactions must be understood in terms of the structure and properties of that very small sample of material.

Solutions require different communities of researchers to work together on these problems and, unavoidably, expensive engineering. "There's already work going on with some new tools, but model development will require significant and extensive engineering that examines the physics of these various interactions to develop models that can be used in this metrology."

Thin-film magnetoresistor structures read hard disks and magnetic memory. Spin is injected from a magnetic source into silicon and manipulated by the gate before accepted or rejected by the magnetic drain. Metrology must determine electron and spin concentration, and identify factors that change spin orientation. (Source: Intel Corp.)

Garner views the sample probe interaction problem as one of the most significant issues. "Devoting effort to this would probably bring the most benefits as we migrate to nanometer scale structures," he said. "Clearly, we're looking for new tools and the ability to characterize different state variables, such as spin. The bigger issue is that if we were to get probe signals out, we would not necessarily understand what they mean, so having a model that can provide insight into the correlations of structure to concentration of carriers, concentration of polarized electrons, for example, and decouple it from the probe interaction would be very valuable."

In the electron microscopy arena, even TEMs now get all sorts of diffraction patterns in some case. Modeling is a must to understand whether this originates from the actual structure of the material itself or if it is an artifact of the electron beam being diffracted by the crystalline structure of a nanowire, for instance.

Garner's message is simple: The semiconductor metrology community must collaborate more closely with government and university researchers to develop models that will enable the decoupling of sample probe interactions at the nanometer scale for both structure and properties (including charge), and push the limits of resolution for structure and properties.