Calibrated AFM Offers High Accuracy, Repeatability

Aaron J. Hand

Atomic force microscopes (AFM) have come a long way since they were developed in the mid-1980s, with advances bringing more repeatability and stability. Engineers at the National Institute of Standards and Technology (NIST) are taking the technology to the next level, devising an AFM that is calibrated with an interferometer.

NIST's calibrated atomic force microscope (right) measured a silicon-on-insulator linewidth specimen (below) prepared by Mike Cresswell and Richard Allen of NIST and collaborators at Sandia National Laboratories. The line is about 500 nm wide, and the area imaged is about 4 × 8 µm. It is being used to compare linewidth measurement techniques with scanning electron microscopy, atomic force microscopy and electrical resistance.

"Early instruments had no inherent accuracy in their scales," said Ronald Dixson, a metrologist at NIST involved in the development of the calibrated AFM. A few more recent commercial designs and several AFMs used by national metrology institutes have integrated position sensors, often capacitance sensors, which make the instruments repeatable and reasonably stable, but they must still be calibrated. "What distinguishes our instrument from most others is that we're not relying on anyone else's calibration."

Displacement interferometry, with a 633-nm HeNe laser, is performed in all three axes. This is important because it is used as the ruler for measuring length. The fact that the wavelength of light is the de facto standard for length measurements makes interferometry a natural for accurate scale calibration, he said.

NIST's calibrated AFM makes very accurate, nanometer-scale measurements. It has a lateral measurement resolution of better than 1 nm and a vertical measurement resolution down to 0.04 Å. Step height measurement uncertainty is 0.4 percent of the measured interval, and pitch measurement uncertainty is about 0.1 percent.

The instrument's accuracy has much to do with the scanner, which has little tilt or off-axis motion, Dixson said. Special stages are used in the scan apparatus, but the trade-off is that the scanning range is only 50 µm. NIST hopes to have an X-Y stage with a 100-µm range and even less parasitic tilt by next summer.

Developers are still working to reach the instrument's maximum capabilities. It operates in a contact or intermittent-contact basis, and the group is also trying other force-sensing mechanisms.

Linewidth measurement capabilities also are progressing.

A major problem is the size and stability of the probe. If the tip is blunt or becomes damaged, it's less likely to affect pitch and step height measurements.

However, for width measurements, the width of the tip shows up as an error, he said, so the tip must be characterized and must remain constant to be minimized as a source of error.

NIST's goal is to provide an AFM calibration service for manufacturers of integrated circuits or other nanotechnology devices. That capability is at least a year away, after the technology has a history of accuracy, Dixson said. "We guard that status pretty carefully," he said.