New Probe Preps AFM for Inline Metrology
Alexander E. Braun, Senior Editor -- Semiconductor International, 5/13/2008 8:25:00 AM
Researchers at the Georgia Institute of Technology (Atlanta) have developed atomic force microscopy (AFM) probes that can quickly and simultaneously measure material properties such as adhesion, stiffness, elasticity and viscosity, in addition to the traditional topography scan. This development promises to move AFM from an offline to an inline metrology and inspection modality.
The Georgia Tech research group is headed by Professor Levent Degertekin of the university’s George W. Woodruff School of Mechanical Engineering. The research for this novel force sensing integrated readout and active tip (FIRAT) was funded by the National Institutes of Health (Bethesda, Md.) and the National Science Foundation (NSF, Arlington, Va.).
Traditional AFM systems scan a sample surface using a cantilever with a sharp tip a few nanometers in diameter at the end. An optical beam is bounced off of the tip to measure the cantilever’s deflection as the tip moves over the surface and interacts with the material being analyzed to determine the surface topography.
 |
| 1. AFM works much like an old-time phonograph. A probe tip attached to a flexible cantilever tracks the surface irregularities of the sample, and the signals produced are used to generate a 3-D image. (Source: IFR) |
Instead of a cantilever, the new probe uses a drum-like membrane from which the tip extends to scan the sample. In one scanning mode, as the tip moves about a surface, it lightly taps it, gathering precise information about the tip’s position and the forces acting on it, while sensing the material’s shape, stiffness and adhesion properties. An output signal is generated only when there is an interaction force on the probe. This means interaction forces can be measured during each tip’s tap with high resolution and without any background signal. Degertekin, who has previously researched membrane-based devices for medical ultrasound imaging — heart arteries, intravascular imaging — decided to investigate whether some of these techniques could be adapted to AFM. The result is the FIRAT. Degertekin reasoned that if he placed a tip on a membrane that can be moved up and down, and its motion can be detected using optical interferometry, it satisfies the requirement of moving the AFM tip very rapidly, is not limited by a piezoactuator, and simultaneously can very accurately detect the tip’s motion using the optical interferometer built into the device.
 |
| 2. The AFM probe designed by the Georgia Tech group. The probe tip is attached to a drum-like membrane from which the tip extends to scan the sample. (Source: Levent Degertekin) |
“We fabricated the probe system using surface micromachining; meaning that the mechanical structure is a couple of microns away from the surface, whereas in a regular AFM, the cantilever wafer sits in air,” Degertekin said. “Thus, by adjusting the chip’s distance from the surface, we can also adjust its dynamic behavior. We can set it so that it has no resonance or give it a resonance behavior of a 10 or 50 quality factor.”
Another advantage is that for material characterization purposes, the stiffness of the membranes can be electrically tuned by changing the spring constant of the FIRAT probe. This provides the capability to use the same probe to identify the mechanical properties of different samples. For instance, if a tip is produced with a stiffness of 40, by applying a bias it can be made 40, 5 or 4 — this becomes yet another adjustable parameter. In general, there is unprecedented control over the tip’s behavior in terms of speed or how it vibrates, because the actuator acts directly on it, unlike a regular cantilever version, which provides direct access to the substrate, but no direct control over the tip’s motion. This gives several extra “knobs” to fine-tune the AFM’s operation.
Degertekin estimates that this development has the effect of speeding up an AFM by two orders of magnitude. “We’re able to get a frame per second; therefore, the limitation now isn’t how fast one can move the tip, but how fast one can do the x-y scanning. Right now, our device can support up to 40 frames per second worth of imaging. However, like with other AFMs, when the x-y scanner is moved left and right at a rate of more than 60 Hz, it starts ringing because it has not been designed for such fast motion. Then the x-y motion begins coupling to the z motion, producing imaging artifacts. This is why we cannot move the speed up further. Others are working to improve the scanner, however,” he said.
These probes could make possible the measurement of overlay errors because, in the same scan, it becomes possible to get topography as well as material property information. Further down the line, the actuators will be able to move the tip in three dimensions. “Then, if you have a suitable tip, you can tap a trench’s sidewalls without scanning the whole substrate,” Degertekin said. “Because it is also possible to bias one side of the beam more than the other, it is possible to tilt the tip and have excursions in x, y and z without actually physically scanning the stages.”
Although an AFM cantilever is produced through a micromachining process, it is not a complete MEMS device with actuation. Degertekin uses more MEMS technology integrated in the AFM probe. The issue the research group now faces is whether these tips can be fabricated in volume with the actuators — the FIRAT devices — required for CD metrology. This requires tips that can easily go in and out of trenches. Work is ongoing to solve these hurdles.
