Can High-Pressure Microscopy Save Mask Metrology?

The 2001 International Technology Roadmap signaled that "Mask linewidth controllability fails to meet the requirements of the chipmakers." This technology shortfall is due to the increasing complexity of masks and reticles (including phase-shift masks and OPC), as well as problems that were not critical at larger device dimensions — such as surface roughness, line-edge roughness and mask error enhancement.

Fortunately, there is a developing technology — in fact, a disruptive technology in SEM metrology — that may be ideally suited to meet the needs of the mask industry, said Michael Postek, group leader of the Nano-scale Metrology Group of the National Institute of Standards and Technology (NIST, Gaithersburg, Md.). The platform is an environmental SEM that has largely been used on biological specimens, not on semiconductors. High-pressure or environmental SEM operation employs a gaseous environment to neutralize sample charging, which causes measurement error on the order of several nanometers. Sample charging depends on the sample, instrument and operating mode, making it very difficult to control and model, Postek said. Models, typically based on Monte Carlo simulation, exist for photoresist, wafer and polysilicon features that approach 1 nm accuracy, but are just being developed for reticles.

Images taken in a state-of-the-art environmental SEM from FEI (Beaverton, Ore.) with high-pressure operation (0.8 Torr) show the better contrast and information content at higher accelerating voltages (3.0 keV in a, 7.0 keV in b, 5.0 keV in c, 10.0 keV in d). The lower SEMs compare a tilted image with 1.2 Torr pressure and 10.0 keV accelerating voltage (e) to top-down imaging at 0.7 Torr and 5 keV (f). (Source: NIST)

In SEMs, sample charging causes significant, non-reproducible error. Charging causes a deceleration of the electron beam that changes the landing energy, and it also deflects the beam. In practice, charge can be neutralized by sample coating, charge balancing, backscattered electron imaging or high-pressure microscopy. Commonly used in lab environments, coating with a conducting material, such as gold, is not only destructive and therefore incompatible with production reticles, but it destroys contrast. Another approach involves balancing the charge by adjusting the accelerating voltage and optimizing measurement conditions such as tilt, extraction field, scanning speed and final lens configuration. However, reducing the voltage also decreases image resolution, and details such as edge roughness can be lost. Charge can also be minimized by filtering out secondary electrons and collecting high-energy backscattered electrons in the SEM. This method reduces the appearance of charge, but does not eliminate it.

One reason Postek favors high-pressure microscopy is its potential to eliminate sample charging, and the challenge of modeling it. In high-pressure microscopy, the injection of air as little as 0.15 Torr (20 Pa) into the chamber has been shown to reduce the charging potential of an insulator by an order of magnitude. At pressure levels of 1.2 Torr (160 Pa) and 0.8 Torr (107 Pa), mask images can be taken at high accelerating voltages for better resolution and definition of edge detail (Figure).

Postek's vision for metrology's future involves combining all of the components needed to accurately measure a structure in an aggregate program embedded in the SEM. "Model-based metrology" would include electron-beam interaction modeling, instrument response modeling, the charging model (if required) and other necessary components into a single measurement technique. Such an approach has already improved the precision of CD-SEMs by 3×.

For additional information on yield management, go to www.semiconductor.net/yield.