Home > Nanotechnology Columns > UAlbany College of Nanoscale Science and Engineering > Critical Dimension Environmental Scanning Electron Microscopy
Critical dimension (CD) metrology is one of the most critical enabling technologies in semiconductor manufacturing. Much media attention is devoted to the ever shrinking feature sizes of microelectronic devices, as this parameter heavily influences speed of operation and power demands.
May 16th, 2007
Critical Dimension Environmental Scanning Electron Microscopy
Critical dimension (CD) metrology is one of the most critical enabling technologies in semiconductor manufacturing. Much media attention is devoted to the ever shrinking feature sizes of microelectronic devices, as this parameter heavily influences speed of operation and power demands. Less appreciated is the fact that while the dimensions are on the nanometer scale, the manufacturing tolerances must be far smaller. This places great demands on the quality assurance assessment techniques.
Up until now, special purpose high vacuum scanning electron microscopes (SEMs) have served admirably, but conventional technology is reaching its limits. In a nutshell, the problem is that an electron microscope directs a beam of electrons at the material being examined. This beam delivers a current to the specimen, just as if a wire were connected to it. However, many of the materials that need to be examined are insulating, so if the incident beam current is not exactly balanced by the emission currents, excess charge accumulates. This built up charge has two main consequences. First, a surface potential develops which can deflect the beam. Knowing the beam position precisely is essential to making accurate CD measurements: random deflections even as small as an Angstrom are unacceptable. Second, a surface potential significantly affects the emission of secondary electrons and therefore the contrast in the image. If enough charge accumulates in a region of the sample, sporadic emission flares can result. Sharp edges tend to have higher emissivities than flat surfaces, so edge highlights are used to delineate structures. Clearly, sporadic emission flares would be very detrimental when trying to identify features. Charging can be reduced by lowering the beam current, but doing that also reduces the signal-to-noise ratio. Unfortunately, under most circumstances the edge highlights get drowned in the beam noise long before the current is low enough to avoid charging effects.
One successful way to deal with the charging issue is through a relatively new form of scanning electron microscopy that places the sample in an environment of a low pressure of gas (around 100 Pa), hence, the term environmental SEM (ESEM). Through interactions with high and low energy electrons, the gas in an ESEM becomes ionized, which allows it to remove excess electronic charge from insulating samples. Additionally, if a positively biased anode is placed near the specimen surface, an ionization gas cascade is initiated by secondary electrons leaving the surface. The cascade results in a current which is an amplification of the secondary electron emissions. This current can be used as a signal for image formation. The upshot is that an ESEM can be used to produce high quality, high resolution secondary electron images of insulator surfaces.
CD-ESEM would appear to be an ideal candidate for next-generation metrology. Indeed, extremely high quality images of very difficult samples (such as chrome-on-quartz photolithographic masks) have been obtained. However, since the image formation process is very different from the high vacuum case, it needs to be understood in order to interpret the images correctly. That is, how does the image of a feature correspond to the actual shape of the feature? Our research team at the College of Nanoscale Science and Engineering is tackling this issue. In high vacuum systems, the characteristics of the electron probe striking the specimen can be known with a high degree of confidence: the electron optics define the probe diameter, convergence angle, and beam current. The secondary electron emission from any given point is determined by the beam energy, angle of incidence with the surface, and the material type. The emissions are collected by a scintillation detector, and amplified by a photomultiplier tube. The noise characteristics of each process are well understood and can be measured and tabulated for a given instrument. In short, it is relatively straightforward to predict the picture that would result from imaging a structure under a given set of instrument conditions.
In an ESEM, all of the above processes are present, but there are some additional complications. First, the probe shape is altered through interactions with the gas before it strikes the sample surface. Although surprisingly this does not degrade the resolution, it does affect the background level and signal-to-noise ratio. Additionally, the amplification and noise characteristics of the gas ionization cascade depend on the exact operating conditions as well as the type of gas used. Our team is using a combination of experimental measurements, theory and Monte Carlo simulations to develop an understanding of all these processes. The goal of our work is to produce analytic descriptions of each stage in the image formation process, showing the dependencies on instrument operating conditions and sample characteristics. Ultimately, we expect that the ESEM will emerge as a powerful tool for CD-Metrology.