NPGS: Lithography Setup
Practice running the SEM until good gold resolution images are obtained. If the images are not comparable with the best the microscope has ever produced, it is pointless to try to do fine lithography, although larger structures can be successfully exposed with sub-optimal SEM setup. See the SEM Optimization page for a discussion on optimizing the SEM.
Scribe several marks extending ~1 mm from the edge of the sample onto a PMMA (~150 nm thick, 950k MW) coated sample, and then mount to a sample holder. (If the ultimate goal is to use a bilayer resist or a very thick resist, it is still recommended to begin with a thin single layer. Once the skills are acquired to optimize the SEM well enough for fine lithography in a single layer resist, it will be reasonably easy to move on to other resist structures. However, starting with a more complicated resist can make the diagnosis of exposure problems more difficult.)
Optimize the SEM on a gold resolution standard which is mounted at a similar height as the sample. The optimization should be done at the beam current which will be used for lithography (typically between 5 pA and 50 pA for the finest patterns). The goal is to get the best possible optimization (least astigmatism) at a magnification* between 100,000x to 500,000x. The actual magnification used will depend on the SEM and the personal preference of the operator.
Move to the edge of the sample and do an initial focus using the fine Z of the SEM stage. Do not focus with the electronic focus of the SEM and do not try to adjust the astigmatism on the sample. Also, rotate the sample to make one edge parallel to the x or y stage axis. (The NPGS Global Correction mode can make this rotation unnecessary, however, in order for these steps to apply to manual systems, that feature is not discussed here. Also, even with a fully automated SEM and stage, it is often best to start in a manual mode during the initial learning cycle, and then add the more advanced features once the basics are understood.) This alignment makes it easier to find the patterns after development.
Find one of the scratches made previously on the surface and follow the scratch to the end away from the edge of the sample. (If the stage flatness has been characterized or when using the NPGS X-Y-Focus feature, the scratches are unnecessary. However, when first learning lithography, the scratches are a simple way to minimize focus problems that may be caused by a tilted sample.)
Do a final focus using the fine Z of the stage (with a field size of ~3 microns or smaller) and then check the focus using the electronic fine focus of the SEM. Do the focusing on the smallest feature that can be seen which is directly on the PMMA. For example, do not focus in the bottom of the scratch or on top of the ridge at the edges of the scratch. Do not adjust the stigmator settings of the SEM.
Optionally, make "Contamination Spots" by leaving the beam stationary on the surface of the PMMA for ~1 to ~30 seconds in order to confirm that the beam is well focused and well astigmated. If not, either reoptimize on the gold standard or try to reoptimize using the contamination spots themselves. Note that the spots will degrade rapidly when viewed at high magnification, consequently, they can be difficult to use for optimizing the astigmatism.
Check which directions on the stage coordinates will move past the end of the scratch.
With the end of the scratch still visible in the imaging area, set the blanker (if available) and SEM to lithography mode. Now**, write "sample0", which will write an array of "wheel" patterns at different doses. This exposure may overlap part of the scratch, but that is the intention. After processing, this pattern will be easily seen right at the end of the scratch, and if it is not seen, you will know not to spend hours looking anywhere else. Also, any area where you have focused on the PMMA at high magnification will have an exposed rectangle (or a square on a few SEM models). Consequently, if these rectangles (squares) are not found, there is something very wrong with the resist or the processing.
Move the stage 0.1 mm away from the end of the scratch and repeat "sample0". (Moving the stage too far from the focus location is a common problem when first learning to do lithography. The flatness of a sample can be checked by focusing at points around the perimeter. After the SEM optimization has been mastered, the issue of sample flatness can be addressed using the NPGS X-Y-Focus mode.)
Repeat the stage movement and write other patterns as desired. The "sample3" run file will write the "fractal" star pattern, which is a good demonstration of complex filled polygons. Whenever writing patterns, be sure that the microscope magnification and beam current match the values entered into the lithography software.
Continue to move the stage and write whatever patterns are desired. When finished writing, check the beam current using the picoammeter to see if it has drifted significantly. (Note that NPGS can have a single run file that will automatically move an automated stage, set parameters on a digital SEM, automatically align to existing registration marks, and write patterns aligned to the registration marks, however, this example just covers the basics which will apply to any system.)
Now, return to the gold standard, randomly change the SEM settings, then completely reoptimize the SEM, and repeat the steps above. This will allow the consistency of your SEM optimization to be checked. With practice, you will be able to bring the SEM from any initial state (within reason) to good optimization within about 15 minutes or so.
After developing the sample, it is recommended to first observe the sample in an optical microscope. This will allow you to see if the patterns are there or not, and with practice, you will be able to do an initial assessment of the quality of the patterns, even though the linewidths are smaller than the wavelength of visible light. (Quite simply, a fine pattern will never be clearly resolved in an optical scope, but will look different if the lines are narrow and smooth versus wide and/or rough.) This confirmation that the patterns exist can also save hours of time looking in the SEM in the cases where some mistake (such as setting the blanker to the beam off state, instead of external control) has resulted in no exposures to be found.
Sputter coat the sample with 10 to 20 nm of Au or AuPd and look at the sample in the SEM without doing liftoff. This is recommended because if the exposure was not done properly and everything is underexposed, liftoff will remove all signs of the patterns, while even underexposed patterns can be seen before liftoff and can give useful insight into the optimization of the SEM during the pattern writing. (If sputtering is not available, evaporated metal can be used.)
*Most numerical references to SEM magnification will assume that the SEM is calibrated for a Polaroid output (which is the only option for older models). However, for PC based SEMs, the displayed magnification number may refer to different output dimensions, such as for a Polaroid output, the screen display, or a videoprinter printout. When the displayed SEM magnification is calibrated for a videoprinter, the values will be similar to a Polaroid calibration (since the two output images are nearly the same size), but when it is calibrated for the SEM display, the magnification value will typically be about 2 to 2.5 times higher (since the typical screen display size is about 2 to 2.5 times larger than a Polaroid image).
Consequently, for the same scanned area, an SEM with the magnification calibrated for the display screen size will show a larger magnification value that many people think indicates a higher "performance" than an SEM with the magnification calibrated for a Polaroid or videoprinter output. In summary, while magnification values were a reasonable indicator of performance in the past when all SEMs were calibrated to Polaroids, the introduction of the variable magnification calibration options in newer PC based SEMs has made magnification less useful as an indicator of SEM performance and anyone comparing "magnification" values between microscopes should also compare the calibration mode used for the magnification.
**If the microscope has a digital interface for column parameters, such as magnification, and an interface for stage control, a single NPGS run file can automatically set the magnification and move the stage for many unique patterns. Consequently, automated exposure of an almost unlimited number of patterns can be done, however, the instructions here describe a simple method that moves the stage manually.