Ebeam – Keep It Simple

Ebeam lithography is a very powerful technique to pattern your designs. It is also a fairly mature technique. Large body of knowledge and practical tricks exists to overcome many of the limitations of both process and tool native performance specifications. For anyone who is trying to create a single “perfect” device, it can be said without much of an exaggeration – success is only a question of patience and persistence. Think about it.

The ebeam tool exposes the pattern one pixel at a time, with a beam as small as 5nm in diameter. The actual size of the pattern in resist will depend on dose, and the dose can make that 5nm spot become smaller or bigger. The 5nm value is a value of a 2 sigma of the Gaussian beam. Small increase in the dose will make the pattern grow a bit, slightly lower dose will make it become a bit smaller. The control of the dose is very, very fine (one controls the dwell time of the beam, something that can be done with high accuracy). Thus, the user has essentially an atomic level control of the pattern image size.

Similar situation is with the pattern placement. The stage is controlled by the laser interferometer, and any position is thus know with better than 1nm. The repeatability of the tool overlay may well be only about 20-30nm. But what is to prevent the user to create several copies of his/her pattern and place them in a Vernier-like fashion on a wafer? With a fine enough gradation in spacing between the patterns, one of them will be in an exact position. Devices requiring nearly perfect (1nm or so) overlay accuracy have been successfully fabricated despite the native spec limitations of the tool, simply using this simple trick.

As much as it can be a powerful technique, ebeam lithography can easily become quite complex. The level of complexity being bounded only by user’s imagination. Perhaps the most important advice one can give to anyone who is just starting to learn ebeam lithography is this:


Start with the basic data fracturing and simple uncorrected exposure and develop process. Do not try from the start to implement all sorts of corrections and more advanced schemes, just because you have seen your older colleague using them, until you have scrutinized the lithographic results of your basic scheme and are able to see the deficiencies and imperfections. Proceed from there by adding one type of correction at a time, always carefully evaluating the results and making sure they agree with your mental image of what is going on. Once you achieve a good enough performance for your needs, stop there, and “lock” your process. Of course it can be made even more perfect, but each additional correction step comes with a cost (setup, calibrations, increased complexity, learning curve, write time…), and it is easy to become fascinated with “reaching lithographic perfection” and “loosing track of the level of lithographic quality actually needed” for the device physical performance.

Remember the old saying: “Every shortcut will only make your journey longer.” Do not proceed with your process development until you are confident you understand exactly where you are. It becomes nearly impossible to troubleshoot your lithography if you have not build the process step by step, and are thus not able to de-convolute the individual process steps and applied corrections contributions to the whole.

But once you determine that you need to improve something, do not hesitate and add the necessary step, new correction, more complex scheme. The beauty of ebeam lithography is that you will always immediately see your results, and most problems can be very quickly identified.