Alignment
If you’re only interested in writing a single-layer pattern on a wafer, then you can safely ignore this entire section. However, i you need to write patterns that align to existing patterns on a wafer, then you’ll need to master this section as well.
The JEOL ebeam system can align one pattern layer to prior pattern layers with an accuracy typically well below 50 nm, and for good conditions, as good as 10-20 nm (mean + 3*sigma). However, getting good alignment does require careful planning of your design and process.
Overview of E-Beam Alignment
First, some background:
The only method to find alignment marks on a wafer is to use the primary electron beam. There is no other imaging system available to "see" the alignment marks. This has several implications:
- Alignment marks need to provide significant contrast in back-scattered electron imaging mode
- Wherever you look for an alignment mark, you expose the resist
- You need to tell the machine where to look for the marks, and you need to be pretty close on this.
Because of the first requirement, you need to spend some planning and possibly some experimentation, just figuring out what marks will work with your process. Marks that look good in an optical alignment system might be completely invisible to 100 KeV backscattered electrons. For example, an alignment mark buried beneath a layer of silicon dioxide makes a fine usable alignment mark for optical lithography, but for even modestly thick oxide films are not usable for EBL, because the oxide film can significantly reduce the contrast from the underlying alignment mark.
Sometimes, you may even have to add an entire process step just to create alignment marks visible to the e-beam. In other processes, you may have to add an additional process step just to clear extraneous material from intervening layers that has covered alignment marks. These are not unusual circumstances; they're often part of the price of getting the extremely good resolution and alignment of e-beam lithography.
Like most every other aspect of e-beam litho, there is some considerable flexibility in the details of your alignment scheme. There are trade-offs in time and complexity versus alignment results. At WNF, we really don't have a reliable way to accurately or precisely measure the overlay that results from a specific alignment scheme -- this generally requires expensive, dedicated equipment that we do not have. The best we can do is to design-in some optical measurement features we which can use to manually estimate alignment, but in general, these are at best cursory checks -- few people have the patience or endurance to manually read enough alignment verniers to adequately characterize an alignment process. I've put quite a bit of work into developing some overlay metrology capability, which can produce very accurate and repeatable measurements for a limited case of materials and thicknesses.
The pages here about alignment include:
Alignment is complicated, and encompasses the entire process: CAD, pattern processing, calibration, jobfile setup, and exposure.
Mark Design - Discussion of CAD design for marks, both the marks themselves and their placement in your design. Also discusses the importance of keeping your pattern origins and extents in mind throughout the process.
Example CAD - Some suggested CAD designs for usable alignment marks, both created by EBL and by optical lithography. Includes a link to a GDS file containing these designs that you can incorporate into your own patterns.
Jobfile Commands - Describes the commands in your SDF & JDF files for implementing alignment.
Calibration Subprograms used for alignment: SETWFR, CHIPAL and MDRG.
Parameter Setup - Describes the steps necessary to define and adjust parameters to successfully locate alignment marks.
Setting up Video Levels
Mark Scan Parameters
MARKS - utility for saving mark parameters