CAD Data Formats


Text Formats


For certain types of patterns, such as simple repeating patterns such as diffraction gratings or arrays of square or circular dots, or patterns best described by algorithmic means, there are several simple ASCII test formats that can be used to generate e-beam patterns.

Pros


  • Very easy to generate simple patterns -- we can create large gratings or dot arrays with less than 10 lines of ASCII text
  • Have precise control, down to the single-pixel level of e-beam exposure
  • Easy to write code in any high-level language to generate complex shapes, such as fractal designs, or diffractive lens elements
  • Very easy to modify simple arrays, for example, to change periodicity or element size

Cons


  • Files can grow very large for certain types of patterns
  • Manually editing large files is cumbersome
  • Generally not appropriate for complex devices such as electronics

There are two common text file formats available.

 JEOL J01  

This is a firmly standardized, documented format from JEOL, and they provide the conversion programs to convert the data directly into JEOL-exposeable files. It is however, a pretty bare-bones format, so might be appropriate only for simple elements such as diffraction gratings. It is also showing its age a bit; everything must be designed in grid units of 100 nm (it is possible to design in a different grid, such as 1 nm increments, but then you must remember to always include a 100:1 scaling factor when you convert to J52 format.) Full format specification can be found here. (login required).

CTXT / LTXT / TXL


TXL is an ASCII-based text layout format that can be used to code up simple patterns, and is a good choice for many sorts of algorithmic patterns, such as gratings or simple arrays. If you’re familiar with the GDS-II pattern data format, the terminology here will seem familiar -- boundaries, paths, structures and references all behave just like in GDS-II. Although not a standard, this is fairly common text file format. It is read and displayed by GenIsys, so it is easy to verify results. It has extensions for circles and arcs, so it can be nice to use for patterns such as regular arrays of dots (although these types of patterns have other issues such as run-time overheads that might cause you to make other choices...) If your project is best served by a text format, I’d recommend using LTXT as your format. I’ve used it for several projects with good success. LTXT is documented in the GenISys LayoutBEAMER Manual. as well as quick reference here

I’ve also developed a Python library to output LTXT format for algorithmic patterns. It supports most, but not yet quite all, features of LTXT, and I’ve used it successfully for several algorithmic patterns. So if you’re familiar with Python and want to use my Python TXL library, let me know and I’ll share it with you.

CAD Formats



GDS-II


Far and away the most common CAD file format for microfabrication of any complexity of GDS-II, (wikipedia) often just called gds. This is an industry-wide standard (although there are also various non-standardized “extensions” to the documented format that have varying degrees of support throughout the industry. I strongly recommend that if possible, you use GDS-II as your cad interchange format. Any serious microelectronics layout program will be capable of both importing and exporting GDS-II format data.

OASIS


A newer, rigidly-standardized format, touted as the successor to GDS-II, is called OASIS (wikipedia). In reality, it has not received much widespread acceptance; GDS-II remains the dominant file format in use. That said, in principle, we can support OASIS data format files, but I have no direct experience.

AutoCAD DXF and CIF


In principle, it is possible to work with pattern data in either of these data formats. However, the big problem with both formats is that they are not designed for microelectronics pattern data, and they are not strongly-standardized. These factors lead to many ambiguities in exactly how various elements in the design should be interpreted, with no standard to dictate the answer. So if you look at a DXF file in 4 different viewers, it’s quite possible that you’ll see 4 unique interpretations of some of the more specialized features within those formats. And no doubt, what our e-beam produces may or may not match any of those 4 interpretations. While it may be possible to use a DXF or CIF file as your data source, I strongly recommend you not do this. If you insist, at least try a small test file before you invest much time. It’s very easy to create files that won’t give you what you want from the e-beam, and that’s a waste of time for everyone.

If you ignore my advice and insist on using AutoCAD or DXF format, be aware that:
1) I will not waste time helping you troubleshoot pattern errors
2) I will not waste time helping you investigate pattern warning messages
3) Any and all exposures done will be charged. There will be no cancelled machine charges or free re-writes if there are problems.


Since a few folks still don’t believe me about why you shouldn’t use DXF, here’s a few quotes from other folks in this business:

“DXF format is bad. Do not use it.” Mike Rooks, Yale University, http://www.yinqe.yale.edu/layout_tutorial.pdf


“DXF is accepted by most mask houses but usually contains numerous errors on conversion and conversion fees may be added.” Bill Martin, Compugraphics https://snf.stanford.edu/SNF/processes/process-modules/photolithography/maskmaking/PhotomaskBasics-General.pdf/view





Bitmap Formats

Sorry, but bitmap formats are not typically acceptable data formats for the type of writing we are doing. There are a variety of reasons, both technical and historical, why this is true. There are some methods that we can use with which we may possibly convert a bitmap image to an exposable pattern file, but it’s generally not a very efficient way to draw data, and certainly not the first choice if you can avoid it. Talk to me if you need to expose a bitmap and we’ll see if we figure it out.

CAD Programs


There are even more CAD programs available than data formats. Many of these will work, but caution is recommended, because every program has at least some vagaries about how it works and how it draws various elements. It’s problematic when you CAD program displays your pattern in one way, but GenISys or the e-beam interprets your pattern data differently.

If you are in an established fabrication arena that has existing CAD, chances are, we can work with that. As long as it outputs GDS-II, we should be good, but talk with me about it before you invest much time. Examples of commercial CAD programs that should work just fine include: CADENCE, SYNOPSIS, L-Edit (Tanner), LayoutEditor or DW-2000. There are also some freeware or shareware CAD programs out there; I recommend caution, because you might just get what you paid for....

If you are a University of Washington student, staff or faculty doing university research, the WNF purchased and supports a site license for the full version of LayoutEditor. See more details on this program here. LayoutEditor is available for Mac, Windows, and Linux users, and has many very powerful features. Once you have the program, I've written an introductory tutorial, and there a multiple other tutorial documents out there on the web.

I’ve also had good luck using a Freeware CAD package called: KLAYOUT. It is very fast, even with large patterns, and is also available for Mac, Windows, or Linux. I have KLAYOUT installed on the ebeam workstation; talk with me if you’d like to use it there. You can read a bit about using KLAYOUT on this page. And I've also written an introductory CAD tutorial using KLayout.