BACKGROUND:

"Lift-off" is a simple, easy method for patterning films that are deposited. A pattern is defined on a substrate using photoresist. A film, usually metallic, is blanket-deposited all over the substrate, covering the photoresist and areas in which the photoresist has been cleared. During the actual lifting-off, the photoresist under the film is removed with solvent, taking the film with it, and leaving only the film which was deposited directly on the substrate.

Depending on the type of lift-off process used, patterns can be defined with extremely high fidelity and for very fine geometries. Lift-off, for example, is the process of choice for patterning e-beam written metal lines. Because film sticks only where photoresist is cleared, the defect modes are opposite what one might expect for etching films (for example, particles lead to opens, scratches lead to shorts in metal lift-off.)

Any deposited film can be lifted-off, provided:

1. During film deposition, the substrate does not reach temperatures high enough to burn the photoresist.
2. The film quality is not absolutely critical. Photoresist will outgas very slightly in vacuum systems, which may adversely affect the quality of the deposited film.
3. Adhesion of the deposited film on the substrate is very good.
4. The film can be easily wetted by the solvent.
5. The film is thin enough and/or grainy enough to allow solvent to seep underneath.
6. The film is not elastic and is thin and/or brittle enough to tear along adhesion lines.

There are three basic ways in which lift-off can be performed: 1. Standard photoresist processing; 2. LOL 2000 processing; 3. Surface-modified photoresist processing. The method you choose will depend on your process requirements. For more info about lift-off processes, check out chapter 6.1.7 of Micromachined Transducers Sourcebook by Greg Kovacs. A good website (which also lists primary references) is the Swedish Nanometer Lab.

STANDARD PHOTORESIST PROCESSING:

This is the easiest method, because it involves only one mask step and the photolithography is completely standard. The main disadvantage of this method is that film is deposited on the sidewall of the photoresist and will generally continue to adhere to the substrate following resist removal. This sidewall may peel off in subsequent processing, resulting in particulates and shorts, or it may flop over and interfere with etches or depositions that follow.

Prior to film deposition, particularly for sputtering or evaporation processes, a post-develop bake is recommended. This will drive off excess solvent so that there will be less outgassing during the film deposition. However, bake should not too long or at too high a temperature, otherwise resist will reflow slightly.

The film should be deposited as usual. Lift-off can be accomplished by immersing in acetone. The length of time for lift-off will depend on the film quality (generally, the higher the film quality, the more impermeable it is and the longer it will take to lift-off.) Depending on how robust the film and substrate are, sidewalls from deposited film can be removed using a gentle swipe of a clean-room swab or a directed stream of acetone from a squeeze bottle. As a rule, keep the substrate immersed in acetone until all the film has been lifted-off and there are no traces of film particulates -- once particles dry on the substrate, they are notoriously difficult to remove.

a.) Substrate with AZ3612 resist. b.) Following expose and develop. c.) Following film deposition. d.) Following lift-off. e.) Following mechanical "scrub".

*Modification of Standard Photoresist Processing (from Yahong Yao, useful if feature size is not critical):

• Photoresist (PR): 1.6uµm (AZ3612)
• Over expose to get a negative slope profile (exposure time longer than 5s)
• PR descum
• Evaporate metal
• Remove PR by 1165

LOL2000 DUAL LAYER RESIST PROCESSING:

LOL2000 is an inert, non-UV-sensitive polymer, which can be etched with most standard developers. The LOL2000 is first spun on the wafer and baked, then standard photoresist, such as AZ3612, is spun on and baked. The AZ3612 is exposed as usual, and then the substrate is developed. The standard LDD26W developer will clear the exposed AZ3612 areas, but will also "etch" away the LOL2000, leading to undercutting of the photoresist. It is this overhang that prevents sidewall deposition of film. The resist can be lifted-off using acetone or a resist removal solvent (such as Microposit 1165). Acetone will not completely remove LOL2000, so additional polymer removal measures should be taken. Because the overhang profile is caused by undercutting the resist, care must be taken to avoid completely undercutting (and lifting off) very narrow geometries.

a.) Substrate with underlying LOL2000 layer (purple) and overlying layer of AZ3612 (brown.) b.) Area of AZ3612 resist exposed to UV (light purple). c.) Following develop in standard developer. The developer will etch the underlying LOL2000. d.) Following directional sputter film deposition. e.) Following lift-off and clean.

The procedure for LOL2000 lift-off processing is as follows:

• Make sure to obtain and use dedicated LOL2000 boats. These are labeled and stored near the Litho cassettes. Please do not use standard Litho cassettes for handling LOL2000 coated wafers -- this polymer will contaminate the cassettes (it is not removed with Acetone).
• Singe for 30 minutes at 150°C (unless your wafers came out of a furnace or deposition system within the previous one hour).
• Prime with HMDS before LOL2000 (not absolutely necessary, but results are consistently better.)
• Apply LOL2000 on the Headway ONLY at 3000 rpm's for 60 seconds (=2000Å). There is a spin speed chart in the notebook in litho (range 1300Å to 3500Å)
• Bake for 5 minutes on a 150°C or 170°C hotplate, or 30 minutes oven bake (for oven see Mahnaz Mansourpour or Uli Thumser). There is a chart on the effects of bake temp and time on dissolution rates
• Coat with photoresist, you may use the SVG coater. AZ3612 will work fine. If your devices have topography, use at least 1.6 microns of photoresist to ensure coverage.
• Bake for 60 seconds on a 90°C hotplate.
• Expose with standard exposure times for the photoresist used.
• Develop for 60 seconds in LDD26W, you may use the SVG developer track.
• Bake 60 seconds on a 110°C hotplate
• Lift-off can be done using either:
  • Acetone. However, Acetone alone does not completely remove LOL2000 polymer. Additional clean should be done using O2 plasma ash, or photoresist developer, or Microposit Remover 1165.
  • Ultrasonic immersion in Microposit Remover 1165 at 50°C. This should be done at the solvent bench ultrasonic bath. Microposit waste should be collected locally in the labeled carboy at this wet bench, according to chemical waste handling procedures. Carefully monitor the temperature, as the flash point of Microposit 1165 is around 85°C. If you choose to use PRS1000 or SVC127 resist strippers to do the final clean, make sure your substrate is completely dry before immersing in these solutions (trace moisture will result in highly corrosive conditions, which may etch your metal.)

SURFACE-MODIFIED RESIST PROCESSING:

The top surface of the photoresist can be chemically modified to develop at a slower rate than the underlying resist. The result is something similar to the LOL2000 process, where there is undercutting of the top layer. This can be accomplished by soaking in chlorobenzene (the traditional method -- for more info, check out the Berkeley Microlab website on the Chlorobenzene process or the SNL process) or toluene. Basically, the standard photoresist procedure is used, but the wafer is soaked in chlorobenzene (for 10 minutes, typically) or toluene (for 1 minute, usually) either prior to UV-expose or just after, but prior to develop. In either case, the wafer is blown dry and baked briefly. Following standard develop, the result should be an undercut profile (similar to figure c., above.)