Microfabrication: Photolithography, Metalization etc.

To learn more about microfabrication processes look at http://en.wikipedia.org/wiki/Photolithography. The most important source of information is always the photoresist manufacture's data sheet (see the T: drive). There are also some good textbooks in the OSU library that cover all aspects of microfabrication:

• “Fundamentals of Microfabrication” 2nd Ed. by Madou (excellent level of detail)
• “The science and engineering of microelectronic fabrication” 2nd Ed. by Campbell (good overview)
• “Introduction to microelectronic fabrication” by Jaeger

For some philosophy on how to make a good recipe you might enjoy reading the zen of device making.

We use DesignCAD (2D version) for CAD design and for exporting .dxf files to the direct write laser lithography system. Previously we used Layout Editor and exported .gds files.

Chrome masks are patterned using the direct write lithography system on campus. Chrome masks allow us to reach the 3-4 micron resolution limit of the contact aligner. There are two write lenses (labeled 10 mm and 2 mm), with a 2 µm or 0.5 µm spot size respectively. For the 2 µm spot size a mask can be exposed at a rate of 1 square inch every 20 minutes regardless of the amount of exposed area. Using the 0.5 um spot size the write time is on the order of 1 square inch every 1.5 hrs. The write times are fairly insensitive to the complexity of the pattern data. However, large arrays of repetitive shapes can take up to 5 times longer to expose than normally expected. Cost: The chrome plate costs $12. Write time on the DWL costs about $50 per hour. Previously we sent CAD drawings to a mask making company. Commerically produced chrome masks cost $400 - 800 each.

Image of a chrome mask made with the DWL 2 µm spot size. The pink is chrome, the grey is transparent. More mask images.

For less demanding resolution, for example 10-20 micron minimum feature size, a printed transparency can be used. These masks cost about $60 each and are purchased from a company in Bandon. Now we prefer to make chrome masks in the DWL.

We are using shared equipment in John Wager's lab to process our chips. The lab is in Owen Hall 4th floor West Wing. You can look through the big windows to see the impressive equipment. The lab is run by Chris Tasker, chris@eecs.oregonstate.edu. To use equipment in this lab you must be trained by someone in our group and then certified. Rick Presley is our main contact for certification, presley@engr.orst.edu.

The most important source of processing information is always the photoresist manufacture's data sheet (see the T: drive). For more technical advice you can also call from the photoresist company, MicroChem, (the company that distributes Shipley products). We have worked hard to get good/reliable recipes - some of the trials are documented.

Contact Aligner Walk Thru

Check with Josh to see if these are current -

(updated 11-12-2009)

• 5 min prebake 115°C
• spin S1813 photoresist at 4000 rpm for 30 sec
• 3 min bake 115°C
• 6 sec exposure
• 40 sec develop
  • developer solution: 4 parts DI H20, 1 part MF-351 developer

(updated 11-12-2009)

• spin LOR3B photoresist at 2500 rpm for 45 sec
• 2 min bake 190°C
• spin S1813 photoresist at 4000 rpm for 45 sec
• 2 min bake 115°C
• 5 sec exposure
• 20 sec develop
  • Developing solution: CD-26
• deposit metal
• remove underlayer with mircoposit 1165 (located in Weniger 306 lab)

• Eric Sundholm's Recipe

Matt has documented that hot PR remover leaves less PR residue than any other method we have tried.

Matt and Ethan have used the ebeam system at CAMCOR (University of Oregon, 50 minute drive). A basic recipe is available.

The initial photolithography steps are done on 3” wafers. To grow nanotubes, however, we have to cleave the Si [1 0 0] wafer or ST-cut quartz into smaller pieces that will fit in our furnace. Matt is the expert when it comes to cleaving.

The only way to get truly square quartz pieces is to use a wafer dicing saw. Pallavi Dhagat has used American Precision Dicing Inc. The cost is a minimum of $150.

After doing photolithography, a thin layer of metal (tens of nanometers thick) is deposited on a nanotube chip. The interface between nanotube and metal is critical for device operation (for example, we do not store the chip in a gelpak before evaporation. This will lead to hydrocarbon contamination).

Our workhorse deposition system is the thermal evaporator in Janet Tate's lab, 4th floor Weniger Hall. We have used this system for chrome, gold and iron. Please see Matt Leyden for training. Always follow the standard operating procedure in a step by step fashion - missing one step can ruin your sample and the equipment. Matt and Josh renovated this evaporator in summer 2008.

Evaporation boats are made Tungstun and cost $6-8 each (R. D. Mathis). The boats should be kept clean and reused when possible.

Thin film deposition is monitored by a quartz crystal microbalance (QCM). The operator enters the density of the metal, the z-ratio of the metal and the tooling factor of the evaporator. Density and z-ratio can be looked up on a table. The tooling factor is a geometric factor, basically the ratio of the distance between source and QCM and the distance between the source and sample. When the QCM readings do not agree with AFM characterization of film thickness, users typically adjust the tooling factor.

For thermal evapoartion, metal is usually held in a boat made out of Tungsten (W) because the melting point of W is 3422 °C. It is possible to break the boat by heating or cooling too fast. It is also bad for the sample if the temperature inside the evaporator gets too hot.

This evaporation table gives useful advice about which metals can be thermally evaporated. Similar information is on the Kurt Lesker website. For example,

• Iron attacks W so thermal evaporation may not be possible.
• Ti outgases when first heated.
• Cr is available in rod form, so a boat is not necessary. Cr is a great sticking layer.
• Pt and Au are both suitable for thermal evporation in a W boat, these metals do not stick well to an oxide surface.

Sputtering is an alternative to thermal evaporation that we have not yet explored. The deposition rate should be more stable. Milo Koretski has a beautiful new sputtering system set up in the chemical engineering department which he used for CNT catalyst deposition.

John Wager's group has an ebeam evaporator. E-beam evaporation is especially useful for metals that are hard to control or hard to melt, like iron and molybdenum.

When making devices from aligned CNTs the unused tubes need to be etched. We currently use the O2 plasma in the Owen Hall clean room. This O2 plasma removes photoresist at a rate of about 100 nm/minute. The recipe from Rogers Nature Nanotech 2007 is 50 mTorr, 20 sccm O2, 30 W, 30 s.

More oxygen plasma details.

Other etching options