Introduction

Lithography is a very important tool which when used correctly is a great method for precision micro/nano machining. There are two main varieties of lithographic techniques, parallel and serial. In a parallel process, such as ultra-violet microlithography, a mask or reticle is illuminated in front of the specimen being exposed. This is done so that the entire specimen is exposed at the same time. In serial lithography, such as electron beam lithography, the specimen is exposed over a period of time as an electron beam is scanned deterministically across it. Once the specimen has been exposed, it must be developed, coated, and lifted until the final pattern is complete.

Goal

The goal of this project was to use electron beam lithography to create sub-micron structures in polymethylmethacralate (PMMA). This is done so that they could then be coated with a gold-palladium alloy, and the structures would go through a lift off procedure such that the remaining structure be PMMA free.

Specimen Preparation

The substrate used in this project was silicon. This was chosen because silicon wafers are relatively cheap, and more conductive then glass (no charging problems). In order to coat the substrate the wafer was sonicated for half an hour in methanol. This sonication is done to remove any extraneous particles. If these particles are not removed, the result is a non-uniform PMMA coating. The PMMA used for this project had a molecular weight of 495k, and was dissolved in a an anisole solution, 4% PMMA by weight (Solution: A4). This solution was then spun onto the silicon wafer using a spin coating machine. The solution was measured with a micropipette and spun at 3000 rpm's for 90 seconds. This combination of parameters should yield about a 200nm coating, as seen in the graph below. If the sample was not properly cleaned in the PMMA spinning procedure will produce a"comet" as seen below.

In order to use the Nanometer Pattern Generation System (NPGS), that deterministically moved the beam during exposure, a pattern must be created to draw. This was done using DesignCAD Express 16. Using this software a .dc2 file was saved and loaded with the NPGS software. Using this .dc2 file a run file is created, in which the SEM magnification and beam dwell times are specified. Once the specimen is prepared and the run file is created, the specimen can now be loaded into the SEM sample chamber. After the chamber is pumped down, the electron beam must be focused and corrected for astigmatism. Once this is done the beam must be turned off to prevent premature exposure of the resist, this is accomplished by turning on the beam blanker. Next the sample must be moved so that the desired target area is located in the beam path. Before the run file can be executed, SEM must be placed in to external control mode. After the run file is finished executing the sample may be moved to execute another run file.

Development

After the sample has be exposed, it must be developed. Since PMMA is a positive resist, the long PMMA chains that get exposed are broken into smaller chains. It is this exposed region that will be removed by the developer. The developer used was a 2:1 isopropyl alcohol to methyl ethyl ketone (MEK). The developing times tried ranged from three to seventy-five seconds.

Coating

Two different coating methods were proposed for use. First sputter coating was used. Sputter coating is a high vacuum, omni-directional coating technique. This makes it difficult to lift-off the remaining PMMA if there is no undercut. This undercut should be achievable with a dual layer of PMMA. By having a higher molecular weight PMMA layer on top of a low molecular weight PMMA a "dove tail-esc" undercut is created. Evaporation on the other hand is a directional coating technique. This means that lift-off should be able to be achieved without an actual undercut. The metal used for sputtering was a gold-palladium alloy, whereas for evaporation, aluminum was used. The reason a directional coating technique allows for lift-off is that the coating direction can be oriented such that there is a gap in the metal coating in the shadow of the microstructure. This gap will allow the stripper to attack the remaining PMMA. Unfortuately due to time constraints evaporation was not able to be done on any samples.

Results

Spin Coating

Depending on the size of the sample being spin coated, the amount of PMMA used was varied. The majority of the samples that were done were coated evenly throughout the center of the sample. Due to edge effects, the outside of the sample was unusable since the PMMA would bunch up, and an uneven thick layer was produced as seen below. One of the samples spun was not properly cleaned since a "comet" was produced.

Pattern Writing (Post Development & Pre Coating)

Linear Gratings

Circular Gratings

TL Lithography Test Pattern

Pattern Writing (Post Sputter Coating)

Linear Gratings

Circular Gratings

TL Lithography Test Pattern

 

Pattern Lift-Off

Linear Gratings

Circular Gratings

TL Lithography Test Pattern

Discussion

The majority of the linear gratings that were produced had been driven negative. Normally PMMA is a positive resist; this means that the area under exposure is removed by the developer since the polymer chains are broken up. If too much dosage is applied (~beam current and dwell time) the resist can be driven negative. When this happens, the polymer chains start to link together, which makes them unable to be removed by the developer. This process can be seen in the sputter coated TL sample. In the upper left, the proper dosage was applied, and each successive pattern has a slightly longer dwell time (greater doseage). As the dwell time is increased you can see how the pattern goes from usable to nonexistent, to negatively driven. If a sample is negatively driven, it makes it practically impossible to lift off. This is because the metal coating is not in direct contact with the substrate, and since no gap is in the coating the developer cannot attack the PMMA. Even though some of the TL pattern wern't negatively driven, the lift off was unsuccessful. After an hour of sonication in acetone, it was placed in chlorobenzene for a few minutes. Even after this the lift off would happen. On the Circular gratings however, the lift off started to happen in the middle. After more sonication however complete lift off was still unsuccessful.

Conclusions

It was difficult to try to find the proper dose and development time for each of the different samples. Now that the specimen current has been recorded for the patterns written along with what development times worked, proper conditions should be determined quicker for new patterns. In order to get proper lift off, a gap must be made in the metal coating. This may either be done by using two molecular weights of PMMA to get an undercut or through directional coating such as evaporation.

Interesting Lithographic Mishaps

I'm not too sure what happened here. It should be a linear grating with 200nm period and 50 percent duty cycle.

The rings of this circular grating were some how ejected out.