Final Project - Lithography Exposure Tool

About My Final Project

For my final project, I have decided to make a crude photolithography exposure tool. Photolithography is the most important and critically controlled step in semiconductor manufacturing as it is the process that determines the actual size of the features that get created. I have used many direct write (serially draws a pattern like 3D printing) and stepper (projects an already existing pattern) exposure tools. They each have their own benefits and drawbacks, but when it comes to passion projects these tools are extremely cost prohibitive and require a lot of support steps. I have some projects in mind that I would love to do with the use of photolithography, and I wouldn't even need a top-of-the-line high-resolution tightly-controlled machine. I could very likely get by with a very simple near-contact exposure tool and crude printed masks on transparent film. As such, I have decided to make the photolithography exposure tool I would need to be able to do my own basic photolithography processing!

Sections:

• Initial Planning and Sketching
• Creating a Proof of Concept Model
• Refining the Design and Creating a Realistic Assembly
• Printed Parts
• Laser Cutting Acrylic Parts
• Water Jet Cutting the Base Plate
• CNC Milling Aluminum Parts

Initial Planning and Sketching

The below images are of the initial sketches that I created when starting to think about the different components that would be needed for this idea. I initially wrote out some of the main characteristics for this project, namely the compatibility with printed masks and the ability to take multiple different substrates. Following the initial sketch, I came up with some ideas for sub-components that are needed like the chuck frame and the light source column. As I was drawing these sections, I was able to further break out how I wanted to design individual components to accomplish the goal of the sub-assembly. A prime example of this would be the gears and linear toothed rail/slide thing. Initially, I planned on using a set screw for the alignment. But after trying to figure out how to translate the rotational motion into a different direction I realized that I could keep the rotation in the same plane. The result is the sketch of what I called the linear element (because I have no idea if there is an actual technical term for it) that consists of a gear that will turn and move the slide based on the interlocking teeth. The last things to point out are some initial ideas for how to slide the light box out of its housing after the substrate has been aligned, and a checklist I came up with for parts that I need to CAD.

Creating a Proof of Concept Model

Once I had some initial sketches of what I wanted to do, I created a very rudimentary CAD model with approximate dimensions. Not all the components are included, for example there is no gear system in this model, but this helped me get an idea of the dimensions and spacing I would be working with when I actually start improving these designs. As part of the Week 1 assignment, I also used this model to texture parts, render a high quality image, and create an animation.

In the above images, the top left picture is of an alignment chuck for a 6in wafer. The central pocket is slightly larger to allow the alignment pads some room to come in from the cutouts on the sides to actually hold the substrate in place. The top right image is a crude drawing of the light box, initially I was thinking about putting bulbs in but now I am going to make my own board with surface mount white or UV LEDs pending some experimentation. The bottom left is the raised frame that will have magnets inserted into the holes to hold the printed mask in place above the substrate, this piece is transparent because I am planning on making it out of a plastic/acrylic material and was playing with material features. And finally, the bottom right is the main body/plate that everything will be attached to. Ultimately the light column will be made separately and attached mechanically, but for simplicity I modeled it all together. This component is also currently made from aluminum in the CAD, but I am thinking that I will likely make it out of acrylic if possible due to material costs. When combined, these components create the assembly seen below. I then imported the assembly into Fusion 360 to handle the rendering and animation.

Refining the Design and Creating a Realistic Assembly

A lot of the CAD work to turn the initial assembly into an actual assembly revolved around separating the main body/plate into its individual components. This needed to be split into the bottom plate, 4 sidewalls and a top to enclose the light source, 2 sidewalls and top for holding the light source, and the linear rail system to allow the light source to slide in and out. The plate, walls, and top pieces were fairly simple to create, they were just rectangular plates with holes positioned at particular locations for assembling them together. I had CAD-ed up some brackets that I plan on 3D printing, so I just needed to make sure that the holes lined up properly. The rail system was a bit more complicated to design though. I had purchased a cheap set of ball bearings off Amazon and I designed inserts for them so they could be attached to the sidewalls of the light column with screws/bolts. I also created a part that would snap onto the outside of the ball bearings with a groove cut into it such that the rail I was designing would seat properly. These parts took a bit of back and forth to make sure I was getting the dimensions correct and that there would be sufficient travel to expose the entire substrate. I could probably have used a double rail to get more travel with less space used, but I didn't know how I would go about designing that at this point in time.

Another area that required a couple of design iterations was the gear system I had implemented to align the substrate. I had CAD-ed all the parts for the assembly in Week 1, and screenshots of these parts are in the Week 1 page. There is the linear toothed component, a large and small gear, and then a mount for the gears. I did not update these designs while I finished making the rest of the parts, but when I assembled them all together somewhere along the way my math was off. The large gear physically collided with the other parts which obviously is a problem, a visual inspection also revealed that the teeth of the gears/rail could potentially jump or catch because the spacing was off, and the light source would crash into gears because the clearance was too low. After a few design iterations, I ended up removing the large gear entirely in favor of two of the same small gear. This allowed me to also reduce the height of the mount, so that I only needed to raise the height of the light source by a tiny bit. This distance was pretty important to me because without expensive controls and optical components the light will spread out more from the point of origin as the light source gets further away from the mask. For the best results and clearest image transfer, it is important to make sure that the light source is as directional as possible, meaning the light has as little time to spread out as possible and therefore should be close to the mask. In the images below the old design is on the left and the new design is on the right.

When I started the assembly, I had anticipated using 6 ball bearings and I had not yet updated the gear system. Thats why in the picture containing all the loose parts some may not have actually made it into the final assembly picture to follow (also not all the light source parts were included at that point either).

Even with this final assembly however, there are a few additional details worth pointing out that will be addressed after I have a consultation with my TA to make sure things are on the right track. The first of which is that there is no handle to turn the gear system at the moment. I do plan on putting in a knob that the user can turn, but I am not sure if the best way to do this is with a keyed rod that would insert into the gear or to just glue it. The second issue is that there is no counterbalance on the sliding rail to prevent the light box from disconnecting from the ball bearings. I could add a bottom protrusion on the rail to make sure it can't come off, but then getting the rail on the bearing in the first place will be a challenge. There are methods I know I could implement that I know would work, but I would rather get some input from the TA first to make sure I do it in a good way. And lastly, I have not yet included any plans for the electronics. I will be using a microcontroller and screen to set up a "job" that would run an exposure at a power setting for a set period of time. This will be done through a visual interface that I will likely attach at the front left of the tool, meaning wires will have to connect the interface, the microcontroller, and the LED board. I am confident that I can design and 3D print a screen housing to fit in the free space left over, and the light box should have more than enough real estate to fit the microcontroller and LED array on a custom PCB. Depending on the performance and limitations of the UV LEDs that I find, I may also need some form of power converter but I will figure that out in the electronics design portion of class!

After talking with Dan about the design, we came up with a plan to handle the areas of concern I highlighted above. To handle turning the gears, he thought from the drawing the user would basically use their thumb to turn the gears from the back one. I hadn't thought about this, and at least tentatively will try that method since I can always design and print a knob if it doesn't work well. For the rail, the method I was thinking of turned out to be the simplest and that would consit of adding a protrusion on the lower part of the rail to prevent it from lifting off the ball bearing. I would then need to screw or glue on an endcap to then prevent the rail from sliding off the bearings entirely. We didn't discuss anything about the electronics, which is fine because I have a lot of space to fit them into later. Dan however did show me a scrap device that was set to be disposed of with a very nice aluminum bottom plate. It fit the dimensions I was planning for my base, so I grabbed it and will start machining it soon!

Add updated rail design

Printing Parts (Additive Manufacturing Technique)

There are a number of parts in my design that are great candidates for 3D printing, largely because the print time will be much faster than machining them and they do not require a high amount of strength. So far I have created two batches of parts through printing. The first set are the parts that will go on the inside and outside of the bearings which I did on the Prusa out of PETG. I mostly just wanted to make some progress and check the fits of the parts on the bearings so I may remake these parts later. I also don't have pictures of them but that is on the way. The second batch of parts I printed on the Fuse 1+ SLS printer and consistes of a wide variety of parts. I included all of the brackets, the gears, the gear mounts, plate standoffs, the linear slides, and the alignment contact pads for the 6in wafer plate. The build area was pretty packed, but the Formlabs software did a great job optimizing the layout of the parts and was super easy to use. The print would take about 3 hours plus cooling time, so I started it on a Tue evening and came in before class on Wed to unload and finish my parts. I cleaned them up pretty well in the hood next to the fuse to recover as much poweder as possible, but then bead blasted them with fine glass pellets to get rid of any remaining nylon. The process did take a bit of time, but it was very simple and the parts came out great. I had done a bunch of traditional 3D printing before, but I am a huge fan of the SLS process, so I may be printing a few more parts on the Fuse down the line.

Aside from the Fuse, I did also make use of the Prusa printers in the CBA shop. I ran a very quick job on an MK4S to print the parts that would go on the inside and outside of the ball bearings. This was intended to primarily check the fit and the dimensions, but they turned out pretty well so unless I have to I am not going to re-print them. The main job I ran on the Prusas however, was the ball bearing rail because of the footprint needed. This could not be made in the Fuse since the print volume was not big enough, and it just fit into the Prusa XL print bed. I set up a job to run one evening and came in the next day to remove and clean my parts. Both jobs were at 15% infill and out of PETG, with supports enabled because I did have some recessed features.

Laser Cutting Acrylic Parts (Subtractive Manufacturing Technique)

I decided to form most of my structural elements out of acrylic because it is cheap, easy to shape, and strong enough for my application. None of these pieces required any special processing methods or techniques, so I laser cut all of them on the XTool. I exported all of the sketches as DXFs, loaded the pieces of stock, and used the very easy interface to set up my cutting paths. I specified the material was ~6mm (1/4in) in the software and it handled all of the laser settings automatically. I did cut a quick test structure to make sure the settings would get through my stock and that was successful without any changes. With that test complete, I just had to sequentially put in my stock and cut all the pieces I needed.

Water Jet Cutting the Base Plate (Subtractive Manufacturing Technique)

I chose to make my base plate out of aluminum so that it would be more sturdy than acrylic when it comes to supporting everything mounted to it. There are also a few pockets that would be cut into it that I want to be rigid and they would be exposed to UV light, so it made sense to go with a metal. Luckily, Dan showed me a 1/4in thick plate that was about to be scrapped that nearly fit my dimensions perfectly. We determined it would be best to make the intiial cuts on the water jet first to get the footprint and any through holes needed, before going to milling. I exported a DXF of the back side of the plate so only the through holes and outside footprint would be captured, and then set up the job on the water jet. I had water jet a few parts before, but this was very simple to start. We clamped it down outside the expected path of the nozzle, set the file, selected a starting position, and zeroed the XYZ coordinates of the tool. Then it ran really smoothly and clean up at the end was simple enough!

CNC Milling Aluminum Parts (Subtractive Manufacturing Technique)