# Is it possible to build a grid of silicon photosites to build a digital camera at home?

So I know a digital camera takes light and focuses it via the lens onto a sensor made out of silicon. It is made up of a grid of tiny photosites that are sensitive to light. Each photosite is usually called a pixel, a contraction of "picture element". I was wondering, even if it is as small as 100x100 pixels or even smaller, is it possible to build my own grid of tiny photosites at home out of silicon and use that to make a digital camera? I know you can go out and buy them but I am trying not to use premade parts for this project only raw materials, machinery and other thing(transistors, diodes, resistors etc.) Is it possible to build one of these silicon grids at home? If so how?

• maybe if you build a sensor the size of a wall: bit.ly/1TRWC8Y – Wesley Lee Jun 2 '16 at 20:57

Because discrete photo-sensitive components (LDR or photo-transistor, or PV cells or whatever) have significant physical size, you will end up with VERY LOW effective resolution because the space between the sensors will likely be much larger than the sensitive areas themselves.

Not to mention the issue of scanning/multiplexing, etc. 10,000 or more sensors.

If you mean actually creating an array of photo-sensitive diodes (or whatever) into a MONOLITHIC silicon substrate, then it would appear that the old rule applies: If you have to ask, you probably can't do it. It takes millions of $worth of very specialized equipment to process IC wafers. Not to mention cleanroom environment, and knowledge of design and manufacturing techniques. Not to mention raw materials (wafers, dopants, gases) that would be very difficult to source if you are not in the industry and have big piles of money. • Second Hand Stepper$1.5M, rest is cheaper. Definitely not billions. – Asmyldof Jun 2 '16 at 20:59
• @Asmyldof Masks cost hundreds of $. You need an evaporation chamber which will be$100k at least. You need pure phosphor and boron dopants samples to form diodes (pure samples cost a lot). You will need to form transistors and amplifiers in order to get anything out of the photo diodes (which also need to be formed themselves). That require a lot of careful design. There is a whole lot more in IC fab than just a stepper. Plus you absolutely need a clean room as any dust particles will result in defects in any devices you form rendering them useless. – Tom Carpenter Jun 2 '16 at 22:08
• @Asmyldof - And then there's the question of wafer source. Start with very pure silicon and ultra refine it using the zone purification process (need a really nice induction furnace operating in a vacuum). Grow your own silicon monocrystal boule, saw it into wafers, polish the wafers. Might take some work and money, don't you think? – WhatRoughBeast Jun 2 '16 at 22:56
• You guys are aware you can get wafers everywhere, right? I have 10 viable 6inch doped ones right here. Even so, definitely not billions. And everything is still cheaper than the stepper. So as long as you guys don't dream up some outlandishly stupid chain of 800 machines, I see no flaw in my troll. (PS Masks for simple shit like photodiodes can be gotten from China for your second hand stepper at $25 ~$75) – Asmyldof Jun 3 '16 at 9:15

The most practical approach is to find small SMD photo-diodes and build a grid of them. You then need to find a way of reading them.

Building your own wafer from scratch is not feasible. Surprisingly you can buy blank ones on ebay, but the chemical processing is nasty and requires a clean room and at least a chemistry degree.

• You don't need a clean-room if you have no quality requirements. People need clean-rooms because dust makes spots in the resist and hampers air-bearings and all such, not because the processing will explode when there's dust near. – Asmyldof Jun 2 '16 at 20:57
• still, photo resist is freaky. – b degnan Jun 2 '16 at 21:51
• Many years ago I knew a guy who had acquired nearly all the fab tools needed to process 4 inch (100mm) wafers. His intent was to make his own Photo-Voltaic ("solar") Cells. PV doesn't require very sophisticated processing and almost-trivial patterning. Dunno whether he ever succeeded. – Richard Crowley Jun 2 '16 at 22:21

As an alternative approach you could consider using one decent sensor and moving it with an X-Y motion to "scan" the focal plane behind the lens. This would, of course, be rather slow and, so, only suitable for static scenes. The beauty of it is that you read only one pixel at any time.

How to:

• Make a "camera" box big enough to hold your X-Y motion. Paint the inside black to prevent reflections.
• Position a lens the correct distance away to focus the image on the plane of the sensor. You will need a view port to check this.
• Mount the sensor on the X-Y and connect it to your micro analog port.
• Have your micro move the X-Y motion in a back and forth scanning motion and record the brightness reading from the sensor to memory.

It sounds like a lot of work.

• Or just study the very earliest days of inventing television back when they were still using mechanical scanning. There are some interesting documentaries on YouTube. – Richard Crowley Jun 2 '16 at 22:43
• "Mechanical scanner" is the phrase I was looking for. Thanks. – Transistor Jun 2 '16 at 22:51

DIY Photolithography in your garage Guide.

1. You can build plasma-induced Hg machine for photolithography by yourself. Nothing ultra-hardcore about it! You need Hg lamp. Make appropriate skills in chemistry and glass blowing or... better just buy one. 180nm is sufficient.
2. Building machine for injection of metal into surface is the real problem, because you need to modulate magnetic field for that. A lot of physics is required from you for that part! Some solenoids and you are good.
3. Etching is the real task... even more.... You need to make strong laser for evaporation. But the best way is to use plasma. Create your own plasma controlled by field. Hardcore? This one is NIGHTMARE. If you didn't finish QED course.