Can EPROMs Be Effective UV Light Detectors?

[ruudlustig]

Using an EPROM as a UV detector.

I was planning to make a UV meter using an EPROM. The idea was that ultraviolet light erases an EPROM. When a fully programmed EPROM (all zero's) is exposed to uv light I should clearly see the number of bits (one's) increasing.
So I have built a small device that counts the number of bits (one's) in an EPROM.
I had guessed that the speed of increase of bits would certainly be a measure of the UV intensity.
But...
When exposing the EPROM to uv-light I expected a smooth and stable increase of bits, but that didn't happen. In stead, after some time of exposure to uv-light the bitcount begins to vary very much and it seems that the bits (all Floating Gate devices) become unstable.

Knowing that the bits become unstable, I covered the EPROM window with aluminium and watched what would happen. Not to my surprise the bitcount kept varying (I measure 64k bit 4 times per second). I have a built in threshold of 50 counts, and everytime 2 measurements have a difference of 50 of more I show a flash on a LED.
And it is flasing very randomly between 20 to 60 flashes per minute. Now my question is: What is it exactly what I measure? I don't know much about uv or radiation, so perhaps you can help me out?

 

[berkeman]

Using an EPROM as a UV detector.

I was planning to make a UV meter using an EPROM. The idea was that ultraviolet light erases an EPROM. When a fully programmed EPROM (all zero's) is exposed to uv light I should clearly see the number of bits (one's) increasing.
So I have built a small device that counts the number of bits (one's) in an EPROM.
I had guessed that the speed of increase of bits would certainly be a measure of the UV intensity.
But...
When exposing the EPROM to uv-light I expected a smooth and stable increase of bits, but that didn't happen. In stead, after some time of exposure to uv-light the bitcount begins to vary very much and it seems that the bits (all Floating Gate devices) become unstable.

Knowing that the bits become unstable, I covered the EPROM window with aluminium and watched what would happen. Not to my surprise the bitcount kept varying (I measure 64k bit 4 times per second). I have a built in threshold of 50 counts, and everytime 2 measurements have a difference of 50 of more I show a flash on a LED.
And it is flasing very randomly between 20 to 60 flashes per minute. Now my question is: What is it exactly what I measure? I don't know much about uv or radiation, so perhaps you can help me out?

Welcome to the PF.

Interesting experiment. If you expose it to enough UV for long enough, do all the bits erase? What is your UV source? If you look at the datasheet for the windowed EPROM, does it give data for the amount of UV radiation needed to ensure erasure?

 

[Rive]

I think you are measuring some random fluctuations and the noise of the power source.

As the floating gate of a bit depletes, the output of the FET will become undetermined and extreme sensitive to any (fast) external effect during readout.

EPROMs are not the best choice for this kind of experiment, as the capacity of the floating gate is relatively high (causes low sensitivity) and the programming voltage gives a wavelength barrier for the sensitivity (through the thickness of the insulation of the floating gate). Also, the state-sensing is not linear: it has a kind of comparator like characteristics.

Sometimes it's possible to get some old-old DRAM chips on ceramic carrier with metal caps. After removing the cap they might be sensitive for light and their bits has lower electric capacity than the floating gates of the EPROMs. (In exchange they might be sensitive to visible light too, and the readout also works as a refresh cycle - it makes things a bit difficult.) Maybe they worth a try.

Or, maybe a linear CCD sensor from some old scanner might be useful too...

 

[ruudlustig]

Hi Berkeman,
Normally, a programmed EPROM will show a steady and constant bitcount (it should be of course).
I use my EPROM eraser with a 4W uv lamp. Normally an EPROM is erased in about 10 minutes. When I expose long enough, the EPROM erases completely.
Currently, I have erased about 40k bits of my 65k bit EPROM.
The measurement shows that the bitcount is also sensitive for normal sunlight, that is, the the more light, the more heavily the bitcounts fluctuate, but on average the number of erased bits stays constant.

I am also thinking of measuring the effects of using just one bit, or a number of adjacent bits. and Hi Rive:
It is just an experiment to see if uv measurement is possible using an EPROM and I don't have very high expectations. But because of the instability of the bits I would imagine that any form of radiation would affect the bitcount. On the other hand, I am sure you are right, also power fluctuations will have effects. So, maybe I have just built myself a (bad) random generator?

However, I am convinced that with the use of statistics, it must be possible to extract data from the measurements that make some sense.