11

Flashes of UV light can create electron-hole pairs in transistors. Or at least reports of the "UV flash bug" claim this happens in microprocessors. People observed raspberry pies reset when subjected to a bright flash.

I'd like to do a demonstration of triggering generic transistors (e.g. 2n3904) directly with light rather than a base current. I'd like to see how a discrete transistor behaves under the light influx.

However, there appears to be a dearth of information about this on the internet. e.g. what light frequency to use (green for deeper penetration vs. UV for direct band gap excitation), how bright, which types of transistors are affected, whether you need to take the casing off the transistors ...

Any help would be appreciated, be it a website link, or your personal experience.

ions me
  • 341
  • 1
  • 8
  • This is the reverse process (current -> light) https://lab.whitequark.org/notes/2014-06-14/transistor-as-a-light-source/ – ions me Apr 30 '22 at 03:55
  • 1
    Remove the encapsulation material and expose the silicon die. – user57037 Apr 30 '22 at 04:04
  • Do you have recommendations on how to remove the encapsulation? – ions me Apr 30 '22 at 04:06
  • Maybe dissolve it in acetone. Not sure if that will work, though. The encapsulation material seems to be very resistant to most solvents. – user57037 Apr 30 '22 at 04:10
  • 7
    Usual way to remove epoxy encapsulation is very strong acids such as white fuming nitric acid or nitric-sulfuric mix, also at elevated temperatures. It still takes some time. You can find instructions on the net, but I wouldn't recommend it (the chemicals are pretty nasty), and the reagents can be hard to source because of liability and association with explosives – Spehro Pefhany Apr 30 '22 at 04:30
  • There are transistors (I have a bunch of them) with transparent TO-92 packaging. They were (are?) intended for light detection. (Might be some in older computer mice?) Perhaps they are still around? – jonk Apr 30 '22 at 04:41
  • @jonk Even LEDs were [available](http://www.decadecounter.com/vta/pic/to92led.jpg) in TO-92. More common THT parts these days seems to be either cheap LED packages (2 leads) or expensive hermetic TO-18 parts with a glass lens. And SMT parts in huge variety. – Spehro Pefhany Apr 30 '22 at 04:45
  • @SpehroPefhany Mine are BJTs. In case that was being called into question. If not, I missed the lead. – jonk Apr 30 '22 at 04:54
  • 1
    Acetone is unlikely to work. It's good for cleaning epoxy while it's still liquid or soft; not so much once it's cured. –  Apr 30 '22 at 13:28

3 Answers3

12

Easiest would be to buy a TO-5 or TO-18 (or even TO-3) transistor and grind or saw the top off the can. Eg. 2N2222, a pretty common type even today.

I would expect similar spectral response to an ordinary silicon phototransistor (peaking around 800 or 900nm with tails out to blue and IR, but very little in UV).

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • I'll have to try that. That's interesting that ordinary silicon photo transistors have that spectral response, because the direct bandgap transition for silicon is ~4 eV which is in the UV range. – ions me Apr 30 '22 at 04:27
  • AFAIU, the light sensitivity peak is at somewhat shorter wavelength than the bandgap corresponds to. In silicon that's 1.14eV which is ~1100nm. – Spehro Pefhany Apr 30 '22 at 04:36
  • 1
    I remember doing this with 2N3055, and I was wondering at the time (early highschool) whether it's possible to make an amplifier as a gift for my neighbour, and then interrupt or override his music with flashlights, from a distance. Oh, the memories... – a concerned citizen Apr 30 '22 at 16:18
  • 4
    I don't know if it's true of all plastic encapsulated transistors, but at least some of them have enough light leakage through the plastic that their performance is affected. I've seen it happen with JFETs (2N5555) exposed to ordinary room lighting. It's nothing you'll find on a spec sheet: it's generally not known. If the plastic could somehow be removed the effect should be greater. – stretch Apr 30 '22 at 17:38
11

The old OC71 germanium transistor used to be packaged in a painted glass tube. All you needed to get a photo-transistor was to scrape the black paint off.

De-lidding a T018 or T05 is probably the simplest route for silicon transistors these days.

Neil_UK
  • 158,152
  • 3
  • 173
  • 387
  • 3
    I'm old enough to remember doing this. I think Mullard used to sell an OCP71 phototransistor which was more expensive than the OC71, so scraping the paint off saved money. – abligh Apr 30 '22 at 15:29
  • 2
    And they eventually changed the gel that surrounded the actual device from a clear one to an opaque (Blue IIRC) version to stop people doing that! – Dan Mills Apr 30 '22 at 19:18
  • @DanMills - I remember that. The early ones were with a gel similar to petroleum jelly before it changed to blue. My OC44/5s were always blue. – Kevin White May 01 '22 at 01:37
4

You can do this. Teenage me made phototransistors over half a century ago by grinding the cans open. The limits to Si photosensitivity are that its transparency increases with increasing wavelength. Go too far into the infrared, and light goes straight through without interacting. Go too far into the ultraviolet, and light only interacts at the surface, rather than in the active silicon. What the limits are depends on the device construction, but if the light's visible to your eyes, most silicon devices will see it. You may "cover your bets" by using white light.

John Doty
  • 2,227
  • 5
  • 12