Diode Construction

[planters]

I have been thinking about exposing some diodes, such as the red AlGaAs etc., to some extreme environmental conditions. I'm interested in vacuum as well as high and low temperatures. I need to know where the Achilles's heal might be, but I don't know enough about the details of their construction. Other than the flanged base with it's insulated electrical feed-thrus are there any other organics? What holds the leads to the chip and the chip to the copper post? If canned, is the can glued down or ventilated? Guesses are OK. Links would be nice. Any experience with pushing these limits would help also.

[-bart-]

What holds the leads

[Solonar]

The wire bonds shouldn't be effected by temp changes, the semiconductor shouldn't have any problems being chilled ( as evidenced by extreme overclockers chilling cpu's with LN2). High heat could cause the die to fail, but only after getting well above reasonable temps.

My guess would be that the output facet would be the weak link.

[LaNeK779]

i do remember mixedgas mentioning something about the output facet "exploding" in extreme cooling, but i can't find the reference...

[mixedgas]

Don't know about modern diodes, but serious attempts to chill 650s about 10-15 years ago by myself, and by a commercially minded friend, resulted in blown output faces with TE cooling. The idea was to chill down about 20 nm using stacked TECs. Intracavity energy density went way up for some reason, so diode current had to go way down. He called it "Tipping the tail of the Dragon", and teh reduction in drive current nicely canceled the gain in visual effective power.

My experience with red LEDs in liquid nitrogen was not so great. Most went LONGER in wavelength.

Green Nichia leds circa year 2000 lost visual emission at just dry ice temperatures, for example.

Get Krazer in here to tell us about Bandgap vs Temperature.

Not sure this will apply to all chemistries.

Steve

[planters]

Several years ago a Russian group at Unipress had temp tuned as well as pressure tuned some 650nm and 635 nm diodes with LN2 and pressures to something like 15,000 bar. They commented on the complexities of pressure tuning, but did not mention any failures with either technique. The wavelength shift with temp was similar to my own experiments last year with a lab chiller backed up single stage TEC down to -50C. The shift is about 0.2nm/C. I currently run my LOCs in my big projector at -20C for the wavelength shift. They still die as frequently as LOCs do, but not more often than my RT LOC's.

I am going to test the G71 with a poor man's cold finger based on dry ice (step one). Just wondering, how did you conclude that the intracavity energy density increased? But,I hear you Steve. I'll use a slow drop and moderate drive currents. Hopefully the can won't pop off. And if it fails I'll post the low temp limit.



I am having trouble linking to the Youtube " Liquid nitrogen cooling of laser diode" and Liquid nitrogen cooling of LED" but check them out.

[mixedgas]

No scientist in his right mind mentions a hardware failure in a paper except to warn others where it could cost them funding or result in a failure of replication.
Once in a while I see "We tested XX devices, N failed".
Steve .

[planters]

Good point.

An interesting feature of the multitude of you tube LED cooling "dips" is their significant wavelength shift while the lone LN2 diode laser seems to show a much smaller change. My theory is that this has less to do with the chemistry than that the cavity with the laser is probably optimized for the RT wavelength and so will operate, but with decreasing performance as the band gap increases. These competing characteristics might explain the counter-intuitive results when cooling laser diodes.

There has also been a lot of talk about boosting the efficiency of 940nm pump diodes at LN2 temperatures. Of course they make it clear that these diodes have been optimized for operation at these low temperatures. What interests me is just what did they do ? Was it extensive and so "apples and oranges" or minimal and so they might work, just not quite as well at these low temps?

I found when chilling to -50C that the wavelength decreased as I stated above and the output power increased at a given drive current. The drive voltage needed to be raised to maintain the same current flow. The relationships were linear and relatively small, on the order of 10%. I was testing the LOC 815 at 300ma. How did you decide that you thought that the intracavity energy density went way up?

Because the G71 costs a little more, based on your experience I guess I will do a run with some LOC 826's first.

[mixedgas]

I found when chilling to -50C that the wavelength decreased as I stated above and the output power increased at a given drive current. The drive voltage needed to be raised to maintain the same current flow. The relationships were linear and relatively small, on the order of 10%. I was testing the LOC 815 at 300ma. How did you decide that you thought that the intracavity energy density went way up?

end quote.

Chunks blown out of faucets are either: current density too high, or optical density too high. My friend retained the services of a diode expert, who told him it was optical from the cracking patten, so I am led to believe. He also had consulted the factory on the issue.

Said friend had one heck of a market and could have made beaucoup bucks if he could have sold TE cooled red diodes at even 650, back then, krypton ion was your real red choice for more then a few mW. If he could have gotten the high power 670 nm reds down to 650ish, he had the marketing mechanism in place to sell them like hotcakes.

Steve

[planters]

Well we're passed that. I am running 32 LOC's at 648 and so far so good. This should be fun. As soon as Dave's mounts arrive, I don't have even one spare at this point, we'll see what happens at -70-80. If optical density, the only thing I can speculate is that the coating became less leaky off optimum and so the higher Q did the diode in. The reflectivity change with wavelength/temp. might also explain your experience with longer wavelength emission with cooling. The Unipress investigators did not specify power output and might have finessed this, so to speak, with very low drive current. Again we'll see.

There is one subtlety that might help to support your hypothesis. If the power output increases smoothly as the temperature drops and this is proportional both above AND below threshold then the Q probably remains constant. What do you think?