LEDs at MWIR & SWIR

[Ultrapurple]

I was interested to discover that LEDs are readily available (and affordable) down to at least 4.3µm wavelength. In fact there are a range of common 'colours', including lots in the (very) near-IR and

0.98µm
1.05µm
1.07µm
1.2µm
1.3µm
1.45µm
1.5µm
1.65µm
1.72µm
3.6µm
4.3µm
4.5µm
4.8µm

Whilst these emitters aren't as cheap as, say, commodity green LEDs (they're typically GBP 10 or so at Farnell) they are affordable-enough that it's practical to consider making an active wavelength reference 'chart'. Has anyone done anything like this?

I am of course aware that a hot black body (or even a soldering iron) will emit a fairly continuous spectrum across the IR range and that one could use a prism or grating to select particular wavelengths, but you're still left with the question of wavelength calibration, which is quite simply not a problem when you buy LEDs of known wavelength.

Discuss...

[Ultrapurple]

I'm thinking like this, only not so attractive to the human eye (image from Adafruit):

[Fraser]

I was looking at similar SWIR LED’s recently but they were £33 each so I held off buying some. I was thinking about building a test source to see how far into the SWIR band my 7290A could see and also to check the SWIR capabilities of my FLIR SC4000 MWIR-SWIR science camera. I am still considering options for testing the SC4000. I could use filters of known performance either in front of the cameras lens or in front of a broadband energy source. Sadly such filters are very expensive and placing one in front of a powerful broadband energy source is asking a lot of a filters spectral response characteristics. Placing a filter in a  ‘low energy’ scenario where it is in front of the cameras lens sounds a better approach.

The idea of a SWIR or MWIR LED ‘reference’ emitter still appeals to me 

Fraser

[Ultrapurple]

I was genuinely surprised at how far down the spectrum you could get LEDs. However, I do dimly recall seeing for sale (very expensive) LEDs that went down into LWIR. From distant memory they were from someone like Roithner but browsing their site recently shows nothing beyond the figures I quoted above.

On the other hand, page 2 of this paper shows that GaSb-based LEDs can reach as low as 12µm, which is well into LWIR territory.

More interesting still is the knowledge that there is ongoing research into building the LWIR equivalent of a large screen high resolution video projector based on long wave LEDs and other techniques. Though I can't imagine who would want to project realistic, life-size thermal images of tanks, infantry and so on.

[Fraser]

I have seen LWIR emitters for laboratory use but they used an open filament behind a LWIR transparent window. The drive current controls the emissions produced. They looked simple enough to DIY but the proper emitters are ‘calibrated’ whereas a DIY version is not. I bought some of the filament materials and foil strips used in Vaping units to experiment with. Sadly I have not got around to it yet. The emissivity of the filament is surprisingly good. Various metals and alloys are used in LWIR filament based emitters and thankfully all are easily sourced.

The commercial LWIR filament emitters come in many sized and I have one from an FTIR that looks like a metal reflector MR10 Halogen lamp but instead of a halogen capsule, it has an open filament. I also have small ceramic reference heaters from FTIR equipment. The small LWIR filament emitters that I saw were enclosed in what looked like a LDR (large) or photo transistor (small) casing and appear a neat little source of LWIR 

http://www.dias-infrared.de/pdf/pyrosens_ir-emitter_eng_mail.pdf

Interior view of filament type emitters....

http://www.hawkeyetechnologies.com/source-selection/ir3x/

Fraser

[DaJMasta]

If you look at the emission charts for such LEDs, they are not especially narrow band when compared to normal wavelength calibration sources.  Something like the normal Mercury Argon reference source is commonly used for visible band spectrometer calibration, but the spectral lines stretch down to 922nm and have a MUCH lower FWHM.  Then, while fewer in number, it's relatively inexpensive to get a hold of low power laser sources in the SWIR band on the used market - I picked up a 1550nm optical fiber verification source for ~$40 recently, and while it's only one wavelength and the choices are fewer, again the FWHM of the peak is much lower, so it would be much cleaner as a spectral reference.

That said, for just a visual indicator of SWIR performance for cameras or something, these sorts of sources may be the only real option.  Of course to cover the band of available wavelength options you'd probably be dropping $100 or more before your driver circuit, but you could get a visual indicator of how far into the SWIR band your tester could see fairly easily.

For me, these LEDs are a tough sell for illumination outside of buying the specific wavelength you need to do your QA fault indication or similar using an SWIR camera, trying to make the equivalent of a LED light with variable color temperature but for SWIR would be an expensive proposition even though it could be fascinating for demonstration purposes or for configurable illumination.

Filtering is a possibility for similar performance when put in front of a broadband source, but it is difficult to find inexpensive bandpass filters well below visible light, so even if there are some good choices for communications applications, it may be tough to get together good choices to filter a broadband source.

For general illumination, my choice (and the cheapest choice) is a low color temperature halogen bulb and a filter.  I picked up a 2900k bulb that accepts mains input and fits in a standard socket (though these should be available down to about 2700k), then you can put a hot mirror with extended range (many only go down to 1200nm or so), a cold mirror, or a long pass filter like RG1000 in front of it and so long as your housing can manage the heat, you have a VERY inexpensive broadband source.

[_Wim_]

For general illumination, my choice (and the cheapest choice) is a low color temperature halogen bulb and a filter.  I picked up a 2900k bulb that accepts mains input and fits in a standard socket (though these should be available down to about 2700k), then you can put a hot mirror with extended range (many only go down to 1200nm or so), a cold mirror, or a long pass filter like RG1000 in front of it and so long as your housing can manage the heat, you have a VERY inexpensive broadband source.

Combine that with a monochromator for the SWIR range (the bulb at the input and the camera at the output of the monochromator) and you can really test the response of the camera by scanning the monochromator. Calibrating is the hard part, as both the grating as well as the bulb have an unknown curve, but it should be possible to find an ingaas photodiode with known transfer characteristic quite cheaply to perform a rough calibration.

I already have a monochromator for the visible range, still looking to a suitable grating to make the above work.  But I did consider the leds also recently. Maybe a group buy? 

[Vipitis]

You should built some cheap prism or refraction grating and do some MWIR and LWIR spectroscopy to see how monochromatic those LEDs are

[_Wim_]

You should built some cheap prism or refraction grating and do some MWIR and LWIR spectroscopy to see how monochromatic those LEDs are

Spectrum of these leds is defined quite good in the datasheet (see example attached). These were the digikey parts numbers I had in my shopping basket for a first test:

Part Number   Manufacturer Part Number   Description   Unit Price
475-SFH4737CT-ND   SFH 4737   OSLON P1616. IR BROADBAND EMITTE   6.11
1125-1496-ND   MTE7110C2   SWIR EMITTER 1070NM 3MM PLASTIC   12.1
1125-1472-ND   MTE1200C2   SWIR EMITTER 1200NM 3MM PLASTIC   12.1
1125-1477-ND   MTE1300D4   SWIR EMITTER 1300NM 3MM PLASTIC   12.1
1125-1493-ND   MTE6114D4   SWIR EMITTER 1450NM 3MM PLASTIC   12.1
1125-1485-ND   MTE5115D4   SWIR EMITTER 1550NM 3MM PLASTIC   12.1
1125-1488-ND   MTE5116C2   SWIR EMITTER 1650NM 3MM PLASTIC   12.1

[ArsenioDev]

WHOAH, I had no idea the emitters went this deep and didn't cost a STUPID amount. I've got a few SWIR telco lasers I use as that band's check sources, have a lot of experience driving them from a project at NASA.

[Max Planck]

I was looking at similar SWIR LED’s recently but they were £33 each so I held off buying some. I was thinking about building a test source to see how far into the SWIR band my 7290A could see and also to check the SWIR capabilities of my FLIR SC4000 MWIR-SWIR science camera. I am still considering options for testing the SC4000. I could use filters of known performance either in front of the cameras lens or in front of a broadband energy source. Sadly such filters are very expensive and placing one in front of a powerful broadband energy source is asking a lot of a filters spectral response characteristics. Placing a filter in a  ‘low energy’ scenario where it is in front of the cameras lens sounds a better approach.

The idea of a SWIR or MWIR LED ‘reference’ emitter still appeals to me 

Fraser

The easiest and cheapest solution is tu use an old spectrometer/monochromator for VIS/NIR and eventually(?) SWIR band. Solution tested "in the field" by the way.   
Max

[bap2703]

Don't forget that spectrometers and monochromators also need to be calibrated.
Just as with thermal imaging it's very easy to get a result, a result you can trust is a bit more difficult

[_Wim_]

Don't forget that spectrometers and monochromators also need to be calibrated.
Just as with thermal imaging it's very easy to get a result, a result you can trust is a bit more difficult

In this case I think is relatively ok, as we do not want the characterize the light source, but want to compare 2 detectors using the same source. So if you fist scan the spectrum with an ingaas photodiode with know characteristic (readily available on Ebay for around 30€), you can then compare the camera response with the know intensities coming out of the monochromator.

This does not tell you anything about the light source however, as the grating has its own spectral response, and data for that will be very hard to find.

[Max Planck]

Don't forget that spectrometers and monochromators also need to be calibrated.
Just as with thermal imaging it's very easy to get a result, a result you can trust is a bit more difficult

Not really. Unless needing 1nm precision, one can use narrow-band light sources (LEDs) to evaluate the sensitivity limits of an InSb FPA. The camera I played with was able to see the light coming from a red LED.

Max