Public Lab Research note

Testing high-brightness 405nm LEDs in fluorescence spectroscopy of oils

by warren | January 05, 2016 16:20 | 408 views | 5 comments | #12561 | 408 views | 5 comments | #12561 05 Jan 16:20

Read more:

What I want to do

After a great 2015 Barnraising session on the topic, @tonyc and I have been trying out a few different designs for a more compact, generalized cuvette holder for the Desktop Spectrometer that could potentially have a dimmable LED with a brightness dial, for either absorption or fluorescence work.

After some discussion, I'd ordered some very bright 405nm blue/UV light emitting diodes (LEDs) from LED Group Buy ($5 each, although we can probably buy the component LED for cheaper, and/or in quantity) but unfortunately they're ~3-3.5v, which means USB power (or a standard Arduino) will burn them out (which I tested :-P).

I looked at using resistors or something more complex, but Adafruit sells a 3.3v "Trinket" which is a tiny Arduino compatible which runs on a 3 volt circuit. Perfect!


My attempt and results

So to start with, I just set up a test using a left over mineral-oil-diluted oil sample from the Oil Testing Kit. I don't have the original samples anymore, but this was either a motor oil or a diluted crude oil. In any case, it fluoresced quite strongly from a 405nm laser.

To my delight, the 405nm LED was very bright, and not only did it clearly create a lot of fluorescence, it did so even when almost 6 inches away, although my smartphone camera wasn't as good at picking this out as my naked eye:


I then took some spectra, and they worked well; the baseline mineral oil only spectrum wasn't as narrow as a laser peak (recorded from a past test), but it was not bad. It extends about 25nm further to the right, so we'll keep that in mind. Below, red is laser, and black (and dimmer RGB channels) is LED:


The comparison of the unknown oil vs the mineral oil:

Here's the same with equalized height (which you can do yourself in the "tools" menu in the embed above):


Questions and next steps

This was all without any dimming at all, by attenuation with a filter strip or by dimming the light. The point was to see if the light was bright enough to produce as strong fluorescence as the laser. So I think this is a success; the next step will be to attach a dimmer dial so the brightness can be tuned, and perhaps reproduce the Oil Testing Kit Public Beta with a full range of samples.

Doing a more thorough test and comparing to other OTK beta tests will help us decide if the broader bandwidth of the LED is an acceptable loss given all the advantages of this method.

Why I'm interested

The advantages of this are many:

  • reproducible dimming with a (numbered) dial
  • potentially pairing this with a light sensor to make the brightness of the dial an absolute measure -- given a known input spectrum, this might be able to give us a reasonable intensity calibration. I.e. set the dial with no cuvette, to a known measured brightness, then insert the sample.
  • more compact design
  • safety: no need to worry about laser eye hits, or as much worry about eye protection
  • easier to order/ship without special permission (as we need with the lasers)
  • adaptable to visible light absorption work too

I'm going to move ahead (really in my spare time, as I have lots of other higher priority coding work to get done) with building one of @tonyc's stacking cuvette prototypes, based on the sketches we did:

I did this Help out by offering feedback!

People who did this (0)

None yet. Be the first to post one!


Just looping in @ethanbass, @ygstcu, @matej, & @tonyc who were part of this discussion/design process, @stevie and @gretchen, and @stoft because of the possibilities I mentioned at the end about pairing this with a light sensor to try to get an absolute measurement of input light brightness, and possibilities re: absolute intensity calibration. What do you think?

Is this a question? Click here to post it to the Questions page.

Reply to this comment...

Just a note that I found this interesting article (not peer-reviewed; Photonics is an industry trade journal, I think) which discusses LED input light:

The UV probe fired a xenon flashlamp at 254 nm, and the visible-range device used an LED to provide excitation at 370 to 460 nm. The researchers tested the probes using phenanthrene as the key standard to help determine the concentration of diesel fuel, gasoline, gear oil and motor oil in a water sample. They found that the LED-based probe did not respond at all to the diesel fuel, whereas the UV sensor picked out diesel at less than 1 mg/l and was sensitive to all of the other potential pollutants as well.

So, the LED we're using is close to what they're calling a "visible-range device." More:

In a paper published in Power Plant Chemistry in 2010, Malkov and his colleague, Dietmar Sievert, also of Hach Co., focused their attention on polycyclic aromatic hydrocarbons (PAHs). They set out to show how well UV fluorescence would provide a correlation between oil concentration and the sensor readings they acquired. They compared the results of the UV probe with those from a device using visible wavelengths.

“The biggest challenges [are] that UV sensors are too specific to PAH and therefore won’t see natural oils in water, while [visible] sensors are not specific enough and will suffer from interference provided by natural organic matter in water,” Malkov said.

So, it sounds like whether we continue to use a 405nm laser or a 405nm LED, we really do have to take organic matter seriously, as we'd expected, as a false positive.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...

Very cool.

If you're going to try to make a cuvette frame, another line of thought is the minimal version, which tucks into the DSK housing itself for "rigidity" [sarcastic quotation marks intentional, but only half kidding, as it does lock it in to some extent).

Anyway, here are some photos of a quick little version I mocked up then almost recycled on accident. It's more tape than craft, but it would be light tight and reusable with a simple lid.

Here are some images, shoudl speak for themselves:






IMG_1123.JPG this is just a tube formed to the cuvette, then pinched at the bottom to set a depth to it for demo purposes. Window cut out of front and side to allow light to come in, then fluoresce out.

IMG_1129.JPG shoves in from the top.

IMG_1130.JPG from inside the DSK

whole think took 5 mins to build. Also, the whole thing fits assembled inside our new DSK box. this may all be off topic, but hopefully cool to someone t

Reply to this comment...

Hey Warren, thanks for your comment on my lab set up. Yeah I'm trying to do oil contamination testing here were I live (city point hopewell va) there are a couple of big chemical plants here right on the appomattox river(shore line,) a lot of people still fish off the shore even with the warning signs posted. I have learned that gas and diesel will fluoresce under a UV light but I want to see if I can detect smaller amounts of contaminants with UV-VIS and Raman spectrophotometry.

Reply to this comment...

Very interesting. A new board from seeed is intriguing- though a bit pricey - seems like it could be turned into an auto-monitoring system.

Reply to this comment...

Login to comment.

Public Lab is open for anyone and will always be free. By signing up you'll join a diverse group of community researchers and tap into a lot of grassroots expertise.

Sign up