<div class="alert">This page is under development.</div>

**How do you test liquid or solid samples with your [DIY Spectrometer](/wiki/spectrometer)?** Read about ways to prepare and scan samples here, and [read about different tests you can do with your spectrometer](/wiki/spectrometry-activities).

###Sample containers

What do you store liquid samples in? A good sample container has flat sides, so you can shine lights (and lasers) through it without lots of reflections. It's also good to have the light travel through a consistent amount of the sample -- many cuvettes (traditional spectrometry sample containers) are 1cm x 1cm, so the light always goes through 1cm of the sample.

[![dropper.jpg](https://i.publiclab.org/system/images/photos/000/001/730/medium/dropper.jpg)](https://i.publiclab.org/system/images/photos/000/001/730/original/dropper.jpg) [![Cuvette_with_penny.jpg](https://i.publiclab.org/system/images/photos/000/001/731/medium/Cuvette_with_penny.jpg)](https://i.publiclab.org/system/images/photos/000/001/731/original/Cuvette_with_penny.jpg)

_A square-sided bottle, left, and a cuvette, right (photo from [Wikipedia](http://simple.wikipedia.org/wiki/Cuvette))._

Unfortunately, we've found that cuvettes with plastic stoppers will leak when filled with oil and not kept upright (for example if you travel with them). A really nice source of completely sealable rectangular 1/4 oz glass jars which are pretty perfect for sampling can be found for $3.50 for a ten-pack here (see image below of 2 on top of a flashlight): 

http://www.sciplus.com/p/WHITCAP-BOTTLE_48212

###Water sampling

Water is usually very clear in small amounts -- even murky water in a small container will look pretty transparent. That makes it hard to measure with spectrometry unless you shine light through a *lot* of it. But some tests have been done -- see this example of a scan of water from the Charles River before and after 7 days of settling, by [Jeff Hecht](https://spectralworkbench.org/sets/show/330):

[![charles-river.png](https://i.publiclab.org/system/images/photos/000/001/732/large/charles-river.png)](https://spectralworkbench.org/sets/show/330)

However, most research in Public Lab to date has focused on oil spectroscopy -- attempting to identify petroleum residue in sediments. Read on to learn more!

![oil fluorescence](http://old.publiclab.org/sites/default/files/imagecache/default/IMG_0786.JPG)

_Ultraviolet light illuminating a sample of extra virgin olive oil (left) and BP crude oil residue dissolved in mineral oil (right)_

###Oil sampling

To identify oil contamination, we [have been attempting](/tag/fluorescence) to illuminate oil samples with UV flashlights and green lasers, which can make some oils fluoresce, or glow, as pictured above.

The basics of sample preparation for oil identification are still being refined, but our best practices to date are: 

1. Collect soil, sediment, tar, or other solids which you suspect contain petroleum contaminants.
2. Put a pea-sized amount of sample in a medium (1 cup or 300ml) jar and fill halfway with unscented mineral or baby oil (from a pharmacy). Stir up or mush until it breaks down.
3. Leave overnight or for up to 48 hours to settle in a dark place.
4. Use an eyedropper to move the clear, hopefully yellowish solution near the top into a new, small (1oz), glass, flat-sided container or cuvette.
5. Try shining a laser or UV light through; if it's too opaque, let it settle again and eyedrop it into another container with more mineral oil. It should look roughly like olive oil in darkness and color.

**Recent successes:**

* preparing tarball samples by Scott Eustis: http://publiclab.org/notes/eustatic/08-01-2013/making-grand-isle-coffee
* [streamlined sample extraction](/notes/warren/12-11-2013/spectrometry-sample-extraction-at-the-parts-crafts-toolshed-raising) (Nov 2013)
* very clear fluorescence and [good measurements with blue lasers](/notes/warren/11-08-2013/fluorescence-of-bp-oil-with-uv-laser-success)
* [extracting fish oil from fish](/notes/warren/11-25-2013/poster-on-issues-with-pah-detection-in-fish)

**Ongoing research questions include:**

* how to get samples to glow brightly enough to capture with a spectrometer, or how to detect only dimly glowing samples (see [pan attempt here](/notes/warren/05-14-2013/laser-oil-fluorescence-proof-of-concept) and [more success here](/notes/warren/11-08-2013/fluorescence-of-bp-oil-with-uv-laser-success))
* what sample containers to use ([see above](#Sample+containers))
* [how to orient the light with regard to the spectrometer and the sample](/notes/warren/11-08-2013/fluorescence-of-bp-oil-with-uv-laser-success) -- keeping the input light perpendicular to the direction of the spectrometer so we measure the glow but not the laser or UV light
* what level of dilution is useful (see [these samples which were too opaque](/notes/eustatic/09-09-2013/nungesser-test-2-uv-and-green-lazer)) and [these which were too dilute](/notes/warren/12-11-2013/spectrometry-sample-extraction-at-the-parts-crafts-toolshed-raising)

[![Flame spectroscopy](https://i.publiclab.org/system/images/photos/000/001/634/medium/IMG_1794.JPG)](https://i.publiclab.org/system/images/photos/000/001/634/original/IMG_1794.JPG)

_Burning potato chips to measure the sodium emission spectrum from the NaCl (salt)._

###Flame spectroscopy

Another type of spectrometry which involves measuring the light of a flame and can detect specific elements (not molecules) as they emit light at very specific "peaks" -- narrow wavelength bands. Besides flames, these "emission lines" can be produced by exposing gases or sometimes liquids to UV light, lasers, or electric fields (as in a fluorescent bulb). The fluorescent bulb spectrum [you get when calibrating](/wiki/spectral-workbench-calibration) is an example of a mercury emission spectrum.

Emission lines are produced by atoms, not whole molecules (the latter produce absorption lines, which we might still be able to detect since we have the flame -- a good broad-spectrum light source -- but that is just a theory at this point). So sulfur and carbon are possible targets, but we won't be able to distinguish CO2 from CO. 

**Basic setup:** For a more complete description, please read about the "flare spectroscopy activity" below, however, the basic setup involves simply pointing a spectrometer at a flame (which can be difficult to line up if the flame is far away), and later comparing any peaks to known peak locations of looked-for elements. We are compiling a collection of such known elements by importing "idealized" spectra from the NIST database, a process which you can [read more about here](/notes/warren/08-13-2013/importing-spectra-from-nist-and-webmineral-com-to-spectral-workbench). 

Read more about flame spectroscopy:

* [Flare spectroscopy activity](/wiki/flare-spectroscopy-activity) (with description of experimental setup)
* Flare spectroscopy research: http://publiclab.org/tag/flare
* A report from user [straylight](/profile/straylight) on a classroom activity involving [measuring different elements in a bunsen burner flame](/notes/straylight/10-14-2012/classroom-flame-spectroscopy)
* A listing of imported emission spectra of elements from the NIST database so far: https://spectralworkbench.org/tag/nist