Neon Spectral Power Distribution Peak Identifier Script

Fig. 1

Pure noble gas spectra have distinct shapes and produce peaks at distinct wavelengths.

Fig. 2

Ruby red glass tubes are filled with different noble gases— producing spectra similar to their respective gases

Fig. 3

60+ experimental tubes created by David Ablon, each representing a common recipe used in his fabrication studio.

I developed an experimental technique to reliably identify the gas-fill of a “neon” light unit using a handheld spectrometer. Spectrometers produce ‘spectral power distributions’ by measuring the energy levels at each wavelength in the visual light spectrum. Spectral power distributions serve as a “fingerprint” of a light source, as it is the key to how all light sources render colors. Fig. 1 shows the pure noble gases ionized in clear glass. As you can see, their spectra have distinct shapes and produce peaks at distinct wavelengths, allowing us to identify the intangible, imperceptible gas fill sealed in a glass envelope.

This technique becomes incredibly valuable for identifying tubes used in artworks or historical signs for documentation and preservation purposes. Fabricators chose certain ‘recipes’ to achieve exact colors, and each have their own aging properties.

The differences become evident when we look at the spectral power distributions. In fig. 2 we see distinct spectral peaks produced by each tube demonstrating that there are two different gas fills. If we compare these spectra to the noble gas references I created we see the distinct pure argon distribution produced by the top tube and the distinct neon spectral distribution on the bottom. So if one of these tubes was in an artwork or a sign that required replacement in the future, we are able to capture spectral power distributions and identify the gas fill before it fails.

Neon Peak Identifier Python Script

I worked with neon fabrication extraordinaire David Ablon in Brooklyn to create 60+ experimental tubes, each representing a common recipe used in the studio (fig. 3). I realized there was so much spectral data to parse through from all the units, so I decided to spend my time making a program that was scalable and sharable.

Creating a python script that plotted and identified spectral power distributions required preparing the raw data from the spectrometer (see figs. 4-9). The raw data was downloaded as a csv file generated by the spectrometer’s software web platform. By using a Python module called pyplot, I cleaned up the raw data of each unit and automated the identification of peaks and their corresponding wavelengths on a graph. Then I extracted those peak wavelengths and wrote them to a text file for each experimental tube. From this master list, I am able to compare the peak wavelengths of all experimental units.

See video for the tool in use!

For access to reference and experimental spectral data click here. The images and wavelength spread sheet can be used to compare your own spectral data and identify gas fills!

Fig. 4

Spectrum Genius is the spectrometer’s web platform. It stores all of the captured spectral data.

Fig. 5

A report for each measurement can be downloaded as a csv or text file.

Fig. 6

The report included a ton of data beyond the wavelength and intensity that I was interested in. I created a python script to clean up the csv— isolating the wavelength and intensity as X and Y axes.

Fig. 7

Excerpt of python script

Fig. 8

The wavelength and intensity were plotted on a graph, and the peaks were identified by using a pyplot function called "find_peaks”. The peak wavelength values were then written to a text file.

Fig. 9

All of the identified peak wavelengths were appended to a spread sheet and organized according to their recipe category