Burning Water Is Not the Same as Seeing It

The copper sulphate burned green, clean and unmistakable. That part worked. I’d dropped a Nichrome wire into a dish of it, waved it through the Bunsen flame, and there it was — the colour I’d read about in a dozen tutorials. Green. That happened.

Then I dipped a wire into a sample of my tap water — the same 180 ppm stuff that kills the Venus flytrap if you bottom-water it for more than a few days — and held it into the flame.

It went yellow-orange and did not stop.

The brightest yellow-orange I’ve ever seen anything produce while not actively on fire. Sodium. Ninety percent of the periodic table was presumably in that sample — calcium, magnesium, potassium, whatever minerals Alberta’s aquifer decides to ship through the pipes — but the flame showed only sodium. The sodium said hello, shook hands, sat down at the table, and ate dinner while everything else stood in the kitchen waiting.

This is apparently the first beginner mistake in flame spectroscopy. Not “hold the sample too long and ignite the wire,” which is what I’d prepared for. Not “forget that halide salts emit better than oxides,” which I understood from brewing water profiles. No — the mistake is discovering that sodium is not one element among many in your sample. Sodium is a bully. Sodium is brighter than every other element’s contribution combined, and if sodium is anywhere in your sample — which it is, because sodium is everywhere in tap water and sweat and dust and the ancient decision to salt the earth — sodium wins.

So I reached for the cobalt blue glass, the optical filter that’s supposed to let you see through the sodium’s glare. I’d read about it: cobalt blue absorbs yellow and lets the other colours through, a simple hack from the Victorian era. It was on its way from the same vendor who shipped the blank DVD I’d already sacrificed for a diffraction grating.

While waiting, I decided to photograph the copper sulphate spectrum anyway — the one thing that had worked. I set up the Bunsen burner, the DVD grating on a makeshift stand, and my smartphone on a tripod focused at the spectrum. The green of the copper should diffract into a clear line. I could photograph it, measure the angle on the sensor, and run through a calibration against known wavelengths. Proof of concept.

The photograph came out grey.

Not “the spectrum is faint,” grey. Not “I need longer exposure,” grey. The kind of grey that means the light I’m trying to photograph is being entirely swallowed by the background illumination of the room. The smartphone camera is stupidly sensitive to the infrared heat pouring off the flame and the far-field reflection of the Bunsen’s broad yellow. The blue-green of the copper is a thin slice of the spectrum, and it’s vanishing into noise. I tried closing the aperture on the phone — did nothing. I tried shooting in a darker room — marginally better, still useless. I tried angling the grating to catch the light at a lower angle to the sensor. Nothing.

The problem is that I’m trying to photograph something a few millimetres across, at distance, with a smartphone camera that has no actual aperture control and a fixed lens. A spectrometer needs to collect the narrow beam of light and magnify it or focus it. A diffraction grating spreads light over a wider angle, which makes the individual wavelengths visible to the human eye — but the light intensity per wavelength drops. The eye integrates over time and area; it sees the spectrum. The camera has to gather enough photons from a specific spot, and I’m not giving it enough.

The tutorials online all show the spectrum photographed either with a proper camera (aperture, shutter speed, optics), a point-source light far away (laser or arc lamp, which concentrates intensity in a narrow beam), or shooting in a darkened room with the flame as the only light source. I’ve got a smartphone, a Bunsen burner on a kitchen counter with ambient afternoon light, and a flat DVD. This is the part where the theory meets the specific constraints of trying this in a real room, and theory loses.

The cobalt blue glass arrived this morning. I’ll try again with the filtered sodium flame — at least that should remove one variable, and I can see with my eye if the filtered spectrum shows any structure at all. But I’m staring at the copper-green photograph I managed to pull from the sensor, and it’s a grey smear, and somewhere between the physics I’d read and the equipment I could afford, the elegant idea of building a personal reference atlas of elemental spectra dissolved into the same problem I’ve hit a dozen times before: the gap between knowing how something works and being able to actually see it with the tools at hand.