I went down a rabbit hole for a couple of weeks investigating the bands in the heron’s sky, exploring dead ends, trying things that didn’t work (many calibration test sheets), and otherwise having many adventures. Part way through, I realized that the lighter blue tracked exactly where there was originally cyan where the layers called for ablating blue (would have seen this sooner, but I was mistakenly looking at the PNGs that report where dots are supposed to end up in the end, not the PNGs that specify where to ablate). This led to the conclusion that the “blue anodizing to cyan instead of blue” effect had something to do with the fact that those areas had been cyan previously. I had noticed that there was a slight brownish tinge in those areas, and that suddenly seemed much more relevant.
I thought that perhaps I was using too much energy to ablate which was causing residue. Lowering power (shorter duration and/or lower power) didn’t fix it, and many settings didn’t ablate at all because there’s a pronounced threshold effect.
So then I thought maybe I was using too little power. Higher power didn’t help either.
After extensive test sheets, I am mostly convinced that there is no sweet spot that properly ablates cyan in preparation for anodizing blue, and this seems to also be true for ablating silver in preparation for anodizing blue. My working theory is:
- The silver and cyan layers have oxide’s strong effect of having a "hreshold below which no ablation happens.
- They are thin so they absorb less energy once the threshold is exceeded.
- Absorbing less energy from the ablation and having enough energy exceed the threshold results in the metal heating enough to react with the air which leaves the brownish tinge.
I assumed the layer was oxide produced by the heat. I was puzzled how having a start on the oxide layer could end up making a different color than bare metal. I’ve learned that hysteresis effects happen from just about anything, but the impact of the brownish tinge was pretty extreme (the equivalent of 29 volts instead of 23 volts). Why didn’t the oxide stop forming roughly when the new layer grew on top of the brownish oxide so the two were at the blue thickness would would have the resistance needed to counter the 23 volts that anodizes blue? It instead went to the depth caused by 29 volts that produces cyan.
I experimented with directly coloring titanium both titanium and 304 stainless with the F1 (was surprised how well that worked on my first attempt doing small squares which I now realize are the easiest shape to color). I was surprised that the best colors resulted from using the blue laser. I’m intrigued by that approach and may see what kinds of images I can produce with direct laser heating (like Povilas has done so well). I might even try anodizing a relatively thin layer (probably blue) and seeing if I can increase the oxide depth by heating the anodized layer in selected areas. I would love to be able to develop a technique people could use to make color images using my separator by directly heating the metal instead of having to set up the anodizing stuff just to get their first image. But there are many possible ways this could have challenges. One I’m sure will arise is that temperature on the metal will be affected by the density of the dots which dithering inevitable varies. And 100mm width means the metal will have time to cool before the laser gets back to any given area. This could be another fun puzzle but for now at least I want to make some more actual images.
I had lots more brainstorming conversations with both Gemini and Claude, and eventually one of them suggested that the brownish tinge might be something other than oxide. It informed me that if the metal gets hot enough in air, nitrogen can also react. It said that if nitrogen was reacting, it was probably forming titanium oxy nitride. It might also be titanium nitride (which makes some drill bits yellow), but that requires higher temperatures and the oxygen present in the air would most likely also react). Importantly, it told me that I could determine if nitrogen was involved by measuring the layer’s resistance. Titanium oxide has high resistance, oxy nitride or nitride have low resistance. So I got out my multimeter and carefully measured resistances (gently because I didn’t want to get low resistance because I scratched through the layer). I consistently got low resistance in the brownish areas and high resistance in the oxide layers.
At that point, I realized that the low resistance areas would anodize the full thickness brown voltage oxide on top of them. Both AIs agreed that this thicker combination would likely have similar optical properties to a pure oxide layer of the same thickness. If the oxy nitride layer’s thickness was roughly the difference in thickness between blue and cyan oxide layers, that would explain the mysterious “anodizing blue makes cyan” behavior I observed.
I had already made a feature for the separator that segregates the “blue over cyan” dots into a separate PNG so I could try ablating them with different settings. I realized I could ablate the two regimes of blue (areas not over cyan and areas over cyan) separately and anodize them separately at different voltages. I would anodize the not over cyan region at the normal voltage and then anodize the over cyan region at a lower voltage to grow the oxide just enough to make the combination match the blue thickness. I tried this, starting at a deliberately too low voltage (got brown at first) and increasing voltage until the boundary between the blue and blue on cyan was minimized. I think I overshot the voltage a little, and the presence of the brown seems to have affected the hue a little, but the result was a big improvement.
Here’s the result:
This investigation took a lot of time away from actually making new art, but it was a fun puzzle, I learned a lot, and I’m feeling good about eventually being able to make a prediction and see it work.
I haven’t tried it yet, but I bought “wine preserver” argon gas and plan to build a frame to hold it on the sheet (argon is denser than air, so it should stay put ok) while I ablate the silver and cyan before anodizing blue. Argon is an inert noble gas that won’t react with the Ti no matter how hot it gets, so this should prevent formation of oxide, nitride, and oxy nitride, leaving nice bare metal for the blue 23 volts to anodize consistently regardless of what colors were under it. The argon is expensive ($18 for a can that AI estimates is enough to fill the frame 6 times), but I can get it for much less per volume if I get a larger amount sold in volume for welding. And I only need it for blue when it’s dithered with cyan and (I think) silver and maybe for purple dithered with those colors and blue. Fortunately, brown anodizes well over those colors even though it’s an even lower voltage than blue and purple. I’m not sure why, but I’m grateful. Brown dithering well with all colors is lucky because it gets dithered a lot on pretty much every piece. And I don’t use purple much because it doesn’t seem to be picked by the separator much.
