Alright, way too much google time necessary to be figuring all that out.
Here's the simple & abridged version, skipping a lot of the details:
Information is carried over fiber optics by laser light, usually in the near IR. If you can divide the laser wavelength into really tiny "slices" you can carry more information. This requires filters.
Spectral filters are made by putting together a "stack" of materials with varying index of refraction. If you zap an optical fiber from the side with an ultraviolet laser in the right way, you can change the fiber core's index. Do this in a controlled manner, and you can get a "stack" of different indices in the core that's really long, with a really narrow bandwidth. (This is called a fiber Bragg grating.)
Now, the fiber itself isn't naturally very sensitve to being laser irradiated, but if you let it "soak" in a high-pressure hydrogen atmosphere for a while, the hydrogen diffuses into the fiber and increases it's sensitivity to UV laser irradiation. This is normal SOP. However, the hydrogen can form OH groups that for
some wavelengths cause absorption, making for a "lossy" filter. Using deuterium rather than hydrogen shifts the absorption peak and reduces losses.