First, how many arsenate ions would it take to completely coat the inner surface of one of the glass tubes I'm using? The tubes are about 10 cm long, with an inner circumference of about 2 cm, so that's about 20 cm^2 of glass surface. This paper says that the ionic radius of arsenate ions is 0.248 nm, so I'll assume a diameter of 0.5 nm. If the ions were to pack squarely side-by-side onto the glass surface, there would be about 2x10^3 per µm, or about 2 x 10^6 per mm. That's about 4 x 10^12 per mm^2, or 4x10^14 per cm^2. So if the inner surface of a glass tube were densely coated with a monolayer of arsenate ions, it would sequester about 8x10^15 ions.
Next, how many arsenate ions are in 5 ml of a 40 mM solution?
6x10^23 ions/mole times 0.04 moles/liter times 0.005 liter = 1.2x10^20 ions.So less than 0.01% of the arsenate ions in my medium could be tightly packed in a monolayer on the inner surface of a glass tubes. The assumption of dense packing is very conservative, so I don't think sequestration of arsenate by the glass can be the explanation for the GFAJ-1 growth I'm seeing. But thank you to the commenters for prompting me to do the calculation.