Bill, The naturally occuring oxide layer is only approximately 50 nm thick. I use an electrochemical method to increase the natural oxide to around 1000 nm. At this thickness the surface is quite durable. We use the mirrors to observe Nitrogen fuorescence in the near ultra-violet region 300 nm - 400 nm. I optimized the oxide thickness for 1/2 wave at 337.1 nm (N2 laser line). With this oxide thickness I get from 85% to 90% reflectivity at 337.1 nm.
You are quite right that alkaline solutions will attack the aluminum. In fact, I use an NaOH solution to strip off the old aluminum coating when I want to re-aluminize a mirror. A bare aluminum mirror will strip very quickly using sodium hydroxide but an Al2O3 protected mirror takes a bit more work. :)
-- Stan Thomas <thomas@cosmic.physics.utah.edu>