Further experiments with depositing pure zinc oxide yielded highly variable results, with slide resistances between 500,000 and 30,000,000 ohms per inch. These were obviously far too high to be useful, and I believe the variability was due to the un-doped nature of the oxide film. Pure semiconductors tend to be insulators, so my assumption is that any conductivity was simply due to defects in the deposition process, which naturally would be different for every slide. At this point I decided to add boric acid to the precursor solution, in order to dope the resulting films with boron. The new solution was made with the following ingredients:
I chose boron as a dopant for two reasons. First and foremost, I chose it because it has three valence electrons while zinc has only two. When added in small quantities (in this case, 2% of the metal ions in solution), it will fit into the ZnO crystal structure, leaving the extra electron available to conduct electricity. Second, I chose it for its non-toxicity. Any group III element can act as an electron donor; however, salts of aluminum, gallium, indium, and especially thallium, are toxic. Boric acid on the other hand is fairly benign, with an LD50 of 5g/kg.
After mixing this solution, I tested a number of variations on the deposition method. I quickly discovered that keeping the slide at or above 400°C was absolutely necessary; below this temperature, the films produced were opaque, rough, and non-conductive. Furthermore, I discovered that longer spray bursts produced more conductive films, when compared to slides that received the same amount of solution in shorter intervals. I then extrapolated this and found that a continuous spray is ideal, however the spray pressure must be very low so that the glass isn't chilled from the blowing air. The slide at the top of the page was produced by this method, receiving approximately 10mL of solution over the course of 4 minutes. It is highly transparent, and has a resistance of 7,500 ohms per inch in the central area. This is an improvement of nearly two orders of magnitude from the previous slide, and routes to further improvement are still available. With this in mind, I fully expect that I will be able to cheaply produce useful conductive glass in the near future.