I once conducted a set of experiments in which I had to form ice by direct vapor deposition (not condensation of liquid water followed by freezing, aka riming) onto a Peltier-cooled glass-coated copper rod.
The experiments were conducted outside at night within a small wind tunnel, with ambient air temperatures ranging from -2 oC to -25 oC and the glass substrate was cooled from 1 oC to 10 oC below the ambient air temperature. Occasionally supercooled water droplets would condense, followed by freezing (forming rime ice) rather than direct deposition of ice crystals from the vapor phase. These events would ruin my time-consuming experiments.
The only relevant parameter under my control was the temperature of the glass-coated copper rod. The only relevant parameters I could measure during the experiment were the air temperature and humidity. I also introduced trace-levels (single to hundreds of ppb) of several C1 to C4 alcohols in order to later measure their incorporation into ice during ice deposition from the vapor phase. However, the low levels of these compounds did not affect whether riming or direct vapor-ice deposition would occur, as based on a series of 'blank' experiments without the trace gases.
If I were having trouble with liquid water condensation, in what way would I want to alter the glass substrate temperature so that only the direct deposition of solid ice could occur?