The results of a study conducted in the mid-1950s show that it's not high concentrations of deicers that cause the heaviest scaling, but relatively low amounts in the range of 2% to 4%.
In the study, non-air-entrained concrete was tested after exposure to 25 freeze-thaw cycles, and air-entrained concrete was tested after exposure to 50 and 200 freeze-thaw cycles. During freeze-thaw cycling, both concretes were exposed to different concentrations of various deicers, including sodium chloride, calcium chloride, urea and ethyl alcohol. Graphs of degree of scaling vs. deicer concentrations for both concretes show low scaling at zero deicer, a rapid rise in scaling as deicer concentrations reach 2% to 4%, and then, depending on the deicer involved, a moderately slow to rapid drop-off in scaling as deicer concentrations continue to rise. According to the researchers who conducted the study, two major destructive mechanisms--hydraulic pressure and osmotic pressure--cause this phenomenon.