Water famously freezes at 0°C although this is strictly speaking not true. If the water is kept in very small quantities and remains very pure then it will only crystallise at below -35°C. In order for ice to begin to form at just under 0°C there needs to be something there to act as a nucleus for the crystals, the thing that most often fills this role is bacteria. To be specific the common bacterium Pseudomonas syringae is the genus that has the ability to produce a certain protein, called the ice nucleation-active protein, in its outer cell wall (other genera such as Erwinia, Xanthomonas and Fusarium also share this ability). This protein is very good at acting as a nucleation centre for ice crystals and so these bacteria are essential for ice to form in the way we understand it. This P. syringae group, being airborne, is also believed to have a role in cloud formation and the eventual precipitation. This last part, however, has been under debate for some time as the ration of biological to inorganic ice nuclei is quite low in the atmosphere and so it is possible there importance has been overestimated.
A recent study has aimed to answer the question by changing focus from cloud chamber experiments and instead look at data gathered from rainfall. The rain was collected in the Hulunbuir (呼伦贝尔市) grassland in Inner Mongolia in Northern China (you might think Inner Mongolia would be in Mongolia, but you’d be wrong). The samples were collected over three years and the concentration of nucleation particles was measured for each sample. Along with this each sample was frozen and the temperature at which they froze was also recorded, they also filtrated the samples and checked freezing points to see how removal of organic nuclei effected it. Finally, because the organic proteins denature with heat, the temperature was raised to boiling for the samples and then refroze with the freezing point recorded once again. Biological nuclei are normally many times better than inorganic nuclei for ice formation, so it would be expected that removing all biological components through filtration would drop the freezing point. This happened but only by a few degrees and not by the large amount that would be expected. The freezing point basically remained the same when this filtered samples was pasteurised and so whatever this this mystery ice nuclei is it is unlikely, basically impossible, for it to be biological. So there is a strange producer of ice that must be smaller then 220 nm and is very resistant to heat. Discovering the nature of this species is a primary goal of atmospheric research.