Tropospheric Mineral Particles as Ice Nuclei

Besides crystallization as salts, atmospheric particles also crystallize as ice at sufficiently cold temperatures and high relative humidities. When cirrus clouds form, approximately 1 particle in 1000 is mineral dust in background upper tropospheric aerosol, which can be compared to 1 in 2 collected cirrus ice particles having a core mineral particle. The implication is that these mineral particles are effective ice heterogeneous nuclei.

Quantitative modeling of the role of mineral dust in cirrus cloud formation would require heterogeneous nucleation rates (j, #-freezing-events per cm² per sec). Using the aerosol generator developed for the crystallization experiments, we studied the ice freezing of aqueous ammonium sulfate layers on Fe₂O₃ and Al₂O₃ cores of variable diameter.

Aerosol Flow Tube (AFT) coupled with FT-IR. In this setup the AFT is temperature controlled with two independent chillers. Also, in the background there are two tube furnaces used to oxidize nebulized precursor solution. The oxides are subsequently coated in condensation furnace before entering the AFT.

An initial challenge we faced was that results had disagreed for the several groups working with the aerosol flow tube (AFT) technique. To explain these contradictory results, we examined the fundamental processes occurring in the AFT. As originally conceived, aerosol particles would freeze independently, and ice freezing observed in the infrared spectra would correspond to 1 in 10 particles freezing (based upon infrared sensitivity). However, we showed through experiments and modeling that in fact 1 particle in 10⁶ can freeze under certain conditions, which is followed by pumping of water vapor from nearby aqueous particles to the ice particle. Large ice particles then form and are observable in the IR spectra. This result has tremendous implications for the accurate inversion of freezing results to obtain rates of homogeneous nucleation (J, #-freezing-events per cm³ per sec). The J values are the quantitative product of the AFT technique, and they serve as the input to models of cirrus cloud formation. Once we had this problem solved, we returned to the study of heterogeneous nucleation rates to obtain j values. These values can now be employed to make quantitative predictions of the effects of mineral dusts on cirrus cloud formation.