Ice clouds play an important role in the Arctic weather and climate system. Consequently, it is essential to fully understand their properties and especially their formation process. Measurements from ground-based sites and satellite over the Arctic have revealed the existence of two types of ice clouds (TIC). The first type, TIC-1, is characterized by a high concentration of small ice crystals and is typically observed in non-polluted areas. The second type, TIC-2b, is characterized by a low concentration of larger precipitating ice crystals. Recent studies suggest that TIC-2b are linked to highly polluted air masses. Past field experiments have shown that most aerosols in the accumulation mode are coated by sulphuric acid in polluted episodes in the Arctic during winter and spring. Recent laboratory experiments have shown that sulphuric acid coating can alter the efficiency of ice nuclei (IN) to nucleate ice crystals. In this study, we hypothesize that the resulting lower IN concentration found in polluted air masses leads to the decrease of the ice crystal concentration. Since there is in less competition for the available moisture, ice crystals quickly reach precipitating sizes leading to the formation of TIC-2b. This research aims to better understand the formation process of these two types of TIC using a new ice nucleation parameterization. This parameterization is based on a combination of the results from laboratory experiments and the classical heterogeneous nucleation theory. It has been implemented into the Limited Area version on the Global Multiscale Environmental Model (GEM-LAM). Other empirical parameterizations are also tested in this study. TIC-1 and TIC-2 observed during the Indirect and Semi-Direct Arctic Cloud (ISDAC) field experiment are simulated with GEM-LAM using respectively the uncoated IN parameterization and the acid-coated IN parameterization. TICs-2b observed on April 15th and 25th during ISDAC (polluted cases) and TICs-1 observed on April 5th (non-polluted cases). For each case, polluted and non-polluted, a statistical analysis is made in order to determine which parameterisation is more accurate for a given case.