Based on the work of the Somorjai group [Magni and Somorjai, Catal. Lett. 35, 205 (1995)] we have prepared a thin well ordered MgCl2(001) film by MgCl2 evaporation from a Knudsen cell. This film does not absorb TiCl4 at room temperature if it is not activated by increasing the defect density via electron or ion bombardment. The nature of some of the defects created is characterized by in situ ESR measurements and Auger spectroscopy. Paramagnetic surface defects are altered by the bonding of TiCl4 to the surface as observed by ESR spectroscopy. Ti3+ centers are detected if particularly severely defected MgCl2 layers are prepared. Reactivity studies show however, that these species are not correlated with polymerization activity. Interaction with aluminum alkyl leads to the formation of the active catalyst and we observe for the first time directly ethyl radicals formed from trimethyl-aluminum in an abstraction process which may be formulated as TiCl4/(surface)+AlMe3-->Me-TiCl3/(surface)+AlMe2Cl, Me-TiCl3/(surface)-->TiCl3/(surface)+Me., and Me.+Me3Al-->C2H5.+AlH(Me)(2). The presence of the aluminum alkyl is observed via in situ IRAS in the same apparatus.