Positronium emission from single crystalline Al2O3, MgO and vitreous a-SiO2 surfaces was studied as a function of the positron implantation energy E by means of Doppler broadening spectroscopy and Compton-to-peak ratio analysis. When the Ge-detector is in-line with the positron beam, the emission of para-positronium yields a red-shifted fly-away peak with intensity I-pPs(e). An analysis of I-pPs(e) versus E for Al2O3 and MgO where no Ps is formed in the bulk (f(Ps)=0) results in positron diffusion lengths L+(Al2O3)=(18+/-1) nm and L+(MgO)=(14+/-1) nm, and efficiencies for the emission of Ps by picking up of a surface electron of f(pu)(Al2O3)=(0.28+/-0.2) and f(pu)(MgO)=(0.24+/-0.2). For a-SiO2 the bulk Ps fraction is f(Ps)(a-SiO2)=(0.72+/-0.01), f(pu)(a-SiO2)=(0.12+/-0.01) and the diffusion lengths of positrons, para-positronium and ortho-positronium are L+(SiO2)=(8+/-2)nm, L-pPs(SiO2)=(14.5+/-2) nm and L-oPs(SiO2)=(11+/-2)=nm. Depending on the specimen-detector geometry the emission of Ps at low implantation energy may cause either an increase or a decrease of the width of the annihilation line shape at low implantation energies