No abstract

An amorphous mixing layer (3.5–4.0 nm in thickness) containing silicon (Si), oxygen (O), molybdenum (Mo) atoms, named <i>α</i>-SiO_<i>x</i>(Mo), is usually formed by evaporating molybdenum trioxide (MoO₃) powder on an n-type Si substrate. In order to investigate the process of adsorption, diffusion and nucleation of MoO₃ in the evaporation process and ascertain the formation mechanism of <i>α</i>-SiO_<i>x</i>(Mo) on a atomic scale, the first principle calculation is used and all the results are obtained by using the Vienna <i>ab initio</i> simulation package. The possible adsorption model of MoO₃ on the Si (100) and the defect formation energy for substitutional defects and vacancy defects in <i>α</i>-SiO₂ and <i>α</i>-MoO₃ are calculated by the density functional theory. The results show that an amorphous layer is formed between MoO₃ film and Si (100) substrate according to <i>ab initio</i> molecular dynamics at 1500 K, which are in good agreement with experimental observations. The O and Mo atoms diffuse into Si substrate and form the bonds of Si—O or Si—O—Mo, and finally, form an <i>α</i>-SiO_<i>x</i>(Mo) layer. The adsorption site of MoO₃ on the reconstructed Si (100) surface, where the two oxygen atoms of MoO₃ bond with two silicon atoms of Si (100) surface, is the most stable and the adsorption energy is －5.36 eV, accompanied by the electrons transport from Si to O. After the adsorption of MoO₃ on the Si substrate, the structure of MoO₃ is changed. Two Mo—O bond lengths of MoO₃ are 1.95 Å and 1.94 Å, respectively, elongated by 0.22 Å and 0.21 Å compared with the those before the adsorption of MoO₃ on Si substrate, while the last bond length of MoO₃ is little changed. The defect formation energy value of neutral oxygen vacancy in <i>α</i>-SiO₂ is 5.11 eV and the defect formation energy values of neutral oxygen vacancy in <i>α</i>-MoO₃ are 0.96 eV, 1.96 eV and 3.19 eV, respectively. So it is easier to form oxygen vacancy in MoO₃ than in SiO₂, which implies that the oxygen atoms will migrate from MoO₃ to SiO₂ and forms a 3.5–4.0-nm-thick <i>α</i>-SiO_<i>x</i>(Mo) layer. As for the substitutional defects in MoO₃ and SiO₂, Mo substitutional defects are most likely to form in SiO₂ in a large range of Mo chemical potential. So based on our obtained results, the forming process of the amorphous mixing layer may be as follows: the O atoms from MoO₃ bond with Si atoms first and form the SiO_<i>x</i>. Then, part of Mo atoms are likely to replace Si atoms in SiO_<i>x</i>. Finally, the ultra-thin buffer layer containing Si, O, Mo atoms is formed at the interface of MoO₃/Si. This work simulates the reaction of MoO₃/Si interface and makes clear the interfacial geometry. It is good for us to further understand the process of adsorption and diffusion of atoms during evaporating, and it also provides a theoretical explanation for the experimental phenomenon and conduces to obtaining better interface passivation and high conversion efficiency of solar cell.

No abstract

Solar-grade p-type Czochralski silicon wafers are doped with phosphorus by single-face and double-face diffusions, and the influence of Fe-B pairs on the minority carrier lifetime, the trapping centers density and the internal quantum efficiency （IQE） of cells （fabricated from the wafers） is analyzed by measuring microwave-detected photo-conductance decay minority carrier lifetime. In the doped wafers with single-face diffusion, the minority carrier lifetime is determined mainly by the density distribution of Fe-B pairs. However, in the doped wafers with double-face diffusion, the minority carrier lifetime is less influenced by the concentration distribution of Fe-B pairs than by other impurities and defects. Numerical calculation based on the combination of the transient voltage signal and the trapping model indicates that the density of trapping centers is reduced by either of diffusion process. On the other hand, detailed analysis of selected specific spots in one wafer with single-face diffusion shows that Fe-B pairs are not the major factor influencing the density of trapping centers. The wafers with different diffusion processes are fabricated into c-Si solar cells and the IQE is measured by using a light beam induced current. The results show that the IQEs of cells with double-face diffusion are higher than those with single-face diffusion, which demonstrates the negative effect of Fe-B pairs on the IQE of solar cells.