Influence of oxygen on the dislocation related luminescence centers in silicon

Abstract

Dislocation related luminescence (DRL) centres in Si have high stability to a thermal treatment of samples and a relatively low temperature quenching. These properties make them an attractive candidate for production of Si based light emitting diodes (LEDs). The low energy part of DRL in the vicinity of the D1 line is the most promising from this point of view due to its highest temperature stability and best coupling to fiber optics. Actually this part of DRL can be divided into several bands. One of them with a position 807 meV is known as D1 line. The substantial effect on D1 luminescence has oxygen. The line becomes broader and several subbands can be identified on the low and high energy side after prolonged annealing of deformed samples. Depending on particular treatment some of these bands could be even more intensive than the original D1 line. The strongest effect is usually observed in oxygen rich CZ crystals. In all cases the presence of dislocations is necessary for the appearance of the low energy bands. Several observations clearly show that the corresponding recombination centres include the dislocation related defect as well as some oxygen complexes. It implies that inclusion and mutual configuration of constituents defines the energy of optical transitions in the region of D1 band. In the present paper a review of previous investigations as well as new results regarding oxygen dislocations interaction are presented.

Time‐resolved measurements of dislocation‐related photoluminescence bands in silicon

“…1). According to [5,7] the dependence of the high energy shoulder on the excitation power implies its DA origin, as has been previously assumed for the low energy shoulder [4]. It should be noted that these two shoulders in general can not be observed simultaneously.…”

“…However, it is not the standard D1 line at 807 meV, but a wide band with a maximum around 813 meV. Moreover, the position of this band is not fixed, but varies due to the particular parameters of the thermal treatment [7]. This point is well illustrated by the spectrum of sample b2-1 obtained after quenching of the as-deformed sample in water.…”

“…Clearly, this shoulder is caused by additional annealing at 950 °C during the gettering procedure. Sample b1-1 demonstrates a typical spectral distribution of DRL intensity for Cz grown samples [7]. However, it is not the standard D1 line at 807 meV, but a wide band with a maximum around 813 meV.…”

“…1. The inset of the figure shows the full DRL spectrum region, from which it is clear that the main changes in spectral distribution of intensity due to the different histories of the samples happen in the vicinity of the D1 band [7]. Samples a2d13 and b2-1 should be free from metal impurities due to gettering in a chlorine atmosphere.…”

“…Although necessary precautions were taken to avoid unintentional metallic contamination of the samples during plastic deformation, two of them were subjected to an additional gettering step in a Cl atmosphere to remove residual metallic impurities. A comparative investigation of Fz and Cz grown samples was stimulated by the strong dependence of the D1 band on the condition and concentration of oxygen in the vicinity of dislocation core [7]. Details of the samples are presented in Table 1.…”