The effect of sample edge recombination on the averaged injection-dependent carrier lifetime in silicon

In semiconductors, the effective excess carrier lifetime, τeff, measured in dependence on the injection density, Δn, is an important parameter. It is frequently observed that τeff decreases with decreasing Δn at low-level injection conditions (where Δn is smaller than the dopant density Ndop), which has been difficult to explain. We compare measurements with numerical device simulations to demonstrate that this observed reduction of τeff is caused by a combination of (i) Shockley-Read-Hall (SRH) recombination at the edges of the sample and (ii) transport effects of the carriers toward the edges. We measure τeff(Δn) of boron-diffused and surface-passivated p+np+ and p+pp+ silicon wafers with the commonly applied photo-conductance decay technique, and we vary the sample size. The photo-conductance is probed by inductive coupling within a sample region of about 3 × 3 cm2; hence, the measurements yield an average value of both τeff,av and Δnav within that region. For a detailed analysis, we determine τeff with a high spatial resolution using the dynamic infrared lifetime mapping technique, which shows a strong decrease of τeff toward the edges of the p+np+ samples at low-level injection. We analyze the measurements by numerical device modeling and circuit simulation. We conclude that the sample size should be at least 6 × 6 cm2 for reliable τeff(Δn) measurements at low injection conditions. However, at high-injection conditions, the recombination usually dominates at the dopant-diffused surfaces. Therefore, the saturation current-density, J0, can be extracted from the τeff(Δn) measurements in samples as small as 3 × 3 cm2, with a measurement error due to edge recombination below 10%.