We present a methodology by which one may distinguish three key contributors to enhanced recombination at pyramidally textured silicon surfaces. First, the impact of increased surface area is trivial and equates to a √3-fold increase in <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">S</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">eff,UL</sub> . Second, the presence of {1 1 1}-oriented facets drives a fivefold increase in <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">S</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">eff,UL</sub> at SiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> -passivated surfaces but a small (1.5-fold) increase for SiN <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</sub> passivation. A third factor, which is often proposed to relate to stress at convex and concave pyramids and edges, is shown to depend on pyramid period (and, hence, vertex/ridge density). This third factor impacts least on <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">S</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">eff,UL</sub> when the pyramid period is 10 μm. At this period, it results in a negligible increase in <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">S</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">eff,UL</sub> at SiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> -passivated textured surfaces but causes at least a sevenfold increase at the Si/SiN <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</sub> interface. Finally, we found that <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">S</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">eff,UL</sub> is 1.5-2.0 times higher at inverted pyramid texture than at surfaces featuring a random arrangement of upright pyramids. The results of this study, particularly for the Si/SiN <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</sub> system, likely depend on process conditions, but the methodology is universally applicable. We believe this to be the first study to distinguish the impact of {1 1 1} facets from those of vertices and edges. Further, we find that {1 1 1} surfaces, rather than vertices and edges, are chiefly responsible for the poor-quality passivation achieved by thick oxides on textured surfaces.
Read full abstract