A combination of electron paramagnetic resonance ͑EPR͒ and minority carrier lifetime measurements is used to unambiguously demonstrate that the presence of a B diffusion layer at the surface of oxidized Si ͑111͒ wafers causes a significant increase in the interface defect density as well as interface recombination, compared to undiffused surfaces. EPR measurements show a nearly three-fold increase in the Pb center density, while the lifetime measurements indicate an increase in surface

more »

... tion activity by a factor of more than two, for B diffused samples with a sheet resistance of ϳ250 ⍀ / ᮀᮀ. Shallow, heavily doped regions play an important role in commercial silicon solar cells, as they are used both for the formation of the p-n junction as well as for high-low junctions in order to reduce recombination at the surfaces. The heavily doped region is most often formed through the diffusion of a suitable impurity ͑phosphorus or boron͒. The importance of these regions necessitates a good understanding of their properties. This includes the effect of the introduction of dopant atoms on the density and properties of defects at the interface of the silicon and the dielectric layer ͑such as SiO 2 or PECVD SiN x ͒ used to passivate the silicon surface. The Si-SiO 2 interface in particular is of considerable interest due to the excellent surface passivation that it affords. The investigation of heavily doped silicon interfaces and, in particular, the determination of interface defect densities and associated capture cross sections, is more difficult than for the case of moderately doped material since capacitance-voltage ͑C-V͒ measurements-the chief characterization technique-only yield limited information for highly doped material and with a greater error margin. In what appears to be the only study of its kind, the effect of B and P diffusions on this interface was investigated by Snel, 1 using C-V measurements at 77 K. He concluded that both types of surface doping lead to an approximately linear increase in interface defect density with surface doping above a certain doping concentration, with the onset occurring at a lower dopant concentration for B than P doped surfaces. However, the results of this study were never replicated. Further, the values obtained for the interface defect density D it for moderately doped silicon are inconsistent with the values from the recent studies. 2 In this paper, we investigate and compare the Si-SiO 2 interface of B diffused and undiffused, thermally oxidized, ͑111͒ oriented samples using electron paramagnetic resonance ͑EPR͒ measurements, and minority carrier lifetime measurements on metal-oxide semiconductor ͑MOS͒ structures as a function of applied voltage. 3 The use of lifetime measurements on MOS structures at a sufficiently large applied bias voltage allows direct comparison of the emitter saturation current J oe of diffused and undiffused samples. For undiffused samples, a thin accumulation or inversion layer is created in the Si surface region when sufficient voltage bias is applied, making the measurement of J oe values possible. The lifetime-voltage technique has the advantages over the more commonly used corona charging method of eliminating the possibility of corona introduced interface damage 4,5 and surface potential fluctuations, 6 which can render the interpretation of results on corona charged samples extremely difficult. EPR measurements allow direct determination of the density of unpassivated P b centers, the chief defect on ͑111͒ oriented Si-SiO 2 interfaces. 7 Boron diffusions were carried out using a liquid boron tribromide source in a diffusion furnace at a temperature of 900°C to obtain a sheet resistance of around 250 ⍀ / ᮀ, with variation across the sample of less than 5%. Figure 1 shows the boron profile resulting from this diffusion. Samples used for EPR measurements were n type, ϳ4000 ⍀ cm, ͑111͒ Cz silicon wafers. Samples were cut with a diamond saw into 25 mmϫ 2.5 mm 2 pieces. They were subsequently etched to remove saw damage from the surfaces. Samples for lifetime-voltage measurements were a͒ Electronic mail: hao.jin@anu.edu.au. FIG. 1. The boron profile of the samples used for measurements. APPLIED PHYSICS LETTERS 92, 122109 ͑2008͒