Surface photovoltage minority carrier lifetime/diffusion length analysis of copper contaminated silicon was performed. It was observed that copper and copper associated defects degrade minority carrier lifetime more in n-type than in p-type silicon. This finding is explained by analysis of copper related defect levels identified by other deep level transient spectroscopy studies. In copper contaminated p-type silicon, an optical or thermal activation procedure significantly degrades the diffusion length. A process similar to that of Fe–B in p-type silicon is proposed. The activation process dissociates the Cu–Cu pairs, a weak recombination center in p-type silicon, and the copper forms extended substitutional defects in silicon, which have much greater recombination activity. No recovery of diffusion length was observed following such an activation procedure. The difference in copper and iron diffusion length recovery properties after activation can be used to differentiate iron contamination from copper contamination.

A quantitative analysis of diffusion of iron in silicon dioxide is presented. A source of iron deposited on the surface of thermally oxidized silicon wafers was diffused at temperatures ranging from 700-1100ЊC in an inert (nitrogen) ambient. The iron concentration in SiO 2 and Si was measured using total reflection X-ray fluorescence, deep level transient spectroscopy, and surface photovoltage techniques. A two-boundary diffusion model was applied to the experimental data to determine the diffusivity and segregation coefficient of iron in SiO 2 . It is observed that iron diffusivity in SiO 2 follows the Arrhenius relationship and has a thermal activation energy of 1.51 eV. Iron exhibits a strong tendency to segregate into silicon dioxide and has a value of k ϭ 1.1 ϫ 10 Ϫ7 at 1000ЊC, where k ϭ N Si /N oxide .

Temperature dependent iron precipitation in float zone grown silicon wafers has been experimentally investigated. Results of iron precipitation experiments over a wide thermal process temperature range and time are presented. Precipitation of iron in silicon was analyzed by a quantitative assessment of change in interstitial iron using a surface photovoltage minority carrier lifetime analysis technique. Contamination levels of iron in the range 10 11 -10 13 atoms/cm 3 are investigated. It is concluded that maximum iron precipitation occurs in the temperature range of 500-600°C. Iron precipitation is rapid in this region where more than 90% of the interstitial iron precipitates in a period of 30 min.

Surface photovoltage minority carrier lifetime/diffusion length analysis of copper contaminated p-type silicon was performed. It was observed that an optical or low-temperature thermal activation procedure on Cu-doped silicon significantly degrades the diffusion length. Unlike iron doped p-type silicon no recovery of diffusion length was observed following such an activation procedure. It is proposed that the activation procedure dissociates interstitial copper agglomerations and forms extended substitutional defects in silicon, which have much greater recombination efficiency. The change in phase of copper and the formation of associated defects is a function of activation light intensity, annealing time, and temperature. An activation energy of 0.419 eV is obtained for the process, which is in good agreement with copper diffusivity value in silicon. It is thus concluded that the change in phase of copper and the formation of extended defects with activation is a diffusion limited process.

Effects of copper contamination on the breakdown and reliability characteristics of thin silicon gate oxides are discussed. Gate oxide integrity is measured for thermal oxides of 45, 75, 120, and 200 Å grown on silicon wafers intentionally contaminated with 1010–1015 cm−3 of copper. Copper doping of silicon was performed according to solubility data considerations. The oxide breakdown voltages as a function of copper concentration for the various oxide thicknesses are reported. For 45 Å oxide, copper concentration cannot exceed 1013 cm−3 without severe degradation in oxide quality. The threshold contamination level for 75 Å oxides is ten times higher. Premature oxide breakdown is proposed to occur due to copper silicide precipitation, which locally enhances the electrical field. It is concluded that the impact of copper contamination on oxide breakdown is not as severe as that of iron contamination on oxide breakdown. This is due to the difference in segregation properties of the two metals at the normalSi/SiO2 interface. © 1999 The Electrochemical Society. All rights reserved.