Studies on Polycrystalline Silicon for Solar Applications, including Characterization of Trace Elements B, C and Al by Charged Particles Activation Analysis

Nuclear methods of analysis were used to study the behaviour of trace elements during the metallurgy of polycrystalline silicon for solar applications: these impurities may have an influence on the efficiency of the solar photocells made out of this silicon. The ingots were prepared in a graphite boat by the Bridgman method without seed, using electronic grade silicon as a starting material. Most impurities were determined by neutron activation, the global level being low: ≃ 3.1015 at/cm3. Charged particles were used to determine B, C and Al. For boron, the 10B (d,n)11C reaction with subsequent chemical separation of 11C was used; for carbon, we used non-destructive analysis and the 12C(d,n)13N reaction. For low levels of aluminum in silicon no suitable method existed; we could solve this problem using fast neutrons from the 9Be (d,n)10B reaction and the reaction 27A1 (n,α)24 Na. Interferences from Mg (undetected) and from Si itself were negligible, the detection limit being 5.1014 at/cm3 in ultra pure silicon. High Carbon levels were found close to the solubility limit (2.1018 at/cm3); there is a significant segregation at the top, at the edges and at the bottom of the ingots. Activation results were used to calibrate infrared absorption at low temperature for the determination of carbon in silicon. Residual B and Al concentrations are low. Ingots doped with B or Al were analyzed, the doping with Al being inefficient. Boron was homogeneously distributed, while Al segregated at the top and at the grain boundaries.