Calibrated contamination spiking method for silicon wafers in the 10(10)-10(12) atom/cm(2) range

Applying diluted iron, nickel, and copper solutions to front and back side spin-on, respectively, and to immersion spiking, we investigated the contam ination mechanism of silicon wafers. The immersion procedure was found to b e advantageous as a batch process, with the drawback that it can only be ap plied at lower concentration levels than the spin-on technique and, therefo re, it was much more tedious to control. The immersion technique led to a c hemisorption of metal ions on the native oxide. In comparison to iron and n ickel, copper was less readily adsorbed on the wafer surface by immersion. A thermodynamic model interprets the conspicuous results of copper contamin ation. Spin-on spiking led to a dried film of the contamination solution on the wafer. Chemisorption processes could not bi:confirmed under the given spin-on conditions. Thus, the contamination level on the wafer was the same for nickel, iron, and copper when spiking solutions of the same concentrat ion were used. After the metal drive-in step for nickel and copper at 800 d egrees C and for iron at 1000 degrees C, a recovery rate of 50-100% was fou nd by chemical analysis. In order to avoid cross contamination, different c oncentration levels or different spiking elements should not be treated in the same thermal drive-in batch process. Charge-to-breakdown measurements o f capacitors were used to evaluate the influence of surface metal contamina tion and, after drive-in, for gettering studies. (C) 1999 The Electrochemic al Society. All rights reserved.