Analysis of the etch-back of the front emitter caused by the rear-etch process on czochralski monocrystylline P-type silicon and development of a getter process

Abstract

A lightly doped emitter can be manufactured through the formation of a heavy emitter diffusion followed by an homogeneous etch-back of the emitter. This route to form a light emitter is particularly interesting because it provides the extra benefit of phosphorus diffusion gettering (PDG) of metallic impurities, like iron. The possibility of using the rear-etch process to also perform an etch-back of the heavy diffused front emitter while removing the back emitter is analyzed in the present work, together with an analysis of the gettering efficiency of four different phosphorus emitter diffusion recipes. 156 mm x 156 mm pseudo-square Czochralski p-type silicon wafers with a bulk resistivity of 1.6 .cm were used. The samples were textured by alkaline etching, RCA cleaned and then split in four groups and each group had the phosphorus emitter diffused through a different recipe. The standard Getter and Screen-print recipes used in the Solar Industrial Research Facility were used, together with two modified version of the recipes where an additional drive-in step without oxygen was added after the deposition step. After the emitter diffusions, the PSG layer formed was removed in a 2.5 % hydrofluoric acid solution and a few samples from each group were submitted to the rear-etch process 1, 2 and 3 times and then a 75 nm thick silicon nitride (SiNx) coating was deposited by plasma-enhanced chemical vapor deposition on both sides of all the wafers. The wafers were characterized in order to verify the their interstitial iron concentration before and after the fast-firing process and the effect of the rear-etch process in the front side emitter. Interstitial iron was only observed in the samples with a Screen-print diffusion. The extra oxygen-free drive-in added to the standard Screen-printing recipe resulted in a deeper junction and a higher surface concentration of active dopants, while the addition of the same drive-in step to the standard Getter diffusion didn’t cause a significant impact on the formed junction. It was estimated through the EDNA 2 software that a 100 m etch-back of the emitter would vii viii be necessary to reduce its sheet resistance from 24.7 to the target sheet resistance of 100 =sq. Even though the rear-etch process did performed an etch-back of the front emitter it was not homogeneous, and not strong enough the reach the target sheet resistance even when the process was applied three times in a row, suggesting that the etch-back process is not suitable for the proposed application of etching back a heavily doped front emitter without adding an extra step to the manufacturing process.