Effects of Deliberate Metal Contamination on CCD Imagers
- PDF / 1,723,655 Bytes
- 6 Pages / 420.48 x 639 pts Page_size
- 48 Downloads / 210 Views
EFFECTS OF DELIBERATE METAL CONTAMINATION ON CCD IMAGERS
WILLIAM C. MCCOLGIN, J.P. LAVINE, J. KYAN, D.N. NICHOLS, J.B. RUSSELL, AND C.V. STANCAMPIANO, Eastman Kodak Company, Microelectronics Technology Division, Rochester, NY 14650-2019
ABSTRACT The effects of intentional metal contamination on silicon charge-coupled device imagers are reported. Such imagers are both sensitive to and provide sensitive measures of the presence of metals in the fabrication process. High-purity iron, cobalt, nickel, copper, palladium, and gold were deliberately introduced into the device wafers just before the last high temperature step. Metals were found to cause both electrical defects and distinctive imaging defects. We find that transition metals can be effectively removed from device regions by internal gettering, but that this gettering can be defeated by a fast cool-down. Gold, however, is poorly gettered. INTRODUCTION Image defects in silicon charge-coupled device (CCD) sensors tend to be highly visible and are, accordingly, a serious problem for device yield. Some defects, such as "white spots", have been attributed to contamination from metals either through decoration of crystalline defects [1,2] or directly [3]. However, numerous questions about metal contaminants remain. For example, which metals cause problems, do different metals exhibit different defect "signatures", and can gettering eliminate metal-induced defects. In an attempt to answer these questions we deliberately added contaminant metals during fabrication to wafers in one run of a commercially available CCD image sensor. EXPERIMENTAL The imager chosen is an Eastman Kodak Company 2/3", 370,000 pixel, video interline-transfer device [4]. Details of the pixel are shown in plan view and cross section in Fig. 1. The die size of the imager is 9.9 mm (H) x 7.7 mm (V). Light is converted to charge and integrated in the photodiodes, then transferred to the CCDs and read out to form an image. The devices are built in n-type epi on CZ wafers that are denuded of surface oxygen and precipitated to provide internal gettering sites. The fabrication process and contamination experiment are shown schematically in Fig. 2. Metal doping was performed near the end of the device fabrication, just before the last high temperature step. High purity metal wires (0.5 mm) were scribed onto the exposed silicon of the wafer backs behind one column and one row of devices, as shown in Fig. 3. Wafers were then placed in an expendable furnace tube and given a 950* C, 30 min planarization reflow step. This thermal treatment is more than sufficient to cause metal diffusion through the wafer thickness, 525 pm. Half the contaminated wafers received a "fast" cool, an automated pull-out from process temperature to room temperature, in an attempt to defeat the internal gettering and observe metal-induced defects. The other half received a segregation-type anneal [5,6] intended to help getter the metals. Additional device wafers (not shown in Fig. 2) were not metal-doped and received the reflow an
Data Loading...
