The antifuse devices with planar Al/$SiO_2$/Poly-Si/$SiO_2$/Al structure were fabricated using 1.2 ㎛ CMOS process. The poly-Si layer was doped by boron, the thickness of two $SiO_2$ layers was 100Å. The length and the width of the devices were 1.6-18㎛ 1.2-6.0㎛, respectively. The Ⅰ-Ⅴ characteristics of the devices were analyzed using Hewlett Packard 4145B semiconductor analyzer, and the material transport appearance was investigated using optical microscope and scanning electron microscope.
The resistance of $SiO_2$ layer was a few giga-ohms and the breakdown voltage of $SiO_2$ layer was 7.6V on the average. After the oxide breakdown, one or two holes were formed on the $SiO_2$ layer and Al flowed through the holes into the poly-Si layer. With further application of voltage after breakdown, Al from one of electrode flowed into the poly-Si layer and reached to the opposite electrode. As a result, two electrodes were connected by a conducting line consist of Al and Si. With the formation of an Al-Si alloy line, the resistance of the device was dropped to nearly 30 ohms, which is much lower than the resistance of poly-Si layer (~kΩ). The direction of Al flow coincided with that of electric field, and the cause of Al flow was supposed to be either $Al^+$ ion conduction or hole migration. After two electrodes are connected, the amount of Al flow decreased rapidly but the width of Al line increased slightly. Above a critical voltage the velocity of Al flow was larger than several cm/sec and below which the velocity was smaller than several tens nm/sec. The critical voltage was proportional to the length of devices ; it meant that the existence of a critical electric current. The measured critical current was about 8mA when the width of the device was 2.4㎛.
When a critical voltage was applied without serial resistor to a device with long poly-Si layer, the poly-Si layer is partially melted. This result indicated that the temperature of the devices was very high and current clouding occurred when critical voltage is applied. This could be prevented by attaching serial resistor of larger than a few hundreds ohms. It is supposed that the poly-Si in front of the Al line was partially melted by Joule heating caused by current clouding, when the critical voltage was applied. The melting temperature is supposed to be the eutectic temperature (577℃) of Al-Si, which is much smaller than the melting temperature of pure Si.