Electrostatic Discharges and Semiconductor Testing

We've all had the shocking experience of walking across a carpeted floor and touching something metallic, such as the refrigerator handle, stove knob, or thermostat. Zap!...then that sting at your fingertip. Welcome to static discharge, the release of an electric charge that built up on your body from the action of your shoes scraping across the carpet surface. The more official name of that nice little surprise is electrostatic discharge, or ESD. And that nice little surprise can do a lot more damage than just stinging your fingertip.

Walking across a carpet or rubbing an inflated baloon on your hair is one of the everyday ways to generate a static charge. In either of these cases friction is the agent that causes the charge to accumulate. The friction causes electrons to taken from one surface and transferred to the other. One surface has a surplus of electrons; this surface now has a "negative" net charge. The other surface has a deficit of electrons, or a "positive" net charge. When one of these surfaces comes close to or in contact with an electrical conductor (most commonly, metal) that has a different amount of charge (most commonly, no charge at all, or grounded), the excess charge transfers to the lesser charged conductor. Zap!

The sting from that zap is pretty obvious. What isn't at all obvious is the amount of electricity in that zap. The charge jumps across the small space between the two conductors because the air between the conductors ionizes, or breaks down into ions (charged particles) that conduct the charge from the body with the excess to the neutral or lesser charged body. To break down just a 1 cm gap of air takes 33,000 volts! There are 2.54 cm per inch, so 1 cm is 0.3937", or less than 1/2-inch. 33,000 volts across less than 1/2"!

Microprocessors and the accompanying chips that make up todays PCs, cell phones, PDAs, iPODs, and other electronic gadgets continually get smaller, faster, and cheaper. This continual shrinking has come to be known as Moore's Law, named for Gordon Moore, one of the founders of Intel Corporation. In the still early days of semiconductor devices, Moore stated that the number of transistors that could be packed onto a piece of silicon would double every 18 months. That progression still holds true today because of advances in semiconductor fabrication technology that shrink circuit dimensions to unimaginable smallness. The newest devices have circiut lines that are 45 nanometers wide, or slightly smaller than two-millionths of an inch. Compare that to a human hair, the smallest of which is about 670-millionths of an inch.

The small device geometries allow for pretty amazing processing speed. But in this world, nothing comes for free. One cost of shrinking those sizes is a decreasing ability of those circuits to withstand ESD. Just as you got a shock after walking across the carpet, workers in computer chip factories can build up the same static charge from walking across the floor. If they pick up a semiconductor device and touch the metallic pins or solder balls that connect that device to traces on a circuit board, they will cause an electrostatic discharge, or ESD event, into the device circuitry. Those very small metal lines in the device cannot handle the current spike from the ESD event, and they melt! That device is now permanently damaged and cannot be sold. That's not the worst of it. Sometimes the damage from an ESD event doesn't completely damage the device, but only weakens it; the weakness is not enough to cause the device to malfunction, so it appears to be a good device. It is only when that device is built into a product and used for a while that the weakness degrades into full failure. The product that was working just fine now stops working...right in the middle of that important cell phone call, or just when you're filing your tax return at 11:59 PM on April 14th. Nice!

Semiconductor manufacturers must design features into their chips to protect those small devices from the large voltages and currents that arise from ESD. How does the manufacturer know that this protection circuity works? They test it, just as they test the other circuitry in the chip. There is special test equipment to apply the ESD events and measure the circuit behavior to determine that the circuitry did its job. This test equipment applies ESD events to simulate the different types of events that are likely to happen during the device's lifetime, from the product manufacturing process to its end use by the customer.

The Human Body Model (HBM) derives from the 19th century investigations into mine gas mixture explosions. This model attempts to emulate the vast majority of anticipated human-generated ESD events; the model assumes the charge originates from the fingertip of a standing individual. This model charges a 100pF capacitor through a large series resistance (in the Megohm, or millions of Ohms, range). When the capacitor is charged, the circuit switches to discharge the capacitor through a 1.5K-ohm resistor into the semiconductor device input.

As automated manufacturing processes emerged, a new source of ESD events also emerged: the manufacturing equipment itself! As material moved through the machines, static charge would build up at various points in the machines and eventually discharge into the product being handled. This model is similar to the HBM above in that a capacitor is charged through a series resistance and then switched to discharge into the device under test. However, the MM uses a 200pF capacitor and there is no 1.5K-ohm resistor in the discharge path. The capacitive charge is discharged directly into the device under test input.

This model also emulates ESD events that occur during manufacturing from the motion of the device moving through a piece of manufacturing or test equipment. As the device moves through a piece of equipment, charge accumulates on the device. The device then gets inserted into the test site or socket, which has no charge, and an ESD event occurs. Such an event can be more catastrophic for some devices than the HBM. This test orients the device under test on a grounded metallic plate with its conductive leads exposed (e.g.: pointing up in the air for a device with pins or solder balls). The device is charged through a large resistance (Megohms) similarly to the above models, then subsequently discharged into the low-impedance test circuit. This method is still being refined to reduce the variations that arise from the different test equipment designs.