Контрольная работа по "Английскому языку"

October 2002 issue

Lightning Rods for Nanoelectronics

Electrostatic discharges threaten to halt further shrinking and acceleration of electronic devices in the future

By Steven H. Voldman  

We're all familiar with electrostatic charge: shuffle across a shag carpet in sneakers, touch a piece of metal, and zap. The slight prick we feel--caused when the electric charge built up by the shuffling suddenly leaps to another object--is nothing compared with what modern electronic equipment experiences.

On a dry winter day, walking on a new carpet can generate a whopping 35,000-volt discharge. We are not harmed by this high voltage, because the amount of charge that flows is puny. Still, it is large enough to destroy sensitive micro-electronic components. Researchers have come up with clever ways to prevent such damage. But as circuits get smaller, they become more sensitive to electrostatic discharge (ESD) and the old tricks no longer work. Can we continue to find new ways to prevent electrostatic damage and thereby maintain the pace of innovation?

People who like to tinker with their computers know that when they open up their machines, they should "ground" themselves--perhaps by touching the metal radiator panel or attaching a wire from their fingers to a metal fixture. This grounding diverts any built-up charge into another object. Microprocessors and other chips have built-in protection circuits, but for tomorrow's equipment such precautions will be of even greater concern. ESD is an issue not only for finished products but also during their manufacture, from wafer fabrication to packaging to the assembly of complete systems. Each step has its own electrostatic hazards.

In general, electrostatics poses the greatest threat during manufacturing and handling to install the devices and is less of a concern once the components are safely ensconced inside machines such as the computer on your desk. Some protection methods rely explicitly on this assumption. The hazards start at the earliest stages of manufacture: even photolithographic masks, whose function is entirely mechanical and not electrical, are at risk. The chief danger to microelectronics is damage to the active elements caused by heating and by electrical breakdown of insulating layers. Magnetic disk drive heads, however, face their own unique problems, including magnetic aspects of discharges and considerations of aerodynamics.

ESD protection devices have been incorporated into microchips since the 1960s and have evolved over the decades according to technological necessity and corporate strategy. The main goal with each new generation of microelectronics is to perpetuate Moore's Law by building elements such as transistors that are smaller and faster. Someday in the not too distant future the industry will hit a wall that blocks further progress. We will reach the point where we cannot design and build a smaller, faster transistor. But there is a second wall, because even if we can build the next transistor, it will be useless if we have no way to protect it adequately against ESD. No one knows which wall is coming first.

Thermal Runaway  
 
What causes electronics to fail when ESD occurs? The main culprit is heat generated by the electric current of the discharge, which can be enough to melt the material. Internal temperatures from ESD events can exceed 1,500 degrees Celsius (2,700 degrees Fahrenheit), above the melting points of aluminum, copper and silicon. Damage occurs even without melting. The properties of diodes and transistors are determined by the doping of the semiconductor: carefully introduced impurity atoms, or dopants, produce regions having specific electronic properties. Excessive heating can allow dopants to migrate, ruining the precise pattern of regions that is essential for the device to function properly.

Processes known as electro-current constriction and thermal runaway make matters worse by concentrating the heating in a hot spot: when one location of a semiconductor heats up significantly, its resistance falls, so that more of the current flows through the hottest place, heating it even more. Geometry and design symmetry play a key role in distributing current evenly through a device to forestall the onset of thermal runaway. The material's thermal conductivity, heat capacity and melting temperature are all important in determining its ability to store the heat or diffuse it evenly.

Just as important as the sophisticated transistors of modern devices are the electrical connections between different elements. These include interconnect wires along the surface of the chip's semiconductor layers and "vias" that connect vertically from one layer to another. All of these are reduced in size along with the rest of the device in the effort to improve the speed and computational power of high-performance semiconductor chips. For many years, aluminum was the metal of choice for interconnects, but aluminum melts at only 660 degrees C. Beginning around 1997 (after 10 years of research), the microelectronics industry migrated to copper interconnects, primarily because of copper's superior electrical conductivity, enabling smaller and faster circuits. An additional benefit is copper's higher melting point, 1,083 degrees C, which gives interconnects higher tolerance to heating.

In contrast to the transition to copper, a generational change in insulating materials has had a minor negative side effect on sensitivity to ESD. These "low-k" materials form the insulating regions between metal lines in devices being rolled out in the marketplace. The materials' low dielectric constant (k) reduces the capacitance between the lines, which in turn reduces cross talk (interference between the lines) and increases the travel speed of high-frequency signals and short pulses. Unfortunately, low-k materials have lower thermal conductivity than silicon dioxide (the traditional insulator, or dielectric), so they are not as effective at dissipating energy from electrostatic events. This has to be compensated for by careful electrical design, wider interconnects or other techniques to reduce heating. Still, the net effect of introducing copper and low-k materials together is to improve sturdiness against ESD. Their introduction helped to bring about the transition to one-gigahertz (GHz) applications.

We now turn to the transistors, the workhorses of microchips. The primary digital technology today is the MOSFET device, named after the metal-oxide semiconductor field-effect transistors that they contain. The basic MOSFET structure consists of two doped regions called the source and the drain, separated by a region called the channel. An electrode called the gate sits above the channel, separated from it by a thin layer of silicon dioxide dielectric. The voltage applied to the gate controls how current flows in the channel between source and drain.

Such devices have entered the nanostructure age in recent generations. In August, for example, Intel announced plans to manufacture chips with gates 50 nanometers long and gate oxides 1.2 nanometers thick--a mere five atomic layers. The thinner the dielectric, the lower the voltage needed to cause breakdown. Dielectric breakdown is caused not by heating but by the electrical carriers (electrons or holes) breaking molecular bonds and cutting a path through the insulator like a tiny bolt of lightning. The defects formed by oxide failure are known as pinholes. With very thin oxide layers, mere handling of microelectronic chips can produce pinholes in the gates.

The source and drain of a MOSFET are also sensitive, and ESD on those regions leads to MOSFET thermal breakdown. When the high voltage of the discharge arrives at the drain, say, it increases the electric field there. This strong field accelerates the current-carrying electrons, making them energetic enough to knock other electrons free. These secondary electrons (and corresponding holes) increase the current flow even more and are themselves accelerated enough to knock more electrons free, and so on. Called avalanche multiplication, this process causes current to flow from the transistor into the nearby substrate, which puts the transistor into an unstable "negative resistance" state, further exacerbating the situation. As the current increases, heating leads to the thermal runaway, or thermal breakdown, described earlier.