The Big Bad Wolfdale: Intel Core 2 Duo E8500 Review- Page 2/7
February 29, 2008
By
Thomas Soderstrom
Transistor Basics: Defining the "Problem"
Progress in integrated circuit design often begins with reduced transistor size in order to maximize the number of transistors in any given area. The primary goal of such "die shrinks" is to increase the number of parts produced per silicon wafer, which reduces cost per part. A secondary advantage is that smaller transistors typically "switch" faster, allowing higher stable frequencies and improved performance. Another important benefit is reduced power consumption, but the word "efficiency" has only recently become part of the consumer lexicon.
The challenge Intel faced is that the third point, reduced power consumption, quit working when the company ran smack into the laws of physics. One of those laws is that the smallest particle that maintains the characteristics of a substance is a molecule, or in the case of pure elements, an atom. The second problem is that insulation value decreases whenever the thickness of the insulator is decreased.
To understand the transistor, it helps to understand an even simpler device, the Diode. Long ago, it was discovered that certain material could pass electrical current in one direction and block it in another. But nothing's perfect. What really happens in Diodes is that electrical resistance is low in one polarity (very much current flows) and high in the other polarity (very little current flows). The materials used are called semiconductors, and transistors use this property in a different way.
The magic for turning a diode into a transistor came with the discovery that if the diode was charged with low current on one side, it would allow a higher current to flow across it with very little resistance. The function is similar to that of a mechanical relay, but without the moving parts, and the part of the transistor that was born of the Diode is called the "gate oxide". Transistor radio and audio amplifiers were born of this ability. Configured a little differently, computers use it to determine high and low voltage (or resistance) states, the 1's and 0's the make up a data stream.
As transistors get smaller, they use less voltage and carry less current. The gate oxide becomes thinner, because it requires even lower resistance when "open", but making it too thin prevents it from effectively working as an insulator when "closed". When the "lower voltage" state is nearly as high as the "higher voltage" state, the wasted energy that passes through the closed gate is called "leakage current".
The problem with leakage isn't just that the difference between lower and higher voltage is smaller, but that the waste energy is given off as heat. Heat prevents the transistor from operating at optimal frequency, and the best way to reduce it is to find a gate oxide material that either conducts better when "open" or insulates better when "closed". The ability of a semiconductor to act as both an insulator and a conductor is called its relative dielectric constant, or "K" value, and materials that function better than the classic silicon-dioxide gate are referred to as "High-K".
Events
IE 7
Firefox 2.0
