In 1930s Bell Labs scientists chose to focus on Siand Ge, rather than better known semiconductors like Ag2S and Cu2S, mostly because of their reliable performance. Their choice was rewarded with the invention of bipolar transistors several years later. In 1960s, scientists at Fairchild worked hard to solve a mysterious reliability problem of Si/SiO2 interface called Bias TemperatureInstability. Their successful solution of the problem allowed introduction of MOSFET based circuits that would revolutionize electronics industry for the next thirty years. And in late 1990s, when most companies assumed gate oxides can not be scaled below 3.0 nm and therefore Moore’s law can not continue, our demonstration that oxides can be thinned down to 1.0 nm allowed continued scaling of CMOS circuits. In fact, the very existence of the PC you work on (oxide thickness ~ 1.5 nm) validates that our reliability model must have been correct!
For many of us, analyzing a semiconductor device means computing its I-V characteristics, determine its circuit speed, etc. Of course, without such understanding, a VLSI circuit can not be made. However, once made, the transistors will now have to survive trillions of switching cycles over several years of operation. How do you know that if you make a transistor with a certain specifications, that it will be able to operate reliability over a specified period of time? The people who work on the physics of reliability try to answer this question with a combination of accelerated tests and fundamental studies of “How things break"? I will demonstrate how this done by using examples based on our recent research, hoping that you will catch a glimpse of the complexity of the reliability problems and excitement and implications of solving them.
Professor Alam joined Purdue University as a faculty member of the Electrical and Computer Engineering Department in 2004 after spending nearly a decade in industry, first at Bell Labs and then at Agere Systems. His research interest involves physics of carrier transport in semiconductor devices, and he has worked on theory of electron transport models, quasi-ballistic transport in bipolar transistors, MOCVD and ALD crystal growth, laser dynamics, and most recent recently, on the theory of oxide reliability, transport in nanocomposite materials, and response of Nano-Bio sensors.
EE 317, Purdue University, West Lafayette, IN