Daisuke KOBAYASHI

Profile

What? I'm studying space electron devices, especially forcusing on semiconductor-chip malfunctions caused by cosmic rays called "soft errors." I run my laboratory as an associate professor at Institute of Space and Astronautical Science (ISAS), Japan Aerospace Exploration Agency (JAXA) and the Department of Electrical Engineering, Graduate School of Engineering, The University of Tokyo. This portal also serves as the website for my laboratory.

How? We conduct experiments to discover new phenomena and perfoming simulations to analyze them, but what we are focusing on most is "modeling," writing phenomena in simple equations. Right now, we are mainly focusing on developing "soft-error reliability equations," currently consisited of the following three equations (I will skip the details but they are quite simple and physically intuitive).

$ \sigma = \sigma_\infty \exp{\left(-\frac{L_e}{L_\mathrm{eff } }\right)} $

$ \Phi_\mathrm{eff} = \frac{K_\mathrm{eff } }{L_\mathrm{eff}^2}$

$ \lambda = 2K_\mathrm{eff}\frac{\sigma_\infty}{L_e^2}$

Why? We use accelerators to reproduce cosmic rays. We expose semiconductor chips to them and examine what happens. This “expose and examine” approach is the standard method and is indispensable, especially for reliability assurance of semicondctor chips. However, accelerators are limited in number, so accelerator tests are extremely valuable. We often repeat tests; for example, “what happens if we lower the voltage fed to semiconductor chips a bit from the earlier experiment?” Although those condition changes are certainly meaningful, I feel it’s a bit waste to repeat almost the same tests over and over in accelerator experiments. If we can describe how each parameter affects the result with simple equations, we can say something about the results much more without actually irradiating chips. We will then be able to use the valuable accelerator time for other purposes. We would also be able to cut down the time to wait for a chance of accelerator testing to inspect chips we just made. Our goal is zero! We’re trying to create "magical" equations that allow us to say, without even one acclerator exposure, “This chip is fine!”