home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.mac.hardware
- Path: sparky!uunet!gatech!concert!samba!bigmax.bme.unc.edu!frey
- From: Eric Frey <frey@bme.unc.edu>
- Subject: Re: 25MHz IIsi and my engineering friend
- Message-ID: <1992Nov20.201011.209@samba.oit.unc.edu>
- X-Xxmessage-Id: <A732B050E7016506@bigmax.bme.unc.edu>
- X-Xxdate: Fri, 20 Nov 92 20:10:08 GMT
- Sender: usenet@samba.oit.unc.edu
- Nntp-Posting-Host: bigmax.bme.unc.edu
- Organization: UNC Deptartment of Biomedical Engineering
- X-Useragent: Nuntius v1.1.1d11
- References: <1992Nov15.121708.24693@news.acns.nwu.edu> <lgo13jINNid0@lion.cs.utexas.edu> <1992Nov20.143347.10328@mcc.com> <168A48EFD.LILMARA@YaleVM.YCC.Yale.Edu>
- Date: Fri, 20 Nov 1992 20:10:11 GMT
- Lines: 52
-
- Subject: Re: Lifetime of 25MHz IIsi (was: 25MHz IIsi and my engineering
- friend)
- From: Martin Lilly, LILMARA@YaleVM.YCC.Yale.Edu
- Date: Thu, 19 Nov 92 10:09:58 EST
- In article <168A48EFD.LILMARA@YaleVM.YCC.Yale.Edu> Martin Lilly,
- LILMARA@YaleVM.YCC.Yale.Edu writes:
- >In article <1992Nov16.181708.22881@news.uni-stuttgart.de>
- >skok@itwds1.energietechnik.uni-stuttgart.de (Holger Skok) writes:
- >
- >>>"The candle which burns twice as bright burns half as long."
- >>
- >>(discussion about heat deleted)
- >>It's diffusion that kills the CPU. All those nifty PN or other junctions
- >>tend to even out over time. The concentration of the donor and acceptor
- >>alloying (dotation ?) materials is different on the different sides of
- >>the junctions which amounts to a driving potential for diffusion. The
- >>coefficient of molecular diffusion is temperature dependent - it is
- >>higher at higher temperatures. Conclusion: the higher the temperature
- >>the faster all concentration differences in the IC even out, thereby
- >>destroying it.
- >>
- >My P-chem text says that diffusion is a function of temperature in
- degrees
- >Kelvin. Even if the cpu is running 50 degrees hotter- which it won't
- with a
- >heat sink installed -is there a big difference between 400 and 450
- degrees K?
- >Any physicists out there to comment?
-
- Well, I used to be one. Let's see if I can remember my semiconductor
- processing physics. The rate of a process with an activation energy Ea is
- given proportional to
- exp(-Ea/(k*T)), where k is Boltzmann's constant ant T is the temperature
- in Kelvin. To compare the rate of two processes at different
- temperatures,
- T1 and T2, you take the ratio of this factor and get:
- Rate(T1)/Rate(T2) = exp(-Ea/k*(1/T1-1/T2)) = exp(-Ea/k*(T2-T1)/(T2*T1)).
- For T2=400 and T1=450K, the term (T2-T1)/(T1*T2) is -~3e-4 so,
- Rate(450)/Rate(400)=exp(3e-4*Ea/k). As I recall,
- activation energies for dopant (B, P, As) diffusion in Si are on the
- order of a
- 3 eV. Boltzmann's constant is 8.62e-5 eV/K. So, for a 50K
- difference in temperature and Ea=3eV,
- Rate(450)/Rate(400)=exp(3*3e-4/8.62e-5) = 34,000! Because of the
- exponential dependence, relatively small differences in temperature make
- a big difference in rate!
-
- From pratical experience, think about a pressure
- cooker. Even though the increase in temperature is relatively modest --
- maybe 10 F -- the cooking speed increases a lot!
-
- Eric
-