Xref: utzoo rec.ham-radio:4020 sci.electronics:2247 Path: utzoo!mnetor!uunet!husc6!cmcl2!brl-adm!umd5!purdue!i.cc.purdue.edu!j.cc.purdue.edu!pur-ee!iuvax!inuxc!ihnp4!ihlpf!straka From: straka@ihlpf.ATT.COM (Straka) Newsgroups: rec.ham-radio,sci.electronics Subject: Re: build-it-yourself EPROM erasers (Just the facts) Message-ID: <3711@ihlpf.ATT.COM> Date: 16 Feb 88 19:05:36 GMT References: <8802091255.AA23298@ucbvax.Berkeley.EDU> <12@ucsd.EDU> <21651@clyde.ATT.COM> <698@uthub.toronto.edu> Reply-To: straka@ihlpf.UUCP (55223-Straka,R.J.) Organization: AT&T Bell Laboratories - Naperville, Illinois Lines: 75 In article <698@uthub.toronto.edu> koko@uthub.toronto.edu (M. Kokodyniak) writes: >In article <21651@clyde.ATT.COM>, gwu@clyde.ATT.COM (George Wu) writes: >> ... >> Anyways, the reasoning goes something like this: the UV light is just >> some extra energy you're giving to release trapped electrons. So why not use >> some other form of energy, like heat? >> >> This has actually been successfully done, although I can't remember the >> temperature or time. I think it was half an hour at 300 degrees. ... WAIT A MINUTE!!! All of the above is WAY off base! Let's get back to physics! ***************************************************************************** Let me preface my remarks by stating that 10 years ago, my job was to provide test strategies and reliability for the first N-channel EPROMs, the 2708 and 2716. The chips have gotten better over the years, but the physics remains the same. Summary: Heat alone won't do it. A combination of heat and short wave UV may or may not be significantly better than short wave UV alone. The electrons are stuck up in a floating gate structure in the EPROM. To get them off, they have to "tunnel" across the bandgap represented by the oxide layer to the surrounding layer(s). There is a specific threshold energy that must be exceeded (which can vary as a result of processing parameters) in order to tunnel. The number of electrons which can tunnel is based on the distribution of energy that the subject electrons posess. This is, of course, some sort of (probably) normal distribution given a base ambient temperature. Raising the temperature raises the proportion of electrons that exceed that tunneling threshold and can tunnel across the oxide barrier. The degree to which the device is sensitive to temperature is known as its activation energy, which is measured in units of electron volts. This activation energy is empirically determined, and typically is in the range of 0.7 to 0.8 ev. A very small change in activation energy has a tremendous effect on the acceleration factor, as the relationship is exponential with the ratio of the two temperatures (with regard to absolute zero). See some basic college physics (or chemistry?) text for more details. This phenomenon is basic, and not just related to EPROMs. Manufacturers routinely subject EPROMs to long bakes at elevated temperatures (usually SUBSTANTIALLY higher than the above mentioned 300 degrees F) to ensure proper data retention of each individual part. The object is to simulate worst case high temperature operation for a number of years. Any EPROM worth its salt (so to speak :-)) will in actuality last at least 10 times the standard 5-year specification. Good ones are quite bullet-proof. The reason that short wave (and not long wave) UV light does the trick, whereas heat doesn't is that the UV adds quite a kick to those electrons, yes, enough to get them over that tunneling barrier. That is why long wave, which is of lower energy, doesn't work too well (if at all). It doesn't add enough energy to kick the electrons over the threshold. Heat might add a little bit, perhaps enough to make a bit of difference. I don't know. Sorry for getting a little bit long, but the non-semiconductor-device-physics people need to get some solid information on this one. PS: > >I would not want to try this, or at least not to often, since the heat may >destroy the device. Components in the silicon wafer are formed during the >manufacturing process by controlled diffusion of impurities into selected >areas in the wafer. Further heating might cause further diffusion which >would make these components gradually disappear. Diffusion is not an issue below 900 degrees C. Alloying of the aliminum contacts into the silicon and shorting out junctions is an issue at anything above about 425-450 degrees C. If you have gold-aluminum junctions in the packaging, the temperatures get even lower (but most manufacturers have fixed this particular problem (purple plague)). -- Rich Straka ihnp4!ihlpf!straka Advice for the day: "MSDOS - just say no."