Subject: Re: radiation and computers
From: thrash@io.com (Christopher Thrash)
Date: 19 Nov 2000 14:06:00 GMT
Message-ID: 6507
Newsgroups: sjgames.gurps.traveller

On 16 Nov 2000 08:15:36 GMT, ajackson@iii.com (Anthony Jackson) wrote:
> Realistically, what level of ionizing radiation will cause significant
> software errors (and possible a soft reboot) in hardened electronics?
> For that matter, what level of radiation will cause permanent damage,
> I know that some of the jupiter probes were fried by multiple passes
> through the jovian radiation belts.

From _The Effects of Nuclear Weapons_, Glasstone and Dolan, 1977, Sec. 8.73-
8.88:

"The name commonly applied to the class of effects under consideration 
is "transient-radiation effects on electronics," commonly appreviated to the 
acronym TREE. In general, TREE means those effects occurring in an 
electronics system as a result of exposure to the initial radiation from an 
nuclear weapon explosion. The adjective "transient" applies to the radiation 
since it persists for a short time, i.e. less than 1 minute. The response, 
however, is not necessarily transient... 
"Radiation effects on electronics may be temporary or more-or-less 
permanent... The component responses of short duration are usually the result 
of ionization caused by gamma radiation and are dependent on the dose rate, 
e.g., in rads per second, rather than the dose. The more permanent effects 
are generally -- but not always -- due to the displacement of atoms in a 
crystal lattice by high-energy (fast) neutrons. In such cases the extent of 
the damage is determined by the neutron fluence, expressed in neutrons/cm2. 
When a permanent effect is produced in an electronic component by gamma 
radiation, the important quantity is usually the dose in rads."

Neutron fluence (Fn) at a distance R from a nuclear detonation is 
approximately given by:

Fn = 1.4 x 10^12 Y/R^2

where Y is in kilotons, R is in km, and Fn is in neutron/cm2.

Dose (Dg) from prompt radiation of an explosion is approximately:

Dg = 4 x 10^5 Y^(2/3)/R^2

where Y is in kilotons, R is in km, and Dg is in rads (Si).

Damage Thresholds (gleaned from the text):

Transistors	     10^11-10^15 neutron/cm2
MOS Transisitors     10^4 rads (silicon)
Capacitors	     10^15 neutrons/cm2
Precision Resistors  10^7 rads (carbon)/s
		     10^14 neutron/cm2
NiCd Batteries	     10^7 rads (air)/s
		     10^13 neutron/cm2
Hg Batteries	     10^16 neutron/cm2
Wiring Insulation:
Silicon Rubber	     2x10^15 neutron/cm2
Polyethylene	     10^7 rads (carbon)
Teflon TFE	     10^4 rads (carbon)
Teflon FEB	     2x10^6 rads (carbon)
Polyolefins	     5x10^9 rads (carbon)


From _Space Mission Analysis and Design_, 3d Ed. (SMAD III), Wetz and Larson, 
1999, pp. 214-240:

Commercial Off the Shelf (COTS) and Rad Hard Parts Comparison:

Characteristics       COTS			  Rad Hard

Total Dose	      10^3-10^4 rads		  10^5-10^6
Dose-Rate Upset       10^6-10^8 rads(Si)/s	  >10^9 rads(Si)/s
Dose-Rate Induced
Latchup 	      10^7-10^9 rads(Si)/s	  >10^12 rads(Si)/s
Neutrons	      10^11-10^13 n/cm2 	  10^14-10^15 n/cm2
Single-Event Upset    10^-3-10^-7 error/bit-day   10^-8-10^-10 error/bit-day
Single-Event Latchup/
Single-Event Burnout  <20 MeV-cm2/mg (LET)	     37-80 MeV-cm2/mg (LET)

LET is "linear energy transfer".