Radiation effects on 2n2222 or similar

R

Robert Baer

Guest
Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15 Rem/hour.
Not steady state, but plenty every day.

In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).

Is there any data concerning E-B zener voltage changes, and is there
any material difference WRT transistor type (say 2N3904)?

Thanks.
 
On 6/29/19 2:02 AM, bitrex wrote:
On 6/29/19 2:23 AM, Robert Baer wrote:
   Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15
Rem/hour.
   Not steady state, but plenty every day.

   In the Apollo daze, some data may have been collected concerning
this question (uA709 W-stepping, input pair).

   Is there any data concerning E-B zener voltage changes, and is
there any material difference WRT transistor type (say 2N3904)?

   Thanks.

There are some degredation curves for common components given on p. 235
in this whitepaper from the Voyager program, looks like:

https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19720018918.pdf

one from around the same time for neutron flux, military, stats-math-heavy:

https://apps.dtic.mil/dtic/tr/fulltext/u2/a065162.pdf

Foreign paper (Malaysia), x-rays:

http://www.jcomputers.us/vol9/jcp0909-26.pdf


The first two look somewhat relevant but this data is pushing 40 years
old. how relevant is it to bjts made using modern process? How accurate
was it to begin with? there is probably more recent/detailed info out
there but it sounds like the kind of thing that might require a clearance.

Could always do a FOIA request to the DOE as it's quite specific what
you're looking for see if there's anything unclassified or no longer
classified they might be willing to release.

another one from JPL that's mid 90s vintage, looks like, very topical
but it's pretty bare-bones with respect to data, couple crappy graphs:

https://pdfs.semanticscholar.org/762a/54443600250706f4d73e07aab7b126de9383.pdf


"Recent experience on the Cassini spacecraft project has shown that some
bipolar devices exhibit large decreases in gain at low total dose
levels, severely impacting their use in space. Figure 1 compares gain
degradation of two small signal transistors, measured at the lowest
collector current in the manufacturer’s specifications. The 2N918
transistor exhibits only small changes in gain with total dose, while
the 2N3700 is severely degraded, even at levels below 10
krad(Si).

The gain loss is so severe that it is extremely difficult to use this
device on the project, which must operate at levels between 50-100
krad(Si). This extreme degradation was not observed for earlier lots fl
om the same manufacturer, and was greater than anticipated for any
bipolar transistor with standard construction and normal
breakdown voltage requirements."
Click to expand...

one of the other unfortunate take-aways from the final paper is that
damage profile can apparently vary widely even among the same generic
type from different manufacturers.
 
On 6/29/19 2:23 AM, Robert Baer wrote:
  Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15
Rem/hour.
  Not steady state, but plenty every day.

  In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).

  Is there any data concerning E-B zener voltage changes, and is there
any material difference WRT transistor type (say 2N3904)?

  Thanks.
Click to expand...

There are some degredation curves for common components given on p. 235
in this whitepaper from the Voyager program, looks like:

<https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19720018918.pdf>

one from around the same time for neutron flux, military, stats-math-heavy:

<https://apps.dtic.mil/dtic/tr/fulltext/u2/a065162.pdf>

Foreign paper (Malaysia), x-rays:

<http://www.jcomputers.us/vol9/jcp0909-26.pdf>


The first two look somewhat relevant but this data is pushing 40 years
old. how relevant is it to bjts made using modern process? How accurate
was it to begin with? there is probably more recent/detailed info out
there but it sounds like the kind of thing that might require a clearance.

Could always do a FOIA request to the DOE as it's quite specific what
you're looking for see if there's anything unclassified or no longer
classified they might be willing to release.

another one from JPL that's mid 90s vintage, looks like, very topical
but it's pretty bare-bones with respect to data, couple crappy graphs:

<https://pdfs.semanticscholar.org/762a/54443600250706f4d73e07aab7b126de9383.pdf>

"Recent experience on the Cassini spacecraft project has shown that some
bipolar devices exhibit large decreases in gain at low total dose
levels, severely impacting their use in space. Figure 1 compares gain
degradation of two small signal transistors, measured at the lowest
collector current in the manufacturer’s specifications. The 2N918
transistor exhibits only small changes in gain with total dose, while
the 2N3700 is severely degraded, even at levels below 10
krad(Si).

The gain loss is so severe that it is extremely difficult to use this
device on the project, which must operate at levels between 50-100
krad(Si). This extreme degradation was not observed for earlier lots fl
om the same manufacturer, and was greater than anticipated for any
bipolar transistor with standard construction and normal
breakdown voltage requirements."
 
Robert Baer wrote...
Placing the working unit (open base) in a very
high dose area for 2 months, where it will see
approximately a rate of approximately 15 Rem/hour.
Click to expand...

What's a convenient way to get a "high dose area"?
When in college (long ago) I had ready access to a
nuclear reactor, and all kinds of sources, but ...


--
Thanks,
- Win
 
On Saturday, 29 June 2019 08:46:54 UTC+1, Winfield Hill wrote:
Robert Baer wrote...

Placing the working unit (open base) in a very
high dose area for 2 months, where it will see
approximately a rate of approximately 15 Rem/hour.

What's a convenient way to get a "high dose area"?
When in college (long ago) I had ready access to a
nuclear reactor, and all kinds of sources, but ...
Click to expand...

David Hahn used smoke alarms for alphas. An x-ray tube is also possible.
 
On 6/29/19 3:46 AM, Winfield Hill wrote:
Robert Baer wrote...

Placing the working unit (open base) in a very
high dose area for 2 months, where it will see
approximately a rate of approximately 15 Rem/hour.

What's a convenient way to get a "high dose area"?
When in college (long ago) I had ready access to a
nuclear reactor, and all kinds of sources, but ...
Click to expand...

Part of the takeaway from this paper I posted from the 90s seems to be
that performance can vary widely even among devices of the same generic
type because it's highly dependent on emitter geometry/construction
(does base metallization pass over the emitter-base region in a given
jellybean? sometimes yes and sometimes no apparently, there's no
"reference topology" as it were for those parts) and different
manufacturers have different ways of doing things.

or even the same manufacturer uses somewhat different geometries and
construction on different runs. They're cheap parts they prolly press
into service whatever slack capacity they have at a given time with
whatever process the plant that's slack happens to be set up for, I
don't think e.g. Fairchild has dedicated lines to cranking out jellybean
2N3904s anymore.

<https://pdfs.semanticscholar.org/762a/54443600250706f4d73e07aab7b126de9383.pdf>

There may be data on what particular mfgr as of 2019 makes the most
resilient jellybeans off-the-shelf, it sounds like kind of info that
perhaps reads class-i-fied
 
On 29/06/2019 07:23, Robert Baer wrote:
In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).
Click to expand...

I think the Apollo program used uA702. The uA709 came too late?

piglet
 
On 29/06/2019 18:56, Piglet wrote:
On 29/06/2019 07:23, Robert Baer wrote:
In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).

I think the Apollo program used uA702. The uA709 came too late?

piglet
Click to expand...

Answering myself: the uA709 was indeed used in some of the later lunar
surface science packages. Makes sense as it was much nicer to use :)

piglet
 
On Friday, June 28, 2019 at 10:23:41 PM UTC-7, Robert Baer wrote:
Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15 Rem/hour.

Is there any data concerning E-B zener voltage changes, and is there
any material difference WRT transistor type (say 2N3904)?
Click to expand...

There's not a lot of consistency; a different metallization may
cause an X-ray fluorescence, so the KIND of radiation matters as
much as the ion-pair-production Rem number.
 
On Saturday, June 29, 2019 at 1:23:41 AM UTC-4, Robert Baer wrote:
Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15 Rem/hour.
Not steady state, but plenty every day.

In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).

Is there any data concerning E-B zener voltage changes, and is there
any material difference WRT transistor type (say 2N3904)?

Thanks.

What's a convenient way to get a "high dose area"?
Click to expand...
When in college (long ago) I had ready access to a
nuclear reactor, and all kinds of sources, but ...

Got a friend in the nuclear medicine field?
 
On Sat, 29 Jun 2019 18:20:31 -0700 (PDT), sdy <sdeyoreo@hotmail.com>
wrote:

On Saturday, June 29, 2019 at 1:23:41 AM UTC-4, Robert Baer wrote:
Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15 Rem/hour.
Not steady state, but plenty every day.

In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).

Is there any data concerning E-B zener voltage changes, and is there
any material difference WRT transistor type (say 2N3904)?

Thanks.

What's a convenient way to get a "high dose area"?
When in college (long ago) I had ready access to a
nuclear reactor, and all kinds of sources, but ...

Got a friend in the nuclear medicine field?
Click to expand...

You might also check some amateur radio AMSAT groups. They build
satellites and especially those satellites with highly elliptical
Molnya orbits cross the VanAllen radiation belts twice in an orbit,
four times a day. Check what components they are using.
 
On Saturday, 29 June 2019 03:46:54 UTC-4, Winfield Hill wrote:
Robert Baer wrote...

Placing the working unit (open base) in a very
high dose area for 2 months, where it will see
approximately a rate of approximately 15 Rem/hour.

What's a convenient way to get a "high dose area"?
When in college (long ago) I had ready access to a
nuclear reactor, and all kinds of sources, but ...


--
Thanks,
- Win
Click to expand...

Hi, Win:-

A few years ago I did some rad testing for a potential deep
space project- we used the TRIUMF particle accelerator facility
in BC.

https://www.triumf.ca/

The beam current was set by calling into the control room IIRC (single
digit nA) and you cooked the circuit for as long as you wanted to get the
integrated dosage. I ran long cables out of the high radiation area for
monitoring (and power control in case something glitched). A few PCs, a bunch
of remotely controlled power supplies and test equipment etc. Took that all
in airline baggage.

Our tests were slotted around some Intel and IBM greybeards who were dealing
with some advanced processor.

You do a safety course and briefing and so on, as you might expect.

--Spehro Pefhany
 
On Fri, 28 Jun 2019 22:23:45 -0800, Robert Baer wrote:

Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15
Rem/hour.
Not steady state, but plenty every day.

In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).

Is there any data concerning E-B zener voltage changes, and is there
any material difference WRT transistor type (say 2N3904)?

Thanks.
Click to expand...

Trivially, I would expect the metal-can 2222A to be better than 3904.
 
Wond <lost@the_ether.com> wrote:

> Trivially, I would expect the metal-can 2222A to be better than 3904.

Some radiation may get transformed in the metal-can into a shower of
lower-energy particles with more effect in silicon. So it will depend.
--
Uwe Bonnes bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt
--------- Tel. 06151 1623569 ------- Fax. 06151 1623305 ---------
 
On 7/1/19 12:21 PM, Uwe Bonnes wrote:
Wond <lost@the_ether.com> wrote:

Trivially, I would expect the metal-can 2222A to be better than 3904.

Some radiation may get transformed in the metal-can into a shower of
lower-energy particles with more effect in silicon. So it will depend.
Click to expand...

The main invariant characteristic was that devices with higher than
average B-E operating current densities at both ends of their specified
conditions were more intrinsically hard than others, e.g. the 2N914 as
an example of a more resistant device.
 
On 7/1/19 2:24 PM, bitrex wrote:
On 7/1/19 12:21 PM, Uwe Bonnes wrote:
Wond <lost@the_ether.com> wrote:
    Trivially, I would expect the metal-can 2222A to be better than
3904.

Some radiation may get transformed in the metal-can into a shower of
lower-energy particles with more effect in silicon. So it will depend.


The main invariant characteristic was that devices with higher than
average B-E operating current densities at both ends of their specified
conditions were more intrinsically hard than others, e.g. the 2N914 as
an example of a more resistant device.
Click to expand...

that is to say it's better to have a larger P/A ratio plus a higher
average current density and, up to a point, somewhat smaller volume BE
junction than a smaller P/A ratio, lower current density and larger volume.

the area decreases as the square but the volume decreases as the cube so
up to a point the absorbed dose becomes less because it's a harder
target to "hit." then past that point as the volume decreases further
the damage factor goes up again because what impacts there are hurt the
more fragile small junction a lot more.

I may not have that correct please correct if one can explain the
physics better
 
Am 29.06.19 um 08:23 schrieb Robert Baer:
  Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15
Rem/hour.
  Not steady state, but plenty every day.

  In the Apollo daze, some data may have been collected concerning this
question (uA709 W-stepping, input pair).

  Is there any data concerning E-B zener voltage changes, and is there
any material difference WRT transistor type (say 2N3904)?

  Thanks.
Click to expand...
<
https://gitlab.com/librespacefoundation/sdrmakerspace/radtest/wikis/home
>

That is not a direct answer to your question, but nevertheless
interesting and in the field.

Cheers, Gerhard
 
On Monday, July 1, 2019 at 9:21:11 AM UTC-7, Uwe Bonnes wrote:
Wond <lost@the_ether.com> wrote:

Trivially, I would expect the metal-can 2222A to be better than 3904.

Some radiation may get transformed in the metal-can into a shower of
lower-energy particles with more effect in silicon. So it will depend.
Click to expand...

Yes, exactly: high energy gammas that go through silicon without harm
can hit a gold-plate (or Ni or Cu or Fe) material and make
few-kV X-ray fluorescence that dumps all its energy in a millimeter of Si.
 
bitrex wrote:
On 6/29/19 2:23 AM, Robert Baer wrote:
   Placing the working unit (open base) in a very high dose area for 2
months, where it will see approximately a rate of approximately 15
Rem/hour.
   Not steady state, but plenty every day.

   In the Apollo daze, some data may have been collected concerning
this question (uA709 W-stepping, input pair).

   Is there any data concerning E-B zener voltage changes, and is
there any material difference WRT transistor type (say 2N3904)?

   Thanks.

There are some degredation curves for common components given on p. 235
in this whitepaper from the Voyager program, looks like:

https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19720018918.pdf

one from around the same time for neutron flux, military, stats-math-heavy:

https://apps.dtic.mil/dtic/tr/fulltext/u2/a065162.pdf

Foreign paper (Malaysia), x-rays:

http://www.jcomputers.us/vol9/jcp0909-26.pdf


The first two look somewhat relevant but this data is pushing 40 years
old. how relevant is it to bjts made using modern process? How accurate
was it to begin with? there is probably more recent/detailed info out
there but it sounds like the kind of thing that might require a clearance.

Could always do a FOIA request to the DOE as it's quite specific what
you're looking for see if there's anything unclassified or no longer
classified they might be willing to release.

another one from JPL that's mid 90s vintage, looks like, very topical
but it's pretty bare-bones with respect to data, couple crappy graphs:

https://pdfs.semanticscholar.org/762a/54443600250706f4d73e07aab7b126de9383.pdf


"Recent experience on the Cassini spacecraft project has shown that some
bipolar devices exhibit large decreases in gain at low total dose
levels, severely impacting their use in space. Figure 1 compares gain
degradation of two small signal transistors, measured at the lowest
collector current in the manufacturer’s specifications. The 2N918
transistor exhibits only small changes in gain with total dose, while
the 2N3700 is severely degraded, even at levels below 10
krad(Si).

The gain loss is so severe that it is extremely difficult to use this
device on the project, which must operate at levels between 50-100
krad(Si). This extreme degradation was not observed for earlier lots fl
om the same manufacturer, and was greater than anticipated for any
bipolar transistor with standard construction and normal
breakdown voltage requirements."
Click to expand...
Only thing is "modern processes" for olde/vintage semis is process
cleanliness is better and purities may be better.
So if manufacturer kept the basic process, part "sensitivity" will
remain the same.

**
Thanks for the references.
 
bitrex wrote:
On 6/29/19 2:02 AM, bitrex wrote:
On 6/29/19 2:23 AM, Robert Baer wrote:
   Placing the working unit (open base) in a very high dose area for
2 months, where it will see approximately a rate of approximately 15
Rem/hour.
   Not steady state, but plenty every day.

   In the Apollo daze, some data may have been collected concerning
this question (uA709 W-stepping, input pair).

   Is there any data concerning E-B zener voltage changes, and is
there any material difference WRT transistor type (say 2N3904)?

   Thanks.

There are some degredation curves for common components given on p.
235 in this whitepaper from the Voyager program, looks like:

https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19720018918.pdf

one from around the same time for neutron flux, military,
stats-math-heavy:

https://apps.dtic.mil/dtic/tr/fulltext/u2/a065162.pdf

Foreign paper (Malaysia), x-rays:

http://www.jcomputers.us/vol9/jcp0909-26.pdf


The first two look somewhat relevant but this data is pushing 40 years
old. how relevant is it to bjts made using modern process? How
accurate was it to begin with? there is probably more recent/detailed
info out there but it sounds like the kind of thing that might require
a clearance.

Could always do a FOIA request to the DOE as it's quite specific what
you're looking for see if there's anything unclassified or no longer
classified they might be willing to release.

another one from JPL that's mid 90s vintage, looks like, very topical
but it's pretty bare-bones with respect to data, couple crappy graphs:

https://pdfs.semanticscholar.org/762a/54443600250706f4d73e07aab7b126de9383.pdf


"Recent experience on the Cassini spacecraft project has shown that
some bipolar devices exhibit large decreases in gain at low total dose
levels, severely impacting their use in space. Figure 1 compares gain
degradation of two small signal transistors, measured at the lowest
collector current in the manufacturer’s specifications. The 2N918
transistor exhibits only small changes in gain with total dose, while
the 2N3700 is severely degraded, even at levels below 10
krad(Si).

The gain loss is so severe that it is extremely difficult to use this
device on the project, which must operate at levels between 50-100
krad(Si). This extreme degradation was not observed for earlier lots
fl om the same manufacturer, and was greater than anticipated for any
bipolar transistor with standard construction and normal
breakdown voltage requirements."

one of the other unfortunate take-aways from the final paper is that
damage profile can apparently vary widely even among the same generic
type from different manufacturers.
Click to expand...
Wide variations of even a "simple" part like a 2N2222 seem to have
been a tour-de-force for ages.
Even from the SAME manufacturer over time (different fabs used,
different "formulas", different diffusion containers:new vs used).
Ain't no sech ting as a "standard".
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top