anti-blooming hot defects

[MYPIXEL at aol.com]

In a message dated 97-10-22 11:13:22 EDT, rkohley at astro.uni-bonn.de writes:

<< Hi Derek, Jim and all,
 
Jim here . . .

 I would like to comment again on the 'hot spots in
anti-blooming'-discussion.
 I do not agree with Jim, that the hot pixels seen in anti-blooming pictures
 are generated by the same process as Pocket Pumping, because we do not see
 any neighbouring dark pixels and the anti-blooming clocking should be
 limited to within the pixel. Charge is also not conserved, since they can
 be also seen in dark frames as spikes above the zero level. In my
understanding
 the visible charge in these hot pixels must be generated within the same 
 pixel. 

You can see traps when your'e clocking only two phases as you do with clocked
AB. For example, if there is a slightly higher potential on the side of a
phase (called a spurious potential bump) charge in that pixel can split and
end up in the neighboring pixel (shows up as a bright pixel for a flat field
exposure next to the target pixel which will be darker). Might draw a
potential well plot to show this. This is a common signature . .. therefore,
where are the traps??? Single electron traps are most common and show a very
uniform distribution.

So they are truly hot . . . no dark spots next to them ..  . . mmmmm. 

Bright means charge generation. "Jim dots" occur when you leave one phase low
and clock the other two for a long period of time then read out (just like
clocked AB). Dots were very popular long ago when "all oxide" CCDs were being
fabricated (as opposed to oxide/nitride CCDs today - Kodak and TI still make
these). The effect was a luminescence one (like little LED's). The horizontal
register also glowed in a similar manner since it was usually clocked during
integration (for WF/PC I we had to turn off the horizontal registers during
integration). Anyway, this effect was caused by high fields within the oxide
..  . hot electrons were generated. Oxide nitride CCDs do not show the
problem to my knowledge. You do have an oxide/nitride CCD?

What else generates hot pixels . . .

Dark spikes ..  defects or impurities located at the rising edge of the
potential well (where electric fields are highest). Cooling eliminates these
guys.

Cosmic rays . . . you know when you have one of these.

Bit clipping ..  . anything like that going on?
 
 I agree, that they show the location of traps, but I'm now a bit confused
 about the kind of traps concerned. Maybe, Jim you can help me out:
 I was always believing, that spurious charge is produced by impact
ionisation
 of 'interface trapped holes', 

Yes, spurious charge comes from moving trapped holes from interface back to
channel stops (like a sling shot) causing impact ionization and generating
electron hole pairs.

whereas 'Pockets' are traps far below the
 surface (bulk?), 

Nonuniform potential across a phase . . . a bump or pocket.   Defects or
impurities (that are single electron traps) are uniformily distributed
through out silicon (pumping  or clocked AB shows these clearly if enough
cycles are used (recall charge builds up proportionally to trap size and
number of reversals). There is a physical difference between spurious
potential pockets and defects. Pumping does not differentiate except to say
that pockets are usually much large than defects and impurties.

which show up when moving even low levels of existing charge
 in between two pixels.

Yes
 
 I've looked into our anti-blooming experiments and found some ideas
 to decrease the generation of spurious charge, which should lower the
 hot pixel defects, too:

Why hot pixels too if we don't know what they are . . .?
 
 - increase of inversion voltage (limit for our Loral CCDs: -7 V)
 - decrease of |Vlow - Vhigh|
 - edge shaping and invention of medium voltage step while switching from
   Vlow -> Vhigh
 - enlarging the length of the medium voltage step ( >= 128 microsec )
 - choosing another phase combination than P1 and P2.

Yes, these methods will decrease spurious charge . . . 

I don't think we settled what the hot things really are. However, you should
see light and dark pixels in trap locations when performing clocked AB . . .
I've never seen a CCD that doesn't. The effect is the main complaint about
using clocked AB.

It is difficult to figure things out through the e-mail isn't it . ? drawing
a picture is sometimes the best way to communicate. Keep up conversations on
this ..  .

Jim ..  .

PS  ..  .  Has CCD World worked these last few days. My out going messages
have been bouncing back.


 
 Good luck Derek!


 
 Cheers,
 
 Ralf Kohley
 
 +--------------------------------------------------------------------+
 | Ralf Kohley                                   Tel.: +49 228 735658 |
 | Sternwarte der Universitaet Bonn              Fax.: +49 228 733672 |
 | Auf dem Huegel 71                                                  |
 | D-53121 Bonn                                                       |
 | Germany                           Email: rkohley at astro.uni-bonn.de |
 +--------------------------------------------------------------------+