Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

Dear George,

I also wanted to respond to your question that you posed to Karl from 
Roper. 
I was in a way saddened to see this excellent scientific site being used 
for a product launch. While I am highly reluctant to see this forum being 
turned into a commercial combat arena, I now feel compelled to input into 
the debate which Roper have seen fit to initiate, and make sure the high-
end efforts of other EMCCD manufacturers have been more fairly 
represented. Especially as I think you have been misled with a 
scientifically unsound treatment of your query. 
So let me try to at least remain faithful to the scientific nature of this 
site and address this aspect of what was claimed with an injection of 
technical reality….

DARKCURRENT VS CIC
It is entirely justifiable to be concerned with minimizing Clock Induced 
Charge (CIC), also known as Spurious Noise. This point on its own is fine, 
and one that Andor have in fact addressed quite some time ago with our 
advanced clocking solutions. However, contrary to what was claimed, the 
CIC minimization point is in fact relevant only for  DEEP COOLED EMCCDS, 
such as the vacuum TE cooled iXonEM+.
Are Roper saying that the QuantEM will outperform their own deeper-cooled 
Cascade II 512B, because the Cascade II has higher CIC?? I stand also in 
defense of the new deeper-cooled Hamamatsu C9100-13 system in this regard, 
although I cannot comment on Hamamatsu’s treatment of the remaining CIC. 

Since Roper have been unable to offer cooling below -30 0C with the 
QuantEM, I strongly suspect that the level of amplified darkcurrent 
electrons vastly exceeds the level of CIC (unless the Roper CIC levels 
were extremely high to start with) – even with short exposures! That is 
much more likely to be the primary source of speckled background in their 
system.
Andor have proven this in the past with this very 512x512 CCD97 sensor, by 
showing EM-amplified darkcurrent events at x1000 EM gain @ temperatures 
of -30C, -50C and -80C using 30ms exposures. When you do not cool 
effectively (i.e. -30 C), the amplified noise floor is a forest of 
darkcurrent spikes and the contribution of CIC is the least of your 
concerns – this is not at all suitable for demanding applications such as 
confocal microscopy. The beauty of EMCCDs in low photon background set-ups 
like confocal, is that they can potentially measure ‘zero’ very well, as 
Jim Pawley was very quick to recognize in the ‘early days’ of EMCCD. 
However, this capability is much less likely if you do not cool the sensor 
sufficiently. It is when you push down towards -80 C that the remaining 
CIC (which has been now shown to be temperature independent) becomes the 
true detection limit. 
How would Roper separate CIC from darkcurrent at such relatively high 
temperatures? Is that why they do not quote a CIC spec in the spec sheet? 
Andor have several customers around the world using our iXon/iXon+ cameras 
for single photon counting – something we have shown you can only do when 
both darkcurrent and CIC have been minimized. 

HOW TO MEANINGFULLY COMPARE EMCCDS
The method for comparing sensitivity from the same EMCCD sensor (in this 
case the very popular E2V CCD97 512x512 back-illuminated sensor) in 
cameras from different manufacturers is quite simple, and you don’t even 
need a sample. Since the same sensor is used, the QE will essentially be 
the same (as long as the manufacturer uses only one input window) and the 
read noise floor is overcome through the EM effect. 
In terms of sensitivity performance, all that remains is to compare how 
well the respective manufacturers have reduced remaining dark events 
(darkcurrent and CIC), both of which will be manifest in an intensity 
profile as electron spikes amplified clear of the read noise floor.
Just screw a threaded cap onto the c-mount to make sure there are no 
photons getting to the sensor, turn up the EM gain to x1000 and take, say 
a 30ms image. Notice I don’t even ask you take an especially long 
exposure – significant darkcurrent can be generated during the readout 
process alone. Take a line intensity profile across any row of 512 pixels 
and count the typical number of amplified spikes across that row. 
That gives you a feel for the remaining detection limit of that EMCCD.
I urge anyone looking at high-end EMCCD technology to make sure they 
insist on such a demonstration.

Beyond this simple sensitivity comparison, look to the fundamental 
differences concerning the housing of the sensor. What surprises me more 
than any of the ‘new’ claims made, is that people are still buying 
backfilled systems housing back-illuminated sensors; BI sensors will 
degrade unless all moisture and hydrocarbons are permanently eliminated, 
and if you can ensure such elimination then a permanent vacuum is easy, as 
is deep cooling. It was these compelling reasons that drove Andor to 
develop permanent vacuum technology over 10 years ago.
Vacuum protection also means we can get away with only one input window, 
AR coated on each side – i.e. the sensor QE curve on the spec sheet is 
what you get in reality. A non-vacuum back-filled camera is much less 
protected against these degrading influences, and you would be ill-advised 
not to have an extra protective window in front of the sensor. So, for 
these reasons such a non-vacuum (o-ring seal, back-filled with inert gas) 
system cannot be considered in the same teir 1 EMCCD category as the Andor 
iXon/iXon+ or even the Hamamatsu C9100-13 (about which you originally 
asked).

For anyone else reading this who is thinking of using EMCCD technology, an 
Andor demo is pretty much all we are asking suggest. I am not requiring 
you to immediately believe everything I say while I am casting doubt on 
the assertions of someone else. I simply ask that you let us prove our 
claims. 
Andor pioneered the first scientific EMCCD cameras and have been at this 
game for much longer than anyone else. We have consistently stayed several 
steps ahead of the nearest EMCCD competitors – check out the web site 
www.andor.com for our complete EMCCD line-up and overview of their 
capabilities. You will see that for quite some time our high-end 
iXon/iXon+ solutions have used innovations such as our Baseline Clamp 
(i.e. stable bias over kinetic series or EM gain changes), RealGain (real 
and linearized gain control), EMCAL (patent-pending user initiated 
recalibration of EM Gain – don’t even need a light source) and Anti-Ageing 
technology. 
Apologies, this is turning into a plug too!

Anyway, that’s about all you will hear from me on the matter – I don’t 
wish to clog up this forum any further with commercial righteousness. 

Best Regards,

Colin Coates, PhD
Senior Scientist
Andor Technology Plc.

P.S. For those who feel inclined to read further and find out more about 
what Andor have pioneered in this technology area, go to www.emccd.com 
home page  -  ‘imitation is the sincerest form of flattery’!