Low light amplification
Michael Harwood
I belive that there are two types: infrared -> visible conversion
and amplification of existing dim light.
I am more interested in the physics behind them (eg. semiconductors used)
than how to actually build one.
Thanks.
Michael Harwood. har...@quark.physics.uwo.ca
(Physics Teacher, Ingersoll District Collegiate Institute, Ingersoll, Ontario)
--
Michael Harwood. har...@quark.physics.uwo.ca
(Physics Teacher, Ingersoll District Collegiate Institute, Ingersoll, Ontario)
Morris the Cat
|I belive that there are two types: infrared -> visible conversion
| and amplification of existing dim light.
|I am more interested in the physics behind them (eg. semiconductors used)
|than how to actually build one.
I believe there's a book that was published by Howard Samms' by
I.P. Csorba called "Image Tubes" back in 1985 that would probably
answer your questions. I believe he also edited some sort of selected
papers on image tubes book as well that consisted of reprints of various
journal articles dating back to the 1950s...
If you have access to _Applied_Optics_, you can check there as well,
as I think they've published some articles on 3rd Generation image
intensifier tube technology in the past. These are devices that actually
use a flat piece of semiconductor material like GaAs, InAsP, GaInAs,
etc. A good overview night vision article entitled: "Image Intensifiers
and the Scotoscope" that describes the basic technology up to 1964
shows up in Vol. 3(6), 651-672 (June 1964). Note that this is before
the introduction of the 3-stage fiber-optically coupled image intensifier
tubes into the U.S. military via the AN/PVS-1, etc.
In general, infrared image convertor tubes for night vision goggles
used what is called the "S-1" photocathode, a semi-transparent semi-
conductor material that is a combination of Silver-Oxygen-Cesium. Most
Western designs used the 8598 infrared image convertor tube, which
was produced by ITT and others in Europe. They were powered by a 12KV
source, incorporate an inverting electron-lens, and use a yellow-green
P-20 phosphor. Typical NV goggle: AN/PAS-5. Stochiametric formula of
S-1 is Ag + Cs2O.
Larger versions of the tube were the 6929, used in handheld metascope
IR viewers, and the 6914, used in weaponsights like the AN/PAS-4/Varo
9903.
For the image intensifier designs intended for night vision, there are
basically two types of semi-transparent photocathodes: S-20 and S-25
S-20 was the earliest one; I think it dates back to the mid-1950s. It's
stochiametric formula is Na2KSb + Cs. It's spectral response is visible
to near-infrared. This is called a bialkali photocathode.
S-25 is the later one, and apparently dates back to research in the
1970s. It's stochiametric formula is Na2KSb + Cs2Sb, and has a far
higher luminous sensitivity and extended response into the near
infrared region.
There is also what is called a "Super S-25" photocathode that has been
perfected by the Europeans that is marketed by Delft and others; it's
luminous sensitivity is not that much worst than a 3rd Generation
Negative Electron Affinity photocathode like GaAs.
Most Western design night vision goggles use what is called a "proximity
focussed" image intensifier that basically uses a flat photocathode
that is adjacent to a Micro-Channel Plate which is in turn adjacent
to a P-20 or other material phosphor screen.
Low-cost Russian "1st Generation" goggles typically use an S-25 photo-
cathode single-stage image intensifier tube with inverting electron-
optics.
Ian D. Leedom
(Michael Harwood) wrote:
I won't claim to be an expert but the night vision glasses with which I am
familiar are based on multi-channel plates (ITT Electro-optical for
instance).
They use a standard bi-alkalai photocathode with a typical quantum
efficiency of 20% or so. Electrons ejected from it are accelerated a few
millimeters through a drop of about 1 KV where they strike a plate which
is etched with a series of holes the potential difference between the top
and bottom of the plate/hole is again ~1KV. If the electron strikes the
plate it is lost. If it strikes the top of a hole, however, it rattles
down to the bottom creating a larger and larger shower with every
collision with the side of the hole. In that sense it is like a
photomultiplier's set of dynodes.
The advantages of it are that it provides large (~10**5) amplification
for each photoelectron ejected. That is much larger than can be achieved
with any solid state device. Thus the electronics needed are relatively
standard. None of the ultra-low noise sort of thing is necessary.
Secondly, unlike the photomultiplier the shower is localized and thus
provides spatial information about where the incident photon hit. That
allows one to build pixelated anodes to receive the signal and voila,
high gain + position resolution = night vision.
The disadvantages are low overall quantum efficiency because the plate
can not be all "holes". Only ~60% is and thus the effective quantum
efficiency falls to 10-12%. If one is looking for single photon electrons
(as I was) life becomes very tough. Additionally they are susceptible (as
are photomultipliers) to magnetic field effects, though not to the same
extent as PMT's since they are much shorter. The total length can be
anywhere from 2.5-5.0 cm. In real devices there is generally more than
one plate to give sufficient amplification. One just places the second
(or third) plate directly behind the first. I don't know if ITT is still
in the business, but you might try calling them since they are (or were)
one of the experts in the field.
Regards,
Ian Leedom
Zvi Rozensher
Harwood) writes:
It would be difficult to explain here the entire theory of operation.
A very good book on the subject is: "Image Tubes" by Illes P. Csorba,
published by SAMS (Indianapolis IN) 1985. ISBN: 0-672-22023-7.
Reccomended.
Zvi
Jim0123
>I need some information on how night vision goggles work.
>I belive that there are two types: infrared -> visible conversion
>and amplification of existing dim light.
> I am more interested in the physics behind them
>(eg. semiconductors used)> than how to actually build one.
>Michael Harwood. har...@quark.physics.uwo.ca
>(Physics Teacher, Ingersoll District Collegiate Institute, Ingersoll,
>Ontario)
Ian did a good job of describing modern 'microchannel' devices.
The older night-vision devices did not use these plates at
all. Instead, they had a photocathode which was coated with
various alkali-metal salts which photons can easily knock
electrons out of (cesium alone will work, but has limited
spectal response, thus the mixtures).
Where the microchannel designs derive gain by creating more
electrons out of the original one emitted from the photocathode,
the older style (1st generation) derived their gain strictly by
accelerating the photocathodes electron through a very high
voltage potential ... usually 10-15,000 volts. The accerated
electron would then "smack" into a phosphor screen with enough
energy to emit a noteworthy blip of light.
The primary limitation to this design was the maximum working
voltage. Go above 15kv and you have an x-ray machine more
than a night-vision device. Also, it's hard to render high voltage
devices immune to real-life military field conditions. The usual
compromise was achieved by linking two or three tubes together
so that the output of the first would feed into the second and
so on. The actual raw light-gain produced by such an arrangement
was equal or better than that achieved by single-stage 2nd gen
microchannel devices. Of course, there are limits ... each stage
makes the image a bit fuzzier, and each stage creates a bit
of random 'noise' which the next stage just makes worse. In
practice, three tube stacks seemed to be favored.
The photoelectrons in each stage are normally focused into an
image via an 'electron lens'. This takes the form of a metal
cone with the tip cut off, attached to the anode (+) end of
the tube. Depending on where on the photocathode the electron
was emitted, it would 'feel' a different vector attraction -
with the net result being a lens effect. This being an
all-static hardware solution which you could build right into
the tube and forget - it was quite practical. Alas, this
style of 'lens', often combined with a spherical-section
photocathode, usually produced an image with a lot of
'pincushion' distortion. The eye ignores this pretty well
during field use, but it looks really weird if one is taking
photographs. Early tubes occasionally tried more dynamic
focusing arrangements, either electrostatic or magnetic,
but this was more trouble than it was worth.
You asked about the infrared tubes vs the visible-light
variety. The only real difference was in the alkali metal
admixtures used on the photocathodes. All modern
photocathode coatings are also sensitive to IR radiation
up to about 900nm. 'Real' IR tubes are sensitive up to
around 1500nm - but fall off sharply at the other end
of the equation, around 750nm, and therefore are fairly
blind to visible light.
The latest (3rd gen) MCP designs use something called a
'negative-electron-affinity' coating for the photocathodes.
I think it contains gallium-arsenide. The quantum effeciency
is very high ... although there is now a more traditional
mix called 'super' S-25 which is nearly as good and apparently
cheaper to make.
Hope this fills in historical perspective a little.
--Jim
[ Language, while a product of thought, is an imprecise
mechanism for the conveyance of thought. ]
Morris the Cat
Fortunately for most night vision applications, just one MCP device
will suffice for a luminous gain mechanism... while the Open-Area
ratio of the MCP is based on the physical dimensions, apparently
MCPs that are uncoated to suppress positive ion feedback that beat
hell out of 3rd Generation photocathodes tend to create an electric
field on the input surface where each microchannel is that draws
photoelectrons toward the microchannel itself, rather than impact
on the adjacent web or surface... this aids the S/N ratio...
| The older night-vision devices did not use these plates at
| all. Instead, they had a photocathode which was coated with
| various alkali-metal salts which photons can easily knock
| electrons out of (cesium alone will work, but has limited
| spectal response, thus the mixtures).
The S-1 infrared photosensitive photocathode is the oldest of them
all, being incorporated into various phototube diodes. Infrared
image convertor designs date to the 1930s at least, with the USA,
UK and Germany all working on them.
| Where the microchannel designs derive gain by creating more
| electrons out of the original one emitted from the photocathode,
| the older style (1st generation) derived their gain strictly by
| accelerating the photocathodes electron through a very high
| voltage potential ... usually 10-15,000 volts. The accerated
| electron would then "smack" into a phosphor screen with enough
| energy to emit a noteworthy blip of light.
Without mature microchannel plate technology, research during the
1950s into image intensification design involved things like
fractional magnification, where the electron flux of a relatively
large photocathode was concentrated via an electron lens to
generate a luminous gain. Other designs used secondary electron
emmission plates, while still others a "sandwich" of glass between
a phosphor and photocathode. Note that the fractional magnification
system of gain is a favorite among X-Ray image intensifier tubes.
During the 1980s, there was sold for a short-time the Lixiscope,
a handheld miniature X-Ray machine that used a shielded radiation
source with a mechanical shutter. For viewing, a 25mm MCP image
intensifier tube with a scintillation phosphor attached to the input
fiberoptic window was employed. (What ever happened to this?)
| The actual raw light-gain produced by such an arrangement
| was equal or better than that achieved by single-stage 2nd gen
| microchannel devices.
Actual brightness of an image intensifier to the human eye is usually
limited in night vision equipment so as to take advantage of the
integration over time attributes of the eye so as to increase the apparent
signal-to-noise ratio...
| The photoelectrons in each stage are normally focused into an
| image via an 'electron lens'. This takes the form of a metal
| cone with the tip cut off, attached to the anode (+) end of
| the tube. Depending on where on the photocathode the electron
| was emitted, it would 'feel' a different vector attraction -
| with the net result being a lens effect. This being an
| all-static hardware solution which you could build right into
| the tube and forget - it was quite practical. Alas, this
| style of 'lens', often combined with a spherical-section
| photocathode, usually produced an image with a lot of
| 'pincushion' distortion.
In such multi-stage image tubes, the use of a fiberoptic plate as the
input window and photocathode deposition surface means that radius
of the photocathode can be reduced, thus giving improved distortion
characteristics, as compared to early single-stage infrared convertor
tubes with only a curved glass photocathode. No compromise of the
electron-optics vis-a-vis optical constraints with the FO input
plate.
| The eye ignores this pretty well
| during field use, but it looks really weird if one is taking
| photographs. Early tubes occasionally tried more dynamic
| focusing arrangements, either electrostatic or magnetic,
| but this was more trouble than it was worth.
This depends on the application; EMI was manufacturing a four-stage
image intensifier with magnetic focussing for astronomical applications
some years back. With magnetic focussing (the photoelectrons traveling
through the lines of magnetic force from the magnetic focussing coils,
have a force vector orthogonal to their travel, which causes them to
spiral) a very linear image is obtained.
| You asked about the infrared tubes vs the visible-light
| variety. The only real difference was in the alkali metal
| admixtures used on the photocathodes. All modern
| photocathode coatings are also sensitive to IR radiation
| up to about 900nm.
No, it depends on the type. Some photocathodes (S-11) don't have
significant IR sensitivity, being optimised for the visible-UV
spectrum. "Solar-blind" applications come to mind; the Canadians
use a UV-sensitive image intensifier system to aid in monitoring
the compliance of nuclear-material treaties by using the device
to observe the ultra-violet Cherenkov radiation of nuclear fuel
stored in water tanks...
Allard Mosk
>
> | Ian did a good job of describing modern 'microchannel' devices.
>
> Fortunately for most night vision applications, just one MCP device
> will suffice for a luminous gain mechanism... while the Open-Area
> ratio of the MCP is based on the physical dimensions, apparently
> MCPs that are uncoated to suppress positive ion feedback that beat
> hell out of 3rd Generation photocathodes tend to create an electric
> field on the input surface where each microchannel is that draws
> photoelectrons toward the microchannel itself, rather than impact
> on the adjacent web or surface... this aids the S/N ratio...
( Lots of interesting info deleted )
Could anyone please give a literature reference on the physics
of microchannel plates ? I'm especially interested in saturation
characteristics and short pulse amplification.
(I was hoping to be able to use a 1-mcp image intensifier to get
a reasonable amplification of 10ns fluorescent images at very low
duty cycle, but alas - it doesn't work, the mcp seems to saturate a
although average intensity is very low. Gain iseems to be less than 1
100(
Thanks,
Allard Mosk
van der Waals- Zeeman laboratory
University of Amsterdam
mo...@phys.uva.nl