New silicon 100-500x more light sensitive

SiOnyx Brings “Black Silicon” into the Light; Material Could Upend
Solar, Imaging Industries
Wade Roush 10/12/08

http://www.xconomy.com/boston/2008/...erial-could-upend-solar-imaging-industries/2/

Silicon is a wonderfully cooperative element. It takes relatively
little energy to promote the electrons in a silicon crystal from their
usual, docile orbits around the atomic nuclei into wild, free
circulation. That’s what makes silicon a semiconductor—valuable for
electronic switching devices such as transistors, sensing devices such
as the CCDs in cameras and X-ray machines, and energy-generating
devices such as photovoltaic cells.

But silicon would be more wondrous if it were even more responsive—if
an incoming photon needed less energy to knock loose an electron, for
example, or if a single photon could kick loose many electrons. In
pursuit of this vision, chemists, physicists, and engineers have spent
decades trying out various ways of modifying silicon crystals—for
example, by doping them with atoms of arsenic or other elements that
put more free electrons into the mix.

Almost ten years ago, graduate students in the laboratory of physics
professor Eric Mazur at Harvard University stumbled across a new way
of making silicon more responsive: they found that if they blasted the
surface of a silicon wafer with an incredibly brief pulse of laser
energy in the presence of gaseous sulfur and other dopants, the
resulting material—which they called “black silicon”—was much better
at absorbing photons and releasing electrons. And this week, after
nearly three years in hyper-stealth mode, a spinoff company with an
exclusive license from Harvard to commercialize the process has begun
talking with reporters.

Executives for the company, called SiOnyx, believe that its technology
will help semiconductor manufacturers build far more sensitive
detectors and far more efficient photovoltaic cells, using essentially
the same silicon-based processes they currently depend on—thereby
revolutionizing areas such as medical imaging, digital photography,
and solar energy generation.

The venture-funded startup has emerged with a bang, securing exclusive
coverage by New York Times technology writer John Markoff in today’s
edition. But SiOnyx CEO Stephen Saylor and principal scientist James
Carey, a PhD graduate of Mazur’s lab, also showed me around their
Beverly, MA, facility last week, on the condition that this post would
appear after Markoff’s story.

SiOnyx principal scientist James Carey (L) and CEO Stephen Saylor (R)
“You’ve never been able to detect light the way this stuff detects
light,” says Saylor, referring to black silicon’s remarkable
sensitivity to incoming photons, especially photons at infrared
energies, which pass through normal silicon as if it were transparent.
That property could make it an ideal, and inexpensive, replacement for
less-sensitive detectors in devices as varied as X-ray and CRT
machines, surveillance satellites, night-vision goggles, and consumer
digital cameras. “It means that you solve a clear and obvious pain
point for a very large number of customers,” Saylor says.

And because black silicon is just silicon that’s been roughed up a bit
by femtosecond laser pulses and chemical treatment, SiOnyx’s
technology could theoretically be integrated into existing
semiconductor fabrication lines without much disruption. “You can do
everything we’re talking about without extraordinary, Herculean
effort, and you can do it in a way that fits with high-volume
manufacturing flows,” says Carey.

SiOnyx was incorporated in 2005, secured the Harvard license in early
2006, and obtained $11 million in venture financing from Harris &
Harris, Polaris Venture Partners, and RedShift Ventures in 2007. The
company is going public with its story because “we have enough
momentum now both with strategic partners and with the technology that
it makes sense at this point to share a little more about what we are
up to,” say Saylor.

Harvard, for its part, is holding up SiOnyx as one early result of the
ongoing overhaul of the university’s technology licensing efforts. The
school gained a reputation early in this decade as being unresponsive,
even hostile, toward faculty and students who wished to commercialize
discoveries made in the university’s labs, especially in areas outside
of biotechnology and drug development. For years after the discovery
of black silicon in Mazur’s lab, the school’s technology transfer
office “wasn’t very excited” about the work, according to Carey.

But in 2005 the university brought in university licensing veteran
Isaac Kohlberg to rebuild its technology transfer operation from
scratch. Saylor and Carey say it was Kohlberg and his staff who
finally understood black silicon’s potential and ironed out the
licensing deal that made SiOnyx possible.

“The exciting steps being taken to develop [black silicon] for
commercial application serve as even more evidence of the
entrepreneurial energy that continues to gel and accelerate at
Harvard,” Kohlberg says in a press release set to be issued tomorrow
by SiOnyx and Harvard’s Office of Technology Development.

Bob Metcalfe, a general partner at Polaris Ventures who sits on
SiOnyx’s board, thinks Kohlberg is right: “Harvard seems to be getting
its act together in patent licensing,” he says.

Exactly what makes black silicon such an effective absorber of photons
is a question that even Mazur and Carey couldn’t answer at first. The
material is one of many offshoots of work going on in Mazur’s lab in
the late 1990s using femtosecond lasers—devices that can emit an
intense pulse of light lasting only a millionth of a billionth of a
second. Mazur lab researchers found that zapping a silicon wafer with
such pulses in the presence of sulfur hexafluoride gas—an experiment
initially carried out on a whim—left the wafer festooned with tiny
cones. Silicon roughened in this way soaks up almost all of the light
that strikes it in visible wavelengths, appearing black—hence the
name.

“It took several years for us to begin thinking properly about what we
had,” says Carey. “The original thought was that the surface
roughening process was what created the advantage.” The researchers
hypothesized that photons were bouncing from cone to cone—and that the
more times they bounced, the higher the likelihood that they’d be
absorbed, thus dislodging electrons. But then Carey and his coworkers
realized that black silicon was also absorbing infrared light, “which
you can’t explain just by roughening it.” It takes photons of a
certain energy to bump electrons in silicon’s outermost layer of
electrons, called the “valence band,” into the so-called “conduction
band,” where they’re free to circulate between atoms—and infrared
photons just don’t have enough. So by all rights, these photons should
have been passing right through without interacting with the material,
just as if it were frosted glass.

“That was the real discovery point,” says Carey. The genesis of
SiOnyx, he explains, came when the Mazur lab dug into the changes
caused by the femtosecond laser pulses at the atomic level. And as it
turned out, he says, “the cones weren’t really paramount at all”—
although they certainly look cool (electron micrographs of the cone
forests, like the one below, still appear alongside almost any
discussion of black silicon).

What’s really going on—though this is where Carey and Saylor start to
get cagey, since it gets at the proprietary heart of SiOnys’x
technology—is that the laser pulses force unusually large numbers of
dopant atoms into a thin layer of silicon on the surface of the cones.
“The laser allows you to put in a million times more sulfur than you
would normally get in if you just combined and heated them,” says
Carey. “In that millionth of a billionth of a second you get
structural arrangements frozen at the atomic level.”

Black siliconWith its new structure, the “band gap” in this thin
silicon layer—the difference in energy between the valence band and
the conduction band—is smaller. That means less energy is required to
knock electrons into the conduction band, which explains why infrared
photons can do the job. Another fringe benefit: applying a small
voltage to black silicon (engineers call this “bias”) creates
conditions in which a single incoming photon can knock loose dozens of
electrons. So, not only is the material responsive to wavelengths that
silicon-based devices simply couldn’t detect in the past—it also
produces a much stronger signal in response to a weak stimulus. Black
silicon is between 100 and 500 times more sensitive to light than
untreated silicon, the company says.

These properties mean that SiOnyx is in a position to pioneer new
types of solar cells that could capture the sun’s energy across a
broader spectrum, achieving greater efficiency than today’s
photovoltaic cells.
“Harnessing nuclear fusion energy arriving from Sol—solar energy at
1366 Watts per square meter—is the most promising technology for
meeting accelerating world needs for cheap and clean energy,” says
Polaris’s Metcalfe. Black siliicon “promises to dramatically increase
the photo-response (Amps per Watt) of silicon, and not just in the
visible spectrum, but also in the infrared, where silicon currently
misses half of Sol’s energy. Delivering on that promise is very
exciting.”

But that’s the “long shot” applicaiton for the material, Metcalfe
acknowledges. Closer in is the possibility of major sensitivity
improvements in imaging applications such as night vision,
surveillance, digital cameras, and medical imaging. Saylor says that
the company has negotiated strategic partnerships with two “industry
leaders,” and though he won’t name names, he says one of them is
active in the medical imaging area.

The attraction of black silicon in medical imaging is obvious: If you
could build a more sensitive detector for a CT or mammography machine,
you could expose patients to a lower dose of X-rays. (Black silicon,
of course, can’t detect X-rays directly; modern digital X-ray machines
include a component called a scintillator that emits visible light
when struck by X-rays, and that light is what’s recorded by a sensor.)
“If we can do something that allows women to get risk-free mammograms
twice a year or reduce the number of chest-X-ray equivalents that you
get from a CT scan, or address other pain points, we will have an
immediate path to market,” says Saylor.

While SiOnyx is telling some of its story, it’s keeping big pieces of
it under wraps. Asked how many employees the company has, Saylor says
it’s more than 10 and fewer than 50. (Significantly fewer, from what I
could see around SiOnxy’s offices—a space in the former United Shoe
Machinery factory in Beverly, far outside of Boston, that the company
picked because the previous tenant had installed a clean room.) The
company won’t build semiconductors or even semiconductor fabrication
equipment, but will instead work with as-yet-unnamed partners to
develop specifications for machines that can treat isolated areas of
silicon wafers to create black silicon.

SiOnyx engineers were using an automated testing device to examine
sections of such a wafer when I visited. “We are a process engineering
company, not a product engineering company,” says Saylor. “Our job is
to make a transferable process that conforms to [our partners']
manufacturing flow. We are doing a tremendous amount of development
around what are the optimal conditions for making this black silicon—
how do you do it uniformly, how do you make it massively scalable, and
how do you transfer it to a foundry.”

Metcalfe says the biggest challenges before SiOnyx right now are “to
move the black silicon process from labs to fabs, from experimental
facilities/processes at Harvard to production facilities/processes at
SiOnyx” and “to navigate through black silicon’s many opportunities to
the right go-to-market products.”

Saylor says he hopes the company won’t have to raise any more venture
capital to do that. “The first strategic relationships are going to be
with very well-aligned industry leaders, so those will lead to
development relationships and eventually product-revenue
relationships,” he says. The company will be “careful with cash” until
it can grow to the point that it “becomes interesting to someone
outside the venture investing community,” he says.

There’s an interesting irony to SiOnyx’s business: a large chunk of
the semiconductor industry’s effort over the past 50 years has gone
toward making silicon as pure as possible. But now SiOnyx and other
companies are showing how useful—and perhaps profitable—it can be to
craft silicon devices with impurities, defects, and unconventional
structures.

“We are messing up perfectly good silicon,” Carey admits. “But in the
end, the properties speak for themselves.

 

SiOnyx Brings “Black Silicon” into the Light; Material Could Upend
Solar, Imaging Industries
Wade Roush 10/12/08

http://www.xconomy.com/boston/2008/...erial-could-upend-solar-imaging-industries/2/

Silicon is a wonderfully cooperative element. It takes relatively
little energy to promote the electrons in a silicon crystal from their
usual, docile orbits around the atomic nuclei into wild, free
circulation. That’s what makes silicon a semiconductor—valuable for
electronic switching devices such as transistors, sensing devices such
as the CCDs in cameras and X-ray machines, and energy-generating
devices such as photovoltaic cells.

But silicon would be more wondrous if it were even more responsive—if
an incoming photon needed less energy to knock loose an electron, for
example, or if a single photon could kick loose many electrons. In
pursuit of this vision, chemists, physicists, and engineers have spent
decades trying out various ways of modifying silicon crystals—for
example, by doping them with atoms of arsenic or other elements that
put more free electrons into the mix.

Almost ten years ago, graduate students in the laboratory of physics
professor Eric Mazur at Harvard University stumbled across a new way
of making silicon more responsive: they found that if they blasted the
surface of a silicon wafer with an incredibly brief pulse of laser
energy in the presence of gaseous sulfur and other dopants, the
resulting material—which they called “black silicon”—was much better
at absorbing photons and releasing electrons. And this week, after
nearly three years in hyper-stealth mode, a spinoff company with an
exclusive license from Harvard to commercialize the process has begun
talking with reporters.

Executives for the company, called SiOnyx, believe that its technology
will help semiconductor manufacturers build far more sensitive
detectors and far more efficient photovoltaic cells, using essentially
the same silicon-based processes they currently depend on—thereby
revolutionizing areas such as medical imaging, digital photography,
and solar energy generation.

The venture-funded startup has emerged with a bang, securing exclusive
coverage by New York Times technology writer John Markoff in today’s
edition. But SiOnyx CEO Stephen Saylor and principal scientist James
Carey, a PhD graduate of Mazur’s lab, also showed me around their
Beverly, MA, facility last week, on the condition that this post would
appear after Markoff’s story.

SiOnyx principal scientist James Carey (L) and CEO Stephen Saylor (R)
“You’ve never been able to detect light the way this stuff detects
light,” says Saylor, referring to black silicon’s remarkable
sensitivity to incoming photons, especially photons at infrared
energies, which pass through normal silicon as if it were transparent.
That property could make it an ideal, and inexpensive, replacement for
less-sensitive detectors in devices as varied as X-ray and CRT
machines, surveillance satellites, night-vision goggles, and consumer
digital cameras. “It means that you solve a clear and obvious pain
point for a very large number of customers,” Saylor says.

And because black silicon is just silicon that’s been roughed up a bit
by femtosecond laser pulses and chemical treatment, SiOnyx’s
technology could theoretically be integrated into existing
semiconductor fabrication lines without much disruption. “You can do
everything we’re talking about without extraordinary, Herculean
effort, and you can do it in a way that fits with high-volume
manufacturing flows,” says Carey.

SiOnyx was incorporated in 2005, secured the Harvard license in early
2006, and obtained $11 million in venture financing from Harris &
Harris, Polaris Venture Partners, and RedShift Ventures in 2007. The
company is going public with its story because “we have enough
momentum now both with strategic partners and with the technology that
it makes sense at this point to share a little more about what we are
up to,” say Saylor.

Harvard, for its part, is holding up SiOnyx as one early result of the
ongoing overhaul of the university’s technology licensing efforts. The
school gained a reputation early in this decade as being unresponsive,
even hostile, toward faculty and students who wished to commercialize
discoveries made in the university’s labs, especially in areas outside
of biotechnology and drug development. For years after the discovery
of black silicon in Mazur’s lab, the school’s technology transfer
office “wasn’t very excited” about the work, according to Carey.

But in 2005 the university brought in university licensing veteran
Isaac Kohlberg to rebuild its technology transfer operation from
scratch. Saylor and Carey say it was Kohlberg and his staff who
finally understood black silicon’s potential and ironed out the
licensing deal that made SiOnyx possible.

“The exciting steps being taken to develop [black silicon] for
commercial application serve as even more evidence of the
entrepreneurial energy that continues to gel and accelerate at
Harvard,” Kohlberg says in a press release set to be issued tomorrow
by SiOnyx and Harvard’s Office of Technology Development.

Bob Metcalfe, a general partner at Polaris Ventures who sits on
SiOnyx’s board, thinks Kohlberg is right: “Harvard seems to be getting
its act together in patent licensing,” he says.

Exactly what makes black silicon such an effective absorber of photons
is a question that even Mazur and Carey couldn’t answer at first. The
material is one of many offshoots of work going on in Mazur’s lab in
the late 1990s using femtosecond lasers—devices that can emit an
intense pulse of light lasting only a millionth of a billionth of a
second. Mazur lab researchers found that zapping a silicon wafer with
such pulses in the presence of sulfur hexafluoride gas—an experiment
initially carried out on a whim—left the wafer festooned with tiny
cones. Silicon roughened in this way soaks up almost all of the light
that strikes it in visible wavelengths, appearing black—hence the
name.

“It took several years for us to begin thinking properly about what we
had,” says Carey. “The original thought was that the surface
roughening process was what created the advantage.” The researchers
hypothesized that photons were bouncing from cone to cone—and that the
more times they bounced, the higher the likelihood that they’d be
absorbed, thus dislodging electrons. But then Carey and his coworkers
realized that black silicon was also absorbing infrared light, “which
you can’t explain just by roughening it.” It takes photons of a
certain energy to bump electrons in silicon’s outermost layer of
electrons, called the “valence band,” into the so-called “conduction
band,” where they’re free to circulate between atoms—and infrared
photons just don’t have enough. So by all rights, these photons should
have been passing right through without interacting with the material,
just as if it were frosted glass.

“That was the real discovery point,” says Carey. The genesis of
SiOnyx, he explains, came when the Mazur lab dug into the changes
caused by the femtosecond laser pulses at the atomic level. And as it
turned out, he says, “the cones weren’t really paramount at all”—
although they certainly look cool (electron micrographs of the cone
forests, like the one below, still appear alongside almost any
discussion of black silicon).

What’s really going on—though this is where Carey and Saylor start to
get cagey, since it gets at the proprietary heart of SiOnys’x
technology—is that the laser pulses force unusually large numbers of
dopant atoms into a thin layer of silicon on the surface of the cones.
“The laser allows you to put in a million times more sulfur than you
would normally get in if you just combined and heated them,” says
Carey. “In that millionth of a billionth of a second you get
structural arrangements frozen at the atomic level.”

Black siliconWith its new structure, the “band gap” in this thin
silicon layer—the difference in energy between the valence band and
the conduction band—is smaller. That means less energy is required to
knock electrons into the conduction band, which explains why infrared
photons can do the job. Another fringe benefit: applying a small
voltage to black silicon (engineers call this “bias”) creates
conditions in which a single incoming photon can knock loose dozens of
electrons. So, not only is the material responsive to wavelengths that
silicon-based devices simply couldn’t detect in the past—it also
produces a much stronger signal in response to a weak stimulus. Black
silicon is between 100 and 500 times more sensitive to light than
untreated silicon, the company says.

These properties mean that SiOnyx is in a position to pioneer new
types of solar cells that could capture the sun’s energy across a
broader spectrum, achieving greater efficiency than today’s
photovoltaic cells.
“Harnessing nuclear fusion energy arriving from Sol—solar energy at
1366 Watts per square meter—is the most promising technology for
meeting accelerating world needs for cheap and clean energy,” says
Polaris’s Metcalfe. Black siliicon “promises to dramatically increase
the photo-response (Amps per Watt) of silicon, and not just in the
visible spectrum, but also in the infrared, where silicon currently
misses half of Sol’s energy. Delivering on that promise is very
exciting.”

But that’s the “long shot” applicaiton for the material, Metcalfe
acknowledges. Closer in is the possibility of major sensitivity
improvements in imaging applications such as night vision,
surveillance, digital cameras, and medical imaging. Saylor says that
the company has negotiated strategic partnerships with two “industry
leaders,” and though he won’t name names, he says one of them is
active in the medical imaging area.

The attraction of black silicon in medical imaging is obvious: If you
could build a more sensitive detector for a CT or mammography machine,
you could expose patients to a lower dose of X-rays. (Black silicon,
of course, can’t detect X-rays directly; modern digital X-ray machines
include a component called a scintillator that emits visible light
when struck by X-rays, and that light is what’s recorded by a sensor.)
“If we can do something that allows women to get risk-free mammograms
twice a year or reduce the number of chest-X-ray equivalents that you
get from a CT scan, or address other pain points, we will have an
immediate path to market,” says Saylor.

While SiOnyx is telling some of its story, it’s keeping big pieces of
it under wraps. Asked how many employees the company has, Saylor says
it’s more than 10 and fewer than 50. (Significantly fewer, from what I
could see around SiOnxy’s offices—a space in the former United Shoe
Machinery factory in Beverly, far outside of Boston, that the company
picked because the previous tenant had installed a clean room.) The
company won’t build semiconductors or even semiconductor fabrication
equipment, but will instead work with as-yet-unnamed partners to
develop specifications for machines that can treat isolated areas of
silicon wafers to create black silicon.

SiOnyx engineers were using an automated testing device to examine
sections of such a wafer when I visited. “We are a process engineering
company, not a product engineering company,” says Saylor. “Our job is
to make a transferable process that conforms to [our partners']
manufacturing flow. We are doing a tremendous amount of development
around what are the optimal conditions for making this black silicon—
how do you do it uniformly, how do you make it massively scalable, and
how do you transfer it to a foundry.”

Metcalfe says the biggest challenges before SiOnyx right now are “to
move the black silicon process from labs to fabs, from experimental
facilities/processes at Harvard to production facilities/processes at
SiOnyx” and “to navigate through black silicon’s many opportunities to
the right go-to-market products.”

Saylor says he hopes the company won’t have to raise any more venture
capital to do that. “The first strategic relationships are going to be
with very well-aligned industry leaders, so those will lead to
development relationships and eventually product-revenue
relationships,” he says. The company will be “careful with cash” until
it can grow to the point that it “becomes interesting to someone
outside the venture investing community,” he says.

There’s an interesting irony to SiOnyx’s business: a large chunk of
the semiconductor industry’s effort over the past 50 years has gone
toward making silicon as pure as possible. But now SiOnyx and other
companies are showing how useful—and perhaps profitable—it can be to
craft silicon devices with impurities, defects, and unconventional
structures.

“We are messing up perfectly good silicon,” Carey admits. “But in the
end, the properties speak for themselves.

Could be a while, but in the end there are still only so many photons to
collect, this may not help normal photography all that much.

Pete

 

I think they mean photons used to knock electrons out of their
orbits.
I was under the impression that consumer camera sensors were about 40%
efficient, with some thinned, back illuminated sensors used in
scientific fields up to 90% efficient.

 

So 100-500x more efficient, as in the subject line you posted means that
each photon can shift 40 - 200 electrons? (100-500x).
So, even if current sensor were 40%, and 100% could be achieved, then
that's one (...and a bit) photographic stop return for a miracle? And
this is a big deal?

 

[Blubb]

I smell a rat the size of an elephant.
http://clarkvision.com/imagedetail/digital.photons.and.qe/index.html

With a 1.4x converter and 500mm f/4 lens:
Total QE of a 10D (green band) is 9.5%
Total QE of a 1D MkIII (green band) is 15%.

And yes, Total QE includes all the light lost in the lens, in the
bayer filters, between the pixels (less than 100% fill factor)
--- everything!

Without inventing photons and with a *completely* *lossless*
lens, blur filter, green filter, and 100% fill factor you can
not get better than ~6.7 times -- or more than 2.7 stops.

Even against the old 10D, you cannot get betzter than 10.5
times or a tiny bit over 3 stops. Again, with "unobtainium"
lenses.


100 (6 2/3 stops) to 500 times (9 stops)? No way, unless you
use a photomultiplier. Which, since we are already counting
single electrons, doesn't give us anything, except "louder" and
"more noise".

But RichA won't understand that if it bit his behind.

-Wolfgang

 

Thanks, I've used PM tubes and image intensification gear.

 

Not exactly. 100x as much charge for each photon does not multiply the
read noise, so read noise, relative to absolute exposure, is extremely low.
Of course, unless some way of reducing electron gain is available, low ISOs
will not be available without ND filters.

 

You are still photon noise bound --- and it's not like a
photomultiplier was without noise.

Reducing the voltage to the photomultiplier should do that easily.

-Wolfgang

 

That's like saying "but you are still closer to paradise".

There's nothing much more I can ask for a camera than to give me the
photons. The read noise is what turns images to crap.

There is no patterning in photon noise; no blanketing of shadows in a noise
floor. Black is BLACK.

 

| So 100-500x more efficient, as in the subject line you posted means that
| each photon can shift 40 - 200 electrons? (100-500x).
| So, even if current sensor were 40%, and 100% could be achieved, then
| that's one (...and a bit) photographic stop return for a miracle? And
| this is a big deal?

Next thing they will be telling us is they can get 100 to 500 times the heat
out of a 1000 watt electric heater.

 

That isn't a very good analogy.

Whatever the actual photon capture is, it has to compete against noises
added in reading the sensor. Getting multiple electron charges from a
single photo capture means that read noise, relative to signal, is
drastically reduced.

Even reducng read noise to 50% of what it is at high ISOs in the best DSLRs
would be semi-revolutionary. Reducing it to 1% would be astounding, if
dark current noise doesn't become the main issue.

 

Right now, for a short exposure, in the shadows, at high ISO, read noise is
by far, the overwhelming source of noise. Losing most of the read noise,
and still having thermal noise, and shot noise, would be relative nirvana.

 

I presume they are calling this "black" silicon because
that's what it looks like. Light hits it, and doesn't come back.
Normal Si cells are somewhat reflective (photo sensors more
so, due to the AA filter).

Back at the press release:
"Black silicon “promises to dramatically increase the photo-response
(Amps per Watt) of silicon, and not just in the visible spectrum, but
also in the infrared, where silicon currently misses half of Sol’s
energy.
Delivering on that promise is very exciting.”"

IR is heat. Use these things for PV and they are going to
get hot, very hot, is my guess. The promoters seem to sense that...

"But that’s the “long shot” application for the material,
Metcalfe acknowledges."

The article does not explain why.

"Black" could also refer to the hole in which the VC money
will disappear, or not. We'll see.

 

A point which many low noise enthusiasts often forget it that if you
have more pixels than you need for a particular image, you can
sacrifice some of them to reduce noise. I'm not sure how much noise
you lose in practice by downsizing by 50%, but you certainly reduce it
a lot. In other words a high pixel count isn't necessarily to concede
ground in the noise wars, because you can use those extra pixels for
improved resolution or cropping freedom in good light, and for noise
reduction in poor light.

 

Chris Malcolm wrote:
[]

Usually, the signal adds linearly, and the noise in an RMS way. So if you
mean 50% linear scaling (24MP => 6MP), each output pixel would have four
times the signal and twice the noise, and hence a 2:1 improvement in SNR.

David

 

Nope. Do you really believe that photomultipliers are less noisy?
They can pump up the signal level making detection easier, they
can even be used in a logarithmically response curve.

Please do check for yourself how little influence read noise has.

There Is No Black.

-Wolfgang

 

I see. Can you quantify that?

According to clarkvision.com, some read noises are:
D200 7.4 e- at ISO 800 and ISO1600
D50 7.47 e- at ISO 800
D300 4.6 e- at ISO1600
1D MkIII 4.0 e- at ISO3200
1D MkII 3.90 e- at ISO1600
5D (not II) 3.7 e- at ISO1600 and ISO3200
350D 3.7 e- at ISO1600
20D 3.6 e- at ISO1600

-Wolfgang

 

A 4:1 pixel reduction (50% height, 50% width), if done correctly,
will reduce the noise by SQRT(4) == 2. However, at the same
output size that will change nothing. You need to reduce the
output size (or view both versions at 100%, versus viewing the
original at 50%), which'll change the DOF as seen by the viewer.

And because viewing or printing at a size smaller than "100%"
*is* a size reduction, you get the noise reduction for free.
The only way to beat that game is to use pixel binning, where
the charge from a group of pixels is pushed through the reader
as one, in which case you incur read noise only once, instead of
once for each pixel.

-Wolfgang