Observing single electron flow

[fellow]

For a 1pf capacitor storing a single electron, V = e/C = 6.9 x 10(-7) V. So
with careful low noise design, I can't see it being impossible to design a
1pf capacitor and measuring the voltage to determine whether a single
electron is present. Is it worth investigating to see how far I can go?

Thanks.

 

[John Jardine]

For a 1pf capacitor storing a single electron, V = e/C = 6.9 x 10(-7) V. So
with careful low noise design, I can't see it being impossible to design a
1pf capacitor and measuring the voltage to determine whether a single
electron is present. Is it worth investigating to see how far I can go?

Thanks.

An electron charge is 1.6e-19? no?. (i.e. gives 0.16uV)

 

[John Larkin]

For a 1pf capacitor storing a single electron, V = e/C = 6.9 x 10(-7) V. So
with careful low noise design, I can't see it being impossible to design a
1pf capacitor and measuring the voltage to determine whether a single
electron is present. Is it worth investigating to see how far I can go?

Thanks.

I think I posed this question here a while back, but no discussion was
evoked.

If you google "single electron transistor" or "RF-SET" you'll see lots
of hits, but it's mainly custom cryo stuff. People are getting
sensitivities of a few micro-equivalent electron charges per root
Hertz so they can count single e's in roughly the 1KHz rate range.

At room temp or cooled a bit, with a super-bootstrapped jfet
amplifier, I'd guess it's just out of reach, buried in the noise. I
can't think of a statistical way to extract the electron quantization,
(since leakage currents are random) much less count each one. The
trick would be to get the leakage current low enough so that the
average electron rate is low enough that single steps could be seen
for a given amp bandwidth, and you need a narrow bandwidth to keep the
noise down. Microkelvin temperature fluctuations and 1/f noise will be
killers here.

This might work: make a bootstrapped jfet amp, with equivalent
capacitance of a fraction of a pf. Float the gate; it should have
average leakage of not many electrons/second.

Pulse (illuminate) the gate lead with a blue LED or some
short-wavelength source. At each pulse, some small number of electrons
might be ejected by photoelectric emission. It might be possible to
synchronously average the fet output over many shots and verify that
the the resulting level shift is quantized corresponding to a multiple
of e.

An ac-carrier-pumped, possibly resonant, varicap diode thing would be
at least worth analyzing. That would solve some of the noise problems,
maybe.

Let us know if you come up with anything.

John

 

[Winfield Hill]

John Larkin wrote...

This might work: make a bootstrapped jfet amp, with equivalent
capacitance of a fraction of a pf. Float the gate; it should
have average leakage of not many electrons/second.

A painful issue that rises whenever you bootstrap your way to
lower capacitance, is e_n-Cin current-noise, coming from the
JFET's voltage noise and the total unbootstrapped capacitance.
If expressed as a noise density, i_n = e_n C_total 2pi f.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Winfield Hill]

Winfield Hill wrote...

John Larkin wrote...

A painful issue that rises whenever you bootstrap your way to
lower capacitance, is e_n-Cin current-noise, coming from the
JFET's voltage noise and the total unbootstrapped capacitance.
If expressed as a noise density, i_n = e_n C_total 2pi f.

And the other leg to this noise-death chair, is that if you make a
larger JFET to reduce e_n, you increase the JFET's contribution to
C by the square of the improvement in e_n. Sorry, the possibility
of creating a pF or sub pF low-noise JFET is vanishingly small.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[John Larkin]

Winfield Hill wrote...

And the other leg to this noise-death chair, is that if you make a
larger JFET to reduce e_n, you increase the JFET's contribution to
C by the square of the improvement in e_n. Sorry, the possibility
of creating a pF or sub pF low-noise JFET is vanishingly small.

Well, I did suggest that single-e steps would be lost in the noise.
Demonstration of quantization should be possible, however, with signal
averaging *if* you have a trigger to synchronize the analysis with the
charge steps.

I suppose, in this case, bootstraping doesn't fundamentally help.
There are RF-SETs that work at room temp, but they are exotic
semiconductor structures.

Seems like a SEM cantelever might be able to detect single e's if
small charges can be arranged to bend it. They just use a laser and a
photodetector to sense atomic-level forces.

John

 

[Robert Baer]

I think I posed this question here a while back, but no discussion was
evoked.

If you google "single electron transistor" or "RF-SET" you'll see lots
of hits, but it's mainly custom cryo stuff. People are getting
sensitivities of a few micro-equivalent electron charges per root
Hertz so they can count single e's in roughly the 1KHz rate range.

At room temp or cooled a bit, with a super-bootstrapped jfet
amplifier, I'd guess it's just out of reach, buried in the noise. I
can't think of a statistical way to extract the electron quantization,
(since leakage currents are random) much less count each one. The
trick would be to get the leakage current low enough so that the
average electron rate is low enough that single steps could be seen
for a given amp bandwidth, and you need a narrow bandwidth to keep the
noise down. Microkelvin temperature fluctuations and 1/f noise will be
killers here.

This might work: make a bootstrapped jfet amp, with equivalent
capacitance of a fraction of a pf. Float the gate; it should have
average leakage of not many electrons/second.

Pulse (illuminate) the gate lead with a blue LED or some
short-wavelength source. At each pulse, some small number of electrons
might be ejected by photoelectric emission. It might be possible to
synchronously average the fet output over many shots and verify that
the the resulting level shift is quantized corresponding to a multiple
of e.

An ac-carrier-pumped, possibly resonant, varicap diode thing would be
at least worth analyzing. That would solve some of the noise problems,
maybe.

Let us know if you come up with anything.

John

Hasn't anyone thought of the infamous Millikan's Oil Drop
Experiment???

 

[Fred Bloggs]

For a 1pf capacitor storing a single electron, V = e/C = 6.9 x 10(-7) V. So
with careful low noise design, I can't see it being impossible to design a
1pf capacitor and measuring the voltage to determine whether a single
electron is present. Is it worth investigating to see how far I can go?

Thanks.

And how are you going to measure that voltage without removing that
single electron? Take your *nutcase* discussion to sci.phyics- there are
many braindead riffraff there who love to pseudo-intellectualize simple
extremes like this- helps them to get their minds off the realization
that they are otherwise totally useless.

 

[Winfield Hill]

John Larkin wrote...

Well, I did suggest that single-e steps would be lost in the noise.
Demonstration of quantization should be possible, however, with signal
averaging *if* you have a trigger to synchronize the analysis with the
charge steps.

Whoa, that's a very big IF, given that the electrons arrive
randomly and create signals that are *buried* in noise.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Paul Burridge]

Take your *nutcase* discussion to sci.phyics

Yes, SED is reserved solely for the vitriolic mutual exchange of
insults. Any mention of electrons or electronics is OT.

 

[Jim Thompson]

For a 1pf capacitor storing a single electron, V = e/C = 6.9 x 10(-7) V. So
with careful low noise design, I can't see it being impossible to design a
1pf capacitor and measuring the voltage to determine whether a single
electron is present. Is it worth investigating to see how far I can go?

Thanks.

Werner Heisenberg will have his way with you ;-)

...Jim Thompson

 

[Paul Burridge]

this reminds me of something i read in electronics and wirelsss world, where
i think they had a CCD device wich they claimed the wells only had space for
1 electron and so could charge up a capacitor with a known quantity of
electrons and was potentialy a new reference source for voltage.

Only space enough for one electron, eh? A likely story...

 

[John Larkin]

John Larkin wrote...

Whoa, that's a very big IF, given that the electrons arrive
randomly and create signals that are *buried* in noise.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

C'mon, Win, brainstorming is a delicate, easily damped process.

If the background electron leak rate is low, and we can use UV pulses
or something like that to evoke extra electrons at known times, one
could signal average the fet output, using the pulses as the timebase
trigger. So the random leakage guys average out, and the pulsed guys
are quantized so should plot as discrete steps. Since we're not too
far from having a usable signal (0.1 uV step per electron, ballpark)
we might not need an absurd amount of averaging to get this to work. A
good fA opamp might be handy here instead of a discrete fet.

All this sounds quantitatively reasonable to me.

You're right that bootstrapping is silly when the capacitor is
*inside* the fet.

But I think there's a clean, cheap way to clearly demonstrate
non-averaged single-electron leakage using common bench instruments
and a few bucks worth of parts from Digikey.

John

 

[John Larkin]

And how are you going to measure that voltage without removing that
single electron? Take your *nutcase* discussion to sci.phyics- there are
many braindead riffraff there who love to pseudo-intellectualize simple
extremes like this- helps them to get their minds off the realization
that they are otherwise totally useless.

Oh, don't be such a grouch. You're only alive for something like 1e-12
of the expected age of the universe, so why not be happy while you're
here?

Of course you remove the electron. The question is whether leakage
from a capacitive node can be resolved to single-e steps. I think it's
a great question (not the less since I posed it myself here a while
back) that requires some decent math and device knowledge to approach.

I think I have a way to do it... later.

John

 

[colin]

For a 1pf capacitor storing a single electron, V = e/C = 6.9 x 10(-7) V. So
with careful low noise design, I can't see it being impossible to design a
1pf capacitor and measuring the voltage to determine whether a single
electron is present. Is it worth investigating to see how far I can go?

Thanks.

this reminds me of something i read in electronics and wirelsss world, where
i think they had a CCD device wich they claimed the wells only had space for
1 electron and so could charge up a capacitor with a known quantity of
electrons and was potentialy a new reference source for voltage.

Colin =^.^=

 

[John Larkin]

Werner Heisenberg will have his way with you ;-)

...Jim Thompson

RF-SETs (single-electron transistors, basicly tiny fets) can resolve
charge to something like 3e-6 of an electron charge per root Hz
bandwidth, which means they routinely plot single-e current steps with
great s/n. The question is whether we could reliably demonstrate
single steps ar room temp with available parts.

As someone mentioned, Millikan did this very elegantly in 1909.

John

 

[John Woodgate]

I read in sci.electronics.design that Paul Burridge
[email protected]>) about 'Observing single electron flow', on Wed, 14 Jul
2004:

Only space enough for one electron, eh? A likely story...

Electrons have a very low mass. As a consequence (or is it the other way
round?) their wave functions decrease only slowly with distance
(relative to that of the much more massive proton or neutron). Electrons
should be depicted as big fuzzy blobs, not tiny billiard balls - those
are protons and neutrons.

This connection between mass and the wave function indicates that
neutrinos are as big as your butt! (;-)

 

[John Larkin]

Only space enough for one electron, eh? A likely story...

Quantum dot devices have nodes so small that adding a single e changes
the voltage enough that no more electrons can be added. These can be
pulsed to create a stream of *exactly* one electron per pulse,
resulting in a current flow that is *not* random, but entirely
regular. There are people trying to scale the rate up to macroscopic
levels so it can be used as a metrology standard, an absolute
converter of frequency to current. This, with the Josephson junction
for voltage, closes the volts-amps-ohms metrology triangle.

Hmmm, current flow without shot noise.

Quantum dot lasers can dole out light quanta at regular rates, too.

John

 

[Paul Burridge]

Electrons have a very low mass.

You *could* put it that way...

As a consequence (or is it the other way
round?) their wave functions decrease only slowly with distance
(relative to that of the much more massive proton or neutron). Electrons
should be depicted as big fuzzy blobs, not tiny billiard balls - those
are protons and neutrons.

This connection between mass and the wave function indicates that
neutrinos are as big as your butt! (;-)

Hey, my backside passes billions of neutrinos per second (or something
like that) and it's not *that* big. ISTR (from Heisenberg though it's
a long time ago now) that you can't observe a single electron without
interfering with it, so how anyone can be sure they've isolated one
individual particle of this infinitesimal size is beyond me. There
again, some clever-arse reckoned they'd frozen the transmission of
photons a coupla years back, so maybe I need to get fully up-to-date
with this screwball crap. There again, it's not going to help me drink
more beer, so why bother?

 

[Kevin Aylward]

I read in sci.electronics.design that Paul Burridge


Electrons have a very low mass. As a consequence (or is it the other
way round?) their wave functions decrease only slowly with distance
(relative to that of the much more massive proton or neutron).
Electrons should be depicted as big fuzzy blobs,

No they shouldnt. This is sloopy prose that only leads to more
misunderstandings. Experiments put the radious of an electron at <
1e-18M.

The fact that the position of a localised electron may be uncertain
within a region, does *not* imply that an electron physically occupies a
volume of space.

not tiny billiard
balls

All experiments show that they are indeed little tiny buggers.


Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.