Mark Kasevich - Quantum Mechanics at Macroscopic Scales

[music playing]
- Welcome to the NASA Ames 2016 Summer Series.
Two questions:
How does the microscopic world inform the macroscopic world
and vice versa?
What could we learn and how do they interact?
Can we use quantum mechanics
to advance real-world applications?
Today's seminar entitled
"Quantum Mechanics at Macroscopic Scales"
will be given by Dr. Mark Kasevich.
He is a Professor of Physics and Applied Physics
at Stanford University.
He received his Bachelor of Arts
in Physics from Dartmouth,
followed by a Master's in Physics and Philosophy
from Oxford University as a Rhodes Scholar
and then a PhD in Applied Physics from Stanford.
His current research is centered on the development
of quantum sensors of rotation and acceleration
based on cold atoms.
He has published numerous publications
in really relevant journals,
including "Nature" and so on.
And in his spare time,
cofounded AOSense, Incorporated,
a developer and manufacturer
of innovative atom optic sensors
for precision navigation, time and frequency standards,
and gravity measurements,
and currently serves
as the company's consulting chief scientist.
Please join me in welcoming Dr. Mark Kasevich.
[applause]
- Thank you.
Thanks for the introduction.
It's great to be here this morning.
So I'm gonna tell you about experimental work
coming out of my laboratory at Stanford just up the road,
where we're trying to push the boundaries
of quantum mechanics,
to understand to what scales
we can accept that quantum mechanics will be true.
And when I say scales, I'll kind of orient you to the--
to my interest in that
as--as the talk goes on,
but I think it's no secret
that quantum mechanics... probably one of,
arguably, the most successful physical theory we have
that had its birth
in the early 20th century,
has left us puzzled.
And many of us who practice it have both--
by experiment and pencil and paper work,
find points where it just doesn't seem to make sense.
And oftentimes that boundary becomes most puzzling
when you ask, does quantum mechanics manifest
over the distance scales and time scales of everyday life?
Meters and seconds and large numbers of particles.
And so I'm gonna tell you about experiments today
where we're trying to find quantum systems
that we can explore on meter scales
and with large numbers of particles
and so forth.
Our dream would be to explore, you know,
quantum mechanics of humans being in two places at once,
but we're nowhere near that.
But what we have managed to do
is to manipulate atoms in such a way
that they can exhibit
surprising quantum mechanical attributes.
So let me introduce you to the field that I'm in.
I'm an atomic, molecular, and optical physicist,
and back in the early '90s,
there was some incredible work
that came out of the Konstanz atomic physics group.
And what they did, Jurgen Mlynek and his colleagues did,
is they did something
analogous to the Young's double slit experiment
but they used helium atoms.
And the way this experiment worked
is they--they took a beam of helium,
which is at this part of the diagram,
and they-- they took these atoms
and they fired them at a narrow slit.
Now, if you're thinking classically,
the atoms just go through the slit
and the slit kind of defines an atomic beam.
If you're thinking quantum mechanically
and you're a believer in the de Broglie relationship,
and namely that that atom could really be considered a wave
whose wavelength is Planck's constant
divided by its momentum,
then you expect wavelike behavior.
And by design, Mlynek's group made this slit narrow enough
so that the helium atom waves
diffracted into this coherent set of wave fronts
much like you would see a beam of light diffract
as it propagates through a slit in a laboratory experiment
and such that it coherently illuminated
two more slits.
This location and that location.
Well, those two slits
now become sources for diffracted waves,
and those waves overlap
at a detection screen here.
Now, if they were water waves or light waves,
you would have no problem accepting
that those waves would interfere constructively
and destructively
giving you maxima and minima in intensity.
But these are atom waves
and how do the atom waves interfere?
How is that interference manifested?
Well, it's manifested by the constructive
and destructive interference of the atomic wave functions.
And according to the rules of quantum mechanics,
you add the wave functions before you square
and then when you square you get those interference patterns
that you calculate mathematically.
But what it really means,
again following the quantum paradigm,
is that those ripples in the wave function
correspond to ripples in the probability
of detecting the atom at a given position along the screen.
And so as they scanned their detector
along this screen,
they saw the following.
This is the detector position,
and this is the number of atoms that they detected
in five minutes.
They saw this beautiful oscillating pattern,
which is just the kind of picture
that you associate with the Young's double slit experiment.
And when they go back and use the Schrodinger equation
and so forth and calculate the periodicity of the oscillation,
what you would expect from the momentum of the atom
and Planck's constant and the mass of the atom,
turns out it lined up nearly perfectly
with this data,
demonstrating that we had reached a position
or they had reached a position
where they could coherently control
the--the center of mass wave function
of a beam of helium atoms.
Now, this wasn't a surprise to anybody.
I mean, if you believed what was written down
in the early 20th century, of course,
atoms, under certain circumstances, will interfere.
That's the law of quantum mechanics.
Electrons interfere to give us the hydrogen spectrum
and, you know, Davisson-Germer showed us diffraction
from surfaces.
So what's the big deal here?
Well, the big deal was there were some notable physicists
in the, oh, let's say mid-20th century
who said, that's great, you know,
quantum mechanics works just fine for those electrons,
for magnetic spin,
and everything we know about atomic structure,
but ain't nobody ever gonna build
an interferometer for atoms.
Why?
Because, well, every time you build an interferometer,
one of your key engineering challenges
is to stabilize the relative paths
of the two interfering beams.
In other words, if I go back to this slide,
I have to make sure that the phase acquired by the wave
that goes through this slit
and hits the screen there
is stable with respect to the phase
that goes through that slit.
And you could do some simple calculations for atoms
and convince yourself that there's, you know,
maybe you could build the slits,
but there's no way you'd get stable phase.
There'd always be some environmental perturbation
which would shake one wave with respect to the next.
So that pattern wouldn't be stable,
those fringes would shake back and forth,
and as you waited to see the fringe emerge,
you would never see anything,
and that would look like just classical behavior.
And so what happened was
as technology since-- in the kind of mid-20th century
to the early '90s, technology progressed
and we learned, the field learned,
how to control the environmental disturbances
on, in this case, the beam of helium atoms
so that you could have a stable phase.
And this was--this was the paradigmatic experiment
that kind of started our field in the early '90s.
So you could ask, well, technology has progressed
yet again, so where are we?
And I could take you through the entire 25 years,
but I'll take you to a snapshot of our most recent data.
Where are we today?
How--how--how well can we control
the trajectories of atoms, and can we see interference?
And if so, over what types of distance scales
and time scales?
This distance scale here
is measured in millionths of a meter,
ten microns.
So here's some data that came out of my lab
last year.
And what I'm showing you is two ensembles of rubidium atoms.
I'll tell you how we make those atoms in--in a moment.
And these rubidium atoms--
this--this peak here has about 10 to the 5 atoms,
and this peak here has about 10 to the 5 atoms.
and they're separated by a distance
of more than a half a meter.
Basically from here to there, that's--
those atoms are separated.
So it's not a big deal, but what is a big deal
is that I claim each and every atom
that is a member of this peak
also has a partner in that peak.
It's actually the same atom that's been coherently divided
and put in both places at once.
Now, according to laws of quantum mechanics,
when I--when I look at a state like this,
I kill it,
and I collapse the state so that the atom is either
in the left peak or the right peak,
which is what happened when I took this picture.
But before I took that picture,
I claim that I had this amazing state
where each and every one
of those 10 to the 5 atoms was--
the only way I could describe physically,
you know, what it was,
was--was an atom that was in this place and this place
at the same time.
The logical progression would be putting molecules
in two places at once
or human beings separated by a meter scale
and so on.
But an atom is,
I want to remind you, is a pretty complicated object
especially rubidium atom that has, you know,
a grab bag of nucleons
and a bunch of electrons all around it.
It's not--it's not a fundamental particle,
it's a--it's a big collection of fundamental particles.
So obviously, my job now is to convince you
that I've had atoms in two places at once.
This picture alone cannot convince you of that.
I might as well just put 10 to the 5 on the left
and 10 to the 5 on the right.
How do I convince you that they're at two places at once?
I have to do an interference experiment.
I have to bring those clouds back together,
and I have to overlap them
and I have to see them interfere as waves.
And so that's the data I'll show you in the next few slides.
So this is a picture of the apparatus,
and let me just walk you through the apparatus
before I show you some of the data.
So we like to joke in my group
that everything looks better in CAD.
So here's the CAD rendering of the apparatus,
which is in the pit in the basement
of the Varian physics building at Stanford,
and it's about a ten-meter-deep pit.
And what you see here is magnetic shielding
that surrounds an ultrahigh vacuum system
where we've basically pumped everything out
in creating an environment that, if I have a rubidium atom
of the appropriately conditioned state,
it won't interact with or be collided with
other gas elements in the vacuum system.
And what we do is the following.
We--at this location here,
which is at the bottom of that pit,
we used some modern atomic physics techniques
to create a cloud of ultracold,
and I'll explain what ultracold means in a second,
ultracold rubidium atoms, about a million of 'em.
What we do is we take rubidium from a thermal atomic source,
a chunk of metal.
It gets heated up.
You get a stream of atoms
that are moving about 200 meters a second.
You capture them with laser light.
The laser light is configured in such a way
it cools them down to temperatures
of--of millikelvin to microkelvin,
and what we mean by that is that if I look at the kinetic energy
or the velocity of one of those atoms,
I measure it in units of centimeters per second.
The way--this was something, a technique
that was discovered by Steven Chu and others
in the--in the '80s and '90s
for which they received a Nobel Prize.
After we cool the atoms that way,
we further cool them in a-- we--we--they're cold enough now
that I can let them sit in a magnetic field
and that magnetic field serves as a trap,
and in that trap we can do
a further so-called evaporative cooling step
and bring them down to a state of matter
known as a Bose-Einstein condensate.
This was a technique developed by Carl Wieman
and Eric Cornell and Wolfgang Ketterle
in the late '90s, early 2000s,
which also received a Nobel Prize.
And the miracle of this is
after you've done this evaporative cooling,
which works just the same way as when you put your coffee cup
on the counter and the hot stuff jumps out,
leaving your coffee colder,
here we put the atoms in a trap
and the hot atoms jump out, leaving colder atoms behind.
We end up with ensembles of atoms
that have temperatures measured in nanokelvin,
and we further manipulate them so their temperature
is just picokelvins,
at which point their velocity spread is measured in,
you know, units of hundreds of microns per second.
That is, I have an atom, I prepare it this way,
I let go of it, its velocity is defined
to within 100 microns per second.
Slower than an ant will crawl,
which is kind of amazing when you consider that
the atoms in this room are whipping around
at a kilometer per second.
Now, at these low velocities,
per the de Broglie hypothesis or law,
you expect wavelike behavior,
because the velocity is low,
the wavelength, the corresponding wavelength
inversely proportional velocity is long.
And that's the-- getting them cold is key
to observing these interference effects.
So we've cooled them down here.
That's at the bottom of the pit.
And then we have a process whereby we launch them
up this ten-meter tower and they fall back down.
And so that process is-- involves interactions
with beams of light,
and we basically turn on some lasers
in the appropriate way,
and we give the atoms a velocity kick
so that they're moving at about ten meters per second
with respect to the lab frame
and then we let go of them.
They still have that ultracold temperature now,
they're just moving vertically,
and they fly up this tube and fly back down that tube.
It's like if you had a fistful of sand
and you chucked it up.
The sand would go up and would come back down
and--but and the sand has got such a low velocity
that it just kind of doesn't spread out.
These atoms, the size of the clouds when the come back down
to our detection region where we have the cameras
to take pictures of them is, like, hundreds of microns.
Let's see, I've got one more thing to explain
and that is I'm gonna make an interferometer for atoms,
so how do I make it?
Well, that's also done with light and mirrors.
And so we have a beam of-- a bunch of laser beams
that start at this location.
We--we shine them down,
they hit the floor, they bounce back up,
and due to the way we configure the interaction,
and I'll explain this on a subsequent slide,
we precisely manipulate the atomic wave packets.
We coherently divide them in two,
we add momentum to them, we take momentum away,
and we split them up and recombine them
all with pulses of light.
And this is the mirror here that that laser beam reflects off of.
That's the only really special part of this apparatus.
It has to be stable, pointing in the same direction
with respect to the stars.
Inertially it turns out in order to see the interference fringes.
And so what we actively do-- we actually do
is we actively tip the angle of that mirror
so that the beam kind of when it reflects off of it
always points inertially, even though the Earth is--
if the Earth is rotating, a laser beam would rotate with it,
We have to make sure the laser beam,
as the Earth rotates, stays pointed straight,
and so there's a piezo stage there
that makes sure that happens.
how do--this is a little more experimental detail
on how we coherently divide, redirect, and recombine atoms
and make an interferometer,
much as you would use beam splitters and mirrors
and beam splitters to make an interferometer for light.
And so what we do is, as I said,
we shine on sequences of pulses to the atoms.
Here's an example of one of our pulse sequences.
Some of the data, I'll show you, we use sequences
of hundreds of pulses.
And each time the pulses irradiates the atom
in the middle of the sequence here,
what it does is it takes one of the--
the--the wave packets, and I should back up a moment.
This chart here plots the horizontal axis' time,
the vertical axis' position.
And what I'm plotting for you
is the--the center of the wave packet,
the probability distribution,
which tells me where I find the atom.
And the wave packet, of course, has some spatial distribution.
It's about a millimeter wide--
I'm sorry, hundreds of microns wide,
but the center of that wave packet
has a well-defined position in time--at a given time.
And that's what I'm--I'm-- I'm plotting.
And because I'm building an interferometer,
a single atom can--is associated with two wave packets
as it's going through our apparatus.
So here comes the--the-- the cloud of atoms,
and let me just think of one of the atoms in that cloud.
At time T equals zero,
it hits this pulse we know as a pi over 2 pulse,
and what that pulse does is it exchanges momentum
between the photons in the laser beam
and the atoms in this Bose-Einstein condensed
flying cloud in such a way
that it puts the atom in a coherent super position
of its center of mass state
so that the atom has two wave packets,
one which has momentum.
We call this two-photon recoils,
because it absorbed and stimulated--emitted a photon,
and what that amounts to
is about a centimeter per second velocity kick.
And the other part of the wave packet,
which was left behind in this interaction,
which doesn't see that velocity kick.
Well, the atomic physics details are--
you know, I don't want to go into them here,
but it's just standard application
of the Schrodinger equation,
which describes the interaction of a beam of light
with a two-level atom.
And so that's-- that's our beam splitter,
because this part of the atom there
that has a different velocity, if I wait long enough,
it drifts apart from the part
that doesn't have that velocity kick.
Now, we don't have-- we only have so much time
in which we can look at the atoms,
so we want to give 'em a little bit more drift velocity
that just two photon recoils,
and so that's where all these other pulses come in.
Every time I hit the atoms with--
wave packets with one of those pulses,
I add, in the appropriate way,
two photon recoils worth of momentum.
And so if I'm plotting the position as a function of time,
and here I've taken off--
I've subtracted out the overall sag due to gravity.
These atoms are flying up and they're coming down
and I'm saying, forget about the gravitational part of it.
Imagine you're flying with the-- the mean of velocity
as you--as you're going up and coming down.
You just look at the relative velocity of the wave packets.
These wave packets drift apart
and then we-- we bang 'em again
so that they come back together, and at this point, they overlap.
They exit at this point.
This is another beam splitter pulse.
And we look to see interference between those two output paths,
which is experimentally observed
by detecting where the atoms are
at this region of space by just flashing on a beam of light
and taking a picture of 'em.
This is about ten meters.
This splitting here, that's at 54 centimeters.
And now, what I want to show you is
after they've been separated by the 54 centimeters,
when they fall back down here,
hit the final beam splitter,
and then fall into the detection region,
and when I take a picture,
I actually see interference fringes.
So this is our key data,
and what I'm showing you is interference fringes,
which, when I say "fringes,"
what I'm talking about is probability of finding atoms
in one location or the other location
at my detection port,
and those two locations are this location and that location.
And I get excited
when I see that the--the atom's flopping between one location
and the next location.
The only way that's happening is--
the only way that can happen, I claim,
is if I consider the atoms as waves
and if these waves have gone through these two paths
separated by a half a meter.
When they come back together,
if I separate them by just 1.2 centimeters,
I can find relative phases between those waves
where the addition of the wave function is constructive
and the atoms are there,
or destructive and the atoms are there,
and we can--you know, every-- anything in between also
we're capable of detecting,
but I'm just showing you these two extremes.
For fully constructive interference,
all the atoms show up there.
For fully destructive, all the atoms show up there.
Now, it stands to reason, the further I separate them apart,
the harder it is for me to get 'em to all come back together.
So when we separate them by a half a meter,
we don't have what we call perfect contrast.
We have somewhat imperfect contrast where--
but we still see oscillations
between the majority of the atoms being at that position
and the majority of the atoms being at that position.
And for us, this is a smoking gun signature
that these atoms had to be interfering
as quantum mechanical particles.
Now, you might say,
shoot, you know, but quantum mechanics is true,
so what's the big deal about a half meter
versus a centimeter?
And you know, also, what have other people done?
So this data here for massive particles
is--is the world record.
We're about 100 times more spatially separated
than anybody else has-- has--has done
in any other lab in the world.
But you know, so those are kind of nice bragging rights
that you use to maybe get your data into "Nature"
or, you know, some other journal.
But you know, you-- as scientists, we want to know,
is it--was it worth doing an experiment like that?
And so that's--I kind of want to tell you that story
a little bit on the subsequent slides.
You might also be thinking, well, you know,
I also know that light is comprised
of a stream of photons,
and a photon is a quantum mechanical object,
and when I put a photon on the beam splitter
and I separate it and recombine it,
that photon flies over distances which are much larger
than the 54 centimeters.
And I would say, if you're thinking that,
that is definitely, you know, a legitimate and interesting thing
to--to think about, and it's-- it's very interesting,
in my view, to compare, you know,
the quantum mechanics of photons,
you know, massless relativistic particles,
and those of massive particles, like atoms,
and--and how quantum mechanics treats them in,
you know, in terms of--
of what the predictions ought to be.
So what do we learn
when we separate an atom by a half meter?
Well, there's a lot of effort all across the world right now
trying to figure out how macroscopic
we can make a quantum system.
And by macroscopic, how far apart can we separate it
and recombine it?
How massive can the particle be?
You know, I showed you an atom.
People are actually--in Vienna, they're interfering molecules,
fairly substantial molecules,
and one of the experiments they have on their docket
is to interfere a virus, you know,
so you can make something that's maybe alive,
separate it and recombine.
They're doing those big stuff
over much smaller distance scales,
often measured in hundredths of nanometers.
And also, over what time scale they're separated.
There's all these different experiments happening,
and you kind of want to have some way of comparing them all,
you know, and--you know, some notion of how macroscopic,
how big, the quantum state you're making really is.
And so these theorists in Vienna,
Nimmrichter and Hornberger, in 2013
did a really interesting set of theory work
where they created the framework that allows us experimentalists
to kind of compare the macroscopicity
of one kind of quantum experiment with another.
It might be that it's-- it's more interesting
from the point of view of fundamental test of theory
to interfere a molecule over 100 nanometers
than an atom over a half a meter.
And their theory, by the way, only applies to
massive particles, not to photons.
And so taken this way,
I don't really have time to go into the axes on this plot.
My point of showing this is just to kind of show you
people are thinking about this.
And our work, the stuff I just showed you,
is--constrains quantum mechanics in a way
shown by this green line
and what it does for certain parameter ranges on that plot,
which kind of tell you about the separation of the wave packets
and so forth.
We're setting the-- kind of the--the--
the most stringent limits on macroscopicity
by many orders of magnitudes.
So it turns out, it does seem like--
it is interesting to be just exploring the...
The interference patterns of atoms at these distance scales.
Well, you might be asking, well, shoot,
but what--what's the point?
How--you know, what would happen if quantum mechanics was wrong?
You know, how--how might quantum mechanics break down?
A lot of people thinking about that.
There's a lot of kind of crazy theories
that have been proposed for a long time.
I want to tell you about one of them
which resurfaced back just a few months ago
and analyzed our data in the context of this theory.
When you go--when we-- when we do physics,
when we do engineering, we make fundamental assumptions
about space and time.
Homogeneous, isotropic, time progresses uniformly.
but when you look at, you know, a fine-grain scale,
Planck scale, 10 to the minus 34 meters,
you start to think, well, maybe space really isn't homogeneous,
maybe it's grainy.
Maybe time isn't continuous.
Maybe it ticks in a funny way on certain distance scales.
And maybe quantum mechanics of massively separated objects
might be sensitive to these kind of perturbations.
It turns out we have to have extreme control
over the paths of the atoms to get them to come back together
and interfere constructively.
And if the velocity of one of those wave packets changes
by nanometers per second,
that's enough to kind of wipe out the coherence.
And so we make this experiment by the fact that
we see those fringes make very good tests
of anything that's gonna go in there
and shake around the atoms as they're propagating.
And it may be very--the very nature of space-time itself
that's doing this.
This Ellis model imagines a gas of wormholes
that are flying through space,
and this was taken seriously until the early '90s
when some people shot this theory down.
And these wormholes, according to this phenomenology,
collide with the atoms and give small momentum kicks
and ruin the interference pattern.
And so there's a bunch of complicated math these guys did,
and basically, you know, showed that the data
I just showed you dramatically constrains
these sort of wormhole theories.
And to put these theories in context,
there's been a lot of thought about the quantum mechanics
of black holes and things recently
and over the past 20 years,
and this was one of the early types of forays
into quantum and gravity
by people like Lenny Susskind
when they were tying to figure out,
and they still are trying to figure out,
the quantum mechanics of black holes.
I want to change gears a little bit.
Oftentimes when you build an interferometer,
you build it to measure something,
not just to learn something about,
you know, the structure of space and time,
and that's kind of the reason why
we're building these apparatus.
By looking at the interference fringes,
and assuming now the interference fringe
is something stable,
we can learn something about the relative paths
that the two wave packets take.
So what do we learn?
Well, the theory for calculating
the relative phase shifts
between atom wave packets
as they propagate through space and time is well-defined
and it's been articulated since the beginning of this field,
and it just involves, again,
systematic application of the Schrodinger equation.
And again, I'm not here to explain these equations to you,
just to say that those equations exist
and you can go in and actually calculate
what the relative phase shifts for the two paths are.
So you might ask me, well, what drives the relative phase
of those two paths?
Well, anything that changes the velocity of the two paths
is gonna lead to a phase shift
because of the de Broglie relationship.
What changes the velocity?
Well, how about the acceleration due to gravity?
That's something that, as those atoms are flying through
the apparatus, their velocity is changing with time.
Their--their--their phase,
the wavelength is changing with time.
That turns out to lead to a huge phase shift
at the output of the interferometer.
If the interferometer's rotating,
there's another phase shift associated with that,
the Coriolis effect.
And there's a whole list of phase shifts
that you can calculate along these lines.
Bottom line is these-- these devices
are extremely sensitive to inertial forces,
rotations and accelerations.
And that's why we-- we use them--
we build sensors with them.
I'll show you a sensor in just a minute.
This is one of those sensors, which is to say that--
and I want to maybe go to this list first.
When I do that calculation,
you say, well, what do--what do those phase shifts look like?
We--we--we call a table like this
affectionately the term list.
And what I'm-- what I'm describing here
is all the different ways that I can get relative phase
at the output of the interferometer.
And for the highly engineered sensors we build,
these term lists have hundreds of terms,
and I just put in a simplified one here
to show you some of the most dramatic shifts.
This "k" is the propagation vector of the laser beam
that we used to bat around the atoms.
"g" here is the acceleration due to gravity,
and "T" is the time between--
basically the time of flight
between the first beam splitter and the exit beam splitter.
And I'll just focus on this top term
and point out something amazing
that in that data I just showed you,
as the atoms are flying through the interferometer,
about 10 to the 10 radians of phase evolve
between one path and the other
before the wave functions overlap.
So if I measure the phase of the wave function precisely,
I'm basically, and I can count which fringe I'm on,
I basically have an awesome accelerometer,
an awesome way of measuring accelerations,
or if I'm on the Earth, the gravitational field.
Let me just put an order of magnitude in there,
10 to the 10 radians,
and typically we can, on a single experiment,
resolve a phase shift to about a 10 to the minus 3
to 10 to the minus 2 radians.
It says in a single shot
we are, in principle, capable of revolve--
resolving accelerations at the 10 to the minus 12
of little "g" levels.
10 to the minus 12 to 10 to the minus 13
in just a single realization.
How--and that's--that's--I mean, okay, that's a small number.
How small is it?
Well, the gravitational acceleration
of two of you guys sitting next to each other
is 10 to the minus 9 of little "g."
So it means if somebody is drinking a cup of coffee
next to you,
you know, their mass is changed by enough
that the gravitational interaction would be about
10 to the minus 11 of little "g" or something like that.
That still--and that would be detectible
with a sensor like this.
So it kind of makes you want to build these sensors,
maybe not on that grandiose scale.
And so this company, AOSense,
which is literally just down the road,
is building smaller versions of these sensors
for real-world guidance navigation
and control applications,
and also geodetic applications,
studying of the Earth's acceleration
due to gravity and so forth.
And in this can here
is a much smaller atom interferometer
that's measuring the acceleration due to gravity,
and this is some data from that sensor.
This is time in kiloseconds,
and one of these oscillations is about--is about a day.
And these variations in gravity are the well-known variations
in acceleration due to gravity
due to the motion of the Earth and the Moon.
This is about 10 to the minus 7
of the 9.8 meters per second squared of gravity.
This is to point out that,
yeah, these are really sensitive sensors
and when you build them on these grandiose scales,
they become extra sensitive.
So what are we gonna use that extra sensitivity for?
And this is data that's just started rolling in
in the past month.
Well, one of the-- I'm in a physics department
so what basic science might we do with this?
People are interested in answering at a precise level
the age-old question,
does the brick and the feather,
do they--do they fall at the same rate?
The Leaning Tower of Pisa experiment,
the Galileo experiment, in modern times,
you--it's known as equivalence principle measurement
and it's--the equivalence principle is a foundation
for Einstein's theory of general relativity.
And it's theoretically interesting,
I'm told by my theory colleagues,
to probe this principle
at the part in 10 to the 15th
to part in 10 to the 18 level.
Which is to say as the objects are accelerating together
at basically the acceleration due to gravity,
there may be some spurious interaction
due to particles we know nothing about
that make it so a rubidium 85 atom
accelerates at a slightly different rate
than a rubidium 87 atom
because they have different nuclear composition.
And so that's the experiment we're doing here.
We have two ensembles of Bose-Einstein condensed
laser-cooled atoms that have been launched
and subjected to a bunch of pulses
that are flying up and down this tube,
both opening up and closing interferometers,
and then we measure simultaneously
the phase shift from rubidium 85 and 87.
By comparing the phase shifts,
we make a comparative measurement
of their acceleration due to gravity,
and then we seek to understand
whether the observed acceleration due to gravity
is the same for one isotope or the other.
Now, I don't have results to...
You know, formally discuss,
but I do have some preliminary data
to show you kind of what this--
these action shots look like now
when we're making a simultaneous measurement
of the co-acceleration of--of 87 and 85.
This is a picture in false color--
as my colleagues joke,
everything looks better in false color.
So in false color,
rubidium 87 in the upper part of the camera port,
rubidium 85 in the lower part.
This is one output port of the 87 interferometer.
This is the other output port same for 85.
And we introduce a phase-- what we call a phase shear
across the interfering waves
so we can precisely measure the phase.
This is like what happens if you misalign
an optical max into interferometer,
you see fringes.
And basically, by comparing the phase of--
of these--this is a lot of atoms,
and this is a few atoms.
By comparing the phase of those peaks
to the phases of those peaks,
we can measure their relative phase
and then get a handle on the equivalence principle.
The data I'm showing you here right now,
we're--is good to about 12 digits,
and we think that when we get this apparatus all tuned up,
we're gonna have 14 digits of sensitivity or more.
And at that level, our theoretical colleagues
start getting interested in these results.
Thanks.
Let's see, so what's--
You know, where do really interesting things happen?
Well, people think the really interesting stuff
is gonna be at 16 digits.
The world record for equivalence principle right now is in
the University of Washington gravity group,
where they measure equivalence principle
using torsion pendulum to 13 digits,
and they're pushing into the 14-digit level.
And so we're hoping to at least have a measurement
that's competitive with what they're doing
and complementary in the sense that we're using,
you know, this completely quantum mechanical method
for making the measurement.
What else does equivalence principle measurement tell you?
Well, you know, what's the impact
of EP being true or false?
And so here, there's a lot of recent theoretical work
motivated by the fact-- by cosmology
of, basically, there's dark matter out there
and we don't know what it is
and certain flavors of dark matter
lead to equivalence principle violating interactions
that we might see in our lab.
And so this is a chart from one of my colleagues,
Peter Graham, at Stanford, that shows the possible impact
of the measurements we're doing right now
on some of his favorite
equivalence principle violating theories.
And again, I'm not gonna explain the axes,
just to say that this yellow region
is what we know to be excluded right now,
and the kind of data we're taking is--is--
is gonna further push the bounds
on the possible validity or lack of validity
of these obscure theories about dark matter.
I want to change gears to some more wholesome physics.
So... [laughs]
You probably all have seen, in one form or another,
this data, which is the--
the unbelievable data
that was taken by the LIGO Collaboration when--
for their first observation of gravitational radiation.
What they did was they built two conventional
optical interferometers at two regions in the United States.
And the idea is that if...
Black holes merge, spin and merge,
at way far away,
they give rise to perturbations, ripples in space-time,
which propagate at the speed of light
through space-time to us here on planet Earth
and these space-time ripples basically have
the--the effect of stretching and contracting space
as a function of time in such a way
that if you build one of these precision interferometers,
it can be observed.
And so just to take you-- if you haven't seen this data,
take you through some of their signals.
This is the output of one of the interferometers
and this is the output of the other interferometer,
you know, that one there and that one there.
And when you see those fringes, when you see that shaking,
that means that something came and changed the path length
of one of the interferometer arms with respect to the other,
that something being a gravitational wave,
and in such a way that you can observe
the cataclysmic event,
which in this case was way far away,
two black holes coalescing.
And so I just explained
that I'm building interferometers,
and the interferometers I'm building
I think are gonna be very sensitive.
And so you might ask, you--
could we see gravitational radiation
in a way that's similar to LIGO
but now using atom interference?
And if we could, what--
why would you want to build an instrument like that?
And so I'll spend just a few minutes
kind of describing our thinking on that.
We haven't built anything that's nearly as sensitive
as this LIGO instrument,
but we think that a space-based instrument holds promise
for observing certain types of gravitational wave sources.
And here my philosophy is the following:
this was the breakthrough discovery
that taught us that, you know, observing--the physics community
and the astronomy community and the astrophysicist community,
that observing gravitational waves
is really interesting.
And, you know, you go back in time, you say,
well, we know that, you know, using telescopes
to look at the cosmos is also interesting.
And right now, when you do optical astronomy,
there are lots of different telescopes that you build
depending on what you want to look at.
So as gravitational wave astronomy progresses as a field,
I think it stands to reason that there are gonna be
a diversity of telescopes that you're gonna bring to the table
for making interesting--
scientifically interesting observation.
So here's a--here's a plot
generated by Sesana and colleagues
that basically traces the evolution
of the--the black hole merger that I just showed you
through time.
And what's happening in the LIGO signal
is that the LIGO antenna only captures
a transient signal for the--
the very final moments of the merger.
When those black holes are spinning really quick
and then merge,
that's when you see blips in the interferometer.
Well, in this...
Situation, those black holes
are spinning slowly
before they hit this cataclysmic event here
for a long period of time.
And it's interesting to go and look at these black holes
in this region of parameter space
before they finally hit their merger situation.
How might you build an antenna
that is capable of detecting
these low frequency perturbations?
Well, the world has been thinking a lot about that,
and that's where I think our atom interferometer detector
may have something to say.
When I say the world's been thinking a lot about that,
there's a planned ESA mission,
and NASA may--
is looking like it will participate in that,
that is designed to detect
very low frequency gravitational waves.
This is frequency and this is the--
this axis here is the amplitude of the gravitational wave.
And you notice that this antenna, this telescope,
can see the very low frequencies.
LIGO can see the very high frequencies,
but maybe you want to see these intermediate frequencies.
So this is the antenna we propose to build with atoms.
And I'm showing you kind of a theory view
of something that we published in this paper
a number of years ago.
This is the space-time diagram
for the positions of wave packets
at one region in space
and another region of space.
This is-- this region of space now
I want to be separated by about a gigameter,
10 to the 9 meters,
such that if a gravitational wave
comes rolling through this-- this intervening space,
that will lead to a relative phase shift
of this interferometer
with respect to that interferometer,
that I can observe very much like
the differential phase shift that we're observing right now
in our equivalence principle work.
The only difference is these interferometers now
are separated by a long distance
and I have to correlate the phases across that baseline.
It turns out this is a pretty good way
of detecting a gravitational wave.
If you're building instruments,
what we propose--are essentially proposing doing is
there's this LISA instrument, which is very good
for the ultimate low frequencies.
We're proposing replacing a macroscopic proof mass,
which sits inside the satellite
and which has recently been verified
by the LISA Pathfinder collaboration,
spectacularly verified, with a cloud of atoms.
And due to the atomic physics processing,
it turns out we can build this antenna
with just two satellites.
The LISA configuration requires three satellites,
and from a system engineering perspective,
we think this may be favorable.
And if you want to think more deeply about this,
it turns out the atoms are serving as
precision proof masses and position references
and the lasers that we use to interrogate the atoms
are providing a really excellent ruler
by which we're--we're measuring the time evolution
of the two distances
as the gravitational wave comes through.
And if you say, well, what--
how good is this telescope?
The telescope is--I characterize it by its frequency
and its strain response, how big of a wave it can measure.
This is what the LISA antenna can do,
and this is what we think the antennas that we are envisioning
can do.
And if you want to take that a step further,
there's some new ideas we have out there
where the LISA strain curve is sitting up here,
and what we want to do with atoms
is a couple of orders of magnitude better.
So this is not--this is probably
the generation of telescopes--
you know, two generations away.
LISA has to be built first,
but eventually, I hope that we'll be starting to build
these gravitational wave detectors
with these atom sensors.
More practical applications?
Well, I'm measuring--
when I build a gravitational wave detector,
I'm measuring perturbations in space-time
due to gravity very precisely.
In that case, due to a gravitational wave.
If I have a satellite in low-Earth orbit,
I can do something more mundane.
I can look at the perturbations in the relative positions
between two clouds of atoms
this time separated by meters, not 10 to the 9 meters.
In the same way, by building two interferometers
and subtracting, and it turns out,
that makes a very good
differential acceleration sensor,
which I can use to characterize the Earth's gravitational field
and perturbations in this gravitational field.
And this is considered interesting
because the perturbations in the gravitational field,
as observed from orbit,
tell you a lot about the mass distribution on Earth.
And we're interested in the mass distribution
because some of that mass is water,
and as we know, due to climate change,
that water's moving around.
And so ice is melting in one place,
it's going someplace else,
and how can we figure out where it's going?
There's a guy at Goddard,
and there's some other people at JPL looking at this.
This is-- this data here is from--
This analysis is from Scott Luthke's group
at Goddard of where they're actually looking at
the gravity gradient signals of--of water table.
And this is one of the maps they generate,
and, you know, they kind of show you the gravity contrast
from--from water as it's moving around.
And so we're, with NASA Goddard,
are building an instrument which would be a prototype
for one of these-- thanks--
one of these space-based gravity gradiometers.
So I have-- I have five minutes left,
and I want to, just in this last five minutes,
change gears a little bit.
I started the talk by telling you about a--
you know, a story about macroscopicity
and quantum mechanics in terms of distance separation.
I want to finish the talk by telling you
a different type of macroscopicity.
That is, how can we make states
and manipulate states of-- that are quantum mechanical
but contain large ensembles of atoms?
And the state I'm gonna tell you about
is one where we have collections of atoms,
thousands of atoms all glued together
in a fundamentally quantum mechanical way
that--so-called entangled states--
that are doing something very useful for us
in the context of interferometry.
And so my intro to this last subject is
let's talk about noise for a little bit.
When--when we go and build our interferometer
and look at an interference fringe,
we don't measure a perfect phase.
We always get a little noise on top of that fringe,
and I illustrate that schematically here.
If I'm building an interferometer,
I don't care what interferometer it is,
I scan the phase,
I'm detecting some number of particles in an exit port,
and there's always some noise at a particular phase.
Here, I've frozen this-- this cartoon
at this particular phase value
and plotted a distribution
of the number of particles that I detect,
photons or atoms,
at an output port, and it fluctuates.
It fundamentally has to fluctuate
because what I'm doing is I'm collapsing wave packets
when those two interfering photons or atoms
come back together,
and it's a fundamentally statistical process.
Now, that collapse happens often in a way
where the statistics between one particle and the next
are completely uncorrelated,
and it gives us what's called shot noise.
And the--the-- the well-known theory
which predicts the amount of noise you expect to see
for the number of particles you have
coursing through your interferometer.
And to make a long story short,
using kind of laser and atomic physics techniques,
we are now manipulating that noise
in--in a fundamentally quantum mechanical way
to reduce the noise at the output port
of the interferometer.
And just to show you that in data,
this is the output port of an interferometer
which was tuned to operate exactly at the mid-phase point
that I showed you on the previous slide.
And if I use just regular atoms,
not fancily entangled or correlated with some tricks
I don't have time to tell you about,
if I measure the fluctuations, I get a distribution
shown here in blue.
And those fluctuations
for virtually every interferometric sensor
that's built today, those shot-noise fluctuations
fundamentally limit the sensitivity of that instrument.
Now, using quantum entanglement tricks,
we now have manipulated this-- this ensemble of atoms
so that those fluctuations are ten times narrower.
And it stands to reason, if you're associating with
the mean position of that distribution of phase
and you care about measuring the phase,
having a narrower distribution
really helps you in your precision measurement.
And we demonstrate that in this data
by applying a tiny phase shift to this interferometer
that shifts this distribution
from an output of that number of atoms
to this number of atoms.
This is the number of atoms at the output port,
and this is the probability of observing that number of atoms.
This entanglement is really helping us.
Now, this entanglement, this--this correlation
of creating these correlated states of atoms,
which again we do with lights and mirrors,
is doing something amazing
to a cloud of otherwise 10 to the 5 atoms,
which are noninteracting.
What it's doing is they're making them fundamentally linked
in a quantum mechanical way so that when I make a measurement,
I--if I--if I measure an atom here,
somehow that affects the wave function
of an atom over there in--in--in such a way
that when I measure them all, I can't--
they're correlated in such a way that I can't have fluctuations
bigger than what we observe there.
I've been using this word "entanglement" a lot.
This is--I'll skip through this.
We do this with a cavity.
That's what the apparatus looks like.
We use this word entanglement a lot.
What do I mean by that?
Well, I mean when I try and write the wave function
for all those atoms,
and I--if you've had that quantum mechanics course,
you write--one particle you write its wave function down,
and then in the Hilbert space you write the wave function
of the next atom down and the next atom down.
If they're uncorrelated straightforward,
you write the wave function
for all those 10 to the 5 atoms down.
Once you do this, we call it a squeezing operation.
We're squeezing out the noise.
You have trouble writing down that wave function in such a way
that you can separate out the contributions from each atom
and factorize each atom independently.
In fact, for this data, this analysis shown in this plot,
says that the wave functions for these atoms
are in--are correlated in such a way
that I can't think of the atoms in smaller groups than 1,000
independently.
They're--they're--they're correlated in clusters
of at least 1,000 atoms.
This is like a world record
for quantumness of ensembles of atoms.
And what's--what's interesting,
if you've read about quantum computers and things like that,
it's this very entanglement property
which is thought to be-- gonna be useful
for speeding up computation.
Well, for building sensors, it's also useful.
And if by reliably entangling the particles,
I can dramatically reduce the sensor noise,
making a better sensor and actually making
these fancy ideas of quantum mechanics useful.
And so this work is demonstrating macroscopicity
in a different way.
I now have 1,000-atom quantum systems
truly behaving as quantum.
At that, I have to thank the people in my research group
and my theory collaborations-- collaborators,
and you guys for your attention.
[applause]
- Thank you very much. Thank you.
So we have time for a few questions.
If you have a question, please raise your hand,
wait for the microphone, and ask one question only.
Thank you.
This, up front here.
- Thank you.
Thank you for interesting talk.
You mentioned this very sensitive accelerometer
based on atomic interferometry,
but they all required this cryogenic,
very low temperature sort of operation, right?
So you cannot make them very small.
They should be heavy. They should be bulky.
So how can you put something that big
to satellites or...
- Yeah, so when we say ultracold,
the only cryogen I'm using is laser light.
So I need a single-frequency laser
which is, you know, smaller than my--
the semiconductor laser flash
is smaller than my thumbnail, and it's the interaction
of the laser light with the atom cloud
that reduces their temperature.
And so the smallest sensors now
are approaching the size of my fist.
I showed you that-- that gravimeter,
which is about the size of a two-liter Coke bottle.
And as we, you know,
kind of start engineering these
and use--replace kind of bulk optics
with integrated photonics, which we're now manufacturing
with some partners in the valley,
and state-of-the-art semiconductor laser sources,
the size of these sensors is being crushed.
They're really becoming small.
In fact, there are some--
AOSense and my interest aren't in making them tiny, tiny, tiny,
like competing with MEMS sensors.
We want them to be big enough to be high performance.
But we have gyroscopes now that are about--about this big
that have performance figures of merit
that are really pretty extraordinary.
So it's all about, you know, engineering with optics,
and the--
that's--that's the main technical challenge.
There's a little bit of new engineering knowledge
that we have to gain,
figuring out how to make reliable sources of atoms
and all the rest.
Much of that draws on some work from the atomic clock community
in the--in the '50s.
Another way of answering your question is,
you're comfortable with going out and buying an optical--
I'm sorry, an atomic clock from,
used to be Hewlett-Packard
and now it's Microsemi, that's rack mounted.
That, you know, and has a beam of atoms in it
and it's interrogated,
and kind of what we're doing is very similar.
But we do not need liquid helium.
[laughs]
- If you can entangle 1,000 atoms,
any chance of using this
in quantum computing in the near future?
- For what?
- Quantum computing. Quantum computer.
- Oh, that's an interesting question.
So what are the links between--
the field I just discussed
is receiving a great deal of interest right now,
it's called quantum metrology,
and that's using entanglement to improve performance of sensors,
but we all sort of also want to build a quantum computer.
And so what's the link between this kind of entanglement
and the entanglement you need for quantum computation?
Well, I view it as a baby step towards a quantum computer.
So a quantum computer, you have to go in there
and reliably entangle determine--
almost deterministically, I should say,
the way functions of-- of each and every atom
in a way that you-- you sort of understand.
Here I have this grab bag of atoms
that are sort of magically being entangled
but in a way that I have even trouble writing down
what the wave function is.
So there's a big step between
what I've just demonstrated to you and--demonstrated
and the type of entanglement you need for a quantum computer.
Nonetheless, there are--
if you follow this field, there's a lot of discussion
about what makes a good qubit for a quantum computer,
how you're gonna error-correct that qubit and so forth.
And some of these cavity atom ideas
are interesting players in that discussion,
but it gets pretty technical and, you know,
you have to talk about the gate fidelities and things like that,
and it boils down to a bunch of technical details
that maybe aren't, you know,
they're surprising on first blush.
So part of my program is to--to--to figure out
what role this type of entanglement might have
in quantum computation, and I say that.
I'm much more optimistic about other systems than that
for quantum computation.
Thank you.
- Okay, so please join me
in thanking Dr. Kasevich for an excellent talk.
Thank you. - Thank you.
[applause]
[musical tones] [electronic sounds of data]

Mark Kasevich - Quantum Mechanics at Macroscopic Scales

Latest Posts