Hang tight while we fetch the video data and transcripts. This only takes a moment.
Connecting to YouTube player…
Fetching transcript data…
We’ll display the transcript, summary, and all view options as soon as everything loads.
Next steps
Loading transcript tools…
Professor Paola Cappellaro, “Design and control of quantum devices”, Caltech, May 30, 2024 | Farbod khoshnoud | YouTubeToText
YouTube Transcript: Professor Paola Cappellaro, “Design and control of quantum devices”, Caltech, May 30, 2024
Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.
Share:
Video Transcript
Video Summary
Summary
Core Theme
This presentation explores the concept of "Quantum Sensing," highlighting how exploiting quantum properties like superposition and entanglement in quantum systems can lead to sensors with significantly enhanced sensitivity beyond classical limits. It also discusses practical approaches to achieving these quantum advantages, even with limited entanglement, for near-term applications.
it's my great honor and pleasure to
introduce Professor Pao cap caparo from
MIT and Professor cap caparo is the
fourth professor of engineering at MIT
she is a professor of nuclear science
and engineering a professor of physics
and a member of research lab for
electronics where she leads the quantum
Engineering Group Professor caparo
received her PhD from MIT and joined
Harvard University as a postdoctoral
associate in The Institute for
theoretical Atomic molecular and Optical
physics before joining back before going
back to MIT as a faculty okay please
join me to welcome
Professor thank you and and frma you
should stop your sharing otherwise I
cannot start mine thank
okay can you hear me and uh see my
slides yes yes thank you
perfect uh thank you very much for the
invitation to present and uh also for
the uh very nice uh scheduling which had
me just after fre my presentation
because I'm sort of hopefully continuing
on the same topics and uh uh with
somehow a complimentary um view if you
want uh in which I'm I'm really trying
uh to sort of tease out uh what is
quantum in in Quantum sensing if you
want and uh why we want to do Quantum
sensing and what type of Advantage we
can have over a classical sensor uh when
we are exploiting quantum properties and
so the the first message that I really
want to convey is why are we interested
in using Quantum system uh for sensing
and this sort of goes back to uh the
very old thought experiment of shinger
the the famous shinger cut uh that can
be um a awake or or sleeping inside the
box and we don't know uh what it's doing
until we open the box but really uh when
we look at a real experiment well uh we
cannot really a superposition of a
Quantum cut or of if you want a
microscopic cut and that's is because
the cut even inside the box is still
interacting with its environment any
molecule of air any photons Inside the
Box will perur it and immediately
collapse it into other being awake or or
or sleeping and this is because Quantum
system are fragile because they interact
very strongly uh with uh the external
perturbation and this is probably bad if
you're trying to make a quantum computer
because you would like to keep this a
very large superp position but they also
make for very good sensor there a high
sensitive it uh to um any external field
and external perturbation makes Quant
system very good Quantum
sensor and uh of course the uh the goal
that we have is to reach sensitivity
which can be beyond what is possible
with classical sensor and this is a form
of quantum Advantage which sort of
mimics the advantage that you have in
quantum computer so in quantum computer
you have an exponential
Advantage for typical Quantum sensor I
mean there are many flavoring but the
most typical um scaling that we have is
that if you use quantum s systems as a
classical sensor or we just have n
classical sensor the signal to noise
ratio improves as a square root of the
number of sensor themselves however if
we use fanton sensor by exploiting there
a most fundamental property in
particular entanglement we can have a
skin in with the number of sensor which
is proportional to n and so this gives
this if you want polinomial improvement
with respect to classical systems uh
which is what gives the quantum
Advantage however reaching this Quantum
Advantage is uh relatively uh difficult
and so um when we were working actually
together with fredman and also Christian
Gan in trying to analyze and and provide
a review of what we mean by Quantum
sensing we came up with if you want
three classes of quantum sensors uh so
the first one is where we just use a
Quantum object uh uh to sense thing so
if we considering since we were just
discussing few minutes ago uh MRI of
course we're using spins in order to
detect for example the anatomy of brain
and Spins are intrinsically a Quantum
object but we can describe them really
as classical magnetic dipole so we don't
really use the all uh uh if you want
quantum mechanics to describe a very
Basics MRI experiment so we are using
Quantum object but not really exploiting
their Quantum property and this is a
sort of if you want a Quantum 1.0 type
of technology which we're not really uh
so interested in defining at least not
not today for me so what we really
interested or what we really mean when
we Define a Quantum sensor are cases in
which we are using for example at least
some Quantum property for example
coherence superposition in order to have
uh better
sensitivity and uh even more what we
would like to do as I said is to use
entanglement in order to achieve this
Quantum advantage and of course there
has been already some uh demonstration
of this type of sensors so we heard a
lot about V EnV Center so if you use a
single MV or if you use an ensemble but
without using if you want correlation
among the spins this will fall into this
category another good example are atomic
clock so atomic clock are sort of like
in between sometime people complain to
me that they should be in the force
category uh but really we had a lot of
advances for example brought by Juni and
an Maria re in which you're using
Quantum logic to improve your ions at
atomic clocks you're using Ensemble and
correlation to make them even better so
it's sort of like bridging that Gap to
achieve these really exploiting Quantum
properties and uh even ligo in its first
ination was if you want in the second
category like just interferometer but
more recently they've been using a
squeeze light injecting squeeze light in
order to improve the detection of
gravitational W so we already have these
um examples of achieving Quantum
Advantage with this type of systems and
really this is the type things that we
are also trying to do uh to improve
Quantum sensing really to bring the
advantages of quantum sensing uh to real
applications um same as fredman my
favorite Quantum sensor is the EnV
Center in diamond and I think that like
after restock I don't need to really
motivate why uh this system is awesome
uh it can really sort of bridge a lot of
different um modality of sensing and
different applications so uh looking for
example at magnetism and the Nanos scale
looking at effects from Super conductors
effect from ferromagnetic materials uh
but with a spal resolution which are
really down to the nanoc scale um for
example through Vis scanning tip it can
look at biological effects uh for
example some work done uh to detect uh
HIV RNA uh in um like a a bios uh type
of setup and also you can look at just
the Dynamics of other Quantum system for
example correlation of quantum spin
Associated either with the diamond or on
the surface or the Diel so it's really a
very powerful sensor with many different
applications and uh well my manx slide
probably it's a bit superflu since we
already had an introduction of NV Center
much better than mine and by now I keep
it really down to just one slide uh
again these are just optically active
defect uh so we will illuminate with
green light and we see uh the red light
emitted and uh to put in a plug for
tlaps which was the two talks ago they
actually s sell a Quantum diamonds with
Envy defect so you can also purchase
that and start doing uh your own
experiment and of course uh we combine
typically with Optical uh polarization
so if you want cool down by Optical
means and um I used the conventional
Optical rout for MV centers and we
combine that with
microwave control uh where of course
your Envy Center acts a little bit like
the compass needle uh pointing up or
down in response to an external magnetic
field and you can have magnetic resonant
driving OD MV and where you detect a lot
of light if it's pointing up and uh less
light when it's pointing down and you
can drop this very nice coherent
superposition and this is work going
back to a 2006 B children but people
have been uh observing uh even better
coherence time since that time so how
can we achieve this type of quantum
Advantage with this type of defects one
way uh that indeed we propose even back
then uh in some of the first few
proposal is to use large ensembles of NV
centers and trying to exploit it
interaction among them however the force
me that appears when you're having large
ensembles with very very dense and
Venter is that the interaction among the
spin V cells uh sort of make your
coherence time decrease and so you have
less time to do the detection and so the
sensitivity so if you want the minimum
field that you can detect uh is not so
good and you cannot go for very long
sense in time because the coherence
kicks in and you lose your signal so
what you can try to do is to try to uh
if you wanted to couple or cancel out
this interaction and uh some uh very uh
clever people in Misha looking group
they uh devise very powerful sequences
which were able to improve dramatically
how well you can detect your magnetic
field with very high uh dense system
however this does not give you yet um if
you want um the quantum Advantage per se
so uh if you were able to change the
number of mvs uh inside the sample uh
which uh like in this experiment we were
not doing uh you would see that the
sensitivity would just improve as the
one with the square root of the number
of sensor at
most however if we could indeed exploit
instead interaction among the spin and
achieve a type of correlation for
example in the form of a a squeeze state
of the spin then the sensitivity could
improve even
more um we proposed that back in 2009
and we have not really uh been able to
demonstrate an advantage we have been
able to demonstrate a small Advantage
based on some small entanglement but
this powerful scaling with this uh which
improves over the classical level as a
square root of n has not been able to
demonstrate yet so um of course uh I've
been trying to do something else uh in
between and the idea is that well maybe
there is something in between these two
uh limits so we might not just use um
coherence of single uh Spin and we might
not look for the full entanglement uh
enance Metrology but there is maybe some
regime in between The Chew where we use
some small um Quantum system uh some
small amount of entanglement to achieve
some task that a single Cubit sensor a
single Quantum sensor could not achieve
and this is like a very fruitful regime
which sort of uh is the intersection
where engineering um advances come
together with the fundamental quantum
mechanical advances to bring um some
real practical application and this is
what I want to talk in in the rest of my
talk and there are many different things
that we have looked uh over the years
for example one can use some additional
um Cubit sensor to do Quantum eror
correction or to use VMS Quantum memory
in order to both improve the sensing
time and so the coherence time or to
look at correlation what I'm going to
focus today is how to use some
additional uh control and some
additional uh level other some virtual
degrees of freedom or some real degrees
of freedom associated with other spin in
the system in order to achieve frequency
up conversion in order to achieve Vector
sensing and also to enhance
coherency so the first thing that I want
to talk is this vectorial sensing uh
which is achieved uh by doing some modulated
modulated
driving so um before that I want just to
step uh back and sort of describe some
very simple way in which you could do uh
vectorial um sensing so first of all you
could do the typical Ramsey magnetometry
to detect field which are in the Z
direction if you want aligned with the
NV axis with the MV quantization
axis the idea here is to prepare
superposition state of your MV Center
let it evolve for some time under an
external magnetic field which is aligned
in the Z Direction and then detect out
the phase which has been
accumulated uh and this will tell you
how strong of a magnetic field uh it was there
there
if you want to instead detect a
transverse field if you do this you get
some perturbation but it's very small so
it's very insensitive way of detecting
transverse field so what you can instead
do is to detect field which are rapidly
oscillating at the resonance frequency
of your uh spin of your MV Center and
this will uh indeed induce a a pression
uh of the spin now in the um Z let's say
zy plane but this can only detect things
which are uh oscillating at the
resonance frequency or the MV Center
itself so these are gigahertz um for uh
the Z field you could uh indeed also
measure oscillating field okay for
example by using dynamical decoupling
sequences but the typical frequency here
on the order of megaherz while here on
the order of gigahertz so um it's
difficult to bridge between uh these two
uh modalities and also the experiments
are quite different so if you have
system I if you have experimental error
they will be different in the two
experiments and so it's a bit difficult
to really get out some very nicely
consistent way of of detecting SP so how
can we uh Bridge uh these two and find
the single protocol to do indeed Vector
magnetometry so before I explain that I
just want to take a little bit in aide
so if you do want to do this rabi
magnetometry how can we understand that
well we could have this oscillating
field as I said like a transverse field
that we might want to measure or use to
control the spin if you have this
oscillating field you can always
consider it as the Su or two counterpropagating
counterpropagating
field and then if you jump into the
frame rotating with the red uh field uh
this will become of course static in
that rotating frame and the other field
will be rotating very fast so we can
actually ignore it the static field will
then cause a tipping of your Spin and
you can detect how fast it is indeed
rotating and with that you do rabby
magnetometry indeed uh we we did that
actually with my colleague Danielle
Breer of linol lab and achieve some very
good um picotesla level sensitivity for
this fast oscillating field uh by using
their expertise in achieving uh the best
overall sensitivity in Ensemble like
large Ensemble magnetometry so this is
the current sensitivity limit if you
want for this high frequency field
detection now going back to our
oscillating field if the oscillating
field is along the Z Direction and again
you jump into a rotating frame in the XY
Direction well okay nothing really
happens so it's a it's a bit boring but
we can use this fact to achieve this
Vector magnetometry let's say it now I
have an oscillating field okay which is
in some um arbitrary Direction in the uh
in in any direction in Des XY space and
in the lab frame this is that this
frequency Omega s but if I look at it in
the rotating frame rotating at the
frequency of my apply microwave for
example then this splits the two
components so the component which is
longitudinal will stay at its original
frequency but the one which is in the
transverse plane will have a shift of
frequency by the rotating frame uh uh
frequency itself Omega and so one can
sort of detect The Chew in different
ways and indeed this is what we do we
detected transverse component uh by uh
applying for example uh by by looking at
this Shi the frequency while we detect
the longitudinal component at the uh
original frequency and we can indeed see
them uh in this particular case we see
that we were applying a field at about
28 megahertz and this uh can be detected
the longitudinal component can be
detected at the original frequency while
the transverse one is shifted by the NV
frequency and so we are able to use
basically the same uh setup uh the same
NV and the same um experimental um
sequence to detect two components of the
magnetic field so the X component and
the Z
component the limitation here was that
as you can see the difference between
the two is only about 6 mahz and this
was what the frequency of MV we needed
to set and that uh if you know the
technicality that means that you need to
work at a very specific external
magnetic field and this is the if you
want the constraint that you have in
order to be able to detect both of them
simultaneously if we make this a field
frequency with signal frequency much
much higher then it's not really
possible to autod detect the transverse
component with the same type of sequence
so this is sort of a a bit of a
limitation but the advantage is that you
have the same experimental errors and
you can use a single MV uh to do uh both
vectorial magnetometry with all
different axis and so you can also of
course apply this to an ensemble without
needing to use different classes of MV
for different
direction So to avoid this sort of
bandwidth limitation that we had in this
sample uh we decide to use the MV Center
as a Quantum frequency mixer so what you
would need is of course sort of like to
upcon convert or that convert your
frequency to a frequency which your MV
is particularly sensitive to and this
what we wanted indeed to do so again
like if I have H just an arbitrary um
field which is oscillating uh I have
that they are sort of like splitter the
two components are splited and one is
has a frequency shift but this frequency
shift might not be at the frequency
Target which I like and so what we can
do is to add some additional bias field
at an intermediate frequency so that I
will have an effective signal uh which
is now nicely situated at a frequency
typically on the order of a TENS of
megahertz which my MV can detect very
very easily with common uh U um
detection uh um protocols that people
have sort of like deoy them many many
times and indeed we do this and now we
are able to detect for example signal
which are at around 200 150 megahertz
which is typically frequency difficult
for the mvy center but just using a very simple
simple
protocols so how does this work well we
can analyze this by using a FL formalism
which uh What uh it tells us that really
what is happening is that we are
creating some additional virtual degrees
of Freedom additional uh energy levels
uh thanks to uh the drive so thanks to
the drive of our bias field in addition
to the signal field so we are
engineering some additional degrees of
freedom which allows us to do this
better sensing
task and the advantage again is that we
can then use typical um rabi or cpmg
sensing sequence to detect fields which
otherwise we will not at the gigahertz
that we will have needed for Rabbi or
low meah hortz that we would have needed
for cpmg the disadvantages that we have
a heat in the sensitivity uh but of
course if you cannot sens this sort of
like 100 of megahertz or or maybe 100 of
gigahertz frequency in other ways well
you you can sort of survive um to uh to
vis a reduction in
sensitivity so we demonstrated that and
we demonstrate that we can achieve the
same level of sensitivity well we we use
a rabi sensing protocol or a cpmg
sensing protocol cpng is where you apply
periodic pip polies which match uh the
frequency of your target signal and we
can indeed
achieve very good sensitivity with this
produ so this is an example in which we
just use control to engineer if you want
this vition additional energy level but
what we can also do is actually to use
additional degrees of freedom inside the
diamond in particular associated with
the uh nuclear spin in order to uh
improve the sensing task and actually
use the electronic spin now just as a
controller so before I introduce uh the
Envy Center um and also I think that in
the freedom introduction we focus on the
electronic spin but of course we also
have a nuclear spin in particular the
nitrogen 14 nuclear spin which is a spin
one and this has both and hyperfine
interaction which couples the nuclear
spin to the NV and also a quadrupolar
interaction which gives if you want a
local magnetic field on the the nuclear
spin itself and we can control also the
nuclear spin in the same way that we do
the electronic Spin and again by mapping
the nucleus
spin uh um state to the MV Center also
read it out so this is a much slower RAB
sorry this is I think rabi illation it's
better label of the EnV center of the
nitrogen 14 of the MV Center so we can
in principle also use this as a a
magnetic field sensor but it's not very
sensitive and so what people have been
using the um nitrogen spin is for uh
this idea of building a nuclear spin
gyroscope associated with the NV centers
because uh nuclear spin are as sensitive
as electronic spin to rotation and we
can use the electronic spin to mediate
the polariz ation without a
control so uh we uh indeed were looking
at the nitogen 14 and we wanted to test
the fact that nuclear spin in principle
should be uh very very uh uh robust to
any external perturbation and so I have
very long coherence time instead we
found relatively long coherence time
this is almost millisecond time but also
some puzzling behavior in particular if
the MV Center is in the minus one state
or plus one state the coherence time was
much shorter on the order of 2
millisecond instead of almost 1
millisecond so what we figure out very
quickly is that actually variation in
the hyperfine and quadrupolar
interaction were reducing the coherence
time so one can sort of use some Eon on
the Envy Center to correct for
this but we were also interested in
trying to find some other system which
hopefully didn't have this problem at
all and so what we did was to turn to
N15 so N15 is a spin 1 half isotope so
it does not have any quadripolar
interaction so well it should have a
better coherence time and there is no
hyper fine if the NV is in the MS equal
zero state where this spin is inde zero
and so what we expected was that as we
were going from the minus
one state for the EnV Center and looked
in at the coherence time of the nuclear
spin so if we move to the MS equal Z we
should expect much longer coherence time
instead it was a bit longer but not as
much as we were
expecting so again this was a brief
puzzle until we of course figure out if
I have some water I don't have water we
figure out that actually there is some
hyper fine because we have some
transverse component of the hyper fine
and as soon as you have a little bit of
external transverse magnetic
field with this coupled in in a very
very effective
way because that I need to find water
otherwise I'm not going to be able to continue
sorry about that let's see if it is
not so what we found was that transverse
field couple into the
hyperfine giving rise to a shift in the
resonance frequency and if the hyper
fine is in homogeneous then you have
again some Decay and this shift can be
quite large especially around uh the
level The Crossing for the nucleus Spin
and uh and of the electronic nucleus
spin so what can we do well again it
becomes better actually to uh sort of
alternate between the minus one and plus
one states of the electronic Spin and
this will actually improve the coherence
time of the nuclear spin itself because
this will sort of refocus the hyperfine
neogenes that we overwise obsorb and
indeed if you look again at the nuclear
spin coherence Time by doing this
protection scheme we can uh improve the
coherence Time by almost an order of
magnitude so the last uh thing that I
want to say is that we can actually use
the electronic spin also as a Quantum
amplifier again I told you that we have
this announcement of a transverse
magnetic field uh because of the
coupling with the hyper fine between the
nuclear and electronic spin so the
nuclear spin itself will see a much enan
uh uh transverse magnetic field and this
can be used to detect a better a
rotating magnetic field or if you want
if the diamond itself is rotating with
respect to a static magnetic field uh
the nuclear spin will see a much larger
rotation than the electronic spin itself
so what we can do is then to exploit
this fact to have an anouncement in the
sensitivity to rotation sensing so as
you can see the rotation of the diamond
is very slow but the nuclear spin see a
a much larger change and uh its Dynamics
can just be a still like AA following of
its initial state so it's very robust to
additional noise but it can give you a
very sharp advantage in the
sensitivity so what I show you that we
can use the electronic spin uh both as a
Quantum mixer as a a Quantum controller
and as a Quantum amplifier so all type
of quantum Advantage which we can get
from our Quantum sensor we really
needing to go for very large quantum
entanglement so this tells us that
indeed we can already get some Quantum
advantages although of a slightly
different time for hopefully near term
technology so even beyond the isber
limit even if you have some small amount
of entanglement or some additional
degrees of freedom we can make Quantum
and un sensing much more practical so
for example we can have a frequency
conversion of this cerence protection or
amplification of some external
perturbation so what are other
challenges of opportunity for Quantum
sensor I think that what PR describ in
the previous uh uh talk that was
perfectly spot on uh what we want to do
is really to uh try to bring all these
devices from our Labs out in the world
and there is still a lot of work to be
done there in particular in trying to
make uh sensor which are much more
robust so that they can
work in um the condition that people are
really interested in and I think the
other thing which is quite interesting
nowadays is to try to integrate Quantum
sensor in the larger Quantum ecosystem
so really uh integrate Quantum sensor
into Quantum networks integrate Quantum
sensor with quantum computer so that we
can uh really uh both exploit for
example distributed sensing and also uh
try to use a small Quantum compitation
to improve the Sens in task itself with
this I want to thank uh the people who
have uh done uh the work in particularly
minty and Goin Wang who not in this
picture because he already graduated and
uh thank you for your attention and
over one second sorry this is very s you can
can
take okay can can you hear
it uh do we have any questions for Professor
question we are little we they can uh we
can answer the there's a some question
in the Q&A but
maybe Professor can can answer those
offline yes because we a bit behind Okay
and we can start thank you so much thank
you so much I'll stop my sharing thank
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.