On-line SPICE-SPIN+X Seminar: Oleg Tchernyshyov — Transcript
Full transcript
- 0:02Just want to make sure that because it's
- 0:0330 seconds delay, but I have to
- 0:06mute it in the other one if
- 0:0920 are waiting to hear us.
- 0:12Uh because essentially some some are on
- 0:14the on the Zoom and some are on the
- 0:17Let's take a look.
- 0:20Okay.
- 0:21Great. Um
- 0:24I think we have now the
- 0:26the system and I've put it on
- 0:29here.
- 0:31So I would like to welcome
- 0:34everybody for the to the online
- 0:37spintronics seminar series.
- 0:39Uh as you know, this is a webinar
- 0:41format, so please write your questions
- 0:43in the chat
- 0:44uh
- 0:45or the raise your hands and I will give
- 0:47you the floor after the speaker
- 0:50um finishes, but also if it's important
- 0:53I can interrupt and
- 0:55ask for clarification.
- 0:57As always, we we have it on Wednesday at
- 0:593:00 p.m. on the 1500
- 1:02hours German time.
- 1:05But it's not regular every you get a
- 1:07notifications. Next month will be in
- 1:09fact on December the second,
- 1:12but at least it's always on Wednesday at
- 1:14that given time.
- 1:17Uh and of course I do the plan of this
- 1:20seminar with the help of of these folks
- 1:22over there.
- 1:23Today is a pleasure my good old friend
- 1:25Oleg
- 1:27I never pronounce your name.
- 1:35Ternishiov
- 1:36it's just it's just impossible. It's
- 1:38like my my first last name my first name
- 1:40um is very well known in our community
- 1:42in both two communities in the community
- 1:45of spintronics and spintronics as well
- 1:48as
- 1:49systems are quite a proud physicist.
- 1:51He started his PhD in the Columbia
- 1:53University in New York
- 1:54and didn't move that far. Went down
- 1:56south a little bit to go to Princeton
- 1:59for a postdoc and then now I'm in Oak
- 2:00Ridge
- 2:01for quite a while start looking at
- 2:04correlated systems, magnetic systems and
- 2:06the dynamics of solitons and skyrmions
- 2:08and all kind of topological objects
- 2:11as well as spin liquids.
- 2:13Um one of the worst to highlight as well
- 2:15many is the the piece of fellow that may
- 2:16come for the society
- 2:18since 2019.
- 2:20With this I'm going to stop sharing
- 2:23the
- 2:24and allow you know ask you to go ahead
- 2:25and share your
- 2:27slides and
- 2:30just start anytime.
- 2:33>> Thank you Phayro.
- 2:34Um
- 2:36>> Yeah see that.
- 2:38>> Okay.
- 2:39>> Perfect.
- 2:40>> All right.
- 2:41Good.
- 2:43All right. Thank you so much for having
- 2:44me.
- 2:45It's my second talk at
- 2:48Spin X.
- 2:50Wonderful audience.
- 2:52I'm excited to be here.
- 2:55As
- 2:56luck would have it today my brother and
- 2:58his family are visiting Mines
- 3:01and I told them that I will be giving a
- 3:03talk on Mines and they were thrilled
- 3:05but then I told them that this would be
- 3:07online.
- 3:09Nonetheless here we go.
- 3:11I am going to tell you about
- 3:14new field theory for three sublattice
- 3:16antiferromagnets
- 3:18where the order parameter is not a
- 3:20simple vector. It's a rigid body.
- 3:23It's a field theory with
- 3:25four coupling constants
- 3:27and otherwise it's pretty universal.
- 3:31There's there are some experimental
- 3:33consequences that I will touch upon here
- 3:36and we'll see how much I will be able to
- 3:39convey.
- 3:40Let me thank the people with whom I
- 3:43interacted over the years.
- 3:45I want to credit Satoru Nakatsuji for
- 3:49kind of bringing together two of my
- 3:51topics of interest
- 3:55frustrated magnets on which I worked
- 3:58in some distant past
- 4:01and spintronics.
- 4:03So Satoru is credited with
- 4:05uh
- 4:07uh resurrecting interest to manganese
- 4:09three tin, a compound that was
- 4:10discovered in the 1980s uh but that that
- 4:14came to prominence more recently because
- 4:16of the
- 4:17uh
- 4:18uh conducting nature of that magnet and
- 4:21uh uh potential in spintronics
- 4:24through the anomalous Hall effect that
- 4:26it exhibits that's related to its
- 4:27magnetic order.
- 4:29Uh Colin Broholm is my colleague
- 4:31experimentalist here at Hopkins
- 4:34uh who investigated manganese three tin
- 4:35with Satoru.
- 4:37Uh uh
- 4:38Boris Ivanov uh
- 4:41is
- 4:42someone who's been in the field pretty
- 4:44much longer than anybody uh uh
- 4:46else
- 4:47in this audience, I think.
- 4:50Hua Chen was on a sabbatical here at
- 4:52Hopkins well when we discussed things
- 4:54and I have my wonderful former students
- 4:56Sayak Bastian and and Grigor
- 4:59who worked on on on these systems.
- 5:02Here's a list of papers that we
- 5:05put out recently
- 5:07and if you don't want to read all these
- 5:09papers uh I I would suggest reading a
- 5:13piece by Sid Parameswaran
- 5:15in the Condensed Matter Journal Club in
- 5:17May 2025 which gives a very quick and
- 5:20very nice overview of of this
- 5:23spin frame field theory.
- 5:26Uh so let's jump into field theories of
- 5:29of magnets. The first field theory of a
- 5:32ferromagnet was uh
- 5:34uh written down by Landau and Lifshitz
- 5:36who had a very
- 5:38uh practical interest in understanding
- 5:41uh the AC susceptibility of of
- 5:44ferromagnets like iron and nickel.
- 5:47And so they came up with
- 5:49a field theoretic description that would
- 5:51allow them to
- 5:52to deal with the physics of domain
- 5:54walls. Domain walls are magnetic
- 5:56solitons. Dealing with soliton in a
- 5:59lattice is a very difficult problem. If
- 6:01you switch to a continuum description,
- 6:03it's much easier and so that's what what
- 6:05they did.
- 6:07So they introduced the
- 6:10order parameter as a unit vector spin
- 6:12field
- 6:14M.
- 6:15If you multiply that by the spin density
- 6:19number, you get
- 6:22on the left hand side
- 6:24the derivative of
- 6:29the density of angular momentum and on
- 6:31the whatever potential energy
- 6:33functional.
- 6:34And for the Heisenberg ferromagnet, for
- 6:37instance, the functional would be simply
- 6:39the gradient squared
- 6:42of the
- 6:44magnetization or spin
- 6:46field.
- 6:47And so you have a
- 6:49fairly simple equation that
- 6:51you can
- 6:53investigate.
- 6:55Spin waves
- 6:56are elementary excitations in
- 6:58ferromagnets and they look pretty much
- 7:00like this. You have spins that process
- 7:04in the clockwise direction if you look
- 7:06from from above
- 7:08and
- 7:12that's that's a signature of
- 7:14its precessional rather than inertial
- 7:16and it takes some unlearning
- 7:20you know, the intuition we have with
- 7:22particles who whose equations of motion
- 7:24are second order in time.
- 7:27This is completely different.
- 7:29Now, another theory,
- 7:31field theory of of
- 7:32a magnet was
- 7:34written down in the late 1970s and early
- 7:371980s
- 7:39by Boris Ivanov,
- 7:41Andre Ivan Marchenko, Duncan Haldane
- 7:43among others.
- 7:44And
- 7:46their equation of motion,
- 7:48as you can see here, is second order in
- 7:50the time derivatives. Here N is a unit
- 7:52vector field of the staggered spin or
- 7:55nail vector
- 7:56which describes the spins on one
- 7:58lattice, let's say the red one, and the
- 8:00spins on the other sublattice point
- 8:02exactly opposite to that. So you can
- 8:04still encode all of the spins in the
- 8:06antiferromagnet with a single vector.
- 8:09So the equation of motion looks like
- 8:10this. You can see the right hand side
- 8:12barely changed. Uh you still have
- 8:15uh conservative torque density on the
- 8:17right hand side. The left hand side
- 8:19changed. Uh this is now the density of
- 8:22angular momentum differentiated with
- 8:24respect to time. So you have a uh sort
- 8:26of similar equation, but it's now second
- 8:28order in in the time derivative.
- 8:31And in fact, the energy functionals look
- 8:33very similar. Instead of the uh
- 8:36square of the
- 8:38uh derivative of the gradient of
- 8:40magnetization, you have the
- 8:42uh the same for the uh nail vector.
- 8:46So spin waves in an antiferromagnet look
- 8:48a bit different. You can have linear
- 8:50polarizations. Both sublattices would
- 8:52like to precess, but each in its own
- 8:54direction. And so precession basically
- 8:56cancels out. And what remains is this
- 8:59sort of inertial dynamics where spins
- 9:01can have linear polarizations like
- 9:03these, or you can make circular
- 9:05polarizations out of them
- 9:06uh going uh clockwise or
- 9:09counterclockwise. And and these uh spin
- 9:11waves would have the same frequencies.
- 9:15So a field theory uh
- 9:17of uh a nail antiferromagnet
- 9:20uh can be derived from a Lagrangian. And
- 9:22the Lagrangian would have uh
- 9:24the kinetic energy term, which is the
- 9:27time derivative of the nail vector
- 9:30squared times some density.
- 9:32And then it would have a potential
- 9:33energy term, which in the Heisenberg
- 9:35antiferromagnet looks uh like this.
- 9:40Uh you could do this as well uh with uh
- 9:43ferro magnets, although instead of the
- 9:45kinetic energy would have a berry phase
- 9:46term.
- 9:48Uh in the late '80s uh there was
- 9:51interest uh to uh frustrated uh
- 9:54antiferromagnets. Uh
- 9:57in particular, one of the questions that
- 9:58that was uh
- 10:01uh on people's mind whether
- 10:03uh the field theory of of uh
- 10:06antiferromagnets
- 10:07uh had a topological term like it
- 10:10uh in uh
- 10:11one dimension.
- 10:13Haldane gap phenomenon is related to
- 10:15that and we'll discuss that briefly
- 10:18later on in the talk.
- 10:20Also people were
- 10:21trying to see whether topological terms
- 10:23also exist
- 10:25on the square lattice, which turned out
- 10:28not to be the case
- 10:30and then Dubail and Read investigated
- 10:32the question on the triangular lattice
- 10:36whether there is a topological term and
- 10:38maybe a difference between integer and
- 10:40half integer spins, the spectrum might
- 10:42be different and so on and so forth.
- 10:45And so they had to write down a field
- 10:47theory
- 10:48on the triangular lattice
- 10:49antiferromagnet and in this case because
- 10:52the ground state is not collinear you
- 10:54can you cannot use a single vector to
- 10:56describe it, but you have to use
- 10:57something else and that something else
- 11:00happens to be
- 11:02a rigid body. The order parameter which
- 11:04you can think of three
- 11:06spins from the three sublattices
- 11:09that look like a Mercedes-Benz logo,
- 11:12you can think of this as a rigid body
- 11:14and so the dynamics of the order
- 11:16parameter would resemble vibrations of
- 11:18this
- 11:20order parameter
- 11:22in in various directions,
- 11:25perhaps rotations of this order
- 11:26parameter in various directions. So you
- 11:29have to write a field theory in terms of
- 11:31this order parameter and one way to do
- 11:33that is in terms of SO(3) rotation
- 11:35matrices that determine the orientation
- 11:39of this Mercedes-Benz logo.
- 11:41So if you use matrix O
- 11:44an SO(3) matrix then this is your field
- 11:46theory. It looks a bit more complicated
- 11:49and and anharmonic compared to
- 11:53what you saw previously,
- 11:55but you can still look for a topological
- 11:57term in there and
- 12:01the spoiler here is that Dubail and Read
- 12:03did not find a topological term.
- 12:05So that was the end of that
- 12:08and
- 12:10field theory of of triangular
- 12:12antiferromagnets
- 12:14was was pretty much not
- 12:16uh
- 12:17in the vogue anymore.
- 12:18So we resurrected it
- 12:20a few years ago 2024
- 12:23with Bastian Prudenus who is now in
- 12:26Dresden
- 12:28and and Grigora Demian who came later
- 12:31and is now at MIT.
- 12:33And we used a different object
- 12:37mathematical object to describe this
- 12:39field theory
- 12:40instead of using matrix which is kind of
- 12:42unintuitive. If you give me a matrix
- 12:44I have no idea what the orientation it
- 12:46describes. It's you know it's nine
- 12:47numbers that that that I cannot read
- 12:50into.
- 12:51So instead we use a different order
- 12:52parameter.
- 12:53If you have a rigid body
- 12:55a rigid body has principal axes and you
- 12:59can direct three unit vectors along
- 13:01these principal axes and these three
- 13:02unit vectors can serve as as a body
- 13:05frame
- 13:07that rotates along with the body. So
- 13:09here we just define a spin frame
- 13:12which is defined in terms of
- 13:15the three sub lattice magnetizations.
- 13:17You can think of an X and an Y as the
- 13:20two staggered magnetizations that that
- 13:22you have
- 13:23right? You can again take the difference
- 13:25of of two sub lattices
- 13:27and maybe another difference of two sub
- 13:29lattices or the difference between the
- 13:31three sub lattices or something like
- 13:32that
- 13:33and these two vectors will be orthogonal
- 13:34and unit vectors if you properly
- 13:36normalize them.
- 13:38A third vector can be obtained just by
- 13:39taking the cross product of these two
- 13:42and so that gives you
- 13:43a frame.
- 13:45In fact
- 13:46this parametrization relates in a simple
- 13:48way to the SO3 matrix parametrization.
- 13:51If you take the three vectors of the
- 13:53body frame
- 13:55and three vectors of a global fixed
- 13:57frame their scalar products will
- 13:59feature nine numbers and these nine
- 14:01numbers are exactly the SO3 matrix of
- 14:03Dummer and Reed.
- 14:06So why do we call these
- 14:10vectors NX and NY? Well this has to do
- 14:13with directions in the lattice, rather
- 14:16than in spin space.
- 14:18Uh and uh
- 14:20we call them an x and y because they
- 14:21have
- 14:23uh
- 14:23symmetry properties similar to
- 14:25coordinates x and y in the lattice.
- 14:28Uh and
- 14:29here's how it goes. Uh let's take a
- 14:31lattice,
- 14:33uh kagome lattice in this case, uh with
- 14:35the x and y axes in the plane, the z
- 14:38axis perpendicular to it.
- 14:40Uh
- 14:41and the point group of this lattice
- 14:43involves
- 14:45uh several operations of the uh group
- 14:47D3. Among them, let's say this rotation
- 14:51about the y axis, in which uh you swap
- 14:54the two sublattices, one and two, for
- 14:56example, and sublattice three stays uh
- 14:59uh itself.
- 15:00So the magnetization vectors of one and
- 15:03two and three will change into m two, m
- 15:05one, and three.
- 15:07And that would flip the sign of an x but
- 15:09keep the sign of an y.
- 15:11So you can see that an x and an y will
- 15:13transform into minus an x and an y. And
- 15:15it's the same way in which any uh
- 15:18spatial vector transforms under a C2 y
- 15:20rotation.
- 15:21And that's important because we will
- 15:23need the properties of an x and y under
- 15:25these uh point group transformations.
- 15:28And that's in the spirit of Andrei von
- 15:30Marchenko, this famous uh
- 15:33but poorly understood 1980 paper.
- 15:36So knowing the symmetry properties of uh
- 15:39of our
- 15:40uh
- 15:42water parameters uh or the uh frames
- 15:46vectors n x and y and then z enables us
- 15:49to construct uh a simple exchange uh uh
- 15:54energy density
- 15:55which has to be uh of course invariant
- 15:58under uh spatial symmetries uh of the
- 16:02lattice as well as the uh
- 16:05uh global spin rotation symmetry if it's
- 16:08the Heisenberg
- 16:10So what comes out is uh
- 16:13uh something similar to the uh
- 16:15uh
- 16:17collinear antiferromagnets where you had
- 16:19gradients of n squared, but here we have
- 16:22several n's. We have an x and an y.
- 16:25For example, n z
- 16:28is not relevant in this context, but you
- 16:30can
- 16:32include several different
- 16:35pairings between the indices,
- 16:38partial x and x for example, or partial
- 16:41y and y and so on. And it turns out that
- 16:43these are the three possible pairings
- 16:46that survive the symmetry constraints.
- 16:49So it's possible to have three
- 16:51distinct terms in your energy density.
- 16:55So it's slightly more complicated than
- 16:57than the collinear case, but nonetheless
- 16:59manageable.
- 17:00And it's actually similar to the elastic
- 17:02energy of a hexagonal solid
- 17:05written in terms of the atomic
- 17:07displacement field, which also has these
- 17:10indices.
- 17:11And so you can write it in this way. And
- 17:14the constants lambda mu and new would be
- 17:16the combinations of bulk modulus and
- 17:19shear modulus. The difference between mu
- 17:22and new would be zero for a solid
- 17:24because it would describe a rotation.
- 17:27So it's not exactly an exotic
- 17:30energy density. One can understand where
- 17:32it comes from.
- 17:34And so one can add
- 17:35kinetic energy like in a collinear
- 17:38antiferromagnet, which has this form.
- 17:40And this time you have to use
- 17:42all all three vectors of x and y and z.
- 17:45And what this basically is is omega
- 17:48squared, the rotation frequency squared
- 17:52of of the
- 17:54of the order parameter.
- 17:57Okay, so we've got a field theory with
- 18:00four coupling constants. One is the
- 18:02inertia, and there are three
- 18:03stiffnesses.
- 18:05And they determine the the spectrum of
- 18:07spin waves for example.
- 18:09So you can derive the velocity of the
- 18:11spin waves,
- 18:13two of which are
- 18:15rocking your
- 18:17order parameter out of its plane, and
- 18:19the third one is keeping it in the
- 18:21plane. I'll show an illustration of
- 18:23that. What matters here is
- 18:26the speeds can be calculated from
- 18:29the wave equations, and they feature
- 18:32these combinations of the stiffnesses in
- 18:35the numerator and
- 18:38inertia in the denominator.
- 18:40Now what's interesting is
- 18:43the the the first and third stiffness
- 18:45coefficients, lambda and nu,
- 18:47come together as a combination lambda
- 18:49plus nu
- 18:51here as well as here,
- 18:54but never
- 18:55on the road.
- 18:56So it means that effectively lambda and
- 18:58nu do the same thing
- 19:01and instead of having three stiffnesses,
- 19:04we actually have two.
- 19:06And that means that the three spin wave
- 19:08velocities are not independent,
- 19:11but if you square them and adopt the
- 19:13first two,
- 19:14you get the square of the third one. So
- 19:16there's this lovely Pythagorean relation
- 19:18that is experimentally verifiable.
- 19:21So that's one experimental consequence
- 19:24of this field theory, and
- 19:26there will be more.
- 19:27So here are the the spin waves. So you
- 19:30can imagine what what they look like.
- 19:32You have
- 19:33two linear polarizations, this one and
- 19:35that one, and and you have a third
- 19:37linear polarization that
- 19:39basically these are oscillations about
- 19:42different axes,
- 19:45different principal axes in in this
- 19:47rigid body.
- 19:49And of course you can form not only
- 19:51linear
- 19:52polarizations, but you can also form
- 19:54circular polarizations, and that will be
- 19:56a story down the road.
- 19:59Okay. So the exchange energy density has
- 20:01these terms.
- 20:05Lambda and nu terms
- 20:07have the SO(2) rotational symmetry. You
- 20:10can rotate
- 20:12sort of the components of the gradient
- 20:14as well as make combinations of an x and
- 20:17y.
- 20:19And
- 20:21these two terms have just this SO2
- 20:22symmetry
- 20:24because you rotate the gradient
- 20:26components as well as
- 20:28you do the
- 20:30transformation between an X and Y.
- 20:34The mu term actually has more symmetry
- 20:36because here you can rotate separately
- 20:38components of the gradient which are
- 20:40paired
- 20:41and you can separately rotate an X and
- 20:43an Y components
- 20:45of the
- 20:46spin frame. And so this symmetry is
- 20:48higher. It's SO2 cross SO2.
- 20:51It turns out the the triangular lattice
- 20:53only has these terms
- 20:55and the lambda and mu terms are not
- 20:57allowed. So for the triangular lattice
- 20:59the field theory simplifies. You have a
- 21:01much more symmetric form than
- 21:04this field theory allows. And on Kagome
- 21:08if you only include nearest neighbor
- 21:09interactions then it turns out that mu
- 21:11vanishes and lambda plus mu is non-zero.
- 21:15More generally Kagome with longer range
- 21:17couplings will have all of these
- 21:20constants, but the triangular lattice
- 21:22only has mu no matter how many
- 21:25nearest and second neighbors you include
- 21:26and so on.
- 21:30All right. So one consequence of that
- 21:33the higher symmetry of the triangular
- 21:35lattice antiferromagnet is that if you
- 21:37view vertices
- 21:40of the order parameter
- 21:42they actually are nicer and more
- 21:44symmetric. So on the triangular lattice
- 21:46this is what a vortex looks like.
- 21:48Here I'm showing
- 21:50the three sublattices in different
- 21:52colors, blue, red and green.
- 21:55You can see the red and green arrows and
- 21:58the blue arrows are pointing toward you.
- 22:00So this is a rotation about the the
- 22:02direction of the blue arrows.
- 22:04And you see that on the triangular
- 22:06lattice this vortex has a nice round
- 22:08shape.
- 22:09On Kagome
- 22:11with some second
- 22:13well third neighbor coupling
- 22:16we have a different shape. You can see
- 22:18that it's roughly an ellipse.
- 22:21And so generally vertices on the Kagome
- 22:23lattice will have an elliptical shape
- 22:26with the orientation of the major axis
- 22:30that can be arbitrary. They're not stuck
- 22:32to the lattice directions.
- 22:36All right. So
- 22:38one cute thing that came out of this
- 22:40field theory is that you can control
- 22:43the order parameter
- 22:45with spin waves. You can manipulate the
- 22:47direction of your spins by sending a
- 22:51pulse of of magnons
- 22:55which which is something you cannot do
- 22:57in collinear antiferromagnets.
- 22:59And I'll explain why.
- 23:01So here's the
- 23:03expression for the energy of a
- 23:04triangular lattice. You have just the mu
- 23:07term
- 23:08for the stiffnesses
- 23:10and if you work out the speeds of sound
- 23:13waves
- 23:14it turns out that two of the sound waves
- 23:16are degenerate
- 23:18and the third one has a higher speed
- 23:20than than the two.
- 23:22And with degenerate polarizations
- 23:26sorry, with degenerate frequencies of
- 23:28two linear polarizations you can form
- 23:30circular polarizations. So here are
- 23:32linear polarizations, right? You can
- 23:34rock your Mercedes-Benz logo about two
- 23:37directions, but you can also combine
- 23:39them
- 23:40and and here's what what what you're
- 23:42going to have. This is a circularly
- 23:43polarized
- 23:44spin wave made out of
- 23:47the two linear polarizations.
- 23:49Notice something strange as
- 23:53as the circular polarization wave
- 23:56rocks it turns gradually. You can turn
- 23:59either way depending on which circular
- 24:01polarization you excite. So you can
- 24:05gradually rotate clockwise or
- 24:06counterclockwise.
- 24:08And that's an effect that that is not
- 24:09available for collinear spins and we'll
- 24:12we'll we'll see why shortly.
- 24:15Okay. So it turns out
- 24:17that there's a very cute analogy with
- 24:21a wobbling coin,
- 24:22a problem that was popular in the 19th
- 24:24century when people were thinking about
- 24:26spinning tops
- 24:27uh or more recently as the Euler disk
- 24:30that's commercially available.
- 24:57So you can see that this wobbling coin
- 25:00uh keeps rotating.
- 25:02And so after rotation stops, your water
- 25:04parameter will have
- 25:05uh rotated in this direction. And by uh
- 25:09applying a pulse that is uh sufficiently
- 25:11short or long, you can change the uh
- 25:14the amount of rotation in in the system.
- 25:17So why is there a rotation here? What's
- 25:19what's going on? Well,
- 25:20you can think about it this way.
- 25:22Uh at every moment in time,
- 25:25your uh water parameter or the wobbling
- 25:27coin
- 25:28rotates about a different axis. So here
- 25:32the axis is shown as as this gray. The
- 25:34rotation axis is shown uh as this gray
- 25:38straight line.
- 25:39Uh and you can see that the rotation
- 25:41axis changes all the time. That's the
- 25:42instantaneous axis of rotation.
- 25:45It it touches the point
- 25:47uh
- 25:47uh
- 25:48uh
- 25:49it it goes through the center and the
- 25:50point where the the the wobbling coin
- 25:53touches the surface. And so it always
- 25:55changes. And that means that you're
- 25:57rotating your uh
- 25:58coin
- 25:59uh or this uh body frame
- 26:02uh about different axes all the time.
- 26:04And different rotation ex rotations
- 26:06about different axes do not commute.
- 26:09So if you uh simplified down to
- 26:11something
- 26:12uh
- 26:13instead of an infinite number of
- 26:14infinitesimal rotations, you boil it
- 26:16down to four rotations consecutively
- 26:19about the NX and Y NX
- 26:22uh and then Y axes. So you rotate about
- 26:25NX about alpha, then about at Y by beta.
- 26:28And then you do the opposite rotations
- 26:30about minus alpha about an x minus beta
- 26:33about on y.
- 26:35The four rotations don't commute and so
- 26:37if you take the product of these you'll
- 26:39end up with a rotation about the nz axis
- 26:43uh by an angle that's roughly the
- 26:44product of alpha and beta. So that's the
- 26:47non-commutativity of of of rotations
- 26:50uh and and so you end up with a net
- 26:52solid
- 26:53uh with a net rotation
- 26:55by an angle equal to the solid angle
- 26:57subtended by the black nz vector the
- 26:59spin chirality.
- 27:02In fact there's a very simple
- 27:04geometrical picture of why that that
- 27:06works. It's similar to doing parallel
- 27:08transport on a sphere.
- 27:10So suppose I I give you this uh uh
- 27:14Mercedes-Benz
- 27:15depiction of of the magnetic order uh uh
- 27:19in this triangle lattice antiferromagnet
- 27:22and I ask you to carry it around the
- 27:24sphere
- 27:25uh
- 27:26always keeping it horizontal
- 27:28right so it'll be tangential to the
- 27:29sphere everywhere and never turning it
- 27:33never twisting it about the vertical
- 27:36axis your local vertical axis so you
- 27:37cannot twist it about the uh
- 27:40axis perpendicular to the spins so you
- 27:42can only keep it horizontal and so this
- 27:44is what you do you walk down to the
- 27:46equator carrying this Mercedes-Benz logo
- 27:49uh and then you walk
- 27:51along the equator
- 27:53uh
- 27:54and that's what parallel transport looks
- 27:56like and then you come back up the
- 27:58meridian again and voila your order
- 28:01parameter has rotated
- 28:03right and the rotation angle is 90
- 28:04degrees or pi over two which is exactly
- 28:07the solid angle uh that your path takes
- 28:11out on the sphere. So that's basically
- 28:14uh uh
- 28:15you know berry phase geometric rotation
- 28:19uh parallel transport whatever you like
- 28:20to call it.
- 28:22Right so you you rotate it by pi over
- 28:24two and that's exactly what happens
- 28:26here.
- 28:27So why doesn't this happen to collinear
- 28:29spins? Well actually it does we just
- 28:31don't notice it. So here's a circularly
- 28:33polarized wave in the collinear
- 28:35antiferromagnet. This is what it looks
- 28:37like. The spin is always rotated about
- 28:40this
- 28:42ever-changing axis, right? And so after
- 28:45this rotation, you expect that the spin
- 28:48will rotate about its own axis as well.
- 28:51And of course it will, but the spin is
- 28:53round.
- 28:54It's axially symmetric and we don't see
- 28:56any rotation
- 28:59after it stops. It's the same spin. So
- 29:02that's why you cannot see it in the
- 29:03collinear antiferromagnet. The spins do
- 29:05rotate, but they rotate about their own
- 29:07axis and so you cannot observe them.
- 29:10That's why you need another parameter
- 29:11that includes more than one spin and
- 29:14more than two spins. You you need to
- 29:16have a coplanar state
- 29:18as as we had with with this triangular
- 29:20antiferromagnet.
- 29:22So that's a proposal for how one can
- 29:25control the order parameter
- 29:27in an antiferromagnet on a triangular
- 29:29lattice
- 29:30with with spin waves.
- 29:33Okay.
- 29:34Well,
- 29:36topological terms. I started this talk
- 29:38by mentioning that Tummuru and Read went
- 29:40in search of topological terms in the
- 29:43action
- 29:44inspired by Haldane's work on
- 29:47antiferromagnetic spin chains where
- 29:49topological terms are very important and
- 29:51explained the difference between the
- 29:52gapless spectrum of half-integer spins
- 29:55and gap spectrum of integer spins.
- 29:59And so they said, well, can we find
- 30:00topological terms in in this
- 30:04triangular lattice antiferromagnet? And
- 30:06the question the answer was no. The they
- 30:08went looking for a specific topological
- 30:10term
- 30:12with one spatial derivative and two time
- 30:14derivatives and they did not find one.
- 30:17Well, we did.
- 30:19In fact, it turns out that if you take
- 30:22so there are several topological terms
- 30:24in fact that that that could be found
- 30:26and one of them
- 30:28is written here as the difference of the
- 30:31lambda and new terms. Remember the
- 30:33lambda and new terms and the stiffness
- 30:35did the same thing to spin waves.
- 30:38Uh these velocity of the spin waves is
- 30:40controlled by uh the lambda plus new
- 30:44uh combination rather than by lambda and
- 30:46new separately.
- 30:47And there's a simple reason for that. If
- 30:49you take the difference of these two
- 30:50terms,
- 30:52uh in fact by partial integration you
- 30:55can move
- 30:56this spatial derivative to the right and
- 30:58that spatial derivative to the left and
- 31:01you'll get exactly this term.
- 31:03So these terms coincide up to boundary
- 31:06corrections. So in the bulk they
- 31:09actually give zero contribution. So even
- 31:11though this looks like a bulk uh
- 31:14integral it actually boils down to a
- 31:17line integral along the boundary uh and
- 31:20and and it it has uh
- 31:23no influence in the bulk strictly
- 31:25speaking.
- 31:26Uh and so there's there's there's a
- 31:28analogy to the Mermin correlation in
- 31:31theory of P wave superfluids uh where
- 31:34this combination
- 31:36uh can be thought of as a magnetic field
- 31:38of sorts uh of an SO3 gauge shield uh
- 31:42and you can uh express it in terms of a
- 31:45an SO3 gauge potential. And so that uh
- 31:48magnetic field integral uh over a
- 31:51spatial uh
- 31:53of an area boils down to a
- 31:56uh vector potential integral along the
- 31:58boundary.
- 32:00So it's not a bulk term, it's a boundary
- 32:02term.
- 32:03And so as a boundary term it will uh
- 32:06presumably do something interesting on
- 32:07the edge.
- 32:09There's another way to express this term
- 32:11and that's something that Bastian found.
- 32:13I have no idea how he was able to do
- 32:15that but you can show that
- 32:17uh this expression
- 32:19is equivalent to that expression. So
- 32:21alpha and beta here are x and y
- 32:23components. So it only involves an x and
- 32:25an y and the integral here looks like a
- 32:29skyrmion number density for the
- 32:31chirality vector on Z.
- 32:33And so of course if you integrate it
- 32:34over the uh infinite two-dimensional
- 32:37space you get a skyrmion number.
- 32:39Skyrmion number is an integer.
- 32:41And so if you take a an infinitesimal
- 32:43variation of these fields,
- 32:45in an infinite system there will be no
- 32:47change to this integer.
- 32:49And therefore this
- 32:50this is a silent term as far as as far
- 32:53as the classical dynamics goes. So it
- 32:55does not influence the dynamics
- 32:58in an infinite system.
- 32:59But it it does create a boundary term.
- 33:02So if you have a boundary, you have a
- 33:04boundary energy.
- 33:05And so this boundary energy actually
- 33:07makes an interesting effect. It binds
- 33:11spin waves to a boundary,
- 33:13or magnons. And
- 33:15these magnons happen to be chiral.
- 33:19They propagate in only one direction,
- 33:21let's say counterclockwise but not
- 33:23clockwise.
- 33:25A similar term was found in an exotic
- 33:27theory of orbital magnetization
- 33:31by Levitov and his postdocs.
- 33:34And when we saw that paper we
- 33:36immediately realized that we have a
- 33:38similar term and so it should should
- 33:39give rise to
- 33:42these sort of edge modes. So you can
- 33:44analyze these edge modes
- 33:47for a straight edge along the x
- 33:49direction.
- 33:51kx is a conserved momentum
- 33:54of a magnon.
- 33:56And so you can try a spin wave
- 33:59that runs along the edge,
- 34:01but exponentially decays
- 34:04into the bulk along the y direction.
- 34:07And you can figure out what what the
- 34:10the penetration depth is
- 34:13and and show that
- 34:15this wave is only bound to the edge for
- 34:17one sign of kx but not for the other.
- 34:20And so here's what happens. You you make
- 34:23a a circular polarization
- 34:26out of your spin waves, and you discover
- 34:28that there is a mode that travels around
- 34:31the boundary and it only penetrates into
- 34:33the bulk
- 34:36some
- 34:37number of
- 34:38wavelengths which which on on the
- 34:41strength of the topological term.
- 34:43So if you have shorter wavelength, then
- 34:46this wave penetrates less into the bulk
- 34:48and it's better tied tied to the edge.
- 34:51So only waves
- 34:53propagating counterclockwise are bound
- 34:55to the edge
- 34:57and you don't have waves propagating the
- 34:58other way.
- 35:00If you want to make these waves, you
- 35:01have to take
- 35:03the left circular polarization instead
- 35:05of the right circular polarization,
- 35:07which is equivalent to the time reversal
- 35:09and that's why the direction of
- 35:10propagation is reversed. So in fact,
- 35:13there are two counter propagating waves
- 35:15traveling clockwise and counterclockwise
- 35:18and what they differ by is is is the
- 35:20spin projection that that they have
- 35:24that is
- 35:25out of the plane.
- 35:27So it's a field theory and it would be
- 35:29nice to have a lattice model
- 35:31that
- 35:32exhibits this sort of thing.
- 35:36Because the field theory is
- 35:37non-committal. It tells you what's
- 35:38possible. You need an actual lattice
- 35:41model which will realize this exotic
- 35:43effect.
- 35:44And we did find such a lattice model.
- 35:47It's actually not very exotic at all.
- 35:51So take a triangular lattice
- 35:53antiferromagnet.
- 35:54The spins reside in the corners of the
- 35:56triangles
- 35:58and the
- 35:59colored circles are actually
- 36:01non-magnetic ions. These are ions that
- 36:04mediate super exchange.
- 36:06Imagine that you have two kinds of
- 36:09non-magnetic ions.
- 36:11Maybe oxygen shown in red
- 36:14centered on the up triangles and sulfur
- 36:18ions shown in blue centered on the down
- 36:20triangles.
- 36:22And suppose oxygen and sulfur have
- 36:24different super exchange strength
- 36:26mediated
- 36:28between two spins. So the red path
- 36:30through the oxygen gives you
- 36:33this much
- 36:34of
- 36:35super exchange and the blue path gives
- 36:38you that much super exchange. Okay, they
- 36:40differ. Delta measures is uh
- 36:43uh difference between the two
- 36:45superexchange paths.
- 36:47Now, if you take two spins on the bulk,
- 36:49you have two superexchange paths
- 36:51available to them. So you have to add
- 36:53these contributions, and when you add
- 36:55them, delta disappears.
- 36:57So you you just add this plus that, and
- 37:00it becomes simply J, and delta has no
- 37:03effect on the bulk.
- 37:05But it has an effect on the edge.
- 37:07If you are on the edge, these two spins
- 37:09only have a red superexchange path.
- 37:11And if you are on the opposite edge,
- 37:13these two spins have a blue
- 37:14superexchange path.
- 37:16So the bonds on the edge are going to be
- 37:18different from the bonds on the bulk,
- 37:21and they will be influenced by this
- 37:22difference delta. So here this is how we
- 37:25construct
- 37:26what looks like a bulk term,
- 37:28right? It's a bulk superexchange,
- 37:30but the superexchange part, the delta
- 37:32dependence uh does not exist in the
- 37:34bulk, and it only is pushed out to the
- 37:36edge. So that's how this uh field
- 37:39theoretic construction is realized in
- 37:40the lattice.
- 37:42So you can take this model,
- 37:44and you can just numerically determine
- 37:45its spectrum.
- 37:47Uh and here's its spectrum uh in a strip
- 37:50that runs along the x direction.
- 37:52So kx is a conserved momentum, but ky is
- 37:55not conserved because translational
- 37:57symmetry is broken along the y
- 37:59direction. Uh and so for a given kx, you
- 38:02will have not one uh
- 38:04spin wave frequency in the bulk, but a
- 38:06spectrum of frequencies uh
- 38:09that correspond to different values of
- 38:11ky. And because ky can vary continuously
- 38:14in an infinite system, uh you can
- 38:17basically have any frequency uh above a
- 38:19certain value.
- 38:21So the gray area uh are the spin waves
- 38:25residing in the bulk.
- 38:26And you can see that uh near these uh
- 38:29points 2 pi over 3 and minus 2 pi over
- 38:323, you have an interesting spectrum. In
- 38:34addition to the uh
- 38:36bulk spin waves, you have spin waves uh
- 38:39that lie below that bulk continuum, and
- 38:42these are precisely the edge modes uh
- 38:44that live uh uh
- 38:46only near the
- 38:48and they exponentially decay into the
- 38:49bulk.
- 38:51And you can see that the waves propagate
- 38:53one way along the upper edge, the blue
- 38:55curve, and they propagate the other way
- 38:58on the lower edge.
- 39:00And if you
- 39:02And this is for one
- 39:03circular polarization. For the opposite
- 39:05circular polarization, these two will
- 39:07switch and you will have
- 39:10waves propagating right along the upper
- 39:13edge and left along the lower edge.
- 39:16So this is different from the scenario
- 39:18of topological magnons
- 39:21that's discussed
- 39:24in the context of Dzyaloshinskii-Moriya
- 39:26interactions where you can have a
- 39:28spectrum
- 39:30in the bulk
- 39:32represented by these black lines
- 39:34and then there's a gap in the middle and
- 39:37you have
- 39:38topological magnons due to the
- 39:40Dzyaloshinskii-Moriya interaction that
- 39:42reside
- 39:44in between
- 39:45the two
- 39:47upper and lower bands. So our magnons
- 39:50actually live below our edge magnons
- 39:52live below the bulk magnons.
- 39:54And so these magnons here, the
- 39:56topological magnons can decay into bulk
- 39:59magnons of lower energy or pairs of them
- 40:02or something
- 40:03whereas
- 40:04the
- 40:06the topological magnons in this case
- 40:08are stable
- 40:10at least for the smooth edge. If you
- 40:12have a rough edge,
- 40:14there will be scattering of the edge
- 40:15magnons into the bulk. But it turns out
- 40:18that at the longest wavelength the
- 40:20scattering is
- 40:21is not very strong.
- 40:23And so this
- 40:24these edge magnons should survive.
- 40:28All right.
- 40:29So Hira, do you have 5 minutes?
- 40:32>> Yeah, yeah, absolutely. This is really
- 40:34nice.
- 40:35>> Good. So I'll I'll I'll talk about more
- 40:38topological terms.
- 40:40Now Duncan Haldane's topological term
- 40:45Well, okay, so we'll get back to
- 40:46Duncan's topological term. So this this
- 40:48is the topological in the action that I
- 40:50described. It has uh two spatial
- 40:53derivatives and resembles the skyrmion
- 40:55number in the XY plane.
- 40:58You can form other combinations uh of
- 41:01derivatives
- 41:02uh including for example spatial
- 41:04temporal planes. You can have time
- 41:06derivative and a spatial derivative.
- 41:09And it turns out that this uh term here
- 41:11where you sum over alpha which is X and
- 41:14Y and you sum over I which is X, Y and
- 41:17Z. This term is compatible with all the
- 41:20symmetries uh uh
- 41:22uh of the uh uh
- 41:24uh point group.
- 41:26And so this term is in principle allowed
- 41:28in our system. So that's another
- 41:29topological term. It's kind of a
- 41:31skyrmion number in the spatial temporal
- 41:33planes instead of skyrmion number in the
- 41:36spatial plane XY.
- 41:38Right? And again this will be mostly a
- 41:39silent term which which will have no uh
- 41:43influence on on the classical dynamics
- 41:46of spins, but it might do something
- 41:48else.
- 41:49And what it does is it endows uh
- 41:52solitons uh in antiferromagnets with
- 41:55spin.
- 41:56So solitons turn out to be kind of
- 41:58ferromagnetic islands in a way.
- 42:00And there's a well-known
- 42:02uh
- 42:03uh
- 42:04uh counterpart of that in spin chains
- 42:06where the topological term written by
- 42:08Duncan Haldane in 1 + 1 dimension where
- 42:12it's a skyrmion number in the TX plane.
- 42:15Uh this is a stop his topological term.
- 42:18Little S is the length of spin.
- 42:20Uh it turns out that this topological
- 42:22term
- 42:23uh
- 42:24basically uh gives spin
- 42:27uh to inhomogeneities of of the nail
- 42:29vector.
- 42:30Now you can derive that uh
- 42:32from from this form or from that form
- 42:35where you can see that the uh angular
- 42:37frequency of rotation couples uh to the
- 42:41uh
- 42:41uh gradient of the nail vector. And from
- 42:44that you can derive that the spin
- 42:46density has a term proportional to the
- 42:48derivative of the nail vector.
- 42:51So if you have a domain wall where the
- 42:53nail vector changes sign, you'll find uh
- 42:56that a domain wall carries spin plus S
- 42:59or minus S
- 43:00and a chain end has a spin of plus half
- 43:03S or minus half S.
- 43:05So, uh this is why spin ons in the XYZ
- 43:09spin one half chain carry spin plus or
- 43:11minus half
- 43:12and edges uh
- 43:13of spin chains can bind uh a spin equal
- 43:16to one quarter.
- 43:18Uh
- 43:19if you haven't heard of spins one
- 43:20quarter previously, there are actually
- 43:23very recent papers by uh Nathan
- 43:26Nathan Andres group where they uh
- 43:29discuss in detail uh these uh S over two
- 43:32spins uh at the edges.
- 43:36So,
- 43:37in the
- 43:39this is a topological term that's
- 43:41allowed and it also has uh
- 43:44this sort of divergence of of the uh
- 43:48uh
- 43:49of the uh spin frame vectors coupled to
- 43:52the angular uh frequency of rotation and
- 43:55that creates a spin density uh
- 43:57proportional to the divergence of of N.
- 44:00Uh
- 44:01and so there should be linear spin
- 44:02density uh
- 44:04accumulated on domain walls.
- 44:06And
- 44:07if you're not into field theory, there's
- 44:09a very simple uh
- 44:11lattice picture that explains how that
- 44:13happens. Let's start with one dimension.
- 44:15Here's a an antiferromagnetic spin
- 44:17chain. Every spin has length S where in
- 44:21the extreme using limit where domain
- 44:23walls are very sharp, basically once one
- 44:25lattice spacing and they correspond to a
- 44:28couple of spins pointing in the same
- 44:29direction.
- 44:30So, these two spins uh
- 44:33uh
- 44:34are kind of an
- 44:35a ferromagnetic island in the sea of
- 44:37antiferromagnetic spins. So, if you blur
- 44:40your vision and you don't view things on
- 44:42the lattice scale, this is where
- 44:44there'll be a net spin and the amount of
- 44:46spin is minus S here and it's plus S
- 44:49here.
- 44:50Now,
- 44:51picture a a wall uh in a
- 44:55So you have one ground state on the left
- 44:58shown in red and you have another ground
- 45:00state on the right shown in blue
- 45:03and these are time reversed copies of
- 45:05the two ground states. So here I assume
- 45:07a local anisotropy that
- 45:11prefers spins pointing
- 45:13to or from centers of triangles like
- 45:15this. So this is one ground state and
- 45:17that's the other ground state.
- 45:20It's something similar to what happens
- 45:22in manganese 310
- 45:23where it's slightly more complicated.
- 45:27So these two ground states can have
- 45:30domain walls of two varieties between
- 45:32them. One domain wall will have the
- 45:34spins that are oriented
- 45:37more or less to the left and they're
- 45:39kind of part of the red domain if you
- 45:41will, right? So this triangle is is
- 45:43exactly in this ground state and you can
- 45:46time reverse them
- 45:47and get this state where the the spins
- 45:49on the domain wall kind of belong to the
- 45:52to the right domain.
- 45:54And these two
- 45:56domain walls carry a spin that
- 46:00points to the left
- 46:02here and to the right there. You can
- 46:04calculate the linear spin density
- 46:07from from this naive picture and you
- 46:09will get exactly the same result as the
- 46:10field theory predicts.
- 46:12Or here is a horizontal domain wall
- 46:15where you have a domain on the top and
- 46:17domain on the bottom
- 46:19and in between the spins can point all
- 46:21down or they can point all up for
- 46:23example. And again the linear spin
- 46:25density will be the same
- 46:27except it now points in a different
- 46:28direction.
- 46:29And so the prediction of field theory is
- 46:33compatible with with this very simple
- 46:36extreme loss limit.
- 46:39We have numerical evidence in the
- 46:40quantum antiferromagnetic Ising model
- 46:44in a well, sorry, quantum
- 46:47model of this antiferromagnet
- 46:49and its work in progress done in
- 46:52collaboration with Frank Pollmann.
- 46:54This is possible uh it's possible to
- 46:56extend this to 3D plus one dimensions.
- 46:59So pyrochlore, all in and all out,
- 47:01antiferromagnet should also have domain
- 47:03walls that that
- 47:05uh carry spin and work as ferromagnetic
- 47:08islands.
- 47:09So here's a summary of what I've told
- 47:11you. We have a shiny new field theory
- 47:13for three sublattice antiferromagnets.
- 47:15The order parameter is a rigid body. You
- 47:17can describe it in terms of uh spin
- 47:19frame uh which is the analog of the body
- 47:23frame uh in rigid body dynamics. There
- 47:26are four bulk coupling constants,
- 47:28some topological terms, Pythagorean
- 47:31relation for the three magnum
- 47:32velocities, uh elliptical vertices and
- 47:35kagome,
- 47:36and control of the order parameter by by
- 47:38magnets. Thank you very much for your
- 47:40attention.
- 47:41>> Thank you. Thank you, Oleg. I was always
- 47:44It's always a pleasure to listen to. I
- 47:45still remember your black uh
- 47:48your uh your chalkboard talks that I
- 47:50liked to
- 47:51>> it would be hard to give online.
- 47:54>> Yeah, yeah, the the but this one they
- 47:56know it's just amazing.
- 47:58Um one thing uh so I'll start with the
- 47:59questions and please raise your hands uh
- 48:02to ask questions to Oleg um and I'll
- 48:04bring you up to the
- 48:06uh to promote you.
- 48:08Hey, but I'll start with a a great
- 48:09question. Um so particularly looking at
- 48:12these modes, these modes that are on the
- 48:13edge, the extra modes that you've found
- 48:15out with the coin
- 48:17uh
- 48:18I mean, I assumed that you're still
- 48:20working on observing them of how
- 48:21sensitive it would be. How sensitive is
- 48:23that to to edges to essentially uh when
- 48:26you have multi-domain systems, you know,
- 48:28frustrated systems tend to have a small
- 48:29domains.
- 48:31Um
- 48:31>> Right. So so if you have grain
- 48:34boundaries, uh you're going to have uh
- 48:37edge modes uh
- 48:39uh localized on grain boundaries as
- 48:41well.
- 48:42Uh
- 48:43so so these these modes are pretty
- 48:45robust. You can have a sample of any
- 48:47shape.
- 48:48It doesn't have to be a straight edge.
- 48:50As long as as the curve is smooth, it's
- 48:53going to be very stable.
- 48:55And when the curve is not smooth, when
- 48:57you have edge roughness, uh there will
- 49:00be scattering. Uh these these waves
- 49:02would be scattered in the bulk.
- 49:04Uh and they will have a finite lifetime.
- 49:07But the lifetime will increase quite
- 49:09quickly as you go to longer wavelengths.
- 49:11So parts of the spectrum should uh
- 49:14survive, particularly the long
- 49:15wavelength parts.
- 49:17Uh we we we
- 49:19Grigora Deman has a preprint on on
- 49:21archive uh that that uh uh
- 49:24deals with that and he showed that it's
- 49:26pretty stable.
- 49:28>> Do you do you have any material
- 49:29candidates that are the best possible
- 49:31choices for this?
- 49:33>> Yes. Uh
- 49:35when I was visiting the University of
- 49:36Minnesota, uh it was pointed out to me
- 49:38that there's a compound uh with the ABC
- 49:42stacking
- 49:43and uh two different uh
- 49:46uh ligands uh that that alternate
- 49:49uh that could actually do that. Uh
- 49:52I do not have the name of this compound
- 49:55off the top of my head except to say
- 49:56that
- 49:57uh the magnetic ion is rare earth, so it
- 49:59has a DC of like a Kelvin.
- 50:02Uh so I think that's a little bit low.
- 50:06[laughter]
- 50:07So one one should look for for uh
- 50:09transition metal.
- 50:10Um but but yeah.
- 50:12>> difficult with the non-collinear
- 50:13systems. They they don't tend to have
- 50:15very high uh
- 50:16>> Well
- 50:18No, manganese
- 50:19Manganese has good enough, but uh
- 50:22>> Uh but
- 50:22>> not triangular. So you need triangular
- 50:24lattice instead of kagome in that case.
- 50:27>> Um Vladimir
- 50:29Can you please go ahead and ask your
- 50:30question? Uh
- 50:32Okay. Let me just have yeah.
- 50:35>> Uh
- 50:35Hi Oleg. Nice to see you. Hi Kairo.
- 50:38>> Hi.
- 50:40>> Uh
- 50:41I have a couple of questions and all of
- 50:43them are related to topological solitons
- 50:45in this model,
- 50:46in particular vortices which you have
- 50:48shown.
- 50:49Um so in continuum models for
- 50:52ferromagnets or antiferromagnets uh
- 50:55vortices has a core
- 50:57and size of this core is determined by
- 51:00competition between exchange and
- 51:02with the planar anisotropy.
- 51:05So the question is, I think your
- 51:08vortices do not have cores because you
- 51:10have shown the discrete lattice, but in
- 51:13principle, how
- 51:15is it possible to add in addition to
- 51:18exchange interaction which you have in
- 51:20your model to add additional
- 51:22interactions, anisotropy, magnetic
- 51:24field?
- 51:25>> Yeah.
- 51:25>> And this would significantly enrich the
- 51:28zoo of topological solitons which you
- 51:30can obtain in the solutions.
- 51:32>> Of course. Yes. Yeah, indeed, I I showed
- 51:36vortices
- 51:38in the Heisenberg model which is
- 51:40completely isotropic.
- 51:43And these are stable unlike vortices in
- 51:46the
- 51:47ferromagnetic Heisenberg model where
- 51:49they could unwind by going into the
- 51:51third dimension.
- 51:52In this case they can't.
- 51:54They're they're called Z2 vortices. They
- 51:57were introduced by Kawamura in the 1980s
- 52:00and they're stable for topological
- 52:01reasons because
- 52:03in the SO3 manifold a rotation by 2 pi
- 52:08is a path that cannot be continuously
- 52:11shrunk to a point.
- 52:13It it's
- 52:15there are two kinds of closed paths in
- 52:18SO3 and this is a non-contractible one.
- 52:21So so these are these are stable even
- 52:23without
- 52:24XY and anisotropy.
- 52:26Now you can add XY anisotropy, for
- 52:28example, easy plane, and then of course
- 52:30your your vortices will be
- 52:32will stay in the plane
- 52:35but switch out of the plane
- 52:38when you know you come to this toward
- 52:40the center of the core. There will be
- 52:43a certain length scale determined by
- 52:44again the competition of exchange and
- 52:47and anisotropy energy. Yes.
- 52:49>> But even in this continuum model, don't
- 52:52you have a singularity, the divergence
- 52:54of energy in the vortex?
- 52:55>> is there is a singularity and in fact
- 52:58the order parameter cannot survive there
- 53:00just like the order parameter cannot
- 53:02survive in the XY model
- 53:04uh
- 53:05at the center of the core, right? And so
- 53:07the spins have to tilt out of the plane.
- 53:09Uh here it's more interesting because it
- 53:12turns out that
- 53:14this is a true singularity like the
- 53:15singularity at the center of a hedgehog
- 53:18uh in in in three-dimensional
- 53:20ferromagnets. Uh and so the order
- 53:22parameter has to die
- 53:24uh
- 53:25which means means that uh you'll you'll
- 53:27have
- 53:28uh
- 53:29you know you will not have a perfect
- 53:30120° structure at the center of the
- 53:33core.
- 53:34And I can speculate what could happen
- 53:37there.
- 53:38Uh you can think of the three
- 53:40sublattices as forming
- 53:42each will form a vortex of its own.
- 53:45Right? So it's like having three
- 53:47ferromagnetic vortices uh
- 53:50uh on top of one another.
- 53:53And the interesting thing is of course
- 53:55when you have a vortex uh
- 53:58in an easy plane model in a ferromagnet,
- 54:00it has a core where spins point uh in
- 54:03the perpendicular direction.
- 54:04Right? So here the three vortices also
- 54:07will have cores.
- 54:09And curiously it uh
- 54:13the the the three cores cannot point all
- 54:15in opposite directions, right?
- 54:17Two of them will point up and one will
- 54:19point down or vice versa.
- 54:21Uh and that means that there'll be a
- 54:23non-zero net Skyrmion number.
- 54:26And you know what that means. It means
- 54:27that there'll be a gyroscopic force
- 54:29acting on on this vortex.
- 54:33In in an antiferromagnet, in a collinear
- 54:35antiferromagnet, the gyroscopic force
- 54:38cancels out because the the two
- 54:40components of you know the the two
- 54:41sublattices
- 54:42have opposite Skyrmion numbers. And here
- 54:45they will have Skyrmion numbers half,
- 54:46half, and minus half. They will not
- 54:48cancel out. So there should be a net
- 54:50gyroscopic force in this case.
- 54:52>> So this was my next question and this is
- 54:54the last one I promise. So have you
- 54:56already developed the collective
- 54:58variable approach to describe
- 55:00>> we we know each other well enough,
- 55:01Vladimir.
- 55:05Uh
- 55:07I have not developed that approach yet.
- 55:08I was student who should be working on
- 55:11that.
- 55:13And and and so
- 55:15I hope to develop it.
- 55:17But
- 55:19we haven't yet.
- 55:22>> Thank you.
- 55:23>> Okay.
- 55:25Is there any further questions from the
- 55:26audience? I see immediately but you know
- 55:30please
- 55:31contact. I'm going to stop here now the
- 55:33the live streaming for the for the
- 55:36for the 30 people that were on YouTube
- 55:38and then the
- 55:40the other 45 or 50 that you guys were
- 55:42online. Thank you for
- 55:44becoming and thank you like by the way I
- 55:45have another question now that is not
- 55:47being taped
- 55:51so so so the the bottles that you have
About this transcript
This page contains the full transcript of On-line SPICE-SPIN+X Seminar: Oleg Tchernyshyov by SPICE-JGU-Mainz, generated from the public captions YouTube serves with the video. The transcript has 8,190 words across 1,510 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.