YouTube2Text

On-line SPICE-SPIN+X Seminar: Oleg Tchernyshyov — Transcript

by SPICE-JGU-Mainz · 8,190 words · 1,510 segments · language en · Watch on YouTube

Full transcript

  1. 0:02Just want to make sure that because it's
  2. 0:0330 seconds delay, but I have to
  3. 0:06mute it in the other one if
  4. 0:0920 are waiting to hear us.
  5. 0:12Uh because essentially some some are on
  6. 0:14the on the Zoom and some are on the
  7. 0:17Let's take a look.
  8. 0:20Okay.
  9. 0:21Great. Um
  10. 0:24I think we have now the
  11. 0:26the system and I've put it on
  12. 0:29here.
  13. 0:31So I would like to welcome
  14. 0:34everybody for the to the online
  15. 0:37spintronics seminar series.
  16. 0:39Uh as you know, this is a webinar
  17. 0:41format, so please write your questions
  18. 0:43in the chat
  19. 0:44uh
  20. 0:45or the raise your hands and I will give
  21. 0:47you the floor after the speaker
  22. 0:50um finishes, but also if it's important
  23. 0:53I can interrupt and
  24. 0:55ask for clarification.
  25. 0:57As always, we we have it on Wednesday at
  26. 0:593:00 p.m. on the 1500
  27. 1:02hours German time.
  28. 1:05But it's not regular every you get a
  29. 1:07notifications. Next month will be in
  30. 1:09fact on December the second,
  31. 1:12but at least it's always on Wednesday at
  32. 1:14that given time.
  33. 1:17Uh and of course I do the plan of this
  34. 1:20seminar with the help of of these folks
  35. 1:22over there.
  36. 1:23Today is a pleasure my good old friend
  37. 1:25Oleg
  38. 1:27I never pronounce your name.
  39. 1:35Ternishiov
  40. 1:36it's just it's just impossible. It's
  41. 1:38like my my first last name my first name
  42. 1:40um is very well known in our community
  43. 1:42in both two communities in the community
  44. 1:45of spintronics and spintronics as well
  45. 1:48as
  46. 1:49systems are quite a proud physicist.
  47. 1:51He started his PhD in the Columbia
  48. 1:53University in New York
  49. 1:54and didn't move that far. Went down
  50. 1:56south a little bit to go to Princeton
  51. 1:59for a postdoc and then now I'm in Oak
  52. 2:00Ridge
  53. 2:01for quite a while start looking at
  54. 2:04correlated systems, magnetic systems and
  55. 2:06the dynamics of solitons and skyrmions
  56. 2:08and all kind of topological objects
  57. 2:11as well as spin liquids.
  58. 2:13Um one of the worst to highlight as well
  59. 2:15many is the the piece of fellow that may
  60. 2:16come for the society
  61. 2:18since 2019.
  62. 2:20With this I'm going to stop sharing
  63. 2:23the
  64. 2:24and allow you know ask you to go ahead
  65. 2:25and share your
  66. 2:27slides and
  67. 2:30just start anytime.
  68. 2:33>> Thank you Phayro.
  69. 2:34Um
  70. 2:36>> Yeah see that.
  71. 2:38>> Okay.
  72. 2:39>> Perfect.
  73. 2:40>> All right.
  74. 2:41Good.
  75. 2:43All right. Thank you so much for having
  76. 2:44me.
  77. 2:45It's my second talk at
  78. 2:48Spin X.
  79. 2:50Wonderful audience.
  80. 2:52I'm excited to be here.
  81. 2:55As
  82. 2:56luck would have it today my brother and
  83. 2:58his family are visiting Mines
  84. 3:01and I told them that I will be giving a
  85. 3:03talk on Mines and they were thrilled
  86. 3:05but then I told them that this would be
  87. 3:07online.
  88. 3:09Nonetheless here we go.
  89. 3:11I am going to tell you about
  90. 3:14new field theory for three sublattice
  91. 3:16antiferromagnets
  92. 3:18where the order parameter is not a
  93. 3:20simple vector. It's a rigid body.
  94. 3:23It's a field theory with
  95. 3:25four coupling constants
  96. 3:27and otherwise it's pretty universal.
  97. 3:31There's there are some experimental
  98. 3:33consequences that I will touch upon here
  99. 3:36and we'll see how much I will be able to
  100. 3:39convey.
  101. 3:40Let me thank the people with whom I
  102. 3:43interacted over the years.
  103. 3:45I want to credit Satoru Nakatsuji for
  104. 3:49kind of bringing together two of my
  105. 3:51topics of interest
  106. 3:55frustrated magnets on which I worked
  107. 3:58in some distant past
  108. 4:01and spintronics.
  109. 4:03So Satoru is credited with
  110. 4:05uh
  111. 4:07uh resurrecting interest to manganese
  112. 4:09three tin, a compound that was
  113. 4:10discovered in the 1980s uh but that that
  114. 4:14came to prominence more recently because
  115. 4:16of the
  116. 4:17uh
  117. 4:18uh conducting nature of that magnet and
  118. 4:21uh uh potential in spintronics
  119. 4:24through the anomalous Hall effect that
  120. 4:26it exhibits that's related to its
  121. 4:27magnetic order.
  122. 4:29Uh Colin Broholm is my colleague
  123. 4:31experimentalist here at Hopkins
  124. 4:34uh who investigated manganese three tin
  125. 4:35with Satoru.
  126. 4:37Uh uh
  127. 4:38Boris Ivanov uh
  128. 4:41is
  129. 4:42someone who's been in the field pretty
  130. 4:44much longer than anybody uh uh
  131. 4:46else
  132. 4:47in this audience, I think.
  133. 4:50Hua Chen was on a sabbatical here at
  134. 4:52Hopkins well when we discussed things
  135. 4:54and I have my wonderful former students
  136. 4:56Sayak Bastian and and Grigor
  137. 4:59who worked on on on these systems.
  138. 5:02Here's a list of papers that we
  139. 5:05put out recently
  140. 5:07and if you don't want to read all these
  141. 5:09papers uh I I would suggest reading a
  142. 5:13piece by Sid Parameswaran
  143. 5:15in the Condensed Matter Journal Club in
  144. 5:17May 2025 which gives a very quick and
  145. 5:20very nice overview of of this
  146. 5:23spin frame field theory.
  147. 5:26Uh so let's jump into field theories of
  148. 5:29of magnets. The first field theory of a
  149. 5:32ferromagnet was uh
  150. 5:34uh written down by Landau and Lifshitz
  151. 5:36who had a very
  152. 5:38uh practical interest in understanding
  153. 5:41uh the AC susceptibility of of
  154. 5:44ferromagnets like iron and nickel.
  155. 5:47And so they came up with
  156. 5:49a field theoretic description that would
  157. 5:51allow them to
  158. 5:52to deal with the physics of domain
  159. 5:54walls. Domain walls are magnetic
  160. 5:56solitons. Dealing with soliton in a
  161. 5:59lattice is a very difficult problem. If
  162. 6:01you switch to a continuum description,
  163. 6:03it's much easier and so that's what what
  164. 6:05they did.
  165. 6:07So they introduced the
  166. 6:10order parameter as a unit vector spin
  167. 6:12field
  168. 6:14M.
  169. 6:15If you multiply that by the spin density
  170. 6:19number, you get
  171. 6:22on the left hand side
  172. 6:24the derivative of
  173. 6:29the density of angular momentum and on
  174. 6:31the whatever potential energy
  175. 6:33functional.
  176. 6:34And for the Heisenberg ferromagnet, for
  177. 6:37instance, the functional would be simply
  178. 6:39the gradient squared
  179. 6:42of the
  180. 6:44magnetization or spin
  181. 6:46field.
  182. 6:47And so you have a
  183. 6:49fairly simple equation that
  184. 6:51you can
  185. 6:53investigate.
  186. 6:55Spin waves
  187. 6:56are elementary excitations in
  188. 6:58ferromagnets and they look pretty much
  189. 7:00like this. You have spins that process
  190. 7:04in the clockwise direction if you look
  191. 7:06from from above
  192. 7:08and
  193. 7:12that's that's a signature of
  194. 7:14its precessional rather than inertial
  195. 7:16and it takes some unlearning
  196. 7:20you know, the intuition we have with
  197. 7:22particles who whose equations of motion
  198. 7:24are second order in time.
  199. 7:27This is completely different.
  200. 7:29Now, another theory,
  201. 7:31field theory of of
  202. 7:32a magnet was
  203. 7:34written down in the late 1970s and early
  204. 7:371980s
  205. 7:39by Boris Ivanov,
  206. 7:41Andre Ivan Marchenko, Duncan Haldane
  207. 7:43among others.
  208. 7:44And
  209. 7:46their equation of motion,
  210. 7:48as you can see here, is second order in
  211. 7:50the time derivatives. Here N is a unit
  212. 7:52vector field of the staggered spin or
  213. 7:55nail vector
  214. 7:56which describes the spins on one
  215. 7:58lattice, let's say the red one, and the
  216. 8:00spins on the other sublattice point
  217. 8:02exactly opposite to that. So you can
  218. 8:04still encode all of the spins in the
  219. 8:06antiferromagnet with a single vector.
  220. 8:09So the equation of motion looks like
  221. 8:10this. You can see the right hand side
  222. 8:12barely changed. Uh you still have
  223. 8:15uh conservative torque density on the
  224. 8:17right hand side. The left hand side
  225. 8:19changed. Uh this is now the density of
  226. 8:22angular momentum differentiated with
  227. 8:24respect to time. So you have a uh sort
  228. 8:26of similar equation, but it's now second
  229. 8:28order in in the time derivative.
  230. 8:31And in fact, the energy functionals look
  231. 8:33very similar. Instead of the uh
  232. 8:36square of the
  233. 8:38uh derivative of the gradient of
  234. 8:40magnetization, you have the
  235. 8:42uh the same for the uh nail vector.
  236. 8:46So spin waves in an antiferromagnet look
  237. 8:48a bit different. You can have linear
  238. 8:50polarizations. Both sublattices would
  239. 8:52like to precess, but each in its own
  240. 8:54direction. And so precession basically
  241. 8:56cancels out. And what remains is this
  242. 8:59sort of inertial dynamics where spins
  243. 9:01can have linear polarizations like
  244. 9:03these, or you can make circular
  245. 9:05polarizations out of them
  246. 9:06uh going uh clockwise or
  247. 9:09counterclockwise. And and these uh spin
  248. 9:11waves would have the same frequencies.
  249. 9:15So a field theory uh
  250. 9:17of uh a nail antiferromagnet
  251. 9:20uh can be derived from a Lagrangian. And
  252. 9:22the Lagrangian would have uh
  253. 9:24the kinetic energy term, which is the
  254. 9:27time derivative of the nail vector
  255. 9:30squared times some density.
  256. 9:32And then it would have a potential
  257. 9:33energy term, which in the Heisenberg
  258. 9:35antiferromagnet looks uh like this.
  259. 9:40Uh you could do this as well uh with uh
  260. 9:43ferro magnets, although instead of the
  261. 9:45kinetic energy would have a berry phase
  262. 9:46term.
  263. 9:48Uh in the late '80s uh there was
  264. 9:51interest uh to uh frustrated uh
  265. 9:54antiferromagnets. Uh
  266. 9:57in particular, one of the questions that
  267. 9:58that was uh
  268. 10:01uh on people's mind whether
  269. 10:03uh the field theory of of uh
  270. 10:06antiferromagnets
  271. 10:07uh had a topological term like it
  272. 10:10uh in uh
  273. 10:11one dimension.
  274. 10:13Haldane gap phenomenon is related to
  275. 10:15that and we'll discuss that briefly
  276. 10:18later on in the talk.
  277. 10:20Also people were
  278. 10:21trying to see whether topological terms
  279. 10:23also exist
  280. 10:25on the square lattice, which turned out
  281. 10:28not to be the case
  282. 10:30and then Dubail and Read investigated
  283. 10:32the question on the triangular lattice
  284. 10:36whether there is a topological term and
  285. 10:38maybe a difference between integer and
  286. 10:40half integer spins, the spectrum might
  287. 10:42be different and so on and so forth.
  288. 10:45And so they had to write down a field
  289. 10:47theory
  290. 10:48on the triangular lattice
  291. 10:49antiferromagnet and in this case because
  292. 10:52the ground state is not collinear you
  293. 10:54can you cannot use a single vector to
  294. 10:56describe it, but you have to use
  295. 10:57something else and that something else
  296. 11:00happens to be
  297. 11:02a rigid body. The order parameter which
  298. 11:04you can think of three
  299. 11:06spins from the three sublattices
  300. 11:09that look like a Mercedes-Benz logo,
  301. 11:12you can think of this as a rigid body
  302. 11:14and so the dynamics of the order
  303. 11:16parameter would resemble vibrations of
  304. 11:18this
  305. 11:20order parameter
  306. 11:22in in various directions,
  307. 11:25perhaps rotations of this order
  308. 11:26parameter in various directions. So you
  309. 11:29have to write a field theory in terms of
  310. 11:31this order parameter and one way to do
  311. 11:33that is in terms of SO(3) rotation
  312. 11:35matrices that determine the orientation
  313. 11:39of this Mercedes-Benz logo.
  314. 11:41So if you use matrix O
  315. 11:44an SO(3) matrix then this is your field
  316. 11:46theory. It looks a bit more complicated
  317. 11:49and and anharmonic compared to
  318. 11:53what you saw previously,
  319. 11:55but you can still look for a topological
  320. 11:57term in there and
  321. 12:01the spoiler here is that Dubail and Read
  322. 12:03did not find a topological term.
  323. 12:05So that was the end of that
  324. 12:08and
  325. 12:10field theory of of triangular
  326. 12:12antiferromagnets
  327. 12:14was was pretty much not
  328. 12:16uh
  329. 12:17in the vogue anymore.
  330. 12:18So we resurrected it
  331. 12:20a few years ago 2024
  332. 12:23with Bastian Prudenus who is now in
  333. 12:26Dresden
  334. 12:28and and Grigora Demian who came later
  335. 12:31and is now at MIT.
  336. 12:33And we used a different object
  337. 12:37mathematical object to describe this
  338. 12:39field theory
  339. 12:40instead of using matrix which is kind of
  340. 12:42unintuitive. If you give me a matrix
  341. 12:44I have no idea what the orientation it
  342. 12:46describes. It's you know it's nine
  343. 12:47numbers that that that I cannot read
  344. 12:50into.
  345. 12:51So instead we use a different order
  346. 12:52parameter.
  347. 12:53If you have a rigid body
  348. 12:55a rigid body has principal axes and you
  349. 12:59can direct three unit vectors along
  350. 13:01these principal axes and these three
  351. 13:02unit vectors can serve as as a body
  352. 13:05frame
  353. 13:07that rotates along with the body. So
  354. 13:09here we just define a spin frame
  355. 13:12which is defined in terms of
  356. 13:15the three sub lattice magnetizations.
  357. 13:17You can think of an X and an Y as the
  358. 13:20two staggered magnetizations that that
  359. 13:22you have
  360. 13:23right? You can again take the difference
  361. 13:25of of two sub lattices
  362. 13:27and maybe another difference of two sub
  363. 13:29lattices or the difference between the
  364. 13:31three sub lattices or something like
  365. 13:32that
  366. 13:33and these two vectors will be orthogonal
  367. 13:34and unit vectors if you properly
  368. 13:36normalize them.
  369. 13:38A third vector can be obtained just by
  370. 13:39taking the cross product of these two
  371. 13:42and so that gives you
  372. 13:43a frame.
  373. 13:45In fact
  374. 13:46this parametrization relates in a simple
  375. 13:48way to the SO3 matrix parametrization.
  376. 13:51If you take the three vectors of the
  377. 13:53body frame
  378. 13:55and three vectors of a global fixed
  379. 13:57frame their scalar products will
  380. 13:59feature nine numbers and these nine
  381. 14:01numbers are exactly the SO3 matrix of
  382. 14:03Dummer and Reed.
  383. 14:06So why do we call these
  384. 14:10vectors NX and NY? Well this has to do
  385. 14:13with directions in the lattice, rather
  386. 14:16than in spin space.
  387. 14:18Uh and uh
  388. 14:20we call them an x and y because they
  389. 14:21have
  390. 14:23uh
  391. 14:23symmetry properties similar to
  392. 14:25coordinates x and y in the lattice.
  393. 14:28Uh and
  394. 14:29here's how it goes. Uh let's take a
  395. 14:31lattice,
  396. 14:33uh kagome lattice in this case, uh with
  397. 14:35the x and y axes in the plane, the z
  398. 14:38axis perpendicular to it.
  399. 14:40Uh
  400. 14:41and the point group of this lattice
  401. 14:43involves
  402. 14:45uh several operations of the uh group
  403. 14:47D3. Among them, let's say this rotation
  404. 14:51about the y axis, in which uh you swap
  405. 14:54the two sublattices, one and two, for
  406. 14:56example, and sublattice three stays uh
  407. 14:59uh itself.
  408. 15:00So the magnetization vectors of one and
  409. 15:03two and three will change into m two, m
  410. 15:05one, and three.
  411. 15:07And that would flip the sign of an x but
  412. 15:09keep the sign of an y.
  413. 15:11So you can see that an x and an y will
  414. 15:13transform into minus an x and an y. And
  415. 15:15it's the same way in which any uh
  416. 15:18spatial vector transforms under a C2 y
  417. 15:20rotation.
  418. 15:21And that's important because we will
  419. 15:23need the properties of an x and y under
  420. 15:25these uh point group transformations.
  421. 15:28And that's in the spirit of Andrei von
  422. 15:30Marchenko, this famous uh
  423. 15:33but poorly understood 1980 paper.
  424. 15:36So knowing the symmetry properties of uh
  425. 15:39of our
  426. 15:40uh
  427. 15:42water parameters uh or the uh frames
  428. 15:46vectors n x and y and then z enables us
  429. 15:49to construct uh a simple exchange uh uh
  430. 15:54energy density
  431. 15:55which has to be uh of course invariant
  432. 15:58under uh spatial symmetries uh of the
  433. 16:02lattice as well as the uh
  434. 16:05uh global spin rotation symmetry if it's
  435. 16:08the Heisenberg
  436. 16:10So what comes out is uh
  437. 16:13uh something similar to the uh
  438. 16:15uh
  439. 16:17collinear antiferromagnets where you had
  440. 16:19gradients of n squared, but here we have
  441. 16:22several n's. We have an x and an y.
  442. 16:25For example, n z
  443. 16:28is not relevant in this context, but you
  444. 16:30can
  445. 16:32include several different
  446. 16:35pairings between the indices,
  447. 16:38partial x and x for example, or partial
  448. 16:41y and y and so on. And it turns out that
  449. 16:43these are the three possible pairings
  450. 16:46that survive the symmetry constraints.
  451. 16:49So it's possible to have three
  452. 16:51distinct terms in your energy density.
  453. 16:55So it's slightly more complicated than
  454. 16:57than the collinear case, but nonetheless
  455. 16:59manageable.
  456. 17:00And it's actually similar to the elastic
  457. 17:02energy of a hexagonal solid
  458. 17:05written in terms of the atomic
  459. 17:07displacement field, which also has these
  460. 17:10indices.
  461. 17:11And so you can write it in this way. And
  462. 17:14the constants lambda mu and new would be
  463. 17:16the combinations of bulk modulus and
  464. 17:19shear modulus. The difference between mu
  465. 17:22and new would be zero for a solid
  466. 17:24because it would describe a rotation.
  467. 17:27So it's not exactly an exotic
  468. 17:30energy density. One can understand where
  469. 17:32it comes from.
  470. 17:34And so one can add
  471. 17:35kinetic energy like in a collinear
  472. 17:38antiferromagnet, which has this form.
  473. 17:40And this time you have to use
  474. 17:42all all three vectors of x and y and z.
  475. 17:45And what this basically is is omega
  476. 17:48squared, the rotation frequency squared
  477. 17:52of of the
  478. 17:54of the order parameter.
  479. 17:57Okay, so we've got a field theory with
  480. 18:00four coupling constants. One is the
  481. 18:02inertia, and there are three
  482. 18:03stiffnesses.
  483. 18:05And they determine the the spectrum of
  484. 18:07spin waves for example.
  485. 18:09So you can derive the velocity of the
  486. 18:11spin waves,
  487. 18:13two of which are
  488. 18:15rocking your
  489. 18:17order parameter out of its plane, and
  490. 18:19the third one is keeping it in the
  491. 18:21plane. I'll show an illustration of
  492. 18:23that. What matters here is
  493. 18:26the speeds can be calculated from
  494. 18:29the wave equations, and they feature
  495. 18:32these combinations of the stiffnesses in
  496. 18:35the numerator and
  497. 18:38inertia in the denominator.
  498. 18:40Now what's interesting is
  499. 18:43the the the first and third stiffness
  500. 18:45coefficients, lambda and nu,
  501. 18:47come together as a combination lambda
  502. 18:49plus nu
  503. 18:51here as well as here,
  504. 18:54but never
  505. 18:55on the road.
  506. 18:56So it means that effectively lambda and
  507. 18:58nu do the same thing
  508. 19:01and instead of having three stiffnesses,
  509. 19:04we actually have two.
  510. 19:06And that means that the three spin wave
  511. 19:08velocities are not independent,
  512. 19:11but if you square them and adopt the
  513. 19:13first two,
  514. 19:14you get the square of the third one. So
  515. 19:16there's this lovely Pythagorean relation
  516. 19:18that is experimentally verifiable.
  517. 19:21So that's one experimental consequence
  518. 19:24of this field theory, and
  519. 19:26there will be more.
  520. 19:27So here are the the spin waves. So you
  521. 19:30can imagine what what they look like.
  522. 19:32You have
  523. 19:33two linear polarizations, this one and
  524. 19:35that one, and and you have a third
  525. 19:37linear polarization that
  526. 19:39basically these are oscillations about
  527. 19:42different axes,
  528. 19:45different principal axes in in this
  529. 19:47rigid body.
  530. 19:49And of course you can form not only
  531. 19:51linear
  532. 19:52polarizations, but you can also form
  533. 19:54circular polarizations, and that will be
  534. 19:56a story down the road.
  535. 19:59Okay. So the exchange energy density has
  536. 20:01these terms.
  537. 20:05Lambda and nu terms
  538. 20:07have the SO(2) rotational symmetry. You
  539. 20:10can rotate
  540. 20:12sort of the components of the gradient
  541. 20:14as well as make combinations of an x and
  542. 20:17y.
  543. 20:19And
  544. 20:21these two terms have just this SO2
  545. 20:22symmetry
  546. 20:24because you rotate the gradient
  547. 20:26components as well as
  548. 20:28you do the
  549. 20:30transformation between an X and Y.
  550. 20:34The mu term actually has more symmetry
  551. 20:36because here you can rotate separately
  552. 20:38components of the gradient which are
  553. 20:40paired
  554. 20:41and you can separately rotate an X and
  555. 20:43an Y components
  556. 20:45of the
  557. 20:46spin frame. And so this symmetry is
  558. 20:48higher. It's SO2 cross SO2.
  559. 20:51It turns out the the triangular lattice
  560. 20:53only has these terms
  561. 20:55and the lambda and mu terms are not
  562. 20:57allowed. So for the triangular lattice
  563. 20:59the field theory simplifies. You have a
  564. 21:01much more symmetric form than
  565. 21:04this field theory allows. And on Kagome
  566. 21:08if you only include nearest neighbor
  567. 21:09interactions then it turns out that mu
  568. 21:11vanishes and lambda plus mu is non-zero.
  569. 21:15More generally Kagome with longer range
  570. 21:17couplings will have all of these
  571. 21:20constants, but the triangular lattice
  572. 21:22only has mu no matter how many
  573. 21:25nearest and second neighbors you include
  574. 21:26and so on.
  575. 21:30All right. So one consequence of that
  576. 21:33the higher symmetry of the triangular
  577. 21:35lattice antiferromagnet is that if you
  578. 21:37view vertices
  579. 21:40of the order parameter
  580. 21:42they actually are nicer and more
  581. 21:44symmetric. So on the triangular lattice
  582. 21:46this is what a vortex looks like.
  583. 21:48Here I'm showing
  584. 21:50the three sublattices in different
  585. 21:52colors, blue, red and green.
  586. 21:55You can see the red and green arrows and
  587. 21:58the blue arrows are pointing toward you.
  588. 22:00So this is a rotation about the the
  589. 22:02direction of the blue arrows.
  590. 22:04And you see that on the triangular
  591. 22:06lattice this vortex has a nice round
  592. 22:08shape.
  593. 22:09On Kagome
  594. 22:11with some second
  595. 22:13well third neighbor coupling
  596. 22:16we have a different shape. You can see
  597. 22:18that it's roughly an ellipse.
  598. 22:21And so generally vertices on the Kagome
  599. 22:23lattice will have an elliptical shape
  600. 22:26with the orientation of the major axis
  601. 22:30that can be arbitrary. They're not stuck
  602. 22:32to the lattice directions.
  603. 22:36All right. So
  604. 22:38one cute thing that came out of this
  605. 22:40field theory is that you can control
  606. 22:43the order parameter
  607. 22:45with spin waves. You can manipulate the
  608. 22:47direction of your spins by sending a
  609. 22:51pulse of of magnons
  610. 22:55which which is something you cannot do
  611. 22:57in collinear antiferromagnets.
  612. 22:59And I'll explain why.
  613. 23:01So here's the
  614. 23:03expression for the energy of a
  615. 23:04triangular lattice. You have just the mu
  616. 23:07term
  617. 23:08for the stiffnesses
  618. 23:10and if you work out the speeds of sound
  619. 23:13waves
  620. 23:14it turns out that two of the sound waves
  621. 23:16are degenerate
  622. 23:18and the third one has a higher speed
  623. 23:20than than the two.
  624. 23:22And with degenerate polarizations
  625. 23:26sorry, with degenerate frequencies of
  626. 23:28two linear polarizations you can form
  627. 23:30circular polarizations. So here are
  628. 23:32linear polarizations, right? You can
  629. 23:34rock your Mercedes-Benz logo about two
  630. 23:37directions, but you can also combine
  631. 23:39them
  632. 23:40and and here's what what what you're
  633. 23:42going to have. This is a circularly
  634. 23:43polarized
  635. 23:44spin wave made out of
  636. 23:47the two linear polarizations.
  637. 23:49Notice something strange as
  638. 23:53as the circular polarization wave
  639. 23:56rocks it turns gradually. You can turn
  640. 23:59either way depending on which circular
  641. 24:01polarization you excite. So you can
  642. 24:05gradually rotate clockwise or
  643. 24:06counterclockwise.
  644. 24:08And that's an effect that that is not
  645. 24:09available for collinear spins and we'll
  646. 24:12we'll we'll see why shortly.
  647. 24:15Okay. So it turns out
  648. 24:17that there's a very cute analogy with
  649. 24:21a wobbling coin,
  650. 24:22a problem that was popular in the 19th
  651. 24:24century when people were thinking about
  652. 24:26spinning tops
  653. 24:27uh or more recently as the Euler disk
  654. 24:30that's commercially available.
  655. 24:57So you can see that this wobbling coin
  656. 25:00uh keeps rotating.
  657. 25:02And so after rotation stops, your water
  658. 25:04parameter will have
  659. 25:05uh rotated in this direction. And by uh
  660. 25:09applying a pulse that is uh sufficiently
  661. 25:11short or long, you can change the uh
  662. 25:14the amount of rotation in in the system.
  663. 25:17So why is there a rotation here? What's
  664. 25:19what's going on? Well,
  665. 25:20you can think about it this way.
  666. 25:22Uh at every moment in time,
  667. 25:25your uh water parameter or the wobbling
  668. 25:27coin
  669. 25:28rotates about a different axis. So here
  670. 25:32the axis is shown as as this gray. The
  671. 25:34rotation axis is shown uh as this gray
  672. 25:38straight line.
  673. 25:39Uh and you can see that the rotation
  674. 25:41axis changes all the time. That's the
  675. 25:42instantaneous axis of rotation.
  676. 25:45It it touches the point
  677. 25:47uh
  678. 25:47uh
  679. 25:48uh
  680. 25:49it it goes through the center and the
  681. 25:50point where the the the wobbling coin
  682. 25:53touches the surface. And so it always
  683. 25:55changes. And that means that you're
  684. 25:57rotating your uh
  685. 25:58coin
  686. 25:59uh or this uh body frame
  687. 26:02uh about different axes all the time.
  688. 26:04And different rotation ex rotations
  689. 26:06about different axes do not commute.
  690. 26:09So if you uh simplified down to
  691. 26:11something
  692. 26:12uh
  693. 26:13instead of an infinite number of
  694. 26:14infinitesimal rotations, you boil it
  695. 26:16down to four rotations consecutively
  696. 26:19about the NX and Y NX
  697. 26:22uh and then Y axes. So you rotate about
  698. 26:25NX about alpha, then about at Y by beta.
  699. 26:28And then you do the opposite rotations
  700. 26:30about minus alpha about an x minus beta
  701. 26:33about on y.
  702. 26:35The four rotations don't commute and so
  703. 26:37if you take the product of these you'll
  704. 26:39end up with a rotation about the nz axis
  705. 26:43uh by an angle that's roughly the
  706. 26:44product of alpha and beta. So that's the
  707. 26:47non-commutativity of of of rotations
  708. 26:50uh and and so you end up with a net
  709. 26:52solid
  710. 26:53uh with a net rotation
  711. 26:55by an angle equal to the solid angle
  712. 26:57subtended by the black nz vector the
  713. 26:59spin chirality.
  714. 27:02In fact there's a very simple
  715. 27:04geometrical picture of why that that
  716. 27:06works. It's similar to doing parallel
  717. 27:08transport on a sphere.
  718. 27:10So suppose I I give you this uh uh
  719. 27:14Mercedes-Benz
  720. 27:15depiction of of the magnetic order uh uh
  721. 27:19in this triangle lattice antiferromagnet
  722. 27:22and I ask you to carry it around the
  723. 27:24sphere
  724. 27:25uh
  725. 27:26always keeping it horizontal
  726. 27:28right so it'll be tangential to the
  727. 27:29sphere everywhere and never turning it
  728. 27:33never twisting it about the vertical
  729. 27:36axis your local vertical axis so you
  730. 27:37cannot twist it about the uh
  731. 27:40axis perpendicular to the spins so you
  732. 27:42can only keep it horizontal and so this
  733. 27:44is what you do you walk down to the
  734. 27:46equator carrying this Mercedes-Benz logo
  735. 27:49uh and then you walk
  736. 27:51along the equator
  737. 27:53uh
  738. 27:54and that's what parallel transport looks
  739. 27:56like and then you come back up the
  740. 27:58meridian again and voila your order
  741. 28:01parameter has rotated
  742. 28:03right and the rotation angle is 90
  743. 28:04degrees or pi over two which is exactly
  744. 28:07the solid angle uh that your path takes
  745. 28:11out on the sphere. So that's basically
  746. 28:14uh uh
  747. 28:15you know berry phase geometric rotation
  748. 28:19uh parallel transport whatever you like
  749. 28:20to call it.
  750. 28:22Right so you you rotate it by pi over
  751. 28:24two and that's exactly what happens
  752. 28:26here.
  753. 28:27So why doesn't this happen to collinear
  754. 28:29spins? Well actually it does we just
  755. 28:31don't notice it. So here's a circularly
  756. 28:33polarized wave in the collinear
  757. 28:35antiferromagnet. This is what it looks
  758. 28:37like. The spin is always rotated about
  759. 28:40this
  760. 28:42ever-changing axis, right? And so after
  761. 28:45this rotation, you expect that the spin
  762. 28:48will rotate about its own axis as well.
  763. 28:51And of course it will, but the spin is
  764. 28:53round.
  765. 28:54It's axially symmetric and we don't see
  766. 28:56any rotation
  767. 28:59after it stops. It's the same spin. So
  768. 29:02that's why you cannot see it in the
  769. 29:03collinear antiferromagnet. The spins do
  770. 29:05rotate, but they rotate about their own
  771. 29:07axis and so you cannot observe them.
  772. 29:10That's why you need another parameter
  773. 29:11that includes more than one spin and
  774. 29:14more than two spins. You you need to
  775. 29:16have a coplanar state
  776. 29:18as as we had with with this triangular
  777. 29:20antiferromagnet.
  778. 29:22So that's a proposal for how one can
  779. 29:25control the order parameter
  780. 29:27in an antiferromagnet on a triangular
  781. 29:29lattice
  782. 29:30with with spin waves.
  783. 29:33Okay.
  784. 29:34Well,
  785. 29:36topological terms. I started this talk
  786. 29:38by mentioning that Tummuru and Read went
  787. 29:40in search of topological terms in the
  788. 29:43action
  789. 29:44inspired by Haldane's work on
  790. 29:47antiferromagnetic spin chains where
  791. 29:49topological terms are very important and
  792. 29:51explained the difference between the
  793. 29:52gapless spectrum of half-integer spins
  794. 29:55and gap spectrum of integer spins.
  795. 29:59And so they said, well, can we find
  796. 30:00topological terms in in this
  797. 30:04triangular lattice antiferromagnet? And
  798. 30:06the question the answer was no. The they
  799. 30:08went looking for a specific topological
  800. 30:10term
  801. 30:12with one spatial derivative and two time
  802. 30:14derivatives and they did not find one.
  803. 30:17Well, we did.
  804. 30:19In fact, it turns out that if you take
  805. 30:22so there are several topological terms
  806. 30:24in fact that that that could be found
  807. 30:26and one of them
  808. 30:28is written here as the difference of the
  809. 30:31lambda and new terms. Remember the
  810. 30:33lambda and new terms and the stiffness
  811. 30:35did the same thing to spin waves.
  812. 30:38Uh these velocity of the spin waves is
  813. 30:40controlled by uh the lambda plus new
  814. 30:44uh combination rather than by lambda and
  815. 30:46new separately.
  816. 30:47And there's a simple reason for that. If
  817. 30:49you take the difference of these two
  818. 30:50terms,
  819. 30:52uh in fact by partial integration you
  820. 30:55can move
  821. 30:56this spatial derivative to the right and
  822. 30:58that spatial derivative to the left and
  823. 31:01you'll get exactly this term.
  824. 31:03So these terms coincide up to boundary
  825. 31:06corrections. So in the bulk they
  826. 31:09actually give zero contribution. So even
  827. 31:11though this looks like a bulk uh
  828. 31:14integral it actually boils down to a
  829. 31:17line integral along the boundary uh and
  830. 31:20and and it it has uh
  831. 31:23no influence in the bulk strictly
  832. 31:25speaking.
  833. 31:26Uh and so there's there's there's a
  834. 31:28analogy to the Mermin correlation in
  835. 31:31theory of P wave superfluids uh where
  836. 31:34this combination
  837. 31:36uh can be thought of as a magnetic field
  838. 31:38of sorts uh of an SO3 gauge shield uh
  839. 31:42and you can uh express it in terms of a
  840. 31:45an SO3 gauge potential. And so that uh
  841. 31:48magnetic field integral uh over a
  842. 31:51spatial uh
  843. 31:53of an area boils down to a
  844. 31:56uh vector potential integral along the
  845. 31:58boundary.
  846. 32:00So it's not a bulk term, it's a boundary
  847. 32:02term.
  848. 32:03And so as a boundary term it will uh
  849. 32:06presumably do something interesting on
  850. 32:07the edge.
  851. 32:09There's another way to express this term
  852. 32:11and that's something that Bastian found.
  853. 32:13I have no idea how he was able to do
  854. 32:15that but you can show that
  855. 32:17uh this expression
  856. 32:19is equivalent to that expression. So
  857. 32:21alpha and beta here are x and y
  858. 32:23components. So it only involves an x and
  859. 32:25an y and the integral here looks like a
  860. 32:29skyrmion number density for the
  861. 32:31chirality vector on Z.
  862. 32:33And so of course if you integrate it
  863. 32:34over the uh infinite two-dimensional
  864. 32:37space you get a skyrmion number.
  865. 32:39Skyrmion number is an integer.
  866. 32:41And so if you take a an infinitesimal
  867. 32:43variation of these fields,
  868. 32:45in an infinite system there will be no
  869. 32:47change to this integer.
  870. 32:49And therefore this
  871. 32:50this is a silent term as far as as far
  872. 32:53as the classical dynamics goes. So it
  873. 32:55does not influence the dynamics
  874. 32:58in an infinite system.
  875. 32:59But it it does create a boundary term.
  876. 33:02So if you have a boundary, you have a
  877. 33:04boundary energy.
  878. 33:05And so this boundary energy actually
  879. 33:07makes an interesting effect. It binds
  880. 33:11spin waves to a boundary,
  881. 33:13or magnons. And
  882. 33:15these magnons happen to be chiral.
  883. 33:19They propagate in only one direction,
  884. 33:21let's say counterclockwise but not
  885. 33:23clockwise.
  886. 33:25A similar term was found in an exotic
  887. 33:27theory of orbital magnetization
  888. 33:31by Levitov and his postdocs.
  889. 33:34And when we saw that paper we
  890. 33:36immediately realized that we have a
  891. 33:38similar term and so it should should
  892. 33:39give rise to
  893. 33:42these sort of edge modes. So you can
  894. 33:44analyze these edge modes
  895. 33:47for a straight edge along the x
  896. 33:49direction.
  897. 33:51kx is a conserved momentum
  898. 33:54of a magnon.
  899. 33:56And so you can try a spin wave
  900. 33:59that runs along the edge,
  901. 34:01but exponentially decays
  902. 34:04into the bulk along the y direction.
  903. 34:07And you can figure out what what the
  904. 34:10the penetration depth is
  905. 34:13and and show that
  906. 34:15this wave is only bound to the edge for
  907. 34:17one sign of kx but not for the other.
  908. 34:20And so here's what happens. You you make
  909. 34:23a a circular polarization
  910. 34:26out of your spin waves, and you discover
  911. 34:28that there is a mode that travels around
  912. 34:31the boundary and it only penetrates into
  913. 34:33the bulk
  914. 34:36some
  915. 34:37number of
  916. 34:38wavelengths which which on on the
  917. 34:41strength of the topological term.
  918. 34:43So if you have shorter wavelength, then
  919. 34:46this wave penetrates less into the bulk
  920. 34:48and it's better tied tied to the edge.
  921. 34:51So only waves
  922. 34:53propagating counterclockwise are bound
  923. 34:55to the edge
  924. 34:57and you don't have waves propagating the
  925. 34:58other way.
  926. 35:00If you want to make these waves, you
  927. 35:01have to take
  928. 35:03the left circular polarization instead
  929. 35:05of the right circular polarization,
  930. 35:07which is equivalent to the time reversal
  931. 35:09and that's why the direction of
  932. 35:10propagation is reversed. So in fact,
  933. 35:13there are two counter propagating waves
  934. 35:15traveling clockwise and counterclockwise
  935. 35:18and what they differ by is is is the
  936. 35:20spin projection that that they have
  937. 35:24that is
  938. 35:25out of the plane.
  939. 35:27So it's a field theory and it would be
  940. 35:29nice to have a lattice model
  941. 35:31that
  942. 35:32exhibits this sort of thing.
  943. 35:36Because the field theory is
  944. 35:37non-committal. It tells you what's
  945. 35:38possible. You need an actual lattice
  946. 35:41model which will realize this exotic
  947. 35:43effect.
  948. 35:44And we did find such a lattice model.
  949. 35:47It's actually not very exotic at all.
  950. 35:51So take a triangular lattice
  951. 35:53antiferromagnet.
  952. 35:54The spins reside in the corners of the
  953. 35:56triangles
  954. 35:58and the
  955. 35:59colored circles are actually
  956. 36:01non-magnetic ions. These are ions that
  957. 36:04mediate super exchange.
  958. 36:06Imagine that you have two kinds of
  959. 36:09non-magnetic ions.
  960. 36:11Maybe oxygen shown in red
  961. 36:14centered on the up triangles and sulfur
  962. 36:18ions shown in blue centered on the down
  963. 36:20triangles.
  964. 36:22And suppose oxygen and sulfur have
  965. 36:24different super exchange strength
  966. 36:26mediated
  967. 36:28between two spins. So the red path
  968. 36:30through the oxygen gives you
  969. 36:33this much
  970. 36:34of
  971. 36:35super exchange and the blue path gives
  972. 36:38you that much super exchange. Okay, they
  973. 36:40differ. Delta measures is uh
  974. 36:43uh difference between the two
  975. 36:45superexchange paths.
  976. 36:47Now, if you take two spins on the bulk,
  977. 36:49you have two superexchange paths
  978. 36:51available to them. So you have to add
  979. 36:53these contributions, and when you add
  980. 36:55them, delta disappears.
  981. 36:57So you you just add this plus that, and
  982. 37:00it becomes simply J, and delta has no
  983. 37:03effect on the bulk.
  984. 37:05But it has an effect on the edge.
  985. 37:07If you are on the edge, these two spins
  986. 37:09only have a red superexchange path.
  987. 37:11And if you are on the opposite edge,
  988. 37:13these two spins have a blue
  989. 37:14superexchange path.
  990. 37:16So the bonds on the edge are going to be
  991. 37:18different from the bonds on the bulk,
  992. 37:21and they will be influenced by this
  993. 37:22difference delta. So here this is how we
  994. 37:25construct
  995. 37:26what looks like a bulk term,
  996. 37:28right? It's a bulk superexchange,
  997. 37:30but the superexchange part, the delta
  998. 37:32dependence uh does not exist in the
  999. 37:34bulk, and it only is pushed out to the
  1000. 37:36edge. So that's how this uh field
  1001. 37:39theoretic construction is realized in
  1002. 37:40the lattice.
  1003. 37:42So you can take this model,
  1004. 37:44and you can just numerically determine
  1005. 37:45its spectrum.
  1006. 37:47Uh and here's its spectrum uh in a strip
  1007. 37:50that runs along the x direction.
  1008. 37:52So kx is a conserved momentum, but ky is
  1009. 37:55not conserved because translational
  1010. 37:57symmetry is broken along the y
  1011. 37:59direction. Uh and so for a given kx, you
  1012. 38:02will have not one uh
  1013. 38:04spin wave frequency in the bulk, but a
  1014. 38:06spectrum of frequencies uh
  1015. 38:09that correspond to different values of
  1016. 38:11ky. And because ky can vary continuously
  1017. 38:14in an infinite system, uh you can
  1018. 38:17basically have any frequency uh above a
  1019. 38:19certain value.
  1020. 38:21So the gray area uh are the spin waves
  1021. 38:25residing in the bulk.
  1022. 38:26And you can see that uh near these uh
  1023. 38:29points 2 pi over 3 and minus 2 pi over
  1024. 38:323, you have an interesting spectrum. In
  1025. 38:34addition to the uh
  1026. 38:36bulk spin waves, you have spin waves uh
  1027. 38:39that lie below that bulk continuum, and
  1028. 38:42these are precisely the edge modes uh
  1029. 38:44that live uh uh
  1030. 38:46only near the
  1031. 38:48and they exponentially decay into the
  1032. 38:49bulk.
  1033. 38:51And you can see that the waves propagate
  1034. 38:53one way along the upper edge, the blue
  1035. 38:55curve, and they propagate the other way
  1036. 38:58on the lower edge.
  1037. 39:00And if you
  1038. 39:02And this is for one
  1039. 39:03circular polarization. For the opposite
  1040. 39:05circular polarization, these two will
  1041. 39:07switch and you will have
  1042. 39:10waves propagating right along the upper
  1043. 39:13edge and left along the lower edge.
  1044. 39:16So this is different from the scenario
  1045. 39:18of topological magnons
  1046. 39:21that's discussed
  1047. 39:24in the context of Dzyaloshinskii-Moriya
  1048. 39:26interactions where you can have a
  1049. 39:28spectrum
  1050. 39:30in the bulk
  1051. 39:32represented by these black lines
  1052. 39:34and then there's a gap in the middle and
  1053. 39:37you have
  1054. 39:38topological magnons due to the
  1055. 39:40Dzyaloshinskii-Moriya interaction that
  1056. 39:42reside
  1057. 39:44in between
  1058. 39:45the two
  1059. 39:47upper and lower bands. So our magnons
  1060. 39:50actually live below our edge magnons
  1061. 39:52live below the bulk magnons.
  1062. 39:54And so these magnons here, the
  1063. 39:56topological magnons can decay into bulk
  1064. 39:59magnons of lower energy or pairs of them
  1065. 40:02or something
  1066. 40:03whereas
  1067. 40:04the
  1068. 40:06the topological magnons in this case
  1069. 40:08are stable
  1070. 40:10at least for the smooth edge. If you
  1071. 40:12have a rough edge,
  1072. 40:14there will be scattering of the edge
  1073. 40:15magnons into the bulk. But it turns out
  1074. 40:18that at the longest wavelength the
  1075. 40:20scattering is
  1076. 40:21is not very strong.
  1077. 40:23And so this
  1078. 40:24these edge magnons should survive.
  1079. 40:28All right.
  1080. 40:29So Hira, do you have 5 minutes?
  1081. 40:32>> Yeah, yeah, absolutely. This is really
  1082. 40:34nice.
  1083. 40:35>> Good. So I'll I'll I'll talk about more
  1084. 40:38topological terms.
  1085. 40:40Now Duncan Haldane's topological term
  1086. 40:45Well, okay, so we'll get back to
  1087. 40:46Duncan's topological term. So this this
  1088. 40:48is the topological in the action that I
  1089. 40:50described. It has uh two spatial
  1090. 40:53derivatives and resembles the skyrmion
  1091. 40:55number in the XY plane.
  1092. 40:58You can form other combinations uh of
  1093. 41:01derivatives
  1094. 41:02uh including for example spatial
  1095. 41:04temporal planes. You can have time
  1096. 41:06derivative and a spatial derivative.
  1097. 41:09And it turns out that this uh term here
  1098. 41:11where you sum over alpha which is X and
  1099. 41:14Y and you sum over I which is X, Y and
  1100. 41:17Z. This term is compatible with all the
  1101. 41:20symmetries uh uh
  1102. 41:22uh of the uh uh
  1103. 41:24uh point group.
  1104. 41:26And so this term is in principle allowed
  1105. 41:28in our system. So that's another
  1106. 41:29topological term. It's kind of a
  1107. 41:31skyrmion number in the spatial temporal
  1108. 41:33planes instead of skyrmion number in the
  1109. 41:36spatial plane XY.
  1110. 41:38Right? And again this will be mostly a
  1111. 41:39silent term which which will have no uh
  1112. 41:43influence on on the classical dynamics
  1113. 41:46of spins, but it might do something
  1114. 41:48else.
  1115. 41:49And what it does is it endows uh
  1116. 41:52solitons uh in antiferromagnets with
  1117. 41:55spin.
  1118. 41:56So solitons turn out to be kind of
  1119. 41:58ferromagnetic islands in a way.
  1120. 42:00And there's a well-known
  1121. 42:02uh
  1122. 42:03uh
  1123. 42:04uh counterpart of that in spin chains
  1124. 42:06where the topological term written by
  1125. 42:08Duncan Haldane in 1 + 1 dimension where
  1126. 42:12it's a skyrmion number in the TX plane.
  1127. 42:15Uh this is a stop his topological term.
  1128. 42:18Little S is the length of spin.
  1129. 42:20Uh it turns out that this topological
  1130. 42:22term
  1131. 42:23uh
  1132. 42:24basically uh gives spin
  1133. 42:27uh to inhomogeneities of of the nail
  1134. 42:29vector.
  1135. 42:30Now you can derive that uh
  1136. 42:32from from this form or from that form
  1137. 42:35where you can see that the uh angular
  1138. 42:37frequency of rotation couples uh to the
  1139. 42:41uh
  1140. 42:41uh gradient of the nail vector. And from
  1141. 42:44that you can derive that the spin
  1142. 42:46density has a term proportional to the
  1143. 42:48derivative of the nail vector.
  1144. 42:51So if you have a domain wall where the
  1145. 42:53nail vector changes sign, you'll find uh
  1146. 42:56that a domain wall carries spin plus S
  1147. 42:59or minus S
  1148. 43:00and a chain end has a spin of plus half
  1149. 43:03S or minus half S.
  1150. 43:05So, uh this is why spin ons in the XYZ
  1151. 43:09spin one half chain carry spin plus or
  1152. 43:11minus half
  1153. 43:12and edges uh
  1154. 43:13of spin chains can bind uh a spin equal
  1155. 43:16to one quarter.
  1156. 43:18Uh
  1157. 43:19if you haven't heard of spins one
  1158. 43:20quarter previously, there are actually
  1159. 43:23very recent papers by uh Nathan
  1160. 43:26Nathan Andres group where they uh
  1161. 43:29discuss in detail uh these uh S over two
  1162. 43:32spins uh at the edges.
  1163. 43:36So,
  1164. 43:37in the
  1165. 43:39this is a topological term that's
  1166. 43:41allowed and it also has uh
  1167. 43:44this sort of divergence of of the uh
  1168. 43:48uh
  1169. 43:49of the uh spin frame vectors coupled to
  1170. 43:52the angular uh frequency of rotation and
  1171. 43:55that creates a spin density uh
  1172. 43:57proportional to the divergence of of N.
  1173. 44:00Uh
  1174. 44:01and so there should be linear spin
  1175. 44:02density uh
  1176. 44:04accumulated on domain walls.
  1177. 44:06And
  1178. 44:07if you're not into field theory, there's
  1179. 44:09a very simple uh
  1180. 44:11lattice picture that explains how that
  1181. 44:13happens. Let's start with one dimension.
  1182. 44:15Here's a an antiferromagnetic spin
  1183. 44:17chain. Every spin has length S where in
  1184. 44:21the extreme using limit where domain
  1185. 44:23walls are very sharp, basically once one
  1186. 44:25lattice spacing and they correspond to a
  1187. 44:28couple of spins pointing in the same
  1188. 44:29direction.
  1189. 44:30So, these two spins uh
  1190. 44:33uh
  1191. 44:34are kind of an
  1192. 44:35a ferromagnetic island in the sea of
  1193. 44:37antiferromagnetic spins. So, if you blur
  1194. 44:40your vision and you don't view things on
  1195. 44:42the lattice scale, this is where
  1196. 44:44there'll be a net spin and the amount of
  1197. 44:46spin is minus S here and it's plus S
  1198. 44:49here.
  1199. 44:50Now,
  1200. 44:51picture a a wall uh in a
  1201. 44:55So you have one ground state on the left
  1202. 44:58shown in red and you have another ground
  1203. 45:00state on the right shown in blue
  1204. 45:03and these are time reversed copies of
  1205. 45:05the two ground states. So here I assume
  1206. 45:07a local anisotropy that
  1207. 45:11prefers spins pointing
  1208. 45:13to or from centers of triangles like
  1209. 45:15this. So this is one ground state and
  1210. 45:17that's the other ground state.
  1211. 45:20It's something similar to what happens
  1212. 45:22in manganese 310
  1213. 45:23where it's slightly more complicated.
  1214. 45:27So these two ground states can have
  1215. 45:30domain walls of two varieties between
  1216. 45:32them. One domain wall will have the
  1217. 45:34spins that are oriented
  1218. 45:37more or less to the left and they're
  1219. 45:39kind of part of the red domain if you
  1220. 45:41will, right? So this triangle is is
  1221. 45:43exactly in this ground state and you can
  1222. 45:46time reverse them
  1223. 45:47and get this state where the the spins
  1224. 45:49on the domain wall kind of belong to the
  1225. 45:52to the right domain.
  1226. 45:54And these two
  1227. 45:56domain walls carry a spin that
  1228. 46:00points to the left
  1229. 46:02here and to the right there. You can
  1230. 46:04calculate the linear spin density
  1231. 46:07from from this naive picture and you
  1232. 46:09will get exactly the same result as the
  1233. 46:10field theory predicts.
  1234. 46:12Or here is a horizontal domain wall
  1235. 46:15where you have a domain on the top and
  1236. 46:17domain on the bottom
  1237. 46:19and in between the spins can point all
  1238. 46:21down or they can point all up for
  1239. 46:23example. And again the linear spin
  1240. 46:25density will be the same
  1241. 46:27except it now points in a different
  1242. 46:28direction.
  1243. 46:29And so the prediction of field theory is
  1244. 46:33compatible with with this very simple
  1245. 46:36extreme loss limit.
  1246. 46:39We have numerical evidence in the
  1247. 46:40quantum antiferromagnetic Ising model
  1248. 46:44in a well, sorry, quantum
  1249. 46:47model of this antiferromagnet
  1250. 46:49and its work in progress done in
  1251. 46:52collaboration with Frank Pollmann.
  1252. 46:54This is possible uh it's possible to
  1253. 46:56extend this to 3D plus one dimensions.
  1254. 46:59So pyrochlore, all in and all out,
  1255. 47:01antiferromagnet should also have domain
  1256. 47:03walls that that
  1257. 47:05uh carry spin and work as ferromagnetic
  1258. 47:08islands.
  1259. 47:09So here's a summary of what I've told
  1260. 47:11you. We have a shiny new field theory
  1261. 47:13for three sublattice antiferromagnets.
  1262. 47:15The order parameter is a rigid body. You
  1263. 47:17can describe it in terms of uh spin
  1264. 47:19frame uh which is the analog of the body
  1265. 47:23frame uh in rigid body dynamics. There
  1266. 47:26are four bulk coupling constants,
  1267. 47:28some topological terms, Pythagorean
  1268. 47:31relation for the three magnum
  1269. 47:32velocities, uh elliptical vertices and
  1270. 47:35kagome,
  1271. 47:36and control of the order parameter by by
  1272. 47:38magnets. Thank you very much for your
  1273. 47:40attention.
  1274. 47:41>> Thank you. Thank you, Oleg. I was always
  1275. 47:44It's always a pleasure to listen to. I
  1276. 47:45still remember your black uh
  1277. 47:48your uh your chalkboard talks that I
  1278. 47:50liked to
  1279. 47:51>> it would be hard to give online.
  1280. 47:54>> Yeah, yeah, the the but this one they
  1281. 47:56know it's just amazing.
  1282. 47:58Um one thing uh so I'll start with the
  1283. 47:59questions and please raise your hands uh
  1284. 48:02to ask questions to Oleg um and I'll
  1285. 48:04bring you up to the
  1286. 48:06uh to promote you.
  1287. 48:08Hey, but I'll start with a a great
  1288. 48:09question. Um so particularly looking at
  1289. 48:12these modes, these modes that are on the
  1290. 48:13edge, the extra modes that you've found
  1291. 48:15out with the coin
  1292. 48:17uh
  1293. 48:18I mean, I assumed that you're still
  1294. 48:20working on observing them of how
  1295. 48:21sensitive it would be. How sensitive is
  1296. 48:23that to to edges to essentially uh when
  1297. 48:26you have multi-domain systems, you know,
  1298. 48:28frustrated systems tend to have a small
  1299. 48:29domains.
  1300. 48:31Um
  1301. 48:31>> Right. So so if you have grain
  1302. 48:34boundaries, uh you're going to have uh
  1303. 48:37edge modes uh
  1304. 48:39uh localized on grain boundaries as
  1305. 48:41well.
  1306. 48:42Uh
  1307. 48:43so so these these modes are pretty
  1308. 48:45robust. You can have a sample of any
  1309. 48:47shape.
  1310. 48:48It doesn't have to be a straight edge.
  1311. 48:50As long as as the curve is smooth, it's
  1312. 48:53going to be very stable.
  1313. 48:55And when the curve is not smooth, when
  1314. 48:57you have edge roughness, uh there will
  1315. 49:00be scattering. Uh these these waves
  1316. 49:02would be scattered in the bulk.
  1317. 49:04Uh and they will have a finite lifetime.
  1318. 49:07But the lifetime will increase quite
  1319. 49:09quickly as you go to longer wavelengths.
  1320. 49:11So parts of the spectrum should uh
  1321. 49:14survive, particularly the long
  1322. 49:15wavelength parts.
  1323. 49:17Uh we we we
  1324. 49:19Grigora Deman has a preprint on on
  1325. 49:21archive uh that that uh uh
  1326. 49:24deals with that and he showed that it's
  1327. 49:26pretty stable.
  1328. 49:28>> Do you do you have any material
  1329. 49:29candidates that are the best possible
  1330. 49:31choices for this?
  1331. 49:33>> Yes. Uh
  1332. 49:35when I was visiting the University of
  1333. 49:36Minnesota, uh it was pointed out to me
  1334. 49:38that there's a compound uh with the ABC
  1335. 49:42stacking
  1336. 49:43and uh two different uh
  1337. 49:46uh ligands uh that that alternate
  1338. 49:49uh that could actually do that. Uh
  1339. 49:52I do not have the name of this compound
  1340. 49:55off the top of my head except to say
  1341. 49:56that
  1342. 49:57uh the magnetic ion is rare earth, so it
  1343. 49:59has a DC of like a Kelvin.
  1344. 50:02Uh so I think that's a little bit low.
  1345. 50:06[laughter]
  1346. 50:07So one one should look for for uh
  1347. 50:09transition metal.
  1348. 50:10Um but but yeah.
  1349. 50:12>> difficult with the non-collinear
  1350. 50:13systems. They they don't tend to have
  1351. 50:15very high uh
  1352. 50:16>> Well
  1353. 50:18No, manganese
  1354. 50:19Manganese has good enough, but uh
  1355. 50:22>> Uh but
  1356. 50:22>> not triangular. So you need triangular
  1357. 50:24lattice instead of kagome in that case.
  1358. 50:27>> Um Vladimir
  1359. 50:29Can you please go ahead and ask your
  1360. 50:30question? Uh
  1361. 50:32Okay. Let me just have yeah.
  1362. 50:35>> Uh
  1363. 50:35Hi Oleg. Nice to see you. Hi Kairo.
  1364. 50:38>> Hi.
  1365. 50:40>> Uh
  1366. 50:41I have a couple of questions and all of
  1367. 50:43them are related to topological solitons
  1368. 50:45in this model,
  1369. 50:46in particular vortices which you have
  1370. 50:48shown.
  1371. 50:49Um so in continuum models for
  1372. 50:52ferromagnets or antiferromagnets uh
  1373. 50:55vortices has a core
  1374. 50:57and size of this core is determined by
  1375. 51:00competition between exchange and
  1376. 51:02with the planar anisotropy.
  1377. 51:05So the question is, I think your
  1378. 51:08vortices do not have cores because you
  1379. 51:10have shown the discrete lattice, but in
  1380. 51:13principle, how
  1381. 51:15is it possible to add in addition to
  1382. 51:18exchange interaction which you have in
  1383. 51:20your model to add additional
  1384. 51:22interactions, anisotropy, magnetic
  1385. 51:24field?
  1386. 51:25>> Yeah.
  1387. 51:25>> And this would significantly enrich the
  1388. 51:28zoo of topological solitons which you
  1389. 51:30can obtain in the solutions.
  1390. 51:32>> Of course. Yes. Yeah, indeed, I I showed
  1391. 51:36vortices
  1392. 51:38in the Heisenberg model which is
  1393. 51:40completely isotropic.
  1394. 51:43And these are stable unlike vortices in
  1395. 51:46the
  1396. 51:47ferromagnetic Heisenberg model where
  1397. 51:49they could unwind by going into the
  1398. 51:51third dimension.
  1399. 51:52In this case they can't.
  1400. 51:54They're they're called Z2 vortices. They
  1401. 51:57were introduced by Kawamura in the 1980s
  1402. 52:00and they're stable for topological
  1403. 52:01reasons because
  1404. 52:03in the SO3 manifold a rotation by 2 pi
  1405. 52:08is a path that cannot be continuously
  1406. 52:11shrunk to a point.
  1407. 52:13It it's
  1408. 52:15there are two kinds of closed paths in
  1409. 52:18SO3 and this is a non-contractible one.
  1410. 52:21So so these are these are stable even
  1411. 52:23without
  1412. 52:24XY and anisotropy.
  1413. 52:26Now you can add XY anisotropy, for
  1414. 52:28example, easy plane, and then of course
  1415. 52:30your your vortices will be
  1416. 52:32will stay in the plane
  1417. 52:35but switch out of the plane
  1418. 52:38when you know you come to this toward
  1419. 52:40the center of the core. There will be
  1420. 52:43a certain length scale determined by
  1421. 52:44again the competition of exchange and
  1422. 52:47and anisotropy energy. Yes.
  1423. 52:49>> But even in this continuum model, don't
  1424. 52:52you have a singularity, the divergence
  1425. 52:54of energy in the vortex?
  1426. 52:55>> is there is a singularity and in fact
  1427. 52:58the order parameter cannot survive there
  1428. 53:00just like the order parameter cannot
  1429. 53:02survive in the XY model
  1430. 53:04uh
  1431. 53:05at the center of the core, right? And so
  1432. 53:07the spins have to tilt out of the plane.
  1433. 53:09Uh here it's more interesting because it
  1434. 53:12turns out that
  1435. 53:14this is a true singularity like the
  1436. 53:15singularity at the center of a hedgehog
  1437. 53:18uh in in in three-dimensional
  1438. 53:20ferromagnets. Uh and so the order
  1439. 53:22parameter has to die
  1440. 53:24uh
  1441. 53:25which means means that uh you'll you'll
  1442. 53:27have
  1443. 53:28uh
  1444. 53:29you know you will not have a perfect
  1445. 53:30120° structure at the center of the
  1446. 53:33core.
  1447. 53:34And I can speculate what could happen
  1448. 53:37there.
  1449. 53:38Uh you can think of the three
  1450. 53:40sublattices as forming
  1451. 53:42each will form a vortex of its own.
  1452. 53:45Right? So it's like having three
  1453. 53:47ferromagnetic vortices uh
  1454. 53:50uh on top of one another.
  1455. 53:53And the interesting thing is of course
  1456. 53:55when you have a vortex uh
  1457. 53:58in an easy plane model in a ferromagnet,
  1458. 54:00it has a core where spins point uh in
  1459. 54:03the perpendicular direction.
  1460. 54:04Right? So here the three vortices also
  1461. 54:07will have cores.
  1462. 54:09And curiously it uh
  1463. 54:13the the the three cores cannot point all
  1464. 54:15in opposite directions, right?
  1465. 54:17Two of them will point up and one will
  1466. 54:19point down or vice versa.
  1467. 54:21Uh and that means that there'll be a
  1468. 54:23non-zero net Skyrmion number.
  1469. 54:26And you know what that means. It means
  1470. 54:27that there'll be a gyroscopic force
  1471. 54:29acting on on this vortex.
  1472. 54:33In in an antiferromagnet, in a collinear
  1473. 54:35antiferromagnet, the gyroscopic force
  1474. 54:38cancels out because the the two
  1475. 54:40components of you know the the two
  1476. 54:41sublattices
  1477. 54:42have opposite Skyrmion numbers. And here
  1478. 54:45they will have Skyrmion numbers half,
  1479. 54:46half, and minus half. They will not
  1480. 54:48cancel out. So there should be a net
  1481. 54:50gyroscopic force in this case.
  1482. 54:52>> So this was my next question and this is
  1483. 54:54the last one I promise. So have you
  1484. 54:56already developed the collective
  1485. 54:58variable approach to describe
  1486. 55:00>> we we know each other well enough,
  1487. 55:01Vladimir.
  1488. 55:05Uh
  1489. 55:07I have not developed that approach yet.
  1490. 55:08I was student who should be working on
  1491. 55:11that.
  1492. 55:13And and and so
  1493. 55:15I hope to develop it.
  1494. 55:17But
  1495. 55:19we haven't yet.
  1496. 55:22>> Thank you.
  1497. 55:23>> Okay.
  1498. 55:25Is there any further questions from the
  1499. 55:26audience? I see immediately but you know
  1500. 55:30please
  1501. 55:31contact. I'm going to stop here now the
  1502. 55:33the live streaming for the for the
  1503. 55:36for the 30 people that were on YouTube
  1504. 55:38and then the
  1505. 55:40the other 45 or 50 that you guys were
  1506. 55:42online. Thank you for
  1507. 55:44becoming and thank you like by the way I
  1508. 55:45have another question now that is not
  1509. 55:47being taped
  1510. 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.