Neutron Scattering and Its Application to Strongly Correlated Systems
11Neutron Scattering and Its Application toStrongly Correlated Systems
Igor A. Zaliznyak and John M. Tranquada
Brookhaven National Laboratory, Upton, NY 11973-5000, USA [email protected] (631-344-3761), [email protected] (631-344-7547)
Abstract.
Neutron scattering is a powerful probe of strongly correlated systems.It can directly detect common phenomena such as magnetic order, and can be usedto determine the coupling between magnetic moments through measurements of thespin-wave dispersions. In the absence of magnetic order, one can detect diffuse scat-tering and dynamic correlations. Neutrons are also sensitive to the arrangement ofatoms in a solid (crystal structure) and lattice dynamics (phonons). In this chapter,we provide an introduction to neutrons and neutron sources. The neutron scatter-ing cross section is described and formulas are given for nuclear diffraction, phononscattering, magnetic diffraction, and magnon scattering. As an experimental exam-ple, we describe measurements of antiferromagnetic order, spin dynamics, and theirevolution in the La − x Ba x CuO family of high-temperature superconductors. A common symptom of correlated-electron systems is magnetism, and neu-tron scattering is the premiere technique for measuring magnetic correlationsin solids. With a spin angular momentum of ¯ h , the neutron interacts directlywith the magnetization density of the solid. Elastic scattering can directlyreveal static magnetic order; for example, neutron diffraction provided thefirst experimental evidence for N´eel antiferromagnetism [1]. Through inelasticscattering one can probe dynamic spin-spin correlations; in an ordered anti-ferromagnet, one can measure the precession of the spins about their averageorientations, which show up as dispersing spin waves.Neutrons do not couple to the charge of the electrons, but instead scatterfrom atomic nuclei via the strong force. Despite the name, the small size ofthe nucleus compared to the electronic charge cloud of the atom results ina rather weak scattering cross section. The magnetic and nuclear scatteringcross sections are comparable, so that neutron scattering is very sensitive tomagnetism, in a relative sense. a r X i v : . [ c ond - m a t . s t r- e l ] A p r Igor A. Zaliznyak and John M. Tranquada
A challenge with neutron scattering is that the combination of weak scat-tering cross section and limited source strength means that one needs a rela-tively large sample size compared with many other techniques. The value ofthe information that can be obtained by neutron scattering generally makesworthwhile the effort to grow large samples; nevertheless, in practice it isuseful to take advantage of complementary information obtained from tech-niques such as muon spin rotation spectroscopy and nuclear magnetic reso-nance. The latter techniques yield less information but often provide greaterprecision. There have also been continuing developments in resonant x-rayscattering; nevertheless, neutron scattering will remain an essential techniqueto investigate strongly correlated systems for the foreseeable future.As we have space only for a concise introduction to the field, we note thatthere plenty of more extended references available. A good summary of thetheory of neutron scattering is given by Squires [2], while a more detaileddescription is provided by Lovesey [3]. We have contributed to a technique-oriented book [4] and to book chapters on magnetic neutron scattering [5, 6],and new books on the subject continue to appear.To illustrate some of the concepts and capabilities, we will use examples in-volving copper-oxide compounds, especially from the family La − x Ba x CuO ,which includes phenomena from antiferromagnetic order to high-temperaturesuperconductivity. More details on neutron scattering studies of cuprates aregiven in recent reviews [7–10]. The neutron is an elementary spin-1/2 particle, which, together with itscharged relative, the proton, is a building block of the atomic nucleus. Ac-cording to the “standard model” of the elementary particles, the neutronand proton are fermionic hadrons, or baryons, composed of one “up” andtwo“down” quarks, and two “up” and one “down” quarks, respectively. Thebasic properties of a neutron are summarized in Table 1.1.
Table 1.1.
Basic properties of a neutron. The gyromagnetic ratio, γ n , and the g-factor, g n , are defined by µ n = γ n σ n = − g n µ N S n , where σ n is the neutron’s angularmomentum, S n = σ n / ¯ h is the neutron’s spin ( S n = 1 / µ N = e ¯ h/ (2 m p c ) =5 . × − J/T = 5 . × − erg/Gs is the nuclear magneton [11, 12].Charge Mass Lifetime Magnetic Gyromagnetic g-factormoment µ n ratio γ n g n (kg) (s) (J/T) (s − /T)0 1 . × − ± − . × − − . × Although the neutron is electrically neutral, it has a non-zero magneticmoment, similar in magnitude to that of a proton ( µ n = 0 . µ p ), butdirected opposite to the angular momentum, so that the neutron’s gyromag-netic ratio is negative. The neutron’s mass, m n = 1 . m n = 1 . m H = 1 . β − decay into a proton, an electron, and anantineutrino. Although the free neutron’s lifetime is only about 15 minutes,this is long enough for neutron-scattering experiments. For example, a neutronextracted through the beam-tube in a nuclear reactor has typically reachedthermal equilibrium with the water that cools the reactor in a number ofcollisions on its way out (such neutrons usually are called thermal neutrons).Assuming the water has “standard” temperature of 293 K, the neutron’s mostprobable velocity would be about 2200 m/s. It would spend only a fraction ofa second while it travels along the <
100 m beam path in the spectrometerto be scattered by the sample and arrive in the detector.Neutrons used in scattering experiments are non-relativistic. Therefore,the neutron’s energy, E n , is related to its velocity, v n , wave vector, k n = m n v n / ¯ h , and the (de Broglie) wavelength, λ n = 2 π/k n , through E n = 12 m n v n = ¯ h k n m n = h m n λ n . (1.1)Following the notation accepted in particle physics, the neutron’s energy ismeasured in millielectronvolts (meV). The neutron’s wavelength and its wavevector are usually measured in ˚A (1 ˚A = 0 . − cm) and ˚A − ,respectively. Using these units, we can rewrite the Eq. (1.1) in the following,practical fashion: E n = 5 . · − · v n = 2 . · k n = 81 . λ n , (1.2)where E n is in meV, v n in m/s, k n in ˚A − , and λ n in ˚A.For the sake of comparison with the notations used in other techniquesand in theoretical calculations, we list several different ways of representingtypical neutron energies in Table 1.2. The different energy equivalents shownin the Table can be used interchangeably, as a matter of convenience. Neutrons are especially abundant in nuclei of high atomic number, where theycan significantly exceed the number of protons. To create a neutron beam, thefirst challenge is to extract neutrons from the nuclei. The first practical sourcewas the nuclear reactor, in which neutron bombardment of
U nuclei induces
Igor A. Zaliznyak and John M. Tranquada
Table 1.2.
Different notations used to represent the neutron’s energy. e is theelectron charge, h is the Plank’s constant, c is the velocity of light, µ B = e / m e c =0 . × − J/T is the Bohr’s magneton, k B is Boltzman’s constant [11]. Alsoshown are the corresponding neutron wave vector k n and the deBroglie wavelength λ n . E n E n /e E n /h E n / ( hc ) E n / (2 µ B ) E n /k B k n λ n (10 − J) (meV) (THz) (cm − ) (T) (K) (˚A − ) (˚A)1.60218 1000 241.799 8065.54 8637.99 11604.5 21.968 0.28600.160218 100 24.1799 806.554 863.799 1160.45 6.9469 0.90440.0801088 50 12.0899 403.277 431.900 580.225 4.9122 1.279090.0240326 15 3.62698 120.983 129.570 174.068 2.6905 2.33530.00160218 1 0.241799 8.06554 8.63799 11.6045 0.69469 9.0445 fission, a process that releases several neutrons per incident neutron, thus al-lowing for a self-sustaining chain reaction. The neutrons that are released havea very large energy, whereas the fission cross section is enhanced by slower neu-trons. The slowing of neutrons can be achieved quite effectively by scatteringfrom hydrogen, especially in the form of H O, which can also act to cool thereactor core. In a research reactor, where one would like to extract some of theneutrons, the reactor moderator can be made more transparent to neutronsby replacing H O with D O (heavy water, with D representing deuterium).Cylindrical thimbles poking into the water moderator provide an escape pathfor neutrons, which form the beams that supply neutron spectrometers.Another approach is to knock the neutrons out of heavy nuclei with high-energy protons from an accelerator. Again, the neutrons that can escape thenuclei have very high energies that must be reduced by multiple scatteringin a moderator. In contrast to a reactor, which produces neutron beams thatare continuous in time, the proton beam provided by an accelerator can bepulsed, so that a spallation source typically has pulsed beams of neutrons.Targets can be made of a heavy metal such as tungsten, but newer sourceswith higher power tend to use liquid mercury in order to allow adequate heatremoval.A list of the major operating spallation sources in the world is given in theupper portion of Table 1.3. Information on the available instrumentation andcapabilities can be obtained from the listed web sites. With a pulsed neutronsource, each burst of neutrons is produced in a narrow time window, so thatone can distinguish between neutrons of different velocities by their traveltime, or “time of flight”. Using a rotating shutter, one can select incidentneutrons of a desired energy; the energy of scattered neutrons can then bedetermined by their time of arrival at a detector.The spallation source SINQ at the Paul Scherrer Institut provides a con-tinuous, rather than pulsed, beam, so its instrumentation has more in common
Neutron Scattering 5 T a b l e . . M a j o r n e u t r o nu s e r f a c ili t i e s p r e s e n t l y i n o p e r a t i o n . F a c ili t y L a b o r a t o r y L o c a t i o n W e b a dd r e ss S pa ll a t i o nS o u r ce s Sp a ll a t i o n N e u t r o nS o u r c e O a k R i d g e N a t i o n a l L a b O a k R i d g e , T N , U S A h tt p : // n e u t r o n s . o r n l. go v / L u j a n N e u t r o nS c a tt e r i n g C e n t e r L o s A l a m o s N a t i o n a l L a b L o s A l a m o s , N M , U S A h tt p : // l a n s c e .l a n l. go v / l u j a n / I S I S R u t h e r f o r d A pp l e t o n L a b D i d c o t , U K h tt p : // .i s i s . s t f c . a c . u k / J - P A R C J a p a n A t o m i c E n e r g y A g e n c y T o k a i, J a p a nh tt p : // j - p a r c . j p / M a t L i f e / e n / S I N Q P a u l S c h e rr e r I n s t i t u t V illi g e n , S w i t z e r l a ndh tt p : // . p s i. c h / s i n q / R e a c t o r F a c i l i t i e s H i g h F l u x I s o t o p e R e a c t o r O a k R i d g e N a t i o n a l L a b O a k R i d g e , T N , U S A h tt p : // n e u t r o n s . o r n l. go v / N I S TC e n t e r f o r N e u t r o n R e s e a r c h N I S T G a i t h e r s bu r g , M D , U S A h tt p : // . n c n r . n i s t . go v / I n s t i t u t L a u e L a n g e v i n I LL G r e n o b l e , F r a n c e h tt p : // .ill. e u / F R M - II T e c hn i s c h e U n i v e r s i t ¨a t M un i c h , G e r m a n y h tt p : // . f r m . t u m . d e / e n / M ¨ un c h e n L a b o r a t o i r e L ´ e o n B r ill o u i n C E A S a c l a y S a c l a y , F r a n c e h tt p : // - ll b . c e a . f r / e n / J RR - J a p a n A t o m i c E n e r g y T o k a i, J a p a nh tt p : // q ub s . j a e a . go . j p / e n i nd e x . h t m l A g e n c y O P A L AN S T O L u c a s H e i g h t s , N S W , h tt p : // . a n s t o . go v . a u / A u s t r a li a HANA R O K A E R I D a e j e o n , S o u t h K o r e a h tt p : // h a n a r o4 u . k a e r i. r e . k r / Igor A. Zaliznyak and John M. Tranquada with reactor facilities, which are listed in the lower portion of Table 1.3. Witha continuous source, it is common to select the desired energy of incident neu-trons by Bragg diffraction from a crystal (or array of crystals). In a triple-axisspectrometer [4], one also uses Bragg diffraction to analyze the energy andmomentum of neutrons scattered by a sample. Again, many of the facility websites provide a wealth of information on spectrometers and capabilities.
Many of the fundamental advantages of neutron scattering techniques arisefrom the fact that the neutron’s interactions with matter are usually weak andare extremely well understood. Hence, neutrons afford direct experimental in-sight into dynamical properties of the material system of interest, unperturbedby the probe and essentially undistorted by the details of its interaction withmatter. These properties contrast favorably with X-ray or charged-particle(electron, muon) techniques, where the probe could significantly perturb thesystem, and the interaction matrix elements between the system and the probeare often very complicated and profoundly impact the physics measured in theexperiment.The scattering of neutrons by an atomic system is governed by two fun-damental interactions. The residual strong interaction (nuclear force) givesrise to scattering by the atomic nuclei (nuclear scattering). The electromag-netic interaction of the neutron’s magnetic moment with the sample’s internalmagnetic fields, mainly originating from the unpaired electrons in the atomicshells, gives rise to magnetic scattering [2, 3, 13–15].Magnetic interaction of a neutron with a single atom is of relativistic originand is very weak, so that magnetic neutron scattering can be treated using theBorn approximation. The interaction potential consists of the dipole-dipoleinteraction with the magnetic moment associated with the electronic spin, µ se = g s s e ≈ − s e ( g s ≈ − . g -factor),ˆ V se ( r ) = − π µ n · µ se ) δ ( r ) − ( µ n · µ se ) r + 3( µ n · r )( µ se · r ) r , (1.3)and the interaction with the electric current associated with the electron’sorbital motion ˆ V sl ( r ) = 2 µ B ( µ n · l e ) r . (1.4)Here ¯ h l e = r × p e is the electron’s orbital angular momentum, and r = r e − r n its coordinate in the neutron’s rest frame.While the neutron’s interaction with the atomic nucleus is strong—thenuclear force is responsible for holding together protons and neutrons in thenucleus—it has extremely short range, < − cm, comparable to the sizeof the nuclei, and is much smaller than the typical neutron’s wavelength.Hence, to describe the neutron’s interaction with the system of atomic nuclei Neutron Scattering 7 in which the typical distances are about 1 ˚A= 10 − cm, a highly accurateapproximation is obtained by using a delta-function for the nuclear scatteringlength operator in the coordinate representation,ˆ b N ( r ) = b δ ( r n − R N ) . (1.5)Here R N is the position of the nucleus and b is the nuclear scattering length,which is usually treated as a phenomenological parameter [16,17] that has beendetermined experimentally and tabulated [18–20]. In the Born approximation,the scattering length would correspond to the neutron-nucleus interactiondescribed by the Fermi’s pseudo-potential [21],ˆ V N ( r n , R N ) = − π ¯ h m n b δ ( r n − R N ) . (1.6)In general, the bound scattering length (that is, for a nucleus fixed in space)is a complex quantity [2, 13], b = b (cid:48) − ib (cid:48)(cid:48) , defining the total scattering cross-section, σ s , and the absorption cross-section far from the nuclear resonancecapture, σ a , through σ s = 4 π | b (cid:48) | σ a = 4 πk i | b (cid:48)(cid:48) | . (1.7)For the majority of natural elements b (cid:48) is close in magnitude to the character-istic magnetic scattering length, r m = − ( g n / r e = − .
391 fm (1 fm = 10 − cm and r e = e / ( m e c ) is the classical electron radius). In a scattering experiment, the sample is placed in the neutron beam hav-ing a well-defined wave vector k i and known incident flux density Φ i ( k i ),and the detector measures the partial current, δJ f ( k f ), scattered into asmall (ideally infinitesimal) volume of the phase space, d k f = k f dk f dΩ f =( m n k f / ¯ h ) dE f dΩ f , near the wave vector k f , as indicated in Fig. 1.1. Thismeasured partial current, normalized to the appropriate phase space elementcovered by the detector, yields the scattered current density. The double dif-ferential scattering cross-section, which is thus measured, is then defined bythe ratio of this scattered current density to the incident neutron flux density,e.g., d σ ( Q , E ) dEdΩ = 1 Φ i ( k i ) δJ f ( k f ) dEdΩ . (1.8)For each incident neutron in the plane wave state e i k i · r n , the incident fluxdensity is Φ i ( k i ) = ¯ hk i /m n . The scattered current density is determined by Igor A. Zaliznyak and John M. Tranquada (cid:101) = 2 (cid:101) s d (cid:49) f E , i i k E < f f kE , i q k k= - f i sample a) (cid:101) = 2 (cid:101) s d (cid:49) f E , i i kq k k= - f i sample b) (cid:101) = 2 (cid:101) s d (cid:49) f E , i i kq k k= - f i sample c) E > f f kE , i E = f f kE , i Fig. 1.1.
Schematics of the scattering process in a neutron scattering experiment,(a) elastic, (b), inelastic, neutron energy loss, (c), inelastic, neutron energy gain. the transition rate Γ i → f from the initial state | k i , S zn,i , η i (cid:105) , where the neutronis in the plane wave state e i k i · r n with the spin S zn,i and the scattering system isdescribed by the set of variables η i , to the final state, | k f , S zn,f , η f (cid:105) . Accordingto scattering theory [5,13,22], the transition rate is determined by the matrixelements of the transition operator (or T -matrix) ˆ T , satisfying certain operatorequations, which depend on the scattering system’s Hamiltonian, ˆ H , and itsinteraction with the neutron, ˆ V , Γ i → f = 2 π ¯ h (cid:12)(cid:12)(cid:12) (cid:104) k f , S zn,f , η f | ˆ T | k i , S zn,i , η i (cid:105) (cid:12)(cid:12)(cid:12) δ (cid:32) ¯ h k i m n − ¯ h k f m n − E (cid:33) . (1.9)Here E = E f ( η f ) − E i ( η i ) is the scattering system’s energy gain. It is con-venient to introduce the scattering length operator, ˆ b , which convenientlyabsorbs several factors,ˆ b ( r n , S n , η ) = − m n π ¯ h (cid:104) k f , S zn,f | ˆ T | k i , S zn,i (cid:105) , (1.10)and its Fourier transform, ˆ b ( q ),ˆ b ( q ) = (cid:90) e − i q · r ˆ b ( r , S n , η ) d r , (1.11)Summing over all possible final scattering states, we obtain the double differ-ential scattering cross-section for a given initial state, | k i , S zn,i , η i (cid:105) , d σ ( Q , E ) dEdΩ = k f k i (cid:88) S zn,f ,η f (cid:12)(cid:12)(cid:12) (cid:104) η f | ˆ b ( − Q ) | η i (cid:105) (cid:12)(cid:12)(cid:12) δ ( E f ( η f ) − E i ( η i ) − E ) , (1.12)where the dependence on the spin-state of the neutron is implicit in ˆ b ( − Q ).The energy and momentum transfer to the sample are governed by the con-servation laws, Q = k i − k f , E = E f ( η f ) − E i ( η i ) = ¯ h m n ( k i − k f ) . (1.13) Neutron Scattering 9
Finally, following Van Hove [23], one can use the integral representation ofthe delta-function expressing the energy conservation in Eq. (1.12), and thetime-dependent scattering length operator whose evolution is governed by thesystem’s Hamiltonian, ˆ b ( q , t ) = e i ˆ Ht/ ¯ h ˆ b ( q )e − i ˆ Ht/ ¯ h , (1.14)to recast the double differential scattering cross-section in the most usefulform of the two-time correlation function, d σdEdΩ = k f k i (cid:88) S zn,f (cid:90) ∞−∞ e − iωt (cid:104) η i | ˆ b † ( − Q )ˆ b ( − Q , t ) | η i (cid:105) dt π ¯ h . (1.15)Here the sum is over all possible final spin states of the scattered neutron, S zn,f ,since in the general case the scattering length operator, ˆ b ( − Q , t ), depends onthe neutron spin, S n . The sum over the final states of the sample has beenabsorbed into the expectation value of the two-time correlation function of thescattering length operator. The minus sign in front of Q in Eq. (1.15) followsfrom the convention adopted in the conservation laws in Eq. (1.13), where ¯ h Q is the momentum transfer to the sample, which is the opposite of the changein the neutron’s momentum. The total measured scattering cross-section isobtained by taking the proper thermal average of Eq. (1.15) over all possibleinitial states, | η i (cid:105) .While the scattered neutron’s wave vector k f is uniquely determined by k i and Q , by virtue of the conservation laws (1.13), the neutron’s spin statecan be changed by transferring the angular momentum ¯ h ( ∆S zn ) = ± ¯ h tothe sample. In a polarized neutron experiment scattering between differentneutron spin states can be measured. In such a case, the scattering lengthoperator in Eq. (1.15) is a matrix with respect to different initial and finalspin state indices; it determines the various spin-flip and non-spin-flip cross-sections [3, 24]. In the more common case of unpolarized neutron scattering,neutron spin indices should be traced out in Eq. (1.15), so that it determinesa single unpolarized neutron scattering cross-section.Finally, we should mention that the double differential cross-sections inEq. (1.12), (1.15) are general expressions obtained from scattering theory andare valid for scattering of any probe particles. The remarkable advantage ofneutron scattering is in the fact that scattering length operators are rathersimple, very well understood, and are directly related to the fundamentalphysical properties of the scattering sample. For scattering from an individual nucleus, the scattering length operator canbe very accurately approximated by a delta-function, Eq. (1.5). For a col-lection of nuclei in a condensed matter system, the total scattering length
E (meV) o -1 ) q (Å -1 ) E ( m e V ) Fig. 1.2.
Phonon-roton dispersion of the elementary excitations in the superfluid He. The points show the compilation of the experimental neutron data presented inRef. [25]. The solid line is the fit of the low- q part of the spectrum to the Bogolyubovquasiparticle dispersion. operator is obtained by adding scattering lengths of all nuclei,ˆ b N ( r n ) = (cid:88) j b j δ ( r n − r j ) , (1.16)where j indexes the nucleus at position r j with scattering length b j . For asystem of identical nuclei, this is just a particle number density operator inthe scattering system, times b ,ˆ b N ( r n ) = b (cid:88) j δ ( r n − r j ) = b ˆ n ( r n ) , ˆ b N ( q ) = b ˆ n q . (1.17)Substituting this into Eq. (1.15) and summing out the neutron’s spin stateswe obtain, d σdEdΩ = k f k i | b | (cid:90) ∞−∞ e − iωt (cid:104) η i | ˆ n Q ˆ n − Q ( t ) | η i (cid:105) dt π ¯ h . (1.18)Therefore, the nuclear cross-section measures the space-time correlation ofthe atom number density in a condensed matter system. This is exactly thequantity of interest in many theories of strongly-correlated quantum systems.One of the first successes of neutron scattering was the measurement of thephonon-roton dispersion of the elementary excitations in superfluid helium-4.Neutron data have confirmed that the shape of the dispersion is in agreementwith that previously postulated by Landau and Feynman, as illustrated inFig. 1.2. This led to the broad acceptance of the neutron scattering techniqueas a prime tool for studying quantum systems. Neutron Scattering 11
Next we consider the case in which two or more types of nuclear scatter-ers (with distinct scattering lengths b j and frequency of occurrence c j ) arepresent in the sample in a random fashion. For example, an element may havemultiple isotopes, each with a distinct b j , or the nuclei have a spin, so that b j depends on the nuclear and neutron spin orientations, or we have at leasttwo elements that are randomly distributed among equivalent positions. Theaverage product of the scattering lengths for any two sites can then be writtenas ( b j b j (cid:48) ) = ( b ) (1 − δ jj (cid:48) ) + b δ jj (cid:48) , (1.19)where b = (cid:88) j c j b j ,b = (cid:88) j c j b j . (1.20)We can then distinguish between coherent scattering, d σ c dEdΩ = k f k i ( b ) (cid:88) jj (cid:48) (cid:90) ∞−∞ e − iωt (cid:104) e − i Q · r j e i Q · r j (cid:48) ( t ) (cid:105) dt π ¯ h , (1.21)which probes the inter-nuclear correlation, and the incoherent scattering, d σ i dEdΩ = k f k i (cid:16) b − ( b ) (cid:17) (cid:88) j (cid:90) ∞−∞ e − iωt (cid:104) e − i Q · r j e i Q · r j ( t ) (cid:105) dt π ¯ h , (1.22)which probes the local autocorrelation of the nuclear position; the angle brack-ets denote the average over the sample state. In Eqs. (1.21) and (1.22) we haveswitched to the co-ordinate representation of nuclear density operator (1.17)and performed the Fourier integration. As a result, nuclear positions r j and r j (cid:48) ( t ) are quantum-mechanical operators and have to be treated appropriatelyin calculating the cross-section [2, 3, 14]. In a crystal, the equilibrium positions of atomic nuclei are arranged on thesites of a lattice, so that the position of each individual nucleus j can berepresented as r j = R j + u j , (1.23)where R j is the lattice site position, and u j is a small displacement of theatomic nucleus from its equilibrium position at R j .Substituting this into Eq. (1.21), one can show that the coherent nuclearcross-section of a monoatomic crystal is given by d σ c dEdΩ = k f k i N ( b ) e −(cid:104) ( Q · u ) (cid:105) (cid:88) j e − i Q · R j (cid:90) ∞−∞ e − iωt e (cid:104) ( Q · u )( Q · u j ( t )) (cid:105) dt π ¯ h , (1.24)Here (cid:104) ( Q · u ) (cid:105) is the time- or lattice-averaged square of the atomic displace-ment from equilibrium in the direction of Q , and we have taken advantage ofthe fact that the correlation function in Eq. (1.21) depends only on relativecoordinates, which allows one summation over the N lattice sites to be com-pleted. The integral contains an exponentiated correlation function of atomicdisplacements. It is useful to consider the series expansion of this term inpowers of pair displacement correlations.In zeroth order, the exponential factor is just 1, and one obtains the ex-pression for the elastic Bragg scattering in a crystal, d σ B dEdΩ = N ( b ) e − W (cid:88) j e − i Q · R j δ (¯ hω ) , (1.25)where we used the conventional notation for the Debye-Waller factor, with W ≡ (cid:104) ( Q · u ) (cid:105) . Using the lattice Fourier representation, this can be recastin the common form d σ B dEdΩ = N V ∗ ( b ) e − W (cid:88) τ δ ( Q − τ ) δ (¯ hω ) , (1.26)where V ∗ = (2 π ) /V is the reciprocal unit cell’s volume ( V is the volume ofthe unit cell in real space), and τ are the vectors of the reciprocal lattice. Ina non-Bravais crystal, where the unit cell contains several atoms, the sum inEq. (1.24) has to be split into the intra-unit cell and the inter-unit cell parts,leading to d σ B dEdΩ = N V ∗ | F N ( Q ) | (cid:88) τ δ ( Q − τ ) δ (¯ hω ) , (1.27)where the intra-unit cell summation yields the nuclear unit cell structurefactor, F N ( Q ) = (cid:88) µ e − W µ b µ e − i Q · r µ , (1.28)and µ indexes atoms in the unit cell. For some reciprocal lattice points, F N ( τ )can be zero, which gives the Bragg peak extinction rules in a non-Bravaiscrystal.Expanding the exponent in Eq. (1.24) to the first order, we obtain a con-tribution to the cross-section that is proportional to the correlation of dis-placements at two different sites. In cases where static disorder is present inthe crystal, such as dislocations or lattice strain, the time-independent cor-relations of displacements between different sites give rise to elastic diffusescattering. Neutron Scattering 13
Calculation of the time-dependent displacements of atomic nuclei fromtheir equilibrium positions in the lattice is achieved by quantizing their vibra-tions in terms of quantum oscillators, called phonons. A phonon is a normalmode of atomic vibration, a coherent wave of atomic displacements in thecrystal. We distinguish phonons with index s . The polarization vector e s (di-rection of atomic displacements) and the dependence of the energy on thewave vector, ¯ hω q s (dispersion), are determined by the the local inter-atomicpotentials. The total number of such modes depends on the number of atomsin the unit cell of the crystal. Only three phonons, which are all acoustic, arepresent for the Bravais lattice, two transverse and one longitudinal. Taking theproper thermal average over the sample’s equilibrium state, the contributionto the neutron scattering is given by d σ ph dEdΩ = k f k i ( b ) e − W (cid:88) s ( Q · e s ) M ω q s × V ∗ (cid:88) τ [ δ ( Q − q − τ ) δ (¯ hω − ¯ hω q s )( n ( ω ) + 1)+ δ ( Q + q − τ ) δ (¯ hω + ¯ hω q s ) n ( ω )] , (1.29)where M is the mass of each nucleus. The thermal factor n ( ω ) = (e ¯ hω/k B T − − (1.30)is the Bose distribution function describing thermal population of the oscilla-tor states for temperature T of the sample. The first term arises from phononcreation and corresponds to the neutron energy loss, while the second termis from an annihilation of a phonon that has been thermally excited in thecrystal and results in the neutron energy gain.For a non-Bravais crystal lattice, there are also optic phonons, arisingfrom the different intra-unit-cell vibrations. The total number of phonons isequal to 3 ν , the number of vibrational degrees of freedom of the ν atomscomprising the basis of the unit cell of the lattice. The contribution of eachof these phonons to the neutron scattering cross-section is d σ ph dEdΩ = k f k i (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:88) µ b µ e − W µ (cid:112) M µ ω q s ( Q · e sµ )e − i Q · r µ (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) × V ∗ (cid:88) τ [ δ ( Q − q − τ ) δ (¯ hω − ¯ hω q s )( n ( ω ) + 1)+ δ ( Q + q − τ ) δ (¯ hω + ¯ hω q s ) n ( ω )] , (1.31)where e sµ is the polarization vector for site µ in mode s . For an acousticphonon in the hydrodynamic, long-wavelength (small q ) and low-energy limit,this reduces to Eq. (1.29), where the total mass of all atoms in the unit cell, M = (cid:80) µ M µ , should be used and one must multiply by the square of theelastic structure factor, | F N ( Q ) | . The magnetic interaction of a neutron with a single atom is very weak, sothe Born approximation provides an extremely accurate account for magneticneutron scattering by the atomic electrons. In this approximation, the tran-sition matrix is given simply by the interaction potential, ˆ T = ˆ V , where wehave to combine the neutron’s interaction with the electron’s spin and orbitalmagnetic moment, Eqs. (1.3) and (1.4). Accurate accounting for the orbitalcontribution to magnetic scattering presents, in general, a rather difficult andcumbersome task [3]. There are some important cases where the orbital con-tribution is not significant, such as transition-metal atoms in a crystal, wherethe local crystal electric field typically quenches the orbital angular momen-tum, or the case of s -electrons, where l = 0. Nevertheless, under some verygeneral assumptions, the neutron’s interaction with the electron orbital cur-rents can be recast in the same way as its interaction with the spin magneticmoment, yielding for the total magnetic scattering length,ˆ b m ( r ) = − m n π ¯ h (cid:16) ˆ V se ( r ) + ˆ V le ( r ) (cid:17) = m n π ¯ h (cid:32) µ n · (cid:88) e (cid:20) ∇ × (cid:20) ∇ × µ e ( r ) r (cid:21)(cid:21)(cid:33) , (1.32)where µ e ( r ) = µ s,e + µ l,e is the sum of the spin and the orbital magnetiza-tion associated with each electron, e . The Fourier transform of the magneticscattering length (1.32), which determines the scattering cross-section, isˆ b m ( Q ) = − m n π ¯ h πQ ( µ n · [ Q × [ Q × m ( Q )]]) . (1.33)Here m ( Q ) is the Fourier transform of the total magnetization density of theatom, m ( Q ) = m S ( Q ) + m L ( Q )= (cid:90) e − i Q · r (cid:88) e ( − µ B s e δ ( r − r e ) + µ l,e ) d r , s e is the spin operator of e − th electron, µ l,e its orbital magnetic momentoperator.The cross product in Eq. (1.33) ensures the important property that onlymagnetization perpendicular to the wave vector transfer, Q , contributes to themagnetic neutron scattering. Adding the contributions from all atoms in thecrystal and averaging over the neutron polarizations, we obtain the magneticneutron scattering cross-section measured in an experiment with unpolarizedneutrons ( α, β = x, y, z ), Neutron Scattering 15 d σ m dEdΩ = k f k i (cid:18) m n ¯ h µ n (cid:19) (cid:88) α,β (cid:18) δ αβ − Q α Q β Q (cid:19) (cid:90) ∞−∞ e − iωt (cid:104) M α Q M β − Q ( t ) (cid:105) dt π ¯ h . (1.34)Here M Q = (cid:88) j e − i Q · R j m j ( Q )= (cid:90) e − i Q · r (cid:88) j m j ( r + R j ) d r is the Fourier transformed magnetization density operator in the crystal.Hence, magnetic neutron scattering measures the time- and space-dependentcorrelations of the magnetization fluctuations in the sample. Introducing thedynamic correlation function, S αβ ( Q , ω ) = (cid:90) ∞−∞ e − iωt (cid:104) M α Q M β − Q ( t ) (cid:105) dt π ¯ h , (1.35)we can rewrite Eq. (1.34) as d σ m dEdΩ = k f k i r m (cid:88) α,β (cid:18) δ αβ − Q α Q β Q (cid:19) µ B ) S αβ ( Q , ω ) , (1.36)where r m = − µ B µ n (2 m n / ¯ h ) = − . · − cm is the characteristicmagnetic scattering length. The dynamic correlation function defined above by Eq. (1.35) obeys two im-portant relations that are derived in the linear response theory [2,3,15]. First,it is the detailed balance constraint, which relates the energy gain and theenergy loss scattering at a temperature T , S αβ ( Q , ω ) = e ¯ hω/k B T S βα ( − Q , − ω ) . (1.37)The second is the fluctuation-dissipation theorem (FDT), which relates thescattering intensity with the imaginary part of the dynamic magnetic suscep-tibility, ˜ χ (cid:48)(cid:48) αβ ( Q , ω ) = π (cid:16) − e − ¯ hω/k B T (cid:17) ˜ S αβ ( Q , ω ) . (1.38)Here ˜ χ (cid:48)(cid:48) αβ ( Q , ω ) and ˜ S αβ ( Q , ω ) denote χ (cid:48)(cid:48) αβ ( Q , ω ) and S αβ ( Q , ω ) symmetrizedwith respect to { α, β, Q } → { β, α, − Q } . A system with a center of inversionhas symmetry with respect to { Q } → {− Q } , in which case the tildes canbe dropped for the diagonal components in { α, β } indices. This is the casefor which the FDT is most frequently written [5]. The FDT, Eq. (1.38), is a consequence of the detailed balance condition (1.37) and the causality rela-tions, which require that χ (cid:48)(cid:48) αβ ( Q , ω ) is properly asymmetric. The fundamentallaws of nature expressed in Eqs. (1.37) and (1.38) are extremely useful inperforming and analyzing neutron scattering experiments. If there exists a non-zero equilibrium magnetization in the sample, (cid:104) M Q (cid:105) = (cid:104) M Q ( t ) (cid:105) , where the bar over M Q ( t ) denotes the time-averaging, we can intro-duce magnetization fluctuation around this equilibrium, m Q ( t ) = M Q ( t ) −(cid:104) M Q (cid:105) , and write S αβ ( Q , ω ) = (cid:104) M α Q (cid:105)(cid:104) M β − Q (cid:105) δ (¯ hω ) + S αβ inel ( Q , ω ) , (1.39)where the inelastic component S αβ inel ( Q , ω ) is defined similarly to Eq. (1.35),but with M Q replaced by m Q . The first term here leads to elastic scatteringwhich results from static magnetization in the sample, while the second termdescribes the inelastic magnetic scattering arising from its motion. Substitut-ing the first term into Eq. (1.36) we obtain the unpolarized magnetic elasticcross-section, d σ m,el dEdΩ = r m (2 µ B ) (cid:12)(cid:12) (cid:104) M ⊥ Q (cid:105) (cid:12)(cid:12) δ (¯ hω ) , (1.40)where M ⊥ Q is the Fourier transform of the magnetization component perpen-dicular to the wave vector transfer, Q . Eq. (1.40) applies equally well to all cases where static magnetism is presentin a crystal, whether it is a long-range magnetic order leading to Bragg peaks,or a short-range, e. g. nano-scale magnetic correlation, resulting in an appear-ance of a broad magnetic diffuse scattering. In the case of a long-range order,the magnetization density in a crystal typically has an equilibrium static com-ponent, which is modulated with a wave vector Q m , (cid:104) M ( r ) (cid:105) = m ( r ) + m ( r ) e i Q m · r + m ∗ ( r ) e − i Q m · r , (1.41)where m ( r ) is a real vector function that describes the ferromagnetic com-ponent, if present, while m ( r ) can be complex and describes the staggeredmagnetization. These “Bloch amplitudes” are periodic in the crystal lattice,and therefore can be expanded in the Fourier series, m ( r ) = 1 V (cid:88) τ m τ e i τ · r , m τ = (cid:90) V m ( r ) e − i τ · r , (1.42)where the integral is over the unit cell of the nuclear (paramagnetic) crystallattice. Neutron Scattering 17
Substituting (1.41) and (1.42) into Eq. (1.40), we obtain the followingexpression for magnetic Bragg scattering associated with the long-range mag-netic order at a wave vector Q m , d σ m,B dEdΩ = N r m V ∗ (cid:88) τ (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) m ⊥ , τ µ B (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) δ ( Q − τ ) + (cid:12)(cid:12)(cid:12)(cid:12) m ⊥ τ µ B (cid:12)(cid:12)(cid:12)(cid:12) [ δ ( Q − Q m + τ ) + δ ( Q + Q m + τ )] (cid:33) δ (¯ hω ) . (1.43)Here the summation is over the paramagnetic crystal lattice. This is the “largeBrillouin zone” description, which is the most general one, in that it does notrely on the existence of a commensurate magnetic superlattice with a unitcell containing some integer number of nuclear lattice unit cells, and appliesto incommensurate, as well as commensurate magnetic structures. Such adescription is most convenient for stripe phases in the cuprates, which areoften incommensurate.The intensities of magnetic satellites, | m ⊥ τ | , are given by the Fourier am-plitudes of the magnetization, (1.42), which are obtained by performing theFourier integrals over the unit cell of the paramagnetic lattice. In the casewhere the unit cell magnetization could be approximated by a number ofpoint-like magnetic dipoles µ ν located at positions r ν , these amplitudes be-come the conventional unit cell magnetic structure factors, m ( r ) = (cid:88) j,ν µ ν δ ( r − R j − r ν ) , (cid:12)(cid:12) m ⊥ τ (cid:12)(cid:12) = (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:88) ν µ ⊥ ν e − i τ · r ν (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) . (1.44)In discussing magnetic scattering we assume a rigid lattice, neglecting atomicdisplacements due to disorder and vibrations discussed above. The leadingcorrection to this description is obtained by multiplying expressions for mag-netic cross-section with the Debye-Waller factor, e − W . In many important cases the magnetization density in the crystal is carriedby electrons localized on atomic-like orbitals, which are specified by the localatomic variables, such as spin and orbital quantum numbers. In such cases,the matrix element of the atomic magnetization in the magnetic neutron scat-tering cross-section can be factorized into the product of the reduced matrixelement (form factor), which does not depend on the direction of the atom’sangular momentum quantum numbers, and the Wigner 3j-symbol, which en-tirely accounts for such dependence. Hence, the cross-section can be expressedin terms of a product of the Q − dependent form factor, which accounts for theshape of the magnetization cloud associated with the atomic spin and orbital variables, and a dynamical correlation function between these local angularmomentum variables at different lattice sites.For magnetic ions obeying Hund’s rule, neutron scattering usually probesstates belonging to the same multiplets of the angular momentum, ∆L = 0, ∆S = 0 for the Russel-Saunders atoms with weak spin-orbit and strong crystalfield, or ∆J = 0 for the case of strong spin-orbit coupling, where the totalangular momentum J = L + S is a good quantum number, such as in rareearths. Hence, we can write for the Fourier transform of atomic magnetization, (cid:104) η f | M ( Q ) | η i (cid:105) = − µ B F S ( Q ) (cid:104) η f | S | η i (cid:105) − µ B F L ( Q ) (cid:104) η f | L | η i (cid:105) , (1.45)where the spin and the orbital magnetic form factors are, F S ( Q ) = (cid:104) η (cid:48) f , L, S | (cid:80) e e − i Q · r e ( s e · S ) | η (cid:48) i , L, S (cid:105) S ( S + 1) , (1.46) F L ( Q ) = (cid:104) η (cid:48) f , L, S | (cid:80) e e − i Q · r e ( µ e,l · L ) | η (cid:48) i , L, S (cid:105) µ B L ( L + 1) , (1.47)where we made explicit that initial and final states of the sample belong tothe same L and S multiplet. Similar relations hold for the J multiplet in thestrong spin-orbit coupling limit.Typically it is possible to define an effective spin operator, (cid:104) η f | M ( Q ) | η i (cid:105) = − gµ B F ( Q ) (cid:104) η f | ˜ S | η i (cid:105) , (1.48) F ( Q ) = (cid:104) η (cid:48) f , L, S | (cid:16) M ( Q ) · ˜ S (cid:17) | η (cid:48) i , L, S (cid:105) gµ B ˜ S ( ˜ S + 1) = g S g F S ( Q ) + g − g S g F L ( Q ) . (1.49)where g and g S are the effective g − factors, (cid:104) η f | L + 2 S | η i (cid:105) = g (cid:104) η f | ˜ S | η i (cid:105) , (cid:104) η f | S | η i (cid:105) = g S (cid:104) η f | ˜ S | η i (cid:105) . These expressions are exact in the cases of a J -multiplet, where ˜ S = J , g S = 2( g −
1) and g L = 2 − g , or a pure spinmultiplet, where g = g S and the orbital contribution is absent. They give theleading-order approximation in other cases. If the orbital moment is nearlyquenched, as it is for magnetic d -elements in strong crystal field, then ˜ S ≈ S , g S ≈
2, and the orbital contribution to F ( Q ), is small. Assuming this to bethe case, we shall omit tildes and use S for the effective spin.Using the factorization of atomic magnetization provided by Eqs. (1.48)and (1.49), the magnetic neutron scattering cross-section (1.34) can be recastas d σ m dEdΩ = k f k i r m (cid:88) α,β (cid:18) δ αβ − Q α Q β Q (cid:19) (cid:88) j,j (cid:48) g α,j F ∗ j ( Q )2 g β,j (cid:48) F j (cid:48) ( Q )2 × (cid:90) ∞−∞ e − iωt e − i Q · ( R j − R j (cid:48) ) (cid:104) S αj S βj (cid:48) ( t ) (cid:105) dt π ¯ h , (1.50) Neutron Scattering 19 where we allow for the possibility that the g − factor is anisotropic, and thatboth g α,j and F j ( Q ) could be different for different sites j, j (cid:48) of the lattice.Equation (1.50) relates the magnetic cross-section to the dynamic spin struc-ture factor, which is the Fourier transform of the time-dependent two-pointcorrelation function of the atomic spin variables on the sites of the lattice, S αβ ( Q , ω ) = (cid:90) ∞−∞ e − iωt N (cid:88) j,j (cid:48) e − i Q ( R j − R j (cid:48) ) (cid:104) S αj S βj (cid:48) ( t ) (cid:105) dt π ¯ h . (1.51) S αβ ( Q , ω ) is a quantity which is calculated in theoretical models based on thelocal spin Hamiltonians. It also obeys a number of important relations, knownas sum rules, which are extremely useful in analyzing neutron scattering data.The zero moment sum rule is obtained by integrating Eq. (1.51) in Q and ω ,providing the direct connection of the integral neutron intensity with the spinvalue S in the lattice spin Hamiltonian, (cid:88) α (cid:90) ∞−∞ S αα ( Q , ω ) d q d (¯ hω ) = S ( S + 1) . (1.52)The first moment sum rule relates (cid:80) α (cid:82) ∞−∞ ¯ hωS αα ( Q , ω ) d q d (¯ hω ), which isthe integral oscillator strength of the fluctuation spectrum, with the bondenergies in the spin Hamiltonian, and so on. Representing the neutron scattering cross-section via two-point dynamicalspin correlation function, as in Eq. (1.50), is possible in a large number of im-portant magnetic systems, such as cuprates and other 3 d magnetic insulators.Such a representation is extremely useful, as it allows one to connect the mea-sured magnetic neutron intensity with the theoretically predicted propertiesof model spin Hamiltonians, such as the Heisenberg spin Hamiltonian,ˆ H = (cid:88) j,j (cid:48) J jj (cid:48) S j S j (cid:48) = (cid:88) q N J q S q S − q . (1.53)Here J jj (cid:48) = J ( r jj (cid:48) ) is the exchange coupling between sites j and j (cid:48) , and J q and S q are the lattice Fourier transforms, J q = (cid:88) r jj (cid:48) J jj (cid:48) e − i q · r jj (cid:48) , S q = (cid:88) j S j e − i q · r j . (1.54)In many systems with magnetic order, the average value of spin at eachlattice site in the ground state (GS) is “frozen” at nearly the full saturationvalue, (cid:104) S zj (cid:105) ≈ S . In particular, this is a very good approximation for the semi-classical spins, S (cid:29)
1, in more than one dimension (1D). For quantum spins, S = 1 /
2, and/or in the low-dimensional, or frustrated systems, the order may be weak, or absent, and such a picture is inadequate. Nevertheless, in a largenumber of systems magnetic order in the ground state is well developed, andthe semiclassical spin-wave picture applies.Spin excitations in a magnetic system with a well-ordered ground state,such as a ferromagnet, where all spins are parallel, or a semi-classical anti-ferromagnet, where there are two antiparallel sublattices, can be visualizedas small oscillations of classical spin vectors around their equilibrium posi-tions in the GS spin structure. Their wave-like spatial composition resultsfrom the translational symmetry of the system. Frequencies of such spin-waveoscillations can be calculated from the spin Hamiltonian, such as (1.53), to-tally within the classical mechanics, simply by writing the torque equationsof motion for the classical spin angular momenta. For example, in the caseof the Heisenberg Hamiltonian (1.53) for a magnetically ordered system char-acterized by the ordering wave vector Q (this includes ferromagnetism with Q = 0, as well as antiferromagnetism and helimagnetism), one obtains thespin-wave dispersion [26, 27],¯ hω q = 2 S (cid:115) ( J q − J Q ) (cid:18) J q + Q + J q − Q − J Q (cid:19) . (1.55)This can be recast as ω q = √ ω ω Q , where ¯ hω = 2 S ( J q − J Q ).Spin waves are the normal modes of the linearized equations of motion.They involve small spin deviations that are perpendicular to the equilibriumspin direction. Hence, spin waves are transversely polarized, with two mutuallyorthogonal linear polarizations of spin oscillations possible. For a spin systemon a Bravais lattice there are two spin-wave modes.In a quantum-mechanical treatment of spins, the spin-wave calculationproceeds via an approximate mapping of spin operators to Bose creation-annihilation operators, i.e. to local oscillator modes. Hence, the so obtainedspin-wave theory (SWT) describes spin excitations as coherent waves of smalloscillations around the local equilibrium positions, in many ways similar tophonons. The resulting expression for the spin-wave contribution to the neu-tron magnetic scattering cross-section in a sample with a spiral spin structurewith the propagation vector Q is d σ sw dEdΩ = k f k i r m N (cid:12)(cid:12)(cid:12) g F ( Q ) (cid:12)(cid:12)(cid:12) S V ∗ (cid:88) τ ( n ( ω ) + 1) δ (¯ hω − ¯ hω q ) × (cid:20) (cid:18) Q z Q (cid:19) (cid:114) ω ω Q (cid:16) δ ( Q − q − τ − Q ) + δ ( Q − q − τ + Q ) (cid:17) + (cid:18) − Q z Q (cid:19) (cid:114) ω Q ω δ ( Q − q − τ ) (cid:21) , (1.56)where z is the direction normal to the plane of the spiral, and we have re-stricted consideration to the case of a Bravais lattice and retained only the Neutron Scattering 21 contribution corresponding to creation of a single spin wave. The contribu-tion arising from the absorption of a spin wave is written similarly to thatof a phonon in Eq. (1.29). For a ferromagnet, the single spin-wave magneticcross-section simplifies to, [2, 3], d σ sw dEdΩ = k f k i r m N (cid:12)(cid:12)(cid:12) g F ( Q ) (cid:12)(cid:12)(cid:12) S (cid:32) Q (cid:107) Q (cid:33) × V ∗ (cid:88) τ (cid:104) δ ( Q − q − τ ) δ (¯ hω − ¯ hω q )( n ( ω ) + 1) + δ ( Q + q − τ ) δ (¯ hω + ¯ hω q ) n ( ω ) (cid:105) , (1.57)where Q (cid:107) is the wave vector component along the ferromagnetic ordered mo-ment and we have retained the contributions from both the creation and theabsorption of a spin wave. It is clear from Eq. (1.50) that even the exact knowledge of the dynamicalspin structure factor (available from theory in some special cases, such as inone dimension) is insufficient to reproduce the measured magnetic scatteringcross-section. One also has to know the magnetic form factor, which needsto be obtained from an ab initio calculation of the electronic density in thecrystal.In the most common case of a Hund’s ion with 2 S unpaired electronsforming spin (2 S + 1)-multiplet, the spin magnetic form factor (1.46) becomes F S ( Q ) = 12 S S (cid:88) e =1 (cid:90) e − i Q · r | ψ e ( r ) | d r = 12 S S (cid:88) e =1 F S,e ( Q ) , (1.58)where the sum is only over the unpaired electrons. The single-electron density, | ψ e ( r ) | , is determined from the many-electron atomic wave function through | ψ e ( r ) | = (cid:104) η (cid:48) , L, S | δ ( r − r e ) | η (cid:48) , L, S (cid:105) . The magnetic form factor for an atomis therefore simply an average of those for each of the unpaired electrons.Similarly, the orbital form factor is the Fourier-transformed average densityof the uncompensated orbital currents in the atom.If the average Hartree-Fock potential acting on an unpaired electron e in the atom is spherically symmetric, then the effective one-electron wavefunctions in (1.58) are the eigenfunctions of angular momentum and aretagged by the n, l, m = l z quantum numbers, ψ e ( r ) = ψ n,l,m ( r ). The angularand the radial dependencies of the electronic density factorize, ψ n,l,m ( r ) = R n,l ( r ) Y ml ( θ, φ ), where Y ml ( θ, φ ) is the spherical function giving the depen-dence on the polar angles θ, φ . This so-called central field approximation isgood when the contribution to the potential from electrons in the incomplete shell is small. However, it also becomes exact for an almost-filled shell, withonly a single electron, or a single hole, as in the case of Cu , or for a nearlyhalf-filled shell, because the average potential of the closed, or half-filled shell,is spherically symmetric.In the general case, a single-electron wave function can always be expandedin a series in spherical harmonics. In each term of such an expansion, the ra-dial and the angular parts are again factorized, and the magnetic form factoris a sum of Fourier-transformed terms with different l and m . The same kindof an expansion is encountered in calculating the orbital contribution to themagnetic form factor. This is known as a multipole expansion [3]. The calcu-lations are ion-specific and extremely cumbersome. The general expressionscan be obtained only for the leading, isotropic contributions, in the limit ofsmall wave vector transfer, known as the dipole approximation, F S ( Q ) = (cid:104) j ( Q ) (cid:105) , F L ( Q ) = 12 ( (cid:104) j ( Q ) (cid:105) + (cid:104) j ( Q ) (cid:105) ) , (1.59)where j ( Q ) and j ( Q ) are the l, m dependent radial integrals quantifyingthe radial wave function [5, 18]. The radial integrals for most known mag-netic atoms and ions have been calculated numerically from the appropriateHartree-Fock or Fock-Dirac wave functions and are tabulated in Ref [18]. Thefull F ( Q ) is given by Eq. (1.49).Although the dipole approximation (1.59) is the one most commonly used,it is extremely crude. In particular, it does not account for the anisotropy ofthe magnetic form factors, which can be very important for ions with only oneor two unpaired electrons. The anisotropic magnetic form factor of a single5 d hole in a t g orbital of the magnetic Ir ion in the cubic K IrCl wasstudied in Ref. [28]. The authors found that the anisotropy of the magneticform factor is very large, with an additional enhancement coming from thehybridization of the Ir 5 d -orbital with the Cl p -orbitals.The anisotropy of the magnetic form factor is also very pronounced inLa CuO , YBa Cu O y , and related cuprate materials, including the high- T c superconductors, where in the ionic picture a single unpaired magnetic elec-tron occupies a 3 d x − y orbital. In Ref. [29] the authors found that properlyaccounting for the anisotropy of the Cu magnetic form factor is essential forunderstanding the magnetic Bragg intensities measured in YBa Cu O y atlarge wave vectors, and can also explain the peculiar Q -dependence of the in-elastic magnetic cross-section in this material. Accounting for the anisotropicCu form factor was also very important in analyzing neutron scattering byhigh-energy spin waves in La CuO [30, 31], and the chain cuprates SrCuO and Sr CuO [32, 33]. The magnetic excitations in these cuprate materialsextend to several hundreds of meV. Consequently, the measurements requirevery large wave vector transfers, for which the anisotropy of the magneticform factor is very pronounced.The ionic magnetic form factors for 3 d orbitals can be explicitly computedby Fourier transforming the corresponding spherical harmonics. In particular, Neutron Scattering 23 for the d x − y orbital relevant for Cu one obtains [5], F ( Q ) = (cid:104) j ( Q ) (cid:105) − (cid:104) j ( Q ) (cid:105) (cid:0) − cos θ Q (cid:1) + 956 (cid:104) j ( Q ) (cid:105) (cid:18) −
10 cos θ Q + 353 cos θ Q (cid:19) + 158 (cid:104) j ( Q ) (cid:105) sin θ Q cos (4 φ Q ) , (1.60)where θ Q , φ Q are the polar angles of the wave vector Q in the local coordinatesystem used to specify the proper orbital wave functions in the crystal field.Although using the anisotropic ionic magnetic form factor of Cu is muchbetter than using a spherical form factor of the dipole approximation, it is stillnot sufficient for cuprates, as it neglects the effects of covalency (i.e., chargetransfer to the neighboring oxygen) that are expected to be very significant inthese materials. In Ref. [33] it was discovered that covalent bonding results ina marked modification of the magnetic form factor in the quasi-1D antiferro-magnet Sr CuO . The local structure of the planar Cu–O square plaquettesin this material is essentially identical to that in La CuO . Making use ofa precise theoretical result for the excitation spectrum available in 1D, theauthors demonstrated that a good fit to the data requires a form factor thattakes account of hybridization between the half-filled Cu 3 d x − y orbital andthe ligand O 2 p σ orbitals, as given by a density functional calculation. Thehybridization causes the spin density to be extended in real space, resultingin a more rapid fall off in reciprocal space compared to a simple Cu formfactor, as illustrated in Fig. 1.3. Smaller values of magnetic form factor at rel-atively large wave vectors, where the measurement is performed, lead to thesuppression of magnetic intensity, which could be as large as a factor of two ormore [33]. Finally, we note that a study of covalent NMR shifts by Walstedtand Cheong [34] found that barely 2/3 of the spin density in La CuO resideson the copper sites, in excellent agreement with the Sr CuO neutron data ofWalters et al. [33]. The discovery of high-temperature superconductivity in La − x Ba x CuO (LBCO) came as a considerable surprise [35], as ceramic oxides were gen-erally considered to be poor conductors. The structure of LBCO and relatedcuprates involves CuO layers, with the Cu atoms forming a square latticewith bridging O atoms, as shown in Fig. 1.4(a). Anderson [36] predicted thatthe parent compound, La CuO , should have strong antiferromagnetic (AF)superexchange interactions between nearest-neighbor Cu atoms. The occur-rence of antiferromagnetic order was demonstrated by Vaknin et al. [37] us-ing neutron diffraction on a powder sample of La CuO . As illustrated in q (A ° − ) | F ( q ) | q || z (a) Cu Sr CuO La CuO q (A ° − ) q || y (b) Cu Sr CuO La CuO q (A ° − ) q || x (c) Cu Sr CuO La CuO Fig. 1.3.
Wave vector dependence of the ionic magnetic form factor of Cu given byEq. (1.60) (solid line) and the covalent magnetic form factors for Sr CuO (dashedline) and La CuO (dash-dotted line) obtained from the ab initio density functionalcalculations [33]. Panels (a) - (c) show the dependence along three principal direc-tions. Fig. 1.4(b), the antiferromagnetic N´eel order doubles the size of the unit cellin real space, which results in magnetic superlattice peaks, as shown in (c).Thus, the antiferromagnetic order can be detected through the appearanceof superlattice peaks. The challenge in this case is that one must distinguishfrom structural superlattice peaks due to staggered rotations of CuO octa-hedra [38]. Fortunately, the AF and structural peaks appear at inequivalentpositions.The ordered pattern of the octahedral tilts is associated with an or-thorhombic distortion of the crystal structure that makes the diagonal di-rections of a Cu-O plaquette inequivalent [38], as indicated in Fig. 1.5. Byanalyzing the Q dependence of the AF Bragg peak intensities, it was possibleto determine that the magnetic moments on Cu atoms lie within the CuO planes, pointing along the orthorhombic b axis [37]. Furthermore, it was pos-sible to show that the relative arrangement in neighboring planes is as shownin Fig. 1.5. With the magnetic structure determined, one can evaluate themagnitude of the magnetic moments by normalizing the AF peak intensitiesto the nuclear intensities and correcting for the magnetic form factor. Earlystudies yielded a small ordered moment whose magnitude was correlated withthe magnetic ordering, or N´eel, temperature, T N [39]. Neutron scattering stud-ies on carefully prepared single-crystal samples eventually demonstrated theimpact of interstitial oxygen, within the La O layers [40]. Removing the ex-cess oxygen by annealing, one can achieve T N = 325 K [41] and a magneticmoment of 0 . ± . µ B [39].If one assumes a g factor of roughly 2, then the ordered moment yieldsan average ordered spin (cid:104) S (cid:105) ≈ .
3, compared to the expected S = per Neutron Scattering 25 ab Cu O(a) up down(b) 0 101 hk (c) Fig. 1.4. (a) Structure of a CuO plane, with Cu atoms indicated by filled circlesand O atoms by open diamonds. (b) Schematic of antiferromagnetic order, with al-ternating up (filled circles) and down (open circles) spins. The solid line indicates thechemical unit cell, while the dashed line indicates the doubled area of the antiferro-magnetic unit cell. (c) Reciprocal space showing fundamental Bragg peak positions(filled circles) and antiferromagnetic superlattice peak (open circle) at ( , ). Cu atom. The reduction results from the strongly anisotropic structure andthe low value of the spin. For a two-dimensional magnetic system describedby a Heisenberg spin Hamiltonian, long-range order is destroyed at any fi-nite temperature by thermal excitation of spin fluctuations. For La CuO ,weak (nearly-frustrated) couplings between the planes enable the ordering atfinite temperature [43]. Nevertheless, the spin correlations have a stronglytwo-dimensional (2D) character, as demonstrated by neutron scattering stud-ies [44]. The small magnitude of the spin, combined with the enhanced zero-point spin fluctuations in 2D, puts the system close to a quantum criticalpoint [45]. Although the large fluctuations cause problems for perturbationtheory, spin-wave theory nonetheless yields a result, (cid:104) S (cid:105) = 0 .
3, that is veryclose to the value obtained from experiment [46].The exchange couplings between the spins can be determined by analyzingthe dispersion of the spin excitations, which can be obtained by inelasticscattering measurements on a single-crystal sample. Early studies of La CuO with triple-axis spectrometers demonstrated that the superexchange energy J coupling nearest-neighbor spins is greater than 100 meV, and that effects suchas exchange anisotropy and interlayer coupling are very small [43]. Time-of-flight techniques were required to measure the highest-energy spin waves [47], c a b O (cid:60) (cid:0) La (cid:0) Cu (cid:0) Fig. 1.5.
Structure of La CuO , with arrows indicating the arrangement of themagnetic moments in the antiferromagnetic state. Figure reprinted with permissionfrom Lee et al. [42]. Copyright (1999) by the American Physical Society. and these have been refined over time [30, 31]. The most recent results, fromHeadings et al. [31], are shown in Fig. 1.6. The line through the data pointscorresponds to a fit with linear spin-wave theory, which works surprisinglywell in light of the the large zero-point fluctuations. One impact of the latteris the renormalization factor, with a fitted value of Z d = 0 . ± .
04, thatis required to fit the measured intensity. This is somewhat smaller than thevalue Z d ≈ . Q = ( , J = 143 ± J (cid:48) and J (cid:48)(cid:48) , which are relatively weak, and a significant 4-spin cyclic exchange term J c ≈ . J . The overall bandwidth of the magnetic spectrum is ∼ J . A recentanalysis of the couplings, including J c , in terms of a single-band Hubbardmodel has been given by Dalla Piazza et al. [49].To achieve superconductivity, one must dope charge carriers into the CuO planes. Substituting Ba or Sr for La introduces holes. A small densityof holes, p ≈ p ∼ .
055 yields supercon-ductivity. The maximum superconducting transition temperature T c occursfor p ∼ .
16, with T c heading towards zero for p > .
25. Inelastic neutronscattering studies have been performed on single crystal samples across this
Neutron Scattering 27
In general terms, our results show that at the q ¼ð = ; Þ position the spin waves are more strongly coupledto other excitations than at q ¼ ð = ; = Þ . This couplingprovides a decay process and therefore damps the spinwave, reducing the peak height and producing the tail.The question is, What are these other excitations? Aninteresting possibility is that the continuum is a manifes-tation of high-energy spinon quasiparticles proposed intheoretical models of the cuprates [1–3,13,19–21]. Theseassume that Ne´el order coexists with additional spin cor-relations with the magnetic state supporting both low-energy SW fluctuations of the Ne´el order parameter aswell as distinct high-energy spin- = spinon excitationscreated above a finite energy gap [20,21]. Spinons are S ¼ = quasiparticles which can move in a strongly fluctuatingbackground. The anomaly we observe at ð = ; Þ may beexplained naturally in a model where spinons exist at highenergies and have a d -wave dispersion [20,21] with min-ima in energy at q ¼ ð$ = ; = Þ and ð = ; $ = Þ . Underthese circumstances, the lower boundary of the two-spinoncontinuum is lowest in energy at ð = ; Þ and significantlyhigher at ð = ; = Þ . This provides a mechanism for thespin waves at ð = ; Þ to decay into spinons [with ð = ; $ = Þ ] and those at ð = ; = Þ to be stable.The new features in the collective magnetic excitationsobserved in the present study are (i) a q -dependent continuum and (ii) the q dependence to the intensity ofthe SW pole. We estimate the total observed momentsquared (including the Bragg peak) is h M i ¼ : $ : ! B . The continuum scattering accounts for about 29%of the observed inelastic response. The total moment sumrule [15] for S ¼ = implies h M i ¼ g ! B S ð S þ Þ ¼ ! B . We consider two reasons why we fail to observethe full fluctuating moment of the Cu þ ion. First, ourexperiment is limited in energy range to about 450 meV;thus, there may be significant spectral weight outside theenergy window of the present experiment. Raman scatter-ing [22] and optical absorption [23] spectra show excita-tions up to about 750 meV. Recent RIXS measurementsalso show high-energy features [24] which appear to bemagnetic in origin. The second reason why we may fail tosee the full fluctuating moment may be covalency effects[25,26]. The Cu d x & y and O p x orbitals hybridize to yieldthe Wannier orbital relevant to superexchange. This willlead to a reduction in the measured response. However, the(at most) 36% reduction observed in La CuO is substan-tially less than the 60% reduction recently reported in thecuprate chain compound Sr CuO [26].Our results have general implications for the cuprates.Firstly, they show that the collective magnetic excitationsof the cuprate parent compounds cannot be fully describedin terms of the simple SW excitations of a Ne´el ordered h , k ) (r.l.u.) E ne r g y ( m e V ) a E ne r g y ( m e V ) (3/4,1/4)(1/2,1/2) (1/2,0) (3/4,1/4) (1,0) (1/2,0) I S W ( Q ) ( µ B f. u . − ) c Wave vector ( h , k ) (r.l.u.) 0.20.40.60.811.2 (3/4,1/4)(1/2,1/2) (1/2,0) (3/4,1/4) (1,0) (1/2,0) Wave vector ( h , k ) (r.l.u.) I/I S W d Γ χ ′′ ( Q , ω ) ( µ B e V − f. u . − ) hk E ne r g y ( m e V ) e FIG. 2 (color online). q dependence of the magnetic excitations in La CuO . (a) One-magnon dispersion ( T ¼
10 K ) along lines in(c, inset). Symbols indicate E i : 160 meV ( h ), 240 meV ( ), and 450 meV ( ’ ). The solid line is a SWT fit based on Eq. (1).(b) Measured " ð q ; ! Þ . Dashed circle highlights the anomalous scattering near ð = ; Þ . An @ ! -dependent background determinednear ð ; Þ has been subtracted. (c) One-magnon intensity. Line is a fit to SWT with renormalization factor Z d ¼ : $ : . (d) One-magnon intensity divided by SWT prediction. (e) SWT dispersion (color indicates SW intensity). PRL week ending10 DECEMBER 2010
Fig. 1.6.
Spin wave (a) dispersion and (c) intensity measured in antiferromag-netic La CuO at T = 10 K. Lines through data correspond to fits with spin-wavetheory; the fit to the intensity includes a renormalization factor Z d = 0 . ± . et al. [31]. Copyright (2010) by theAmerican Physical Society. entire doping range [7,9]. A couple of the key features are: 1) the bandwidth ofstrong spin-fluctuation scattering decreases linearly with doping, being quanti-tatively similar to the pseudogap energy extracted from various electron spec-troscopies [9, 50], and 2) the wave vector characterizing the low-energy spinexcitations splits about the AF wave vector, becoming incommensurate [7,51].Insight into the cause of the magnetic incommensurability was provided byneutron diffraction measurements on a closely related material,La . Nd . Sr . CuO [52]. The impact of the Nd substitution is to mod-ify the tilt pattern of the CuO octahedra such that the in-plane Cu-O bonddirections become inequivalent [38]. New superlattice peaks were observed inthis low-temperature phase, with in-plane wave vectors Q = ( ± (cid:15), ) and( , ± (cid:15) ) corresponding to spin order and ( ± (cid:15),
0) and (0 , ± (cid:15) ) associatedwith modulations of atomic positions due to charge order, with (cid:15) ≈ .
12. Such (a) (b)
Fig. 1.7.
Cartoons of equivalent domains of (a) vertical and (b) horizontal bond-centered stripe order within a CuO plane (only Cu sites shown). Note that themagnetic period is twice that of the charge period. The charge density along a stripeis one hole for every two sites in length. The registry of the stripes with respect to thelattice (for example, site-centered vs. bond-centered) has not yet been determinedexperimentally. results have been confirmed in the system La − x Ba x CuO [53, 54]. Analysisof the superlattice peaks indicates that they are evidence for spin and chargestripe order [55,56], as illustrated in Fig. 1.7. Because of the crystal symmetry,the orientation of the stripes rotates 90 ◦ from one layer to the next.The occurrence of maximum stripe order corresponds to a strong suppres-sion of the bulk T c at p ≈ [54, 57], suggesting that stripe order competeswith superconductivity; however, recent studies have demonstrated that 2Dsuperconductivity can coexist with stripe order [58]. It now appears that su-perconducting order can intertwine with stripe order [59]. Thus, understand-ing stripe correlations may provide valuable insights into the nature of thesuperconducting mechanism of cuprates.Neutron scattering on a time-of-flight instrument has been used to char-acterize the spin excitation spectrum in La − x Ba x CuO with x = 1 / Q -integrated spectral weight are shown inFig. 1.8. Above 50 meV, the excitations disperse upwards like antiferromag-netic spin waves with an energy gap; the solid line through the points in eachpanel corresponds to a two-leg spin ladder model with J = 100 meV. Below50 meV, the excitations disperse downwards toward the positions of the in-commensurate magnetic superlattice peaks. When the sample is warmed toa state with no static stripe order, the spectrum maintains its essential fea-tures [53, 61]. It appears that stripes, whether static or dynamic, provide away for the superexchange mechanism to survive when the antiferromagneticlayers are doped with holes.The relevance of charge-stripe order is less clear in cuprates families suchas YBa Cu O y and Bi Sr CaCu O δ ; nevertheless, the dispersion of themagnetic excitations in these compounds (measured by neutron scattering)has been shown to be quite similar to that of LBCO [8,9]. The main differenceis that the low-energy excitations tend to be gapped in the superconductingstate, with a pile up of weight (“resonance” peak) appearing above the gap Neutron Scattering 29
The dispersion measured along Q ¼ (1 þ q , q , 0) is presented inFig. 4b. Another interesting quantity to consider is the function S ( q ), obtained by integrating the magnetic scattering intensity S ( Q , q ) over Q . The results are shown in Fig. 4a. With increasingenergy, S ( q ) initially decreases, and then rises to a broad peak near50–60 meV. At higher energies, S ( q ) gradually decreases. Theseresults are qualitatively similar to earlier results on La x Sr x CuO (ref. 23).One generally determines the nature of magnetic fluctuationsfrom the ordered state with which they are associated. In the case ofLa CuO , the high-energy spin waves are clearly associated with theantiferromagnetic order. Although the excitations in our sample areclearly different from semiclassical spin waves, we neverthelessexpect them to be associated with the stripe order indicated bymagnetic and charge-order superlattice peaks .Is there a simple way to interpret our observations? If, for themoment, we ignore the low-energy incommensurate scattering, thefinite-energy peak in S ( q ) suggests that we are measuring singlet–triplet excitations of decoupled spin clusters. Given the stripe orderin our sample, an obvious candidate for such a cluster would be oneof the magnetic domains shown in Fig. 1a and b, corresponding towhat is commonly called a two-leg spin ladder (Fig. 1c). (This namerefers to the pattern formed by the exchange paths between themagnetic ions.) A spin ladder has the following interesting proper-ties . The superexchange J between neighbouring spins keeps themantiparallel, but there is no static order at any temperature. Thisfluctuating, correlated state is said to be quantum disordered. Thereis a substantial energy gap ( , J ) to the first excited state, and theexcitations disperse only along the ladder direction, not along therungs (see Fig. 1e). To compare with experiment, we have calculated simulatedspectra, see Fig. 3b–e, using the single-mode approximation forthe scattering function of a spin ladder with isotropic exchange(I. Zaliznyak, unpublished work). S ð Q ; q Þ < ð " q q k Þ ½ sin ð q k a = Þ þ sin ð q ’ a = Þ& £ ½ d ð q q q k Þ d ð q þ q q k Þ& Here, q k is measured parallel to the ladder, q ’ is along the rungs,and the dispersion q q k , which is proportional to J , is given by ref. 25(see Fig. 1e). Parts of this scattering function have been tested inmeasurements of ladder excitations on Sr Cu O (ref. 26). In thesimulations, we see that the most intense signal has a diamondshape that disperses outward with energy, similar to the right-handside of Fig. 2.The calculated and measured S ( q ) and q ( q ) are compared inFig. 4a and b, respectively. The agreement is remarkable consideringthe simplicity of the model. The energy scale for the dispersion is setby a single parameter, J , and the value of J is only modestly reducedfrom that in the parent compound, La CuO . The downwarddispersion below 50 meV can be modelled by allowing weakcoupling between the ladders, through the charge stripes, asdemonstrated by the unpublished simulations of R. Konik andF. Essler.For completeness, we note that dispersions with similarities toour data have also been obtained in weak-coupling, itinerant-electron calculations . Although this approach does provide apossible alternative explanation of our results, using it to explain theobserved energy scale for the excitations would require fine tuningof parameters , and the weak-coupling approach also does notexplain the charge order in our sample. Thus, we believe that theladder model, within the stripe picture, provides a more compellingexplanation of the results. Given the similarity with recent measure-ments on YBa Cu O þ x , together with the evidence for spatially-anisotropic magnetic excitations in detwinned YBa Cu O þ x (ref. 29), our results provide support for the concept that chargeinhomogeneity, possibly dynamic in nature, is essential to achievesuperconductivity with a high transition temperature in copperoxides . A Received 26 January; accepted 16 April 2004; doi:10.1038/nature02574.
1. Bednorz,J. G.&Mu¨ller, K.A. Possible high T c superconductivity intheBa-La-Cu-Osystem. Z.Phys.B
Phys. Rev.
Science et al.
How to detect fluctuating stripes in the high-temperature superconductors.
Rev.Mod. Phys.
The Physics of Superconductors
Vol. II : Superconductivity in Nanostructures,High-T c and Novel Superconductors, Organic Superconductors (eds Bennemann, K. H. & Ketterson,J. B.) (Springer, Berlin, in the press); preprint at k http://xxx.arxiv.org/pdf/cond-mat/0206217 l (2002).6. Bourges, P. et al. The spin excitation spectrum in superconducting YBa Cu O . Science
J. Phys. Soc. Jpn c superconductors. Phys. Rev. B
Nature et al.
Dispersion of magnetic excitations in superconducting optimally dopedYBa Cu O . Preprint at k http://xxx.arxiv.org/pdf/cond-mat/0307591 l (2003)11. Arrigoni, E., Fradkin, E. & Kivelson, S. A. Mechanism of high temperature superconductivity in astriped Hubbard model. Preprint at k http://xxx.arxiv.org/pdf/cond-mat/0309572 l (2003)12. Moodenbaugh, A. R., Xu, Y., Suenaga, M., Folkerts, T. J. & Shelton, R. N. Superconducting propertiesof La x Ba x CuO . Phys. Rev. B Ba CuO and La Ba Sr CuO . Preprint at k http://xxx.arxiv.org/pdf/cond-mat/0403396 l (2004)14. White, S. R. & Scalapino, D. J. Density matrix renormalization group study of the striped phase in the2D t - J model. Phys. Rev. Lett. N expansion for frustrated and doped quantum antiferromagnets. Int.J. Mod. Phys. B Figure 4
Experimental results for integrated magnetic scattering and dispersion of theexcitations. a , S ( q ), as defined in the text. Circles denote the E i ¼
80 meV data set;squares denote E i ¼
240 meV; diamonds denote E i ¼
500 meV. In distinguishing themagnetic scattering from other signals, care was taken to avoid strong contributions fromphonon branches at 20 and 47 meV. To obtain only the spin-dependent behaviour, wehave corrected for the anisotropic magnetic form factor . Further investigation isrequired to determine whether or not the sharp feature at 42 meV is actually magnetic. b , Dispersion measured along Q ¼ (1 þ q , q ), with the assumption of symmetryabout q ¼
0. Red lines in a and b are calculated from the two-leg spin ladder model withthe same parameters as in Fig. 3. The black dashed line in a is a lorentzian to describethe low energy signal, and the red dot-dashed line is the sum of the other two curves.Vertical ‘error’ bars in a and b indicate the energy range over which data wereintegrated, while horizontal bars in b indicate the half-widths in q of the fitted gaussianpeaks. letters to nature © Nature Publishing Group
Fig. 1.8. (a) Q -integrated spectral weight and (b) effective magnetic dispersion inthe stripe-ordered phase of La − x Ba x CuO with x = 1 /
8, from [60]. The solid linesthrough the data points are described in the text. In (a), the peak at ∼
40 meV isnow know to be due to a phonon mode. In (b), the effective dispersion is plotted for q along a line through the incommensurate magnetic superlattice peaks. for T < T c . The commonality of the dispersions over a broad energy rangesuggests that the charge and spin correlations in superconducting and stripedcuprates are similar. References
1. C.G. Shull, J.S. Smart, Phys. Rev. , 1256 (1949).2. G.L. Squires, Introduction to the Theory of Thermal Neutron Scattering , 3 rd edn. (Cambridge University Press, Cambridge, 2012)3. S.W. Lovesey, Theory of Neutron Scattering from Condensed Matter (OxfordUniversity Press, 1984)4. G. Shirane, S.M. Shapiro, J.M. Tranquada,
Neutron Scattering with a Triple-Axis Spectrometer: Basic Techniques (Cambridge University Press, Cambridge,2002)5. I.A. Zaliznyak, S.H. Lee, in
Modern Techniques for Characterizing MagneticMaterials , ed. by Y. Zhu (Springer, Heidelberg, 2005)6. I.A. Zaliznyak, in
Handbook for Magnetism and Advanced Magnetic Materials ,ed. by H. Kronmuller, S. Parkin (John Wiley & Sons, UK, 2007)7. R.J. Birgeneau, C. Stock, J.M. Tranquada, K. Yamada, J. Phys. Soc. Jpn. ,111003 (2006)8. J.M. Tranquada, in Handbook of High-Temperature Superconductivity , ed. byJ.R. Schrieffer, J.S. Brooks (Springer, New York, 2007), pp. 257–2980 Igor A. Zaliznyak and John M. Tranquada9. M. Fujita, H. Hiraka, M. Matsuda, M. Matsuura, J.M. Tranquada, S. Wakimoto,G. Xu, K. Yamada, J. Phys. Soc. Jpn. , 011007 (2012).10. P. Bourges, Y. Sidis, C. R. Phys. , 461 (2011).11. P.J. Mohr, B.N. Taylor, D.B. Newell. CODATA Recommended Values of theFundamental Physical Constants: 2010. arXiv:1203.542512. F.E. Wietfeldt, G.L. Greene, Rev. Mod. Phys. , 1173 (2011).13. V.F. Sears, Neutron Optics (Oxford University Press, Oxford, 1989)14. Y.A. Izyumov, R.P. Ozerov,
Magnetic Neutron Diffraction (Plenum Press, NewYork, 1970)15. J. Jensen, A.R. Mackintosh,
Rare Earth Magnetism (Clarendon Press, Oxford,1991)16. E. Fermi, L. Marshall, Phys. Rev. , 666 (1947).17. E. Fermi, L. Marshall, Phys. Rev. , 408 (1947).18. A.J.C. Wilson (ed.), International Tables for Crystallography , vol. C (KluwerAcademic Publishers, Dordrecht, 1995)19. V.F. Sears, Neutron News Atlas of Neutron Resonances , 5th edn. (Elsevier Science,2006)21. E. Fermi, Ric. Sci. , 13 (1936)22. B.A. Lippmann, J. Schwinger, Phys. Rev. , 469 (1950).23. L. Van Hove, Phys. Rev. , 1374 (1954).24. M. Blume, Phys. Rev. , 96 (1961).25. R.J. Donnelly, J.A. Donnelly, R.N. Hills, J. Low Temp. Phys. , 471 (1981)26. J. Villain, J. Phys. Chem. Solids , 303 (1959).27. A. Yoshimori, J. Phys. Soc. Jpn. , 807 (1959).28. J.W. Lynn, G. Shirane, M. Blume, Phys. Rev. Lett. , 154 (1976).29. S. Shamoto, M. Sato, J.M. Tranquada, B.J. Sternlieb, G. Shirane, Phys. Rev.B , 13817 (1993)30. R. Coldea, S.M. Hayden, G. Aeppli, T.G. Perring, C.D. Frost, T.E. Mason, S.W.Cheong, Z. Fisk, Phys. Rev. Lett. , 5377 (2001)31. N.S. Headings, S.M. Hayden, R. Coldea, T.G. Perring, Phys. Rev. Lett. ,247001 (2010)32. I.A. Zaliznyak, H. Woo, T.G. Perring, C.L. Broholm, C.D. Frost, H. Takagi,Phys. Rev. Lett. , 087202 (2004)33. A.C. Walters, T.G. Perring, J.S. Caux, A.T. Savici, G.D. Gu, C.C. Lee, W. Ku,I.A. Zaliznyak, Nat. Phys. , 867 (2009)34. R.E. Walstedt, S.W. Cheong, Phys. Rev. B , 014404 (2001).35. J.G. Bednorz, K.A. M¨uller, Z. Phys. B , 189 (1986)36. P.W. Anderson, Science , 1196 (1987)37. D. Vaknin, S.K. Sinha, D.E. Moncton, D.C. Johnston, J.M. Newsam, C.R.Safinya, J. H. E. King, Phys. Rev. Lett. , 2802 (1987)38. J.D. Axe, M.K. Crawford, J. Low Temp. Phys. , 271 (1994)39. K. Yamada, E. Kudo, Y. Endoh, Y. Hidaka, M. Oda, M. Suzuki, T. Murakami,Solid State Commun. , 753 (1987)40. B.O. Wells, Y.S. Lee, M.A. Kastner, R.J. Christianson, R.J. Birgeneau, K. Ya-mada, Y. Endoh, G. Shirane, Science , 1067 (1997)41. B. Keimer, A. Aharony, A. Auerbach, R.J. Birgeneau, A. Cassanho, Y. Endoh,R.W. Erwin, M.A. Kastner, G. Shirane, Phys. Rev. B , 7430 (1992) Neutron Scattering 3142. Y.S. Lee, R.J. Birgeneau, M.A. Kastner, Y. Endoh, S. Wakimoto, K. Yamada,R.W. Erwin, S.H. Lee, G. Shirane, Phys. Rev. B , 3643 (1999)43. M.A. Kastner, R.J. Birgeneau, G. Shirane, Y. Endoh, Rev. Mod. Phys. , 897(1998)44. G. Shirane, Y. Endoh, R.J. Birgeneau, M.A. Kastner, Y. Hidaka, M. Oda,M. Suzuki, T. Murakami, Phys. Rev. Lett. , 1613 (1987)45. S. Chakravarty, B.I. Halperin, D.R. Nelson, Phys. Rev. Lett. , 1057 (1988)46. E. Manousakis, Rev. Mod. Phys. , 1 (1991).47. S.M. Hayden, G. Aeppli, R. Osborn, A.D. Taylor, T.G. Perring, S.W. Cheong,Z. Fisk, Phys. Rev. Lett. , 3622 (1991)48. J. Lorenzana, G. Seibold, R. Coldea, Phys. Rev. B , 224511 (2005)49. B. Dalla Piazza, M. Mourigal, M. Guarise, H. Berger, T. Schmitt, K.J. Zhou,M. Grioni, H.M. Rønnow, Phys. Rev. B , 100508 (2012)50. C. Stock, R.A. Cowley, W.J.L. Buyers, C.D. Frost, J.W. Taylor, D. Peets,R. Liang, D. Bonn, W.N. Hardy, Phys. Rev. B , 174505 (2010)51. K. Yamada, C.H. Lee, K. Kurahashi, J. Wada, S. Wakimoto, S. Ueki, Y. Kimura,Y. Endoh, S. Hosoya, G. Shirane, R.J. Birgeneau, M. Greven, M.A. Kastner,Y.J. Kim, Phys. Rev. B , 6165 (1998)52. J.M. Tranquada, B.J. Sternlieb, J.D. Axe, Y. Nakamura, S. Uchida, Nature ,561 (1995)53. M. Fujita, H. Goka, K. Yamada, J.M. Tranquada, L.P. Regnault, Phys. Rev. B , 104517 (2004)54. M. H¨ucker, M. v. Zimmermann, G.D. Gu, Z.J. Xu, J.S. Wen, G. Xu, H.J. Kang,A. Zheludev, J.M. Tranquada, Phys. Rev. B , 104506 (2011)55. S.A. Kivelson, I.P. Bindloss, E. Fradkin, V. Oganesyan, J.M. Tranquada, A. Ka-pitulnik, C. Howald, Rev. Mod. Phys. , 1201 (2003)56. J. Zaanen, O.Y. Osman, H.V. Kruis, Z. Nussinov, J. Tworzyd(cid:32)lo, Phil. Mag. B , 1485 (2001)57. N. Ichikawa, S. Uchida, J.M. Tranquada, T. Niem¨oller, P.M. Gehring, S.H. Lee,J.R. Schneider, Phys. Rev. Lett. , 1738 (2000)58. Q. Jie, S.J. Han, I. Dimitrov, J. Tranquada, Q. Li, Physica C , 46 (2012).59. E. Berg, E. Fradkin, S.A. Kivelson, J.M. Tranquada, New J. Phys. , 115004(2009)60. J.M. Tranquada, H. Woo, T.G. Perring, H. Goka, G.D. Gu, G. Xu, M. Fujita,K. Yamada, Nature , 534 (2004)61. G.Y. Xu, J.M. Tranquada, T.G. Perring, G.D. Gu, M. Fujita, K. Yamada, Phys.Rev. B76