Spin-charge-lattice coupling near the metal-insulator transition in Ca3Ru2O7
C.S. Nelson, H. Mo, B. Bohnenbuck, J. Strempfer, N. Kikugawa, S.I. Ikeda, Y. Yoshida
aa r X i v : . [ c ond - m a t . s t r- e l ] J un Spin-charge-lattice coupling near the metal-insulator transition in Ca Ru O C.S. Nelson and H. Mo
National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY 11973-5000, USA
B. Bohnenbuck
Max-Planck-Institut f¨ur Festk¨orperforschung, Heisenbergstraße 1, D-70569 Stuttgart, Germany
J. Strempfer
Hamburger Synchrotronstrahlungslabor HASYLAB at DeutschesElektronen-Synchrotron DESY, Notkestraße 85, 22603 Hamburg, Germany
N. Kikugawa
National Institute for Materials Science, Tsukuba, Ibaraki 305-0047, Japan
S.I. Ikeda and Y. Yoshida
National Institute of Advanced Industrial Science and Technology, Tsukuba, Ibaraki 305-8568, Japan (Dated: November 30, 2018)We report x-ray scattering studies of the c-axis lattice parameter in Ca Ru O as a functionof temperature and magnetic field. These structural studies complement published transport andmagnetization data, and therefore elucidate the spin-charge-lattice coupling near the metal-insulatortransition. Strong anisotropy of the structural change for field applied along orthogonal in-planedirections is observed. Competition between a spin-polarized phase that does not couple to thelattice, and an antiferromagnetic metallic phase, which does, gives rise to rich behavior for B k b . PACS numbers: 75.30.Kz, 71.30.+h, 75.80.+q, 61.10.Eq
The interplay of spin, charge, and lattice degrees offreedom is believed to underlie the complicated physicsexhibited by many correlated electron systems, such asthe colossal magnetoresistant manganites.[1] An excit-ing corollary of this interplay is the tunability of groundstate properties— e.g., inducing a metal-insulator tran-sition via application of magnetic field, or controllingmagnetic properties via strain— that has the potentialto turn these systems into technologically useful elec-tronic materials.[2] Optimization of the useful properties,however, requires a thorough understanding of the spin-charge-lattice coupling in these materials.Bilayer Ca Ru O provides a particularly rich exam-ple of spin-charge-lattice coupling in a correlated elec-tron system. Metallic Ca Ru O orders antiferromag-netically at T N = 56 K and exhibits a metal-insulatortransition at T m − i = 48 K,[3] although a quasi-2D metal-lic ground state has been reported for T <
30 K.[4]At T m − i , a collapse of the c-axis lattice parameter isobserved,[5] with a concomitant in-plane expansion.[6]This structural change is surprising in that a smallerc-axis lattice parameter is expected to increase the or-bital overlap, which would result in a stabilization ofthe metallic state. Neutron powder diffraction studiessuggest, however, that a decrease in the Ru-O-Ca bondangle that decreases the interlayer electron transfer mayexplain this unexpected behavior.[6] Finally, a spin re-orientation transition, which was indicated by magneticsusceptibility measurements,[3] occurs in the vicinity of T m − i .[4, 7, 8, 9] Neutron powder diffraction[6] confirmedthe proposed low-temperature magnetic structure[7] asconsisting of ferromagnetic bilayers, antiferromagneti-cally coupled, and with moments along the orthorhombic b axis, using the Bb m notation of space group a < b .[10] The spin reorientation rotates the mo-ments such that they lie along a for T > T m − i .The confluence of spin reorientation, structural change,and a metal-insulator transition clearly demonstrates theinterplay of the spin, charge, and lattice degrees of free-dom in Ca Ru O near T m − i , which has been the focus ofrecent transport and magnetization measurements.[7, 11]In this paper, we further this investigation by report-ing x-ray scattering studies of the c-axis lattice param-eter in Ca Ru O , with applied magnetic fields of upto 10 T. Combining this new information with previ-ously reported measurements results in a more completepicture of spin-charge-lattice coupling near the metal-insulator transition in this material. We observe asharp structural change for magnetic field applied alongthe low-temperature hard axis that is driven by strongmagnetoelastic coupling, and ties the reported colossalmagnetoresistance[9] to an increase in the c-axis latticeparameter. For magnetic field applied along the low-temperature easy axis, we observe a gradual change in thec-axis lattice parameter as a function of temperature, andirreversibility as a function of magnetic field. Correlat-ing this behavior with transport and magnetization datasuggests competition between the spin-polarized and an-tiferromagnetic metallic phases.Single crystal samples were grown at the University ofSt. Andrews and at the National Institute of AdvancedIndustrial Science and Technology (AIST), using floatingzone techniques. Detailed information about the growthtechniques and transport behavior of the AIST-grownsamples have been reported elsewhere.[4] The as-grownsamples are shaped like platelets, with the c-axis alongthe short direction, and relatively flat (001) surfaces. Wenote that samples grown at the two institutions are ob-served to be very similar: the mosaic widths, as char-acterized at the (004) reflections, are 0.1–0.2 ◦ , and thetemperatures at which the zero field structural change areobserved differed by ∼ ± ± m − i = 48.14 K) is used, and 0.22 K has been added tothe measured sample temperature for all data collectedfrom the St. Andrews sample. In addition, the Bb m notation of space group b ( a ) axis.[10]X-ray scattering measurements were carried out onwiggler beamline X21 at the National Synchrotron LightSource. A Si(111) double-crystal monochromator wasused to select the incident energy, and energies of 8.9 and12 keV were used during different experimental runs. Themonochromatic beam was focused to a ∼ spot atthe center of rotation of a 2-circle goniometer. Mountedon top of this goniometer were x , y , and z translationstages, ± ◦ orthogonal tilt stages, and a 13 T, split coil,vertical field, superconducting magnet, which was madeby Oxford Instruments. Scattering was carried out in ahorizontal geometry, and an avalanche photodiode wasused as the detector. The platelet-shaped sample wasglued to a brass holder attached to the end of a sam-ple rod such that the c-axis was in the scattering plane(i.e., Q k c ∗ ). The holder enabled a manual rotation ofthe sample about the (001) surface normal, and thereforethe magnetic field could be applied along either the a or b in-plane axis (see inset to Figure 1).After initially cooling the sample to ∼
25 K, zero fieldmeasurements were carried out while increasing the sam-ple temperature. In Figure 1, longitudinal θ -2 θ scansthrough the (0 0 16) reflection are shown as a functionof temperature as T m − i is approached. A clear shift tolower Q due to the c-axis lattice parameter expansion isobserved between the temperatures of 47.8 and 48.55 K.Although the magnitude of the shift is less than the full-width-at-half-maximum of the (0 0 16) reflection, Gaus-sian fits to these data result in a peak position valuewith σ < . ◦ . This translates into a sensitivity to thechange in the c-axis lattice parameter of ∼ a and b in-plane directions. For all fixed-B measurements, the sam-ple was cooled in zero field to a temperature of ∼
25 K,the field was ramped up to its final value, and data werecollected at fixed B while increasing the sample tempera-ture. These fixed-B data sets are summarized in Figure 2,in which the percent change in the c-axis lattice param-eter from its value at T ≈
25 K is plotted as a functionof temperature. For both field orientations, the temper-ature at which the structural change begins is observedto decrease with increasing field, which indicates thatmagnetic field stabilizes the high-temperature structure.However, both the magnitude of the shift, and more dra-matically, the temperature range over which the c-axislattice parameter changes, exhibit strongly anisotropicbehavior for the two field orientations. Focusing first onFigure 2(a) in which B k a , a step-like change in c ofroughly constant magnitude is observed for B ≤ ∼
25 K: the c-axis resistivity at T = 25 K is reported todecrease from its zero field value for B ≥ B k b , the step-like change in c is absent, and a transi-tion temperature is difficult to determine. For B = 8 Tthis is not unexpected since 8 T is greater than the crit-ical field for the metamagnetic transition observed for B k b .[13] The high-temperature asymptotic approach ofthe B = 8 T data set to the other data sets indicatesthat there is no significant change in the c-axis latticeparameter as the field is ramped up to 8 T, and there-fore that the metamagnetic transition to a spin-polarizedphase is not strongly coupled to the lattice. Note thatthis conclusion is also supported by magnetostriction[14]and Raman scattering[15] studies.The field dependence of the structural change shown inFigure 2 can be used to extract a B-T phase diagram. For B k a , the temperature at which the step-like change isobserved is plotted in closed symbols in Figure 3 to indi-cate the phase boundary. As mentioned above, for B k b a clear phase boundary cannot be determined given thenature of the change in c . Therefore two temperatures—indicative of the beginning and the end of the structuralchange for applied fields less than the critical field for themetamagnetic transition ( ∼ B k a coincides with the structuralchange phase boundary for T < T m − i , which indicatesan intimate coupling between the structure and trans-port. For B k b , the structural change also appears tofollow the phase boundary from transport, although onlyfor the dominant resistivity change (i.e., B c , not B c , inLin et al. [11]). Note that this is true not merely for T < T m − i , but that the field dependence of the tempera-ture corresponding to the end of the structural changemimics the behavior of a second phase boundary deter-mined through transport for T > T m − i . This secondphase boundary is marked by a subtle inflection in thec-axis resistivity at which no change in the magnetizationis observed, and was suggested to be a manifestation ofthe crucial role of the orbital degree of freedom.[11] Ourstructural measurements indicate that, in fact, the latticedegree of freedom may be more relevant. Taken together,our results demonstrate a strong charge-lattice couplingover the field and temperature range investigated, andindependent of the orientation of the in-plane field.One obvious question raised by the phase diagram ofFigure 3 is: what is the nature of the phase between thetwo dashed lines, for B k b ? From our fixed-B measure-ments we know that the c-axis lattice parameter gradu-ally increases with increasing temperature in this phase.Fixed-T measurements can add to our understanding byindicating how magnetic field alone affects the structure.For these measurements, the sample was again cooled inzero field to a temperature of ∼
25 K, the sample temper-ature was then increased to the temperature of interest,and data were collected at fixed T while ramping up thefield. In order to correct for a shift in the position of therod with increasing field, which moves the sample awayfrom the center of rotation and therefore shifts the 2 θ position of the Bragg peaks slightly, data collected at T= 70 K was used to measure the rod shift that was thensubtracted from all T < T m − i measurements. A result ofone such fixed-T measurement— with T = 44.3 K and B k b — is displayed in the inset to Figure 3. The structuralchange is observed to occur between field values of 3–5 T,which is consistent with the leading edge phase bound-ary of the structural change in Figure 3. Intriguingly, thetotal change saturates at a mere 0.04%, which is a factorof ∼ c at T = 44.3 Kis much larger for B = 5 T than for B = 8 T. In theinset to Figure 3 this is not the case, which suggests thatthe path taken to the T = 44.3 K and B = 8 T pointin phase space matters. This will be discussed in more detail below.Additional fixed-T measurements were carried out forboth field orientations, and hysteresis curves from thesemeasurements are shown in Figure 4. The B k a dataset measured just below T m − i displays sharp transitionswith no obvious hysteresis, and at a field consistent withthe Figure 3 phase diagram. Note that a fixed-B hys-teresis curve measured at B = 5 T and displayed in theinset also indicates no obvious hysteresis. For B k b ,however, measurements carried out at T = 46 and 47 Kbehave similarly to each other, and quite differently thanthe B k a measurement. That is, the transitions occurover a broad field range and they are irreversible, as ∆ c is 0.01–0.02% as the magnetic field is ramped down to-ward 0 T. This irreversibility may be related to the firstorder nature of the field-induced transition for this ori-entation. It should be noted that a flopside or spin-flopmagnetic structure with moments aligned transverse tothe applied magnetic field has been proposed based uponmagnetization[7] and Raman scattering studies,[15] anda transition between the antiferromagnetic phase and thespin-flop phase is expected to be first order and possiblyhysteretic.[16] Structural irreversibility seen in our datawould then follow from strong magnetoelastic couplingin this material.[17] A spin-flop magnetic phase mightalso explain the path dependence in phase space thatwas mentioned earlier. That is, in the fixed-B measure-ment the T = 44.3 K and B = 8 T point in phase spaceis reached from within the spin-polarized phase, whilein the fixed-T measurement the path includes the pro-posed spin-flop phase. The spin-flop to spin-polarizedtransition is expected to be second order,[16] and there-fore some of the magnetic moments may remain alignedalong a at B = 8 T.A comparison of our structural measurements withthe reported transport and magnetization measurementsnow provides a more complete picture of the behaviorof Ca Ru O near T m − i . Just below T m − i , magneticfield applied along a drives the spin reorientation, and viastrong magnetoelastic coupling, the c-axis lattice param-eter increases. The “colossal” decrease in the resistivityfollows from the charge-lattice coupling that ties togetherthe expanded c-axis lattice parameter and metallic inter-layer transport. For magnetic field applied along b , twodifferent phases compete in the vicinity of T m − i . One ofthese phases is the spin-polarized phase above the meta-magnetic transition. It does not couple to the lattice,and the moderate reduction in the resistivity exhibitedby this phase can be attributed to tunneling aided by fer-romagnetic interlayer coupling.[4, 5] The other phase isthe antiferromagnetic metallic phase with magnetic mo-ments along a (i.e., above the spin reorientation tran-sition), which does couple to the lattice, as describedabove. This phase exhibits an expanded c-axis latticeparameter and low resistivity. As T m − i is approachedfrom below, these two phases compete and the resultappears to be a compromise— a phase with a slightlyexpanded c-axis lattice parameter and a moderate de-crease in the resistivity. Our measurements suggest thatthe c-axis lattice parameter in this phase is unaffected byincreasing field, but that it grows gradually with increas-ing temperature. This behavior is consistent with theproposed spin-flop magnetic structure. X-ray or neutronscattering measurements would be worthwhile to pursuein order to verify this magnetic structure, as well as itsevolution with temperature and magnetic field.In conclusion, we have used high-field x-ray scatter-ing techniques to characterize the lattice degree of free-dom of Ca Ru O near the metal-insulator transition.The anisotropic behavior of the structural change formagnetic field applied along the a and b in-plane direc-tions results in a phase diagram similar to that reportedfrom transport measurements,[7, 11] and indicates strongcharge-lattice coupling in this material. An intriguingphase arises for B k b , in which competition betweenthe spin-polarized phase above the field-induced meta-magnetic transition, and the metallic antiferromagneticphase that occurs above the spin reorientation transition,gives rise to rich behavior that underscores the delicatebalance of the charge, lattice, and spin degrees of freedomin Ca Ru O .We gratefully acknowledge S.C. LaMarra’s assistancewith the magnet and B. Keimer’s critical reading of thepaper. Use of the NSLS was supported by the U.S. DOE,Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886. [1] See, for example, Physics of Manganites , edited by T.A.Kaplan and S.D. Mahanti (Kluwer Academic, 1999).[2] A.R. Bishop, Synth. Met. , 2203 (1997).[3] G. Cao et al. , Phys. Rev. Lett. , 1751 (1997).[4] Y. Yoshida et al. , Phys. Rev. B , 220411(R) (2004).[5] G. Cao et al. , Phys. Rev. B , 060406(R) (2003).[6] Y. Yoshida et al. , Phys. Rev. B , 054412 (2005).[7] S. McCall, G. Cao, and J.E. Crow, Phys. Rev. B ,094427 (2003).[8] B. Bohnenbuck et al. , unpublished.[9] G. Cao et al. , Phys. Rev. B , 184405 (2003).[10] Cao and co-workers report a as the easy axis at low tem-peratures. However our results, as well as resonant x-raymagnetic scattering studies of St. Andrews-grown sam-ples, are consistent with the results of Yoshida et al. ,Phys. Rev. B , 054412 (2005), with b as the low-temperature easy axis.[11] X.N. Lin et al. , Phys. Rev. Lett. , 017203 (2005).[12] G. Cao et al. , Phys. Rev. B , 014404 (2004).[13] G. Cao et al. , Phys. Rev. B , 998 (2000).[14] E. Ohmichi et al. , Phys. Rev. B , 104414 (2004).[15] J.F. Karpus et al. , Phys. Rev. Lett. , 167205 (2004).[16] L.J. de Jongh and A.R. Miedema, Adv. Phys. , 947(2001).[17] D.J. Singh and S. Auluck, Phys. Rev. Lett. , 097203 (2006). B Q c* FIG. 1: Zero field θ -2 θ scans through the (0 0 16) reflectionas a function of temperature. Inset displays the scatteringgeometry.FIG. 2: Temperature dependence of the c-axis lattice param-eter change— relative to its value at T ≈
25 K— in fixed B k a - (a) and b - (b) axis. Dashed lines in (b) are the limits usedto determine the temperature range of the structural change(see Figure 3). FIG. 3: Structural change phase diagram for B applied alongthe a - ( • ) and b - ( ◦ ) axis, and the lines are merely guides forthe eye. Labels, which are taken from reference 7, are for thesake of comparison, and refer to paramagnetic metallic (PM),spin-flop (SF), antiferromagnetic nonmetallic (AFNM), andantiferromagnetic metallic (AFM) phases. Inset displays thefield dependence of the c-axis lattice parameter change at T= 44.3 K for B k b .FIG. 4: Hysteresis curves as a function of B applied along the b - ( ◦ , △ ) and a - ( • ) axis. Inset displays the hysteresis curvefor fixed B (= 5 T) k aa