Evidence for the charge-density-wave nature of the stripe phase in manganites

Heat capacity and magnetization measurements strongly suggest that the onset of the stripe phase in La0.50Ca0.50MnO3, La0.48Ca0.52MnO3 and Pr0.48Ca0.52MnO3 is due to the formation of a charge density wave. The transition associated with the onset of the stripe phase is second order, and it can be well modelled as a Peierls transition (typically associated with chargedensity-wave formation) in a disordered system (a ‘dirty’ Peierls transition). The entropy change at this transition is very similar for La0.50Ca0.50MnO3, La0.48Ca0.52MnO3 and Pr0.48Ca0.52MnO3, suggesting a common origin. We show that this is not associated primarily with magnetic order since the magnetization varies by a factor of 100 between the compositions while the change in entropy remains roughly constant. (Some figures in this article are in colour only in the electronic version) Many strongly correlated electron systems (e.g., manganites [1, 2], cuprates [3], nickelates [4] and cobaltites [5]) exhibit stripe phases, in which a superstructure forms at low temperatures. The results of transmission electron microscopy (TEM) [1, 2, 6] and neutron diffraction [7, 8] experiments in the manganites led to the suggestion that the superstructure formation was driven by charge separation and localization at atomic sites. However, the interpretation of the high-resolution TEM results has been challenged [9], and more recent experiments with the same technique on layered manganites indicated a smoothly varying superstructure [10]. Recent work has produced conflicting evidence as to the nature of the superstructure, with some studies supporting a model with charge localized at the atomic sites, but with the difference in charge between atomic sites being small [11–15], and others indicating that the superstructure is not tied to the atomic sites [16, 17]. To explain the latter results, it has been proposed that the superstructure resembles a charge density wave (CDW) [10, 16, 18]. In this paper, we show 0953-8984/07/192201+07$30.00 © 2007 IOP Publishing Ltd Printed in the UK 1 J. Phys.: Condens. Matter 19 (2007) 192201 Fast Track Communication Table 1. Average and variance of the radius of the site occupied by rare earth (Re) or alkaline earth (Ae) ions in different compositions of Re1−x Aex MnO3 (the Re/Ae site radius). Here Re is La or Pr, Ae is Ca and x = 0.5 or 0.52 [24]. Average Re/Ae Variance of Re/Ae site radius (Å) site radius (Å2) La0.50Ca0.50MnO3 1.198 3.24 × 10−4 La0.48Ca0.52MnO3 1.197 3.23 × 10−4 Pr0.48Ca0.52MnO3 1.180 2.50 × 10−7 that heat capacity measurements provide strong support for a CDW model of the superstructure in manganites. Previous thermal measurements of La1−x Cax MnO3 with x 0.5 [19–22] have observed two transitions as peaks in the heat capacity when plotted as a function of temperature (T ). The peak at higher T was attributed to order–disorder-type critical fluctuations [19, 20], with a contribution at x = 0.5 from the onset of ferromagnetism (FM). The transition was identified as first order, based on the hysteresis in the resistivity data [21]. The lower T peak was attributed to the transition from a ferromagnetic state to an antiferromagnetism (AFM) [19–22]. Here we use heat capacity and magnetization measurements to gain insight into the nature of these phase transitions. Measurements were made on La0.50Ca0.50MnO3, La0.48Ca0.52MnO3 and Pr0.48Ca0.52MnO3, with the latter two being chosen as compounds with different average cation sizes and variances (see table 1) but in which the superstructure has on average an almost identical wavevector present in all grains, as measured by TEM and neutron diffraction [17]. Since the superstructure is present in all grains with a similar wavevector, it is appropriate to model the system macroscopically as a single ordered phase. The smaller size of the Pr cation is thought to lead to stronger electron–phonon coupling, allowing the superstructure to lock into the lattice in around 25% of the grains [17]. The La0.50Ca0.50MnO3 sample was chosen as it has a nominally commensurate superstructure (though small deviations are seen in TEM measurements [23]) and so provides a contrast between commensurate and incommensurate systems. Samples were prepared by repeated grinding, pressing and sintering of appropriate oxides and carbonates in stoichiometric proportions. The carbonates were decarboxylated by heating for 12 h at 950 ◦C. Each sample was twice reground, repelleted and heated at 1350 ◦C for 4 days. X-ray powder diffraction indicated that the samples were single phase [25]. Heat capacity measurements were made using a Quantum Design Physical Properties Measurement System (PPMS). In order to ensure that the system had reached equilibrium the heat capacity measurements were taken with very dense data points, with a 20 min pause at each T (decreasing the period to 3 min produced substantially different data at of the transitions). The long relaxation times at each T hints at pinning of the superstructure to defects in the system [26]. Magnetic susceptibility measurements were performed using a Quantum Design MPMS. Samples masses were between 30 and 45 mg. The heat capacity data for all three compounds show two transitions (see figure 1), with the transition at higher T exhibiting a much larger change in entropy than the lower transition. In order to make the transitions more visible the background was removed from the data. In the low T (1.8–10 K) range, heat capacity data were fitted to an equation of the form CP = β3T 3 + β5T 5 + γ T + α T 2 + δT , (1) where β3, β5, α, δ and γ are constants; β3 = Nk 12π4 5 1 θ 3 D , where θD is the Debye T . The high T data are modelled with a Debye model and an Einstein mode, where θD has been determined