SPIN-PEIERLS TRANSITION IN CUGEO3 - ELECTRON-PARAMAGNETIC-RESONANCE STUDY

Measurements of electron paramagnetic resonance (EPR) at 9 and 35 GHz between 2 and 300 K in singl,e crystals and powder samples of CuGe0 3 are presented. Below 14 K a !arge decrease in the intensity of the eu 2 + EPR signal is obseIVed. Tue data can be interpreted as due to a spin-Peierls transition. However, an alternative interpretation in terms of a simple structural transition cannot be rule::d out. An opening of an energy gap of -25 K is calculated from the analysis. It is now weil established that, as the temperature is reduced, a one-dimensional (lD) metallic lattice of uni formly spaced atoms with a half -filled conduction band can exhibit a Peierls transition. 1 Such a transition results from a distortion of the lattice in which altemate atoms are dis placed in opposite directions. A Splitting of the conduction band and a reduction in the energy of the electrons occupy ing the Iower band is observed. There is a magnetic analog to the electronic Peierls instability, the so-called spin-Peierls (SP) transition. 2 A uniform antiferromagnetic (AF) quantum chain becomes unstable with respect to an underlying lattice distortion which dimerizes into AF chains.2.3 lt been ar gued that for the SP transition to occur, the 30 lattice, where the 1 D magnetic chains are embedded, needs first to undergo a strong softening. A softening of the phonons at high tem perature in the undimerized state has beeo found in the few organic materials where, al much lower temperature, a SP transition has been observed. 4

It is now weil established that, as the temperature is reduced, a one-dimensional (lD) metallic lattice of uniformly spaced atoms with a half -filled conduction band can exhibit a Peierls transition. 1 Such a transition results from a distortion of the lattice in which altemate atoms are displaced in opposite directions. A Splitting of the conduction band and a reduction in the energy of the electrons occupying the Iower band is observed. There is a magnetic analog to the electronic Peierls instability, the so-called spin-Peierls (SP) transition. 2 A uniform antiferromagnetic (AF) quantum chain becomes unstable with respect to an underlying lattice distortion which dimerizes into AF chains.2.3 lt has been argued that for the SP transition to occur, the 30 lattice, where the 1 D magnetic chains are embedded, needs first to undergo a strong softening. A softening of the phonons at high temperature in the undimerized state has beeo found in the few 10 organic materials where, al much lower temperature, a SP transition has been observed. 4 The AF chains are characterized below the SP transition by an energy gap between a nondegenerate singlet ground state and a band of triple excited states. The energy gap is dependent on the degree of altemation and goes to zero in the uniform chain limit. In zero magnetic field the transition is second order and the degree of altematioo increases as the temperature, T, is lowered, reaching its maximum at T= 0 K. 2 -4 Tue first experimental evideoce confirming the existence of such compounds followed the discovery of the 10 organic materials TTF-CuBDT, TTF-AuBDT, and MUM(TCNQh. Their properties can be satisfactorily explained within the framework of the theory of a SP transition. 5 In many instances tbe spin-Peierls nature has been disputed, questioning if the doubling of the period of the 1 D unit cell is due to a SP instability. Tue observed alternation of the AF coupling between tbe spins can be just a consequence of a simple structural transitioo which results in the doubling of the lattice period. lt is then important to find whether or not the AF interaction is essential to the phase transition. Recently the presence of a SP transition has been reported in an inorganic compound, CuGe0 3 • 6 That study showed that the susceptibility, x. in all the directions of the crystal rapidly decreases to zero below 14 K. Besides, the magnetic field dependence of the transition temperature, Tsp, agrees weil with the theoretical predictions and experimental results reported previously for the organic SP systems. Electron paramagnetic resonance (EPR) is an ideal technique to study CuGe0 3 , as Cu 2 + is one of the easiest ions to detect by EPR. This techllique has been shown to be extremely sensitive for studying the dynamics of low-dimensional spin systems, which includes the already weil characterized SP systems. 5 • 7 · 8 If an energy gap opens in this compound, with a nonmagnetic singlet ground state, a rapid decrease of the intensity of the Cu2+ (S= 1/ 2) EPR signal is expected.
In this paper we present measurements of EPR at 9 and 35 GHz between 2 and 300 K on single crystals and powder samples of CuGe0 3 . Single crystals of about 0.1X0.5X3 mm 3 parallel to the ä, b, and c axes, respectively, were obtained. Powder samples were prepared by the usual sintering method. X-ray studies shows no trace of impurity phases and the data are in good agreement with previous reports Oll this compound. 9 Previous EPR has been reported in CuGe0 3 . 10 However, the authors did not observe a decrease in the intellsity of the EPR signal or at least did not report it. lnstead, they observed for T:s;o7 K an increase of the magnetization and a broadening of the EPR linewidth which they attributed to long-range magnetic ordering.
Magnetization data Oll our samples are similar to the data reported by Hase et al. 6 That is, a rapid decrease of the susceptibility is observed below 14 K, with no significant increase below 7 K, contrary to the measurements reported by Petrakovskii et at. 10 A strong EPR signal of Cu 2 + was· measured for the single crystal as for the powder samples for T -;;;; 14 K. The intensity, / , of the signal diminishes rapidly below this temperature. In Fig. 1 where S= 1/2, g is the gyromagnetic tensor, µ 8 is the Bohr mngneton, nnd H the extemal magnetic field. In Fig. 5 the angular variation of the gyromagnetic factor measured at 35 GHz for the three principal lattice planes is given. The g values for the principal axes, obtained from tbe best fit of the data, are listed in Table 1 for three temperatures.
Tue first question to ask is if CuGe0 3 is weil described by 10 Heisenberg AF. If anisotropic components would be present in the coupling between the Cu spins, shifts in the g values as a function of T would be observed. As seen in Tablt: 1 these shifts are small, so the principal exchange mechanism may be assumed to be isotropic.
In Fig. 1 we see that t:.H increases with T between -40 and 300 K. lt has been suggested that at high temperatures Oseroff et al.  much smaJler than the exchange interaction, T<$.J. As a vaJue of J -90 K is derived for CuGe0 3 from data, 6 Eq. (2) should be a good approximation for T~ T sp. Tue best fitting to tbe data taken in a single crystal measured at 9 and 35 GHz, on powder samples measured at 9 GHz, and suscepti- according to mean-field theory the energy gap ä(T) is given by using values ofJ=90 K, 6 p=1.637, 3 -5 and m=43 K from our measurements, we obtained J 2 /J 1 -0. 7, ~0)-0.17, and ä(0)-25 Kin agreement with Hase et al. 6 Preliminary specific heat measurements show a A-shaped anomaJy at -14 K, similar to that obsetved in organic SP J . Appl. Phys., Vol. 75, No. 10, 15 May 1S94 T at low temperature. 16 In summary. our EPR results, preliminary specific heat and high-field de magnetization oould be interpreted as due to an SP transition in CuGe0 3 • However. a careful analysis of the data is needed before we can rule out a structural transition with the doubling of tbe lattice period as the origin of the alternation of the exchange interaction in the spin chain. We wish to thank Professor R. Lilly, P. Lincoln, and M.