Ferroelectric polarization induced by spin ordering is of current interest, and various compounds exhibiting simultaneous magnetic and ferroelectric phase transitions, called ‘‘multiferroics’’, have been experimentally studied extensively. In most of these studies, pyroelectric-current measurement has been employed to confirm the existence of ferroelectric polarization. However, this measurement can be performed only after a large electric field is applied to the sample for the purpose of aligning ferroelectric domains, and thus, cannot be regarded as the detection of purely spontaneous polarization in a ferroelectric phase. One of the experimental techniques for detecting ferroelectric polarization without aligning ferroelectric domains is second harmonic generation (SHG). SHG is absent from crystals with inversion symmetry, and thus, the detection of the SH signal in multiferroics gives important information on their ferroelectric polarization. The sample investigated in this study is MnWO4. This compound exhibits three different magnetic phases below TN 1⁄4 13:5K, T2 1⁄4 12:7K, and T1 1⁄4 7:6K, and the elliptical spiral spin phase in the intermediate temperature range (T1 < T < T2) was found to show ferroelectric polarization by pyroelectric-current measuement. One of the advantages of studying MnWO4 for SHG is that MnWO4 is optically transparent for the incident light ( 1064 nm) of the SH measurement. Since most multiferroics are opaque at all photon energies, the SH signal can be detected only in the reflection configuration, but this gives a fairly weak SH signal, which is difficult to distinguish from the SHG induced by inversion-symmetry breaking at the sample surface. Using transparent MnWO4, we can detect the SH signal in the transmittance configuration. A single crystal of MnWO4 was grown by the floatingzone method. The sample was easily cleaved along the ac-plane. We checked the sample quality by X-ray Laue reflection analysis and dielectricand pyroelectric-current measurements. The results were consistent with those reported in ref. 5. The 1064 nm light from a Nd:YAG laser (5 ns pulse width, 1mJ pulse energy, and 9Hz repetition) was used as incident light and focused onto the sample with a 1mm spot, and 532 nm light from the sample was detected by a liquid-N2-cooled CCD detector with appropriate optical filters and a grating spectrometer to eliminate stray light. Although the incident light can penetrate into the sample at sufficiently long distance, a small absorption of the incident light causes serious heating of the sample for low-temperature measurement, and thus, the thickness of the sample has to be as small as possible to minimize such absorption. We were able to obtain such a thin sample only by cleaving the crystal along the ac-plane. On the other hand, ferroelectric polarization is expected along the b-axis from the pyroelectric current measurement, Thus, we set the cleaved sample whose b axis (normal to the sample surface) is canted from the laser beam by 30 so that the electric-field vector of the incident light contains a finite b-component. In this polarization configuration, SHG with any third-order tensors is involved in the emitted light. The sample was cooled by a liquid-He-flow cryostat. We also measured the absorption spectrum of the sample between 0.6 and 3 eV at room temperature. Figure 1 shows the absorption spectrum of MnWO4 with the polarization direction along the ac-plane. The increase in absorption coefficient above 2.4 eV corresponds to the onset of an interatomic absorption, presumably between the O 2p level and the Mn 3d level. The small peak at 2.2 eV corresponds to an intraatomic transition within the Mn2þ ion (3d), i.e., a spin-forbidden transition of an electron from the eg orbital to the t2g orbital. 6) In our experiment on SHG, the energy of incident light (1064 nm, 1.16 eV) is far below the absorption edge and the intraatomic absorption, but the SH light (532 nm, 2.33 eV) is on the intraatomic absorption of Mn2þ. Since this absorption is fairly weak, the optical transmittance of the sample with a 0.11mm thickness used for the SHG measurement amounts to 30% at 532 nm. Figure 2 shows the spectrum of the light emitted from the sample when an incident 1064 nm laser was applied. There is a peak at around 532 nm, which indicates the emission of SH light. The fine structures of the peak are due to the irregular shape of the laser spot in real space (caused by the roughness of the cleaved surface) and a wide-open slit of the spectrometer immediately before the CCD camera (wider than the laser spot). As can be seen, the peak is enhanced at 10 and 8K. Figure 3 shows the temperature dependence of the integrated intensity of the peak at around 532 nm, 1 2 0 200 400
Read full abstract