nonlinear magneto-optics in garnets - J-Stage

4 downloads 0 Views 1MB Size Report
sixties, see e.g. [1 ~ 7], nonlinear optical phenomena in . magnetically ordered ..... [4J Y. R. Shen, The Principles of Nonlinear Optics (Wiley,. New York, 1984).
Proceedings of Magneto-Optical Recording International Symposium '96, Magn. Soc. Jpn., Vol. 20, Supplement No. SI (1996), pp. 23-28 © 1996 by The Magnetics Society of Japan

NONLINEAR MAGNETO-OPTICS IN GARNETS ·R.V.Pisarev and V.V,Pavlov

Ioffe Physico-Technicallnstitute of the Russian Academy of Sciences. 194021 St.Petersburg, Russia

A,Kirilyuk and Th.Rasing Research Institute for Materials, KUN, 6525ED Nijmegen, The Netherlands Abstract - We report results on a second harmonic generation study of thin films of ma~netic gan:ets grown on (111)" (2.10),

(001), and (110) substrates. Experimental data on the rotational anisotropy of the SHG using simple assumptions about the symmetry of these magnetic films.

ID

magnetic fields are well descnbed

. . KEYWORDS: NONLINEAR OPTICS, GARNET FILMS, MAGNETIC Al'ID CRYSTALLOGRAPHIC CO~lRlBUTIONS, ROTATIONAL ANISOTROPY OF SHG a separation .of surface and bulk contributions is INTRODUCTION impossible without surface modification [15]. Though in principle this is not true for the magnetic For more than thirt.y years the study of nonllnear contributions, bulk-interface separation can still be optical properties in solids has been restricted quite troublesome in that case as well [16]. exclusively to magnetically disordered media. Though Bulk crystals of magnetic garnets like yttrium different theoretical issues of nonlinear optics of iron garnet Y3FesOt:2, possess a crystallographic as magnetic materials have been discussed since the early well as a magnetic centrosymmetric structure [17,18J. sixties, see e.g. [1 ~7], nonlinear optical phenomena in . Consequently, SHG is forbidden in the electric-dipole magnetically ordered materials have been observed approximation. Thin films of magnetic garnets possess only recently, In the electric dipole approximation~ magnetic and magnetooptical properties different from bulk Second Harmonic Generation (SHG) requires the those in bulk crystals [19J. Moreover, it was shown breaking of -space-inversion symmetry, However, the that in thin films of magnetic garnets the inversion overwhelming majority of magnetically ordered symmetry may be broken due to a distortion of the materials., metallic and dielectric, ferromagnetic and crystal structure (see, e. g. Ref. [20]). Therefore, such antiferromagnetic, are centrosymmetric in their bulk garnet films are very suitable materials to elucidate fonn. However, space inversion is broken at the different mechanisms of nonlinear magneto-optical surface and very recently SHG in reflection has been effects, not only for· this group of materials, but for proven to be a versatile tool for studying magnetized other magnetic materials as well. In this paper we surfaces and interfaces of metallic materials with a report results on SHG in thin films of magnetic garnets centrosymmetric bulk crystal structure [6-12]. grown on substrates of four different orientations. We Monolayer sensitivity, quantum well oscillations and show how the coexisting crystallographic and huge nonlinear magneto··optical effects were observed. magnetic electric dipole contributions can be Interestingly, Pustogowa et at. showed theoretically separated. The magnetic part is found to emerge below that the absence of the symmetry forbidden bulk the transition temperature Tc. and at room temperature dipole contributions actually leads to an enhancement the two contributions are of the same order of of the nonlinear magneto~optical effects [13]. magnitude. Our results show that in transmission Nonlinear Kerr rotation close to 900 were indeed experiments both bulk contributions to the SHG observed for thin Fe films which should be compared dominate the surface one. The SHG intensity in garnet to their linear rotation about 0:040 £14]. (In contrast, films appears to be several orders of magnitude higher Reif et al. found a nonlinear Kerr rotation of 14° for than the SHG from magnetized metallic surfaces. the noncentrosymmetric Heusler alloy PtMnSb, In the electric-dipole approximation the nonlinear compared to 1.1 for its linear equivalent [8].) For optical polarization P(2ro) of a magnetic medium nonlinear magneto-optical surface studies, there is possessing a spontaneous magnetization M(O) can be however one problem~ how much electric-quadrupole written in the form and magnetic dipole allowed bulk signals contribute to. i . (2) P(2ro):::: Xijk (-2ro,ro,ro) Elro) Ek(ro) the SHG relative to the electric-dipole contribution allowed at the surface? It can be shown that, generally, +i XljkP) (-2ro,ro,ro,O) Elro) Ek(ro) M1(0), (l)

of

I)

23

Pisarev, R.V. et al.: NONLlNEAR MAGNETO,OPTICS IN GARNETS

The SHG signal was generated by the output at 0.841 ~m of a mode-locked Ti-Sapphire laser working at a repetition frequency of 82 MHz, a pulse width of about 100 fs at an average power on the sample between 100 and 250 mW. The use of this type of laser gave a significant increase of the SHG output as compared to previously reported experiments performed with longer pulses (10-12 ns) and longer wavelength (1.06 ~m) from Nd-YAG lasers [21-23]. Using the same experimental set-up we could compare the relative SHG intensities from magnetic garnet films and from ferromagnetic metal surfaces. The SHG signals were several orders of magnitude higher in the garnet films. At 0.841 ~m (1.474 eV) the linear absorption coefficient of magnetic garnet films and bulk crystals is of the order of ex = 10-20 cm,l [17], but at the frequency of the second harmonic (2.948 e V), ex = 10-20 cm". Therefore in transmission experiments the detected SHG signal can only escape from a backside layer with a thickness of about 1 ~m. Under such circumstances the phase-matching conditions are unimportant [4]. Most of the experiments were done in transmission geometry with a laser beam propagating along the z-axis and with a magnetic field up to H=2.3 kOe applied along the y-axis in the film plane. Rotating the sample by 3600 around the z-axis we could register the rotational anisotropy of the SHG signal with the magnetization being kept along the yaxis and the incoming and outcoming linear light polarization being fixed along the x or y-axis. As we will show below, such an approach allows unambiguous separation between crystallographic and magnetization induced SHG signals due to their different transformation properties.

where Eiro) and Ek(ro) are the incoming fundamental fields. A polar tensor Xijk(2) of rank 3 describes the crystallographic contribution. It is allowed in crystals lacking inversion symmetry. An axial tensor Xijkl(3) of rank 4 describes the magnetization-induced contribution. It is also allowed in noncentrosymmetric crystals. These two contributions to the nonlinear polarization P(2ro) may coexist in the same medium. They are both spontaneous and no external field is required to observe them in a single-domain state. However,there are important differences between them. First, they possess different transformation properties under the symmetry operations of the medium, and as a consequence they vary differently when the incident polarization Ej(ro). varies with respect to the crystal axes. Secondly, these two contributions should vary differently as a function of temperature. The crystallographic part probes the crystal lattice structure and symmetry. It may depict anomalies at structural phase transitions and in particular at transitions from a noncentrosymmetric to a centrosymmetric phase. The magnetization-induced part should reflect a temperature variation of the spontaneous magnetization and thus should vanish at the transition from a magnetically ordered to a paramagnetic state. EXPERIMENTAL The magnetic films were grown by a liquid phase epitaxial method. Four different types of magnetic garnet thin films with substrate orientations (001), (111), (210), and (110) have been studied. Samples of these types differed in film and substrate compositions (see Table 1). Thin wafers of gadolinium gallium garnet Gd3GasO'2 (GGG) and substituted GGG with a larger lattice parameter have been used as substrates. The largest mismatch between lattice parameters of film and substrate was in the case of the (210) film, and the smallest one in the (110) film. In the (111) film the mismatch was negative.

RESULTS AND DISCUSSIONS Fig. 1 shows the experimental results for the (111) oriented film for different polarization combinations and directions of magnetic field. In a

Table 1. Basic parameters of the samples.

Substrate orientation (001) (111) (210) (110)

Film symmetry 4mm 3m m

mm2

Film composition (YbPrh(FeGa)sO,2 (YLuB iMFeGa)sO I 2 (YPrLuBi)J(FeGa)sOI2 (YBi)3(FeGa)s0I2

24

Film 12.4140 12.3720 12.5276 12.382

Lattice parameter (A) Substrate Misfit 0.0353 12.3787 12.3794 -0.0074 12.4789 0.0487 12.377 0.005

Proceedings of Magneto-Optical Recording International Symposium '96, Magn, Soc, Jpn., Vol. 20, Supplement No. SI (1996), pp. 23-28 © 1996 by The Magnetics Society of Japan

a)

120

?.o".

60





' 0..

90 1(+1\1) I(-M)



a)

fit(-M)

0

180

I(-M) 45 --fit(+M)

135

fit(+M) ......... fit(-M)

,

I(+M)



0

180

Q'O



Q'

qq-

0.; 0.;

0:

0:

0-

0",""

'

,

.0

0:

"

0.

.0

"O"°o'ooti"O"O

225

315

240



b)

0.

270 90

b)

1(+M) I(-M)



4

,,

o

180

240

1(+M) I(-M) --fit(+M) ...... fit(-M) 0.

300

270

Fig. 2. Same as Fig. 1 but for the (210) sample

Fig. 1. The rotational anisotropy of the SHG intensity in (111) film: (a) XX polarization combination, (b) YY polarization combination.

demagnetized sample the rotational an isotropy is characterized by a 60° periodicity, as it was reported previously [21-23]. However, in a magnetized sample the rotational anisotropy is characterized by a 1200 periodicity. In the XX polarization geometry there is a strong magnetic effect that shifts the observed 3-fold symmetry by 60°, whereas in the YY geometry, no magnetic effect is observed: For the (210) film, there are clear magnetic effects for both polarization combinations, though again the XX effects are much stronger (Fig. 2). Fig. 3 shows the rotational anisotropy pattern for the (001) sample. The SHG vanished without application of the magnetic field. In a magnetized sample the SHG intensity has the same value for opposite orientations of magnetization. FigA shows the rotational anisotropy pattern in a (110) film. All these observations can be understood by the following symmetry analysis.

25

(111 )-films have the point group symmetry 3m (C3v) [6,22] and the SHG at normal incidence in transmission is allowed for the crystaIlographic and magnetization-induced parts of Eq. (1). The crystallographic part has been discussed in [22J. The magnetization-induced part depends on the combination of incident-outgoing polarizations, as is shown in Table 2 for the case of the electric-dipole approximation. Taking into account both crystallographic and magnetic contributions we get for the SHG intensity Ixx(2ro,