THE DETECTION OF MAGNETIC PHASE TRANSITIONS IN ER-TM BY ELECTROMAGNETIC ACOUSTIC TRANSDUCERS C M Lim, S Dixon, C Edwards and S B Palmer. Department of Physics University of Warwick Coventry CV 4 7 AL, UK INTRODUCTION The phases in magnetic materials can be studied ultrasonically. Magnetic phase transition points can be determined from discontinuities in ultrasonic velocity as a function of temperature. Conventionally this is done using piezoelectric transducers (normally quartz) pulsed with a tone burst generator. Very accurate velocity measurements are required as the change in velocity is typically 1 part in 10 4 [1]. A common problem experienced with contacting transducers is the fracturing of the acoustic couplant bond at low temperatures. Non-contacting acoustic techniques have no problems with bond failure, electromagnetic acoustic transducers (EMATs) can be used for measurements on rare earth magnetic materials. When an electromagnetic (EM) wave is incident on the surface of a non-magnetic conducting material only a small portion of the wave is absorbed. It then attenuates rapidly within the skin depth, defined by 8 = (npllf)'l!2, where fis the frequency of the EM wave, 11 and p are the permeability and conductivity of the metal respectively [2]. If a spiral coil carrying current Y is placed close to the metal surface, the dynamic magnetic field, H D' from the coil induces an eddy current density which can be approximated to j = V X HD within 8. If a static magnetic field, H s, is present, say normal to the sample surface as shown in figure 1, the conduction electrons experiences a Lorentz force, F = j x B, where B =, lh s. This provide a net displacement of the conduction electrons and in turn displaces the ions from their equilibrium positions. If the coil is pulsed, say with a tone-burst of frequency f, shear waves with this frequency are generated. The polarisation is in the direction of F, i.e. polarised radially. The wave propagation direction is normal to the sample's surface. The acoustic generation efficiency is proportional to H; [2]. The described acoustic generation is referred to as the Lorentz or induction mechanism. In the detection mode, the reverse process applies. Review of Progress in Quantitative Nondestructive Evaluation. Vol. 17 Edited by D.O. Thompson and D.E. Chimenti" Plenum Press, New York, 1998 1451
! H k - wavevector tufnol EMA T spiral coil tt current out of page o current into page Figure 1. Side view of EMAT coil and sample. Applied magnetic field is parallel to the c axis of the sample. The shear wave propagation direction is down the c-axis. Polarisation along the radial direction. In magnetic conductors the acoustic generation not only involves the Lorentz mechanism but also magnetoelastic interactions such as domain wall motions, domain rotations, spin flips in order-order transitions and magnetic phase transitions. Theory on acoustic generation in ferromagnets are discussed in [3-7]. The acoustic efficiency is no longer dependent on Hs but on the magnetic induction B instead. The permeability J.l is not a constant as can be assumed in non-magnetic conductors and the condition where the ultrasound wavelength A» 8 becomes invalid. At magnetic phase transitions an increase in the acoustic generation efficiency has been reported on single crystals of gadolinium and dysprosium [8,9]. EXPERIMENTAL SETUP AND METHOD A MATEC DSP-8000 has been modified for use with electromagnetic acoustic transducers (EMATs) with the aim of studying EMAT acoustic coupling efficiency on single crystal ofer-tm alloy where the atomic percentage oftm is 8.4%. The sample has a hexagonal close packed (hcp) crystal structure. The parallel faces between which the ultrasound propagates are prepared by spark planing and are set 4.301 rom apart. The c axis ofthe crystal is normal to these faces. The magnetic structures of the sample have been determined using neutron diffraction [10]. The Neel temperature is at 85 K where the anti-ferromagnetic ordering is along the c-axis and between 26 K and 47 K the magnetic structure is locked into q=217 c' structure with the basal plane (a-b plane) and c-axis magnetic moments modulated by the same wavevector. Evidence of the decoupling of the basal plane and the c-axis moments is observed between 18 K and 26 K. The q/2 modulation of the basal components persist below 18 K down to 9 K when the wavevector is locked into the q=4/15 c' structure. There is no evidence of a cone phase even at 2 K, the lowest attainable temperature in that experiment. A schematic of the apparatus is shown in figure 2. All the experimental constants and variables can be set using the software except for the attenuator's settings which is manually selected. In addition to the analogue to digital (AID) converter's built-in amplifier a wide band pre-amplifier is required for the detection of the EMAT signal, this is labelled as the EMA T adaptor box in figure 2. This pre-amplifier provides a fixed gain of 50 db at 10 MHz. The original MATEC software is modified to control an Oxford 1452
Temperature Controller PC Running Modified Matec DSP-8000 Variable Attenuator Magnet Power Supply Digitiser Tone-Burst Tuned Amp. Receiver Cryostat and Superconducting Magnet EMAT Adaptor Unit Sample EMAT Figure 2. Schematic of experimental set-up. Modified DSP-8000 software controls the system including signal processing and data storage. Instruments (01) cryostat and superconducting magnet system. The temperature and magnetic field range are; 4 K to 300 K and 0 T to 11 T respectively. Additional modifications to the original software is made on the data storage facilities. Readings from two temperature sensors (carbon-glass type) are stored, where one is used as the control sensor and the other to read the sample's temperature. The captured ultrasound waveform can also be stored in binary format if required. The EMATs used are made with a 0.08 mm diameter copper wire wound into a spiral pancake coil. The coil is glued onto a tufuol base. The spiral coil is placed in light contact with the sample. The applied static magnetic field, H s ' is parallel to the c-axis and the generated shear wave propagates down the c-axis. Since the basal plane of the crystal is isotropic the polarisation of the shear wave does not affect wave propagation down the C-axIS. The EMA T is tuned to resonate at 10 MHz. A 200 V peak to peak voltage is provided by the tone burst generator (MATEC TB-I000) across the EMA T coil. The pulsing rate is 1 khz. The returning signal is first amplified by the pre-amplifier (50 db) and then fed into the 8 bit AID which samples at 100 MHz. The signal is further amplified by a variable gain amplifier, 0 to 70 db, which is software controlled. The AID can capture 64 K of data points per shot but only 8001 points are used and this provided a time window of 80 IlS. An auto-correlation technique is used to extract the time of flight of the echoes. 1453
RESULTS The first set of results, figure 3, shows the EMAT signal amplitude as a function of temperature (4 to 1 00 K) with the presence of static applied magnetic field. Three distinct peaks or increases in the EMA T acoustic coupling efficiency are observed at 20 K, 32 K and 84 K. The signal drops to the noise level between 60 to 70 K. In addition to the shear wave echoes there are also the mode converted echoes plus a longitudinal component observed in the waveform. This complicates the determination of the time of flight of the shear wave echoes and the signal amplitude measurements. Qx) o o t::.. o ast Q75T 10T 15 'ttl 5 o Figure 3 Signal amplitude versus temperature plot for Er916%TmS4%. Radially polarised shear wave propagation down the c-axis. Applied magnetic field parallel to c-axis. Figure 4 shows a contour plot of the EMAT signal amplitude with applied magnetic field and temperature on the x-y plane. It represents a tentative magnetic phase diagram of Er-Tm. The ferromagnetic phase is indicated by the high signal amplitude, labelled F in figure 4. In the area AFI the sample is locked into the q-217c' antiferromagnetic structure. With increasing field the AFI region seems to be destroyed and replaced by the ferromagnetic structure. The EMA T acoustic coupling is efficient before the sample is locked into the c-axis modulated antiferromagnetic phase indicated by AF2. The AF2 area gets reduced as the applied field approaches 2.0 T. 1454
2.0...... E--< -0 1.5 ~ t;:::.;;; u IV c: 00 CIS E 1.0-0.~ c.. 0.. <t: 0.5 4.2 20 40 60 80 100 Figure 4. Contour plot of signal amplitude showing a tentative magnetic phase diagram of Er9L6% Tms.4%. F-ferromagnetic phase, AFl-antiferromagnetic phase described by q=2!7c', AF2- c-axis modulated antiferromagnetic. P-paramagnetic phase. Figure 5 shows shear wave velocity as function of temperature at applied magnetic fields of 0.5 T, 0.75 T and 1.0 T. The velocity increases slowly as the sample is cooled but a change in gradient is observed at 80 K, an indication of a magnetic phase transition. The shear velocity reached a maximum of ~ 1812 ms'] at 60 K and on further cooling the velocity decreases smoothly. CONCLUSION The EMA T acoustic coupling efficiency increases close to the magnetic phase transition temperatures and the EMA T becomes inefficient when the sample locks into the antiferromagnetic phase between - 60 K and ~ 70 K. It is not possible to distinguish conclusively between the Lorentz mechanism and the magnetoelastic interactions, i.e. domain wall motion and domain rotation, from C44 elastic modulus and signal amplitude measurements. However a comparison of the efficiency with non magnetic samples which have high Lorentz coupling (aluminum) indicates magnetic mechanisms are dominant. Additional physical measurements of the sample are needed for a complete analysis of the acoustic generation mechanisms involved. The parameters required are the electrical conductivity, permeability and magnetostriction measurements. However, we have shown here the possibility of using the EMA T acoustic efficiency measurements as a preliminary detection of magnetic phase transition temperatures in single crystals of rare earth. The signal amplitude measurements do not require the sophisticated signal processing needed for the velocity measurements, and have also proved to be more sensitive to phase changes than the shear velocity measurements which only showed small changes in gradient. 1455
1~ r---------------------------~ 11m 1700 1760 "'1740 S 1720.~ 1700 :I) 1~ :> 1000 1640 ~ e 10T "" O.7S T o O.S T 1620 '" 1~ ~~~~r- -. ~~-. ~ 20 40 60 00 100 120 TEJrperat~ (K) Figure 5. EMAT generated shear wave velocity '{ersus temperature plot. Wave propagation down the c-axis. Applied magnetic field parallel to c-axis. REFERENCES 1. R. Truell, C. Elbaum and B.B. Chick 1969, Ultrasonic Methods In Solid State PhYSics, Academic Press, N.Y. and London chap. 2. 2. E.R. Dobbs 1973, PhYSical Acoustic, Vol. X (Ed: Mason W P and Thuston R N), Academic Press, London p. 127-189. 3. MJ.W. Povey, E.R. Dobbs and DJ. Meredith 1973, J. Phys. F. : Metal Phys. 3 p. 234-237. 4. MJ.W. Povey, E.R. Dobbs and DJ. Meredith 1980, J. Phys. F. : Metal Phys. 10 p. 2041-2053. 5. E.A. Turov and V.G. Shavrov 1983, SOy. Phys. Usp. 26(7) p.593-611. 6. V.D. Buchel'nikov and AN. Vasil'ev 1992, SOy. Phys. Usp 35 (3) p. 192-211. 7. V.D. Buchel'nikov and V.G. Sharov 1992, SOy. Phys. Solid State 33 (11) p. 1853-1857. 8. AV. Andrianov, V.D. Buche1'nikov, A.N. Vasil'ev, Yu P. Gaidukov, R.S. Ll'yasov and V.G. Shavrov 1988, SOy. Phys. JETP 67 (II) p. 2331-2337. 9. AV. Andrianov, V.D. Buchel'nikov, AN. Vasil'ev, Yu P. Gaidukov, R.S. Ll'yasov and V.G. Shavrov 1990, SOy. Phys. JETP 70 (5) p. 944-951. 1O.S.R. Pamel, C.M. Lim, R.S. Eccleston, S.B. Palmer, M. Salqueiro da Silva, J.M. Moreira, 1.B. Sousa and GJ. McIntyre To be published in Proceeding of the International Conference on Magnetism, Austrialia (1997). 1456