Nanosecond switching in thin magnetic films

A special pulse equipment including a pulse-sampling oscilloscope with an over-all response time of 0.35 nanosecond sec) for the observation of the nanosecond flux change in thin permalloy films is described. Film switching signals a s short a s 1 nanosecond have been obtained and are discussed with respect to the underlying processes. Inverse switching time versus driving-field curves have been plotted for films of different thicknesses. They show that thinner films switch faster than thicker ones. The slopes of these curves have characteristic values in the nanosecond region of about lo8 per oersted-second. Coherent rotation and oscillation of the magnetization have been clearly detected by picking up the flux change transverse to the driving field. Thin permalloy films with uniaxial anisotropy may behave like a single magnetic domain; flux-changes are then possible by a mere domain rotation. Switching times of about 1 nsec ( sec) for such a “coherent” rotation of the magnetization of the film subjected to a magnetic field pulse have been predicted based on an evaluation of the results of ferromagnetic resonance experiments by the Landau-Lifshitz equation.l> The experimental problem is characterized by the display of the very short signals induced by the rapid flux-change, the length of which lies beyond the time resolution of conventional measuring instruments. By means of oscilloscopes with distributed amplifiers and by travelling wave oscilloscopes, film switching times had been observed down to 10 nsec and 3 nsec.3,4 Even these advanced techniques, however, did not permit a more detailed investigation of the very fast flux-change processes. In this paper, film switching measurements are reported, carried out by an improved technique based on a pulse-sampling oscilloscope with an over-all response time of 0.35 nsec.5 Experimental techniques A magnetic field pulse is generated in a 50-ohm strip transmission line by the discharge of a 50-ohm cable over a coaxial mercury relay (Fig. 1 ) . The rise time of the field pulse is equal to or smaller than the response ime of the oscilloscope. A strip-line, consisting of two qual plates, has been selected because the magnetic field inside the plates is sufficiently homogeneous.6 The stripline is short circuited. Because of the reflected wave, ‘Portions of this material were presented at the Conference on Magnetism and Magnetic Materials, Detroit, November, 1959. the magnetic field is doubled as compared with the matched case. At the other end of the cable this reflected wave is absorbed by a matching network. The diode D serves to disconnect the matching resistor RM from the cable during the charging time between the pulse intervals. In order to compensate for the response time of the diode an rc network is provided. In addition to the pulse field, dc fields and reset fields can be set up in the plane of the film by two pairs of Helmholtz coils. The earth’s magnetic field is cancelled by such a dc field. The longitudinal flux-change of the film, the flux change in the direction of the pulse field, is picked up by a wire placed in the symmetry axis of the strip-line (Fig. 2a). The voltage induced by the air flux is cancelled by symmetrical termination of the wire by four resistors. Disturbances picked up from the electrical field of the pulse are reduced by the short-circuiting of the strip-line. The transverse magnetic flux-change, the flux-change perpendicular to the pulse field, is linked by a wire placed normal to the axis of the strip-line (Fig. 2b). Since no air flux is picked up, one end of the wire can be short circuited for the reduction of capacitive disturbances. The remaining disturbances have been minimized by the use of a thin, resistive pick-up wire. An additional small wire above it has been found useful for capacitive compensation.? The signals are repetitively produced at a rate of about 50 cps and fed to the pulse-sampling oscilloscope over type RG 19 U wideband cable with a delay of about 140 nsec.* The signal delay is necessary because of the delayed generation of the sampling pulse. The synchronization pulse for the sampling oscilloscope is de189 IBM JOURNAL APRIL 1960 S T R I P T R A N S M I S S I O N L I N E 71 90 I r C O A X I A L M E R C U R Y R E L A Y Y I b , L C A B L E C R T A