NIST uses high-frequency noise to characterize commercial recording heads.

A NIST researcher has completed a study of high-frequency magnetic noise in commercial magneto-resistive recording heads. The study was done in collaboration with two private U.S. companies involved in the manufacture of disk drives and recording heads. The magnetic noise in the read sensors was measured over a frequency range from 0.1 GHz to 8 GHz. Magnetic noise is the component of noise due to thermal fluctuations of the magnetic layers used to sense the fields from data bits. These fluctuations scale inversely with the sensor size and, as the sensor dimensions shrink below 100 nm, can become the dominant source of noise. The peak magnetic noise level in 200 nm devices in modern disk drives is 0.2 nV/[Hz.sup.0.5], which is on the same order as the Johnson noise.

The magnetic noise is measured by subtracting a reference spectrum, in which the magnetization is saturated with a large field, from the noise spectrum This subtraction allows the magnetic noise to be separated from other noise sources and gives a noise floor of 5 pV/[Hz.sup.0.5]. The noise was measured as a function of both current through the sensor and a magnetic field applied in a direction to simulate the bit fields. Because electromagnetic pickup from the environment can be considerably larger than the intrinsic noise signal, the high-frequency magnetic noise was measured in a special low-noise environment.

The noise spectra show several resonant peaks. The largest, near 5 GHz, is due to the uniform rotation of the magnetization near the center of the devices. Additional peaks are observed due to magnetic fluctuations at the ends of the devices, which are more strongly pinned by abutted permanent-magnet thin films. Other higher-frequency peaks are observed that may be due to non-uniform modes in the center of the device. A low-frequency noise component, indicative of non-ideal magnetization structure and domain wall motion, varied considerably from head to head. The strength of the peaks, and to a lesser extent, the frequency positions of the peaks, also varied from head to head.

The NIST researcher compared these noise measurements on recording heads with his earlier noise measurements on specially prepared magnetoresistive structures fabricated at NIST. The NIST devices were designed to allow precise measurement of the high-frequency magnetic noise in magnetoresistive sensors. As a function of temperature and bias field, those noise spectra fit theoretical models. Commercial recording heads, however, are considerably more complicated. The microwave properties of the interconnect structures are less ideal, and the magneto-resistive sensor is surrounded by permanent-magnet and shield films that may alter the noise spectra in complex ways. When the noise spectra are fully understood, they will provide a powerful diagnostic of the magnetic structure of the sensors.

CONTACT: Stephen Russek, [phone]; [e-mail].