by Michael A. Fury, Techcet Group
April 20, 2009 – Blogging exclusively for SST, Techcet’s Michael A. Fury sums up papers on the MRS Spring meeting’s final day, with discussion of thin films for magnetic sensors, advances in phase-change memory, and progress in the combinatorial searches for a higher-k dielectric than HfO2 and carbon-doped TaN metal gate compositions for HK.
Ichiro Takeuchi of the U. of Maryland is developing electromechanically coupled bilayer devices of magnetostrictive and piezoelectric thin films for magnetic sensor applications. Freestanding cantilevers of PZT and FeGe films (50μm wide and 500nm thick) have demonstrated sensitivities to nano-tesla levels; with optimization, sensitivity in the pico-tesla range is anticipated. Under the influence of an applied electric field, ferroelastic domain movement in the PZT lattice has been shown to be reversible and stable after removing the field. This mode of operation is the functional opposite of the magnetic field sensor mode, and presents an opportunity for developing this phenomenon as another option for non-volatile memory.
Greg Atwood of Numonyx gave a sweeping overview of recent advances in phase-change memory. These devices are based on a reversible amorphous-to-crystalline phase transition of a chalcogenide circuit element that results in a large resistance change. A common choice is Ge2Sb2Te5, which can become amorphous in 1nsec and recrystallize in 10nsec. Materials of this class are used currently for optical read/write disks; devices storing up to 128Mb have been fabricated and tested. The phase transition of each cell element is driven by two thermal resistance heaters which rapidly drive the material to over 700°C locally. The intended function is programmable ROM and power-down storage, and not laptop RAM.
Martin Green from NIST presented his combinatorial search for ternary high-k dielectrics similar to today’s favorite HfO2, but with a higher k value in the range of 30-50. TiO2 would be a good choice except for the fact that, in contact with silicon, it forms an undesirable silicide. Choosing a Hf-Ti-Y system, Green was able to map dielectric constant and leakage current as a function of composition, and confirmed the presence of a range that exhibited the target high-k and low leakage values. Further characterization work will include a MEMS-based nano-calorimeter that will map thermal stability vs. composition to detect undesirable phase transitions.
In a related study, Kao-Shuo Chang from the same group at NIST used combinatorial methods to screen for carbon-doped TaN metal gate compositions that could meet the target work function for use in direct contact with a high-k gate dielectric, in this case HfO2. In this technique, 50nm metal gate composition libraries were fabricated atop 3nm HfO2 films which were deposited on p+ substrates having SiO2 thicknesses ranging from 4-10nm. Varying the SiO2 thickness while keeping all other dimensions constant enables extraction of the metal gate work function from the electrical data. The technique was shown to have worked, but the group is not yet prepared to make recommendations for high-k and metal gate (HK+MG) compositions.
Abstracts for all of the papers presented over the five days of this symposium can be found online.
Michael A. Fury, Ph.D, is senior technology analyst at Techcet Group, LLC, P.O. Box 29, Del Mar, CA 92014; email [email protected].