Appl Phys A (2013) 111:423–430 DOI 10.1007/s00339-013-7679-2 INVITED PAPER Heusler-alloy films for spintronic devices Atsufumi Hirohata · James Sagar · Leonardo Lari · Luke R. Fleet · Vlado K. Lazarov Received: 13 October 2012 / Accepted: 20 March 2013 / Published online: 29 March 2013 © Springer-Verlag Berlin Heidelberg 2013 Abstract The next generation of magnetic memories re- quires an ideal spin-polarised electron source, achievable by using a half-metallic Heusler-alloy film. For Heusler-alloy film implementation, it is critical to realise both large vol- umes of coherent magnetisation reversal and high interfa- cial atomic ordering. In this review we present solutions to satisfy these requirements by measuring activation volumes and observing cross-sectional atomic structures. We find that polycrystalline thin films possess 10 times larger activation volumes than epitaxial ones and also form the perfectly or- dered crystalline phase. These features are very useful for the application of Heusler-alloy films in a future magnetic memory. 1 Introduction 1.1 Magnetic memory development and requirements As the areal density of a hard disk drive (HDD) has been increasing faster than Moore’s law for semiconductor in- A. Hirohata ( ) Department of Electronics, University of York, York YO10 5DD, UK e-mail: atsufumi.hirohata@york.ac.uk Fax: +44-1904-322335 url: http://www-users.york.ac.uk/~ah566/ A. Hirohata PRESTO, Japan Science and Technology Agency, Kawaguchi 332-0012, Japan J. Sagar · L. Lari · L.R. Fleet · V.K. Lazarov Department of Physics, University of York, York YO10 5DD, UK L. Lari York JEOL Nanocentre, University of York, York YO10 5DD, UK tegrated circuits, the lateral dimension of its read head is now below 100 nm. Such steady miniaturisation raises dif- ficulties in both recording and reading. For the future of recording, focused laser light is expected to be employed to induce localised heating to reach temperatures closer to the Curie point of the recording medium. This allows for the switching of the local magnetic moments by applying a magnetic field, generated using a conventional recording head. This technique is known as heat-assisted magnetic recording (HAMR) and is expected to be installed in a com- mercially available HDD within 5 years. It is therefore criti- cal to develop a new read head, capable of handling a smaller change in the stray fields from the surface of the perpendicu- lar recording medium. In order to achieve such a highly sen- sitive reading, both large magnetoresistance (MR) ratios and small resistance–area products (RA) are required as shown in Fig. 1a[1]. Namely, the requirement for the next-generation HDD is above 2 Tbit/in 2 , achievable by satisfying both MR >50 % and RA < 0.1  μm 2 [1]. Furthermore, as a replacement for both the HDD and dynamic random access memory (DRAM), new universal solid-state memories have been proposed, among which one of the most promising candidates is a magnetic random ac- cess memory (MRAM), which has been intensively devel- oped. In 2010, EverSpin Technologies started to ship a 16- Mbit MRAM, operated at a bias voltage of 3.3 V and a current of 60 (110) mA for readout (write-in) at the speed of 35 ns. Toshiba also successfully demonstrated a 64-Mbit MRAM in 2010 and announced plans to fabricate a com- mercial MRAM with Hynix. Grandis-Samsung has been working on production based on their nearly 200 patents on MRAM technologies. A Gbit MRAM is likely to be pro- duced by 2020. The initial architecture of a MRAM consists of write lines, utilising the ampere field arising from a cur- rent flow, to reverse the magnetisation in the free layer of