Abstract

So far, materials like lead zirconate titanate (PZT) are implemented in ferroelectric field transistors, which possess only limited compatibility with standard semiconductor processing1. Integration into transistor devices is only possible with thick interfacial barriers and ferroelectric layers of almost 200 nm need to be used to achieve a reasonably high memory window2. Hence, no easy scaling of the devices according to current technology node requirements can be realized. This discovery of the ferroelectric properties of doped HfO2 has opened up an interesting path to build highly scalable 28 nm FeFETs for non-volatile data storage.

The polarization hysteresis of HfO2 based ferroelectric capacitors showed a remnant polarization of up to 40 μC/cm² 8. These properties are provoked by a specifically chosen dopant/HfO2 compositions at the structural boundary between the monoclinic and the tetragonal/cubic phase. Here, the existence of a non-centrosymmetric, orthorhombic Pbc21 phase is proposed as the root cause for the ferroelectric properties3. In general, varying the dopant content from 0-10 mol% is causing the formation of different HfO2 phases: a mainly monoclinic lattice for pure HfO2, appearance of the orthorhombic phase for ~5 mol% dopant and the formation of the tetragonal or cubic phase for higher dopant contents. Parallel to the crystallographic transformation, a change of the dielectric properties from paraelectric to ferroelectric is observed. Multiple grains formed ferroelectric domains with a diameter in the 100 nm range. Different dopant atoms in a wide ionic radius range from 50 – 130 pm showed to enable this effect in HfO2 layers4 down to 5 nm film thickness. Switching times in the nanoseconds range were only limited by the RC delay of the measurement set-up5.

Reaching ferroelectricity in CMOS compatible films of only 10nm dielectric thickness enabled scaling of FeFET devices to current 28 nm technology node6. A memory window of 1.2 V was obtained7. Retention characteristics were measured in a wide temperature range and a remaining memory window after 10 years of 0.8-0.9 V was extrapolated at room temperature7. Endurance was verified for more than 104 switching cycles7 and disturb characteristics evaluated.

The work was supported within the scope of technology development by the EFRE fund of the European Community and by the Free State of Saxony (project: Cool Memory).

References

[1] H. P. McAdams at al., IEEE J. of Solid State Circuits, vol. 39, no. 4, pp. 667, 2004

[2] L.V. Hai, M. Takahashi and S. Sakai. IEEE Non-volatile Semiconductor Memory Workshop, June 2011

[3] T.S. Böscke, J. Müller, D. Bräuhaus, U. Schröder and U. Böttger, Appl. Phys. Lett., vol. 99, pp. 102903, 2011.

[4] U. Schroeder et al., ECS Journal of Solid State Science and Technology, 2 (4) N1-N4 (2013)

J. Müller et al., Appl. Phys. Lett., vol. 99, pp. 112901, 2011 and

J. Müller et al., J. Appl. Phys., vol. 110, pp. 114113, 2011 and

S. Mueller et al., Adv. Func. Materials, (2012) 10.1002

[5] J. Müller et al., IEEE Electron Device Letters, vol. 33, 2012, pp. 185–187

[6] J. Müller et al., IEEE Symposia on VLSI Technology and Circuits, 06/2012, Honolulu, Hawaii.

[7] E. Yurchuk et al., IEEE 4th International Memory Workshop, 05/2012, Milano, Italy.

[8] J. Müller et al., IEDM digest of technical papers (2013)

Biography

Uwe Schroeder, Ph.D. has been in a Senior Scientist position at Namlabsince 2009 responsible for all dielectric development activities. The main research topics are DRAM capacitors andthe integration of ferroelectric HfO2-filmsinto memory devices.

Prior to joining Namlab, Schroederwas in a SeniorStaffScientistposition for Qimonda, which was Infineon Technologies Memory Divisionand Siemens Semiconductor before. He jointed Siemensin 1997 for DRAM capacitor development in the DRAM Development Alliance with IBM and Toshiba in Hopewell Jct., NY before transferring to Infineon’s Memory Development Center in Dresden, Germany in 2000. Here, he continued the research on high k dielectricand its integration into DRAM capacitors as a project manager. During this work the so far unknown ferroelectric properties of doped HfO2based dielectrics were found. In 2007 the Infineon Memory Divisionbecame Qimonda. He is (co-)author on more than 100papers and conference contributions and more than 25 patents, including more than 35 publications on ferroelectric HfO2based devices.

Schroederreceived a Master degree in Physics and a PhDdegree in Physical Chemistryfield from University of Bonn, Germanyincluding a research visit at UC California, Berkeley and worked at University of Chicago as a post-doctoral researcher.

▶ 담당교수 : 재료공학부 황 철 성 / 880-7535 ( 문의: 이고은 / 880-5440 ) ◀