(Hf,Zr)O₂‐based Ferroelectrics: From Fundamentals to Applications

It is a great pleasure for us to present this Special Issue in physica status solidi (RRL) – Rapid Research Letters devoted to the rapidly developing field of HfO2/ZrO2-based ferroelectrics, related to the Symposium Q held at the 2021 E-MRS Fall Meeting. (Note that due to the unprecedented COVID-19-related scheduling changes, some of the Symposium's contributions have already been covered in the last year's Special Issue, discussed in Ref. 1). Ever since the discovery of ferroelectricity in Si:HfO2 thin films,[2] both the academic and industrial communities have considered the (Hf,Zr)O2-based (HZO) ferroelectrics as a highly promising material system for non-volatile memory and other applications. The main reason for this is the already widespread adoption of HfO2- and ZrO2-based dielectrics into modern complementary metal-oxide-semiconductor (CMOS) devices sparing these materials from many integration problems hampering adoption of the “conventional” perovskite ferroelectrics. This high promise, together with the ability of HZO ferroelectrics to retain large polarization at ultrathin thicknesses (down to just 1 nm in recent reports[3]) and with the advances in creating devices based on these materials, have earned the HZO the name “ferroelectrics of the future”.[4] As our understanding of these materials has improved, the initial doubts that materials with fluorite-based structure could exhibit true ferroelectricity have been quickly dispersed, thanks to both experimental and theoretical work demonstrating clear structural origins of the ferroelectricity[5] and the previously unsuspected abundance of possible polymorphs within the HZO family.[6-9] However, a full quantitative description of the ferroelectric phase stabilization mechanism, and generally of the HZO phase transformation landscape, is still lacking. Probably this should not come as a complete surprise, given that even the phase stabilization in yttria-stabilized tetragonal and cubic zirconia remains not fully understood at the fundamental level after many decades of research, despite being a simpler problem relating to only a bulk phase, unlike the stabilization of the ferroelectric HZO phase that requires that the electrode and finite-thickness effects are taken into account. Deepening our understanding of the fundamental physics of HZO thus remains an important task, and several papers in this issue address it. On the theoretical side, Azevedo Antunes et al. [2100636] provide further insights into the surprising richness of HZO polymorphism, including a prediction of a ferroelectric monoclinic Pc phase with an energy below that of the “usual” orthorhombic ferroelectric Pca21 phase (thus very likely to be experimentally relevant), as well as highlighting the existence of two inequivalent low-energy phases with the same Pbca space group, among other findings. All these phases have relatively large (24 atom) primitive cells, which may be one of the reasons why they were not discussed in earlier theoretical studies, and which also carries implications on the required complexity of the experimental interpretation of structural data. Addressing the problem of modeling HfO2 phase stability at finite temperatures and focusing on the highest-temperature (cubic) phase, Bichelmaier et al. [2100642] discuss development of temperature-dependent effective harmonic potentials, introducing a procedure to efficiently fit them to first-principles data. On the experimental side, Szyjka et al. [2100582] investigate the doping-induced changes in La-doped HfO2, and the interplay of doping with the role of oxygen vacancies in stabilizing the ferroelectric phase. In particular, their hard X-ray photoelectron spectroscopy data confirms that substitutions of Hf by La are ionically compensated by emergence of oxygen vacancies, at least at high La doping levels, and further indicates that the La–O bond length increases with increased doping. From the applications perspective, the research of ferroelectric HZO extends beyond the capacitor- and transistor-based ferroelectric memory applications that have been studied in the early days,[10, 11] and now includes other classes of binary and multi-level (in particular neuromorphic) memory applications, such as ferroelectric tunnel junctions (FTJs),[12, 13] boosted static random-access memory (SRAM), boosted flash, etc., as well as logic applications making use of the “negative capacitance” ferroelectric layer to improve the transistor characteristics[14, 15] and various devices employing the pyro- and piezoelectric properties of the polar phase.[16] This implies the researchers need to address a wide range of application-specific integration challenges. One of the frequently encountered challenges is the requirement of low-temperature deposition compatible with a back-end-of-line (BEOL) process. Wang et al. [2100572] seek to develop BEOL-compatible physical vapor deposition of ferroelectric HZO capacitors for memory applications such as ferroelectric field-effect transistors (FeFETs). Wang et al. perform a systematic study of the film phase content and ferroelectric properties, demonstrating a clear dependence not only on the chemical composition and the anneal temperature, but also on the sputtering power. By optimizing the sputtering power, Wang et al. achieve a fairly sizeable FE phase fraction (remnant polarization 2Pr = 9 μC cm−2 after cycling) in pure HfO2 films with 400 °C anneal, however, higher anneal temperatures and optimization of HZO composition are still required to further increase Pr. Shimura et al. [2100574] address the challenge of low-temperature deposition and ferroelectric performance for a very different application: micro-electromechanical systems (MEMS), particularly piezo-MEMS. While ferroelectricity is typical in HZO films with a thickness of only a few nanometers, MEMS applications require orders of magnitude larger thicknesses. Shimura et al. demonstrate that with a substantial (≈7%) Y doping and using radio-frequency magnetron sputtering methods, ferroelectric films as thick as 1 μm can be deposited without heating on appropriate substrates. The as-deposited films exhibit good ferroelectric properties and uniform texture epitaxially matched to the substrate. Remarkably, the additional heat treatment is not only reported to lead to no appreciable changes in crystal structure or ferroelectric performance but is actually found to degrade the breakdown performance compared to the as-deposited film. Turning back to memory applications, two papers in this issue focus to improve the performance of FeFETs. Mohit et al. [2100581] study the effects of composition as well as annealing ambients and temperature on electrical characteristics of La-doped HZO ferroelectric-gate thin-film transistors with indium-tin-oxide (ITO) channels deposited via a solution process. They show that La doping at optimal concentrations (≈8%) can lead to improved ferroelectricity, lower leakage, reliable switching and good transistor performance with a large memory window and high on-off drain current ratio. Meanwhile Lee et al. [2100532] seek to tackle one of the problems faced by HZO-based FeFETs, namely that of limited endurance, by optimizing device design. The endurance is limited due to degradation of the dielectric layer that is normally positioned between the FE layer and the channel; the known designs mitigating this problem typically come with an increased device footprint thus decreasing the achievable planar density of the FeFET devices. Lee et al. propose a novel vertical-pillar design that does not increase the lateral device footprint, while still reducing degradation thanks to an increased thickness of the dielectric layer and the absence of a direct dielectric-ferroelectric interface. The technology computer-aided design (TCAD) simulations by Lee et al. indicate that in addition to an improved endurance of over 1010 cycles, this design may also allow fast operation and a wide memory window. Last but not least, the development of FTJs is addressed by two papers in this issue, both focusing on designs introducing a thin layer of a metal with a high affinity to oxidation (such as Ti) between the HZO and the top electrode in a sputtered TiN/HZO/TiN stack. Such FTJs designs benefit from a strong asymmetry (required for high on/off ratio); interestingly, they may also result in an improvement of the ferroelectric properties. Segantini et al. [2100583] study such an increase in remnant polarization due to deposition of an interfacial Ti layer. They present evidence from electron energy loss spectroscopy that the Ti layer undergoes strong oxidation, inducing oxygen vacancies in HZO. The latter, they argue, must be responsible for promoting ferroelectricity (twofold increase in remnant polarization Pr for optimal Ti thickness). Unfortunately, the induced vacancies also speed up the breakdown, and the increased polarization is found to correlate with a lower endurance. While the FTJ performance per se is not investigated by Segantini et al., this is done in a detailed study by Manchon et al. [2100585] who consider a similar FTJ design but with insertion of an Al rather than a Ti interlayer. They arrive at similar qualitative conclusions. Quantitatively, the improvement in Pr is smaller due to an excellent Pr even without the Al interlayer. Additionally, they confirm the Al interlayer leads to an increased tunneling current, which is desirable as it allows to maintain sizeable sensing currents in smaller devices, thus paving the road to successful FTJ miniaturization. As Figure 1 shows, the interest in ferroelectric HZO continues to grow exponentially, with an annual growth of about 35% in the number of publications and almost 60% in the number of citations. As Guest Editors of this Special Issue, we would like to cordially thank the authors for their contributions. Sergey V. Barabash is a Senior Principal Scientist at Intermolecular Inc., a subsidiary of Merck KGaA, Darmstadt, Germany. He has been at Intermolecular since 2011, and previously worked as a postdoc at NREL, research engineer at UCLA, and had visiting appointments at UNL and Stanford University. His Ph.D. in Physics is from OSU (2003). His focus is on simulations relevant to materials development, including first-principles studies of phase identity, stability, and ferroelectricity in HZO. Min Hyuk Park received his B.S. and Ph.D. degrees in Materials Science and Engineering from Seoul National University, Seoul, Korea, in 2008 and 2014, respectively. He worked as a postdoc in Seoul National University (2014–2015) and NaMLab gGmbH in Dresden Germany (2015–2018), and an assistant professor in school of materials science and engineering of Pusan National University (2018–2021). He is currently an assistant professor in the department of materials science and engineering of Seoul National University since 2021. His research interests include ferroelectric and antiferroelectric thin films for neuromorphic computing, memory, energy storage, energy harvesting, and solid-state cooling. Tony Schenk received his B.Eng. and M.Eng. degrees from the Westsaxon University of Applied Sciences Zwickau in 2011 and 2012, respectively. Since 2013, he worked as PhD student and Post-Doc at NaMLab and received his PhD from TU Dresden in 2016 for his work on hafnia/zirconia-based ferroelectrics. After a Post-Doc at Luxembourg Institute of Science and Technology, he joined Ferroelectric Memory GmbH (FMC) in 2020, where he leads the stack development as a Member of Technical Staff.