State of the art and major challenges
Low-temperature studies of transition metal doped III-V and II-VI compounds carried out over the last decade have demonstrated the unprecedented opportunity offered by these systems for exploring physical phenomena and device concepts in previously unavailable combinations of quantum structures and ferromagnetism in semiconductors. It has been found that these systems, particularly (Ga,Mn)As and (In,Mn)As, and their layered structures show functionalities discovered earlier for ferromagnetic metals, albeit often of amplified magnitude, such as giant and tunnelling magnetoresistance as well as planar Hall effect and anisotropic magnetoresistance, and also current-induced magnetisation switching and exchange bias. Importantly, in a few significant cases, research on DMS has been ahead and actually guided studies on ferromagnetic metals, examples being spin-injection, resonant tunnelling, current induced domain-wall displacement, tunnelling anisotropic magnetoresistance, and Coulomb blockade anisotropic magnetoresistance - each of these phenomena combining novel physics and potential for applications in the next generation storage and logic devices. Moreover, it has been suggested that the well-established methods of modulation of the carrier concentration in semiconductor quantum structures could be applied for tailoring the magnetic properties. A control of material properties by external means, as in semiconductors, would thus add a new dimension to possibilities of study and use of magnetic materials. Such an alteration of ferromagnetism by light and electric field has indeed been demonstrated in II-VI and III-V DMS heterostructures of p-type (Cd,Mn)Te or (In,Mn)As, respectively.
An important aspect of research on diluted ferromagnetic semiconductors is that the above experimental accomplishments have been prompted and/or assisted by quantitative theoretical modelling. There is a general consensus now that ferromagnetism of Mn-doped III-V antimonides, arsenides, and phosphides as well as of II-VI tellurides is associated with the presence of holes. The p-d Zener model of hole-mediated ferromagnetism in p-type tetrahedrally coordinated magnetic semiconductors was designed to describe and to predict quantitatively materials' properties and associated functionalities. Accordingly, it has been developed in the spirit of the second-principle method which exploits, whenever possible, experimentally available information. In particular, the host band structure can be thoroughly parametrised by the multi-band kp method or multi-orbital tight binding approximation, the latter particularly useful to describe properties of layered structures. This approach has made it possible to describe pertinent properties and functionalities, particularly in the case of (Ga,Mn)As and related structures, including characteristics usually regarded as the domain of experimental studies such as magnetic anisotropy or the Gilbert dumping constant.
Despite this massive progress in the physics of diluted magnetic semiconductors (DMS), the understanding and control of these materials have emerged as the most controversial and challenging field of today's material science and condensed matter physics. Among the unsettled issues is the question whether the holes mediating the spin-spin coupling in arsenides and phosphides reside in the d band (double-exchange scenario), in an acceptor band split of the valence band (impurity band scenario), or in the valence band (p-d Zener model). This issue is obviously related to the fundamental question why the Curie temperature of (Ga,Mn)As saturates at 150 - 180 K (as found by dozen laboratories world-wide) despite the progress in overcoming the limits of self-compensation and solubility in this model magnetic semiconductor alloy. Higher TC values have recently been predicted for (Ga,Mn)(P,As) but actually lower ones are being observed. Even less obvious is the situation in magnetically doped nitrides and oxides. Here, some groups detect only a paramagnetic response, while others find the ferromagnetism to persist up to well above the room temperature in the same system. The ferromagnetism origin and the possibility of employing it in functional devices are much investigated but no consent emerges. Finally, there are persisting experimental and theoretical reports indicating that high-temperature ferromagnetism can be found in oxides and carbon without any magnetic impurities. This highly unsatisfactory situation has its source in both experimental and theoretical challenges imposed by these materials.
Experimental challenges: The major challenge is to demonstrate a functional semiconductor system with ferromagnetic capabilities persisting to above room temperature. However, as we know now, this challenge has been approached with inadequate tools: The magnetic response of thin DMS films is often inferior to those coming from typical remanent fields, sample holders, substrates, or residual magnetic nanoparticles originating from nominally nonmagnetic source materials or processing procedures. Furthermore, standard x-ray and microscopy characterisation methods show usually neither magnetic nor chemical contrast and, additionally, are often applied to specimens processed in a different way than the SQUID samples. Accordingly, claims about high-temperature ferromagnetism in DMS are persistently subject of uncertainty concerning contamination, spinodal decomposition or nanoscale phase separation. At the same time, the predicted high-temperature hole-mediated ferromagnetism in magnetically doped nitrides and oxides has not yet been verified experimentally because no materials with sufficiently large band hole densities have so far been obtained. While Mg doping of GaN is established, the formation of a 2D hole layer at GaN/(Ga,Al,In) interfaces and p-type doping of ZnO are challenging.
Theoretical challenges: In the carrier-controlled ferromagnetic semiconductors conceptual difficulties of charge transfer magnetic insulators and strongly correlated disordered metals are combined with intricate properties of heavily doped semiconductors and semiconductor alloys, such as Anderson-Mott localisation, defect creation by self-compensation mechanisms, spinodal decomposition, and break-down of the virtual-crystal approximation. Accordingly, these materials have posed an enormous challenge to first principles methods, particularly those involving standard local spin-density approximation (LSDA) and coherent-potential approximation (CPA). It becomes more and more clear that inaccuracies of these approximations, such as improper treatment of strong correlation at transition metal atoms, errors in band-gaps and d-level positions as well as inadequate description of localisation have resulted in incorrect predictions of the magnetic ground state, generally overestimating the tendency towards the ferromagnetism.
In both prehistoric times and today, the technology advances are fuelled by materials developments. For instance, in the present silicon-based technology, more and more silicon and its dioxide are replaced by newly developed compounds with superior performance. In a long run, particularly perspective are multifunctional materials as well composite media, assembled in a bottomup way with nanoscale precision.
Among the multifunctional materials especially attractive are ferromagnetic semiconductors which combine the resources of semiconductors (in the heart of chips, light sources, solar batteries, …) and ferromagnets (employed in hard discs, tapes, electric engines, …). Moreover, these systems show entirely new functionalities, as e.g. the manipulation of magnetism by the electric field. While various families of ferromagnetic semiconductors have already been studied, particular attention is being paid to diluted magnetic semiconductors (DMSs), consisting of technologically important semiconductors doped with magnetic transition metal (TM) impurities, for example (Ga,Fe)N or (Zn,Co)O. Indeed, GaN and its derivatives like (Al,Ga)N are presently second to Si important semiconductor materials, commonly employed in light emitting diodes and high power electronics, so that the incorporation of magnetism-related functionalities to these materials is appealing. However, despite the apparent chemical simplicity of DMSs, the field of their studies has developed into one of the most controversial topic in the condensed matter physics [see, T. Dietl, Nature Mater. 9, 965-974 (2010)].
Over the last couple of years it has become increasingly clear that this unsatisfactory situation reflects unanticipated conceptual and experimental challenges that have been revealed by the research, also ours, on DMSs. The disappointing state of the field is best illustrated by the fact that despite of above four thousand papers on the observation of high temperature ferromagnetism in semiconductors and related oxides, no working device structure incorporating these materials has so far been demonstrated.
The FunDMS project was built on the conviction, which has been substantiated by its execution, that DMSs invalidate the paradigm of the semiconductor and magnetism physics about a uniform distribution of magnetic ions and carries over the crystal volume. Thus, we argued, any meaningful experimental and theoretical studies of DMS samples should be preceded by state-of-the-art element specific and spin sensitive nanocharacterization. Only knowing accurately how particular elements are positioned in the lattice, we may attempt to understand the macroscopic properties. Importantly, as we then noted, even the newly developed powerful nanocharacterization tools are widely available to virtually all researchers via open access to large facilities, like synchrotron radiation centres.
By implementing this methodology into FunDMS activities, we have been able to reveal, to understand, and then to learn how to control nanoscale heterogeneity in the distribution of TM ions and/or carriers. We have understood, in particular, that open atomic d shells, specific to transition metals like Fe, not only provide spins and thus, magnetic moments, but also enhance the strength of the chemical bonding. Accordingly, there exists an attractive force between TM cations (absent in the case of nonmagnetic metals like Ga or Al), which leads to their self-organized aggregation. Nanocomposite magnetic/semiconductor systems formed in this bottom-up fashion constitute complementary media to semiconductor/semiconductor nanocomposites in the form of self-organized quantum dots, extensively investigated over the last two decades. Remarkably, in the most thoroughly studied (Ga,Fe)N we have discovered and elaborated methods to control, by appropriate growth conditions and/or co-doping with other impurities, the aggregation process and, in particular, the chemical composition and, thus the magnetic properties, of the TM-rich nanocrystals [see, A. Navarro-Quezada et al., arXiv:1107.4901]. Our studies have revealed also a highly non-random distribution of nanocrystals, which we are currently exploiting to design functional structures based on these systems.
Furthermore, we have experimentally examined and theoretically described many aspects of the mechanisms accounting for ferromagnetism in DMSs with a uniform Mn distribution. In the case of high quality and well characterized (Ga,Mn)N films we have discovered the presence of ferromagnetism despite the absence of carriers, and identified the relevant mechanism as ferromagnetic superexchange [see, A. Bonanni et al., Phys. Rev. B. 84, 035206 (2011)]. Especially the strong coupling between localized spins is mediated by holes in the valence band, according to the p-d Zener model that works well in the case of p-type DMSs, like (Ga,Mn)As. A challenging question we have addressed concerns the influence of hole localization (occurring in any semiconductor at sufficiently low carrier concentrations) upon hole-mediated ferromagnetism. By changing the hole density in metal-insulator-semiconductor structures, we have demonstrated that critical fluctuations in the carrier density specific to Anderson-Mott localization leads to heterogeneous magnetism consisting of ferromagnetic and superparamagnetic regions, the latter taking gradually over on decreasing the hole concentration [see, M. Sawicki et al., Nature Mater. 6, 22 (2010)].
There are two main objectives of the FunDMS project: (i) by comprehensive program consisting of growth, (nano)characterization, experimental and theoretical investigations resolving controversies concerning the origin of ferromagnetism in various families of dilute magnetic semiconductors and (ii) basing on this insight, to proposed possible functionalities of these materials. Over the last two years three important puzzles have been solved: (i) the nature of magnetism in (Ga,Mn)N ) was elucidated by showing that in high quality samples (no defects compensating Mn3+ ions) the relevant spin-spin coupling mechanism is ferromagnetic superexchange (Phys. Rev. B’ 2012, Appl. Phys. Lett.’2012, Phys. Rev. B Rapid Commun. 2013); (ii) the origin of high temperature ferromagnetism in (Zn,Co)O was explained in terms of dipolar coupling between Co-rich nanocrystals at the interface (Phys. Rev. B’ 2013); (iii) the formation of Mn cation dimmers was found to account for uniaxial anisotropy of (Ga,Mn)As (essential in for functionalities of this system) (Phys. Rev. Lett. 2012). Two functionalities were patented: (i) the controlled in-plane aggregation of ferromagnetic Fe-rich nanocrystals in GaN (also publications in Appl. Phys. Lett. 2012, Phys. Rev. B 2011, 2012); (ii) broad photoluminescence due to Mn-Mgk complexes in GaN that can be exploited for GaN-based infrared lasers in telecommunication windows (also publication in Sci. Reports 2012). There is a delay in the studies of DMSs by the field effect. This topic is investigated vigorously at present.
In order to accomplish the project objectives we will form in Warsaw a team combining six young PostDocs and PhD students with researchers experienced in DMS, oxide, and nitride physics as well as in the materials development and characterisation. We will take advantage of the recent unprecedented investment in equipment and laboratory space at the host institution (IFPAN). This includes a new SQUID system, ALD reactor for DMS and gate oxides, two-chamber MBE system for DMS nitrides and oxides, cathodoluminescence apparatus, clean room facilities, and computer cluster. We will also use new setups for uv, correlation, and time-resolved spectroscopies that have recently been purchased by the University of Warsaw (UW). To exploit the complementary capabilities of different epitaxy methods, in addition to MBE and ALD at IFPAN, we will employ MOVPE via the already successful collaboration with the Johannes Kepler University in Linz (JKU).22,33 For nitride- and oxide-based DMS epitaxy, we will use bulk GaN and ZnO substrates, the world-unique asset of Warsaw's semiconductor milieu. Furthermore, we will take advantage of the existing joint employment between IFPAN and MAX-Lab in Lund and of the collaboration with the University of California in Berkeley to have access to (Ga,Mn)As and (Ga,Mn)(As,P), respectively.
A consensus is now emerging that the progress in DMS research has been severely hampered by inadequate nano-characterisation tools, primarily because corresponding expertise and costs are rarely affordable for materials development laboratories. In order to break this ring of impossibility we will incorporate to our team two PostDoc researchers, who will be based around world leaders in, respectively, synchrotron characterisation of buried quantum dots at JKU and magnetisation-resolved nanoscopy of magnetic nanoparticles at Center for Electron Nanoscopy, Denmark Technical University (DTU).
An important part of the project will be a strong participation of theory which will (i) model charge and spin distribution in layered structures; (ii) assist synchrotron research; (iii) provide magnetic characteristics of embedded nanocrystals and simulate their aggregation.