Вы находитесь на странице: 1из 31

www.advmat.de www.MaterialsViews.

com

REVIEW

Dilute Doping, Defects, and Ferromagnetism in Metal Oxide Systems


By Satishchandra B. Ogale*

Over the past decade intensive research efforts have been carried out by researchers around the globe on exploring the effects of dilute doping of magnetic impurities on the physical properties of functional non-magnetic metal oxides such as TiO2 and ZnO. This effort is aimed at inducing spin functionality (magnetism, spin polarization) and thereby novel magnetotransport and magneto-optic effects in such oxides. After an early excitement and in spite of some very promising results reported in the literature, this eld of diluted magnetic semiconducting oxides (DMSO) has continued to be dogged by concerns regarding uniformity of dopant incorporation, the possibilities of secondary ferromagnetic phases, and contamination issues. The rather sensitive dependence of magnetism of the DMSO systems on growth methods and conditions has led to interesting questions regarding the specic role played by defects in the attendant phenomena. Indeed, it has also led to the rapid re-emergence of the eld of defect ferromagnetism. Many theoretical studies have contributed to the analysis of diverse experimental observations in this eld and in some cases to the predictions of new systems and scenarios. In this review an attempt is made to capture the scope and spirit of this effort highlighting the successes, concerns, and questions.

1. Introduction
The notion of introducing a dopant in a solid to manipulate its properties for the realization of a desired functionality is a highly researched topic for several decades in the context of semiconductor physics, especially in the domains of microelectronics and optoelectronics. The successes of modern microelectronics and related well-known developments in many other elds owe a great deal to the controlled doping effects (charge functionality) in semiconductors. More recently, with the growing interest in the elds of spintronics and magneto-optics,[15]

[] Prof. S. B. Ogale Physical and Materials Chemistry Division National Chemical Laboratory Council of Scientic and Industrial Research Dr. Homi Bhabha Road, Pashan, Pune 411008 (India) Indian Institute of Science Education and Research (IISER), Pune (India) E-mail: sb.ogale@ncl.res.in, satishogale@gmail.com Phone: 9120-25902260, 919822628242 Fax: 9120-25902636

DOI: 10.1002/adma.200903891

signicant research is now being performed on exploring the possibilities of inducing ferromagnetism (spin functionality) in the otherwise non-magnetic semiconductors by doping dilute concentrations of magnetic impurities therein. The corresponding subject is categorized as the eld of diluted magnetic semiconductors (DMS).[619] The signicant interest in this eld stems from the fact that building spin functionality into an already application-worthy material can be envisaged to enhance the novelty and range of its applicability immensely. Moreover dilute doping can ensure that the original applicationspecic non-magnetic set of properties of the undoped parent material could remain broadly unaffected, thereby enabling the usage of the best of both worlds. It must be remembered, however, that the volume magnetization of such diluted systems cannot be as high as bulk ferromagnets such as magnetite, maghemite, ferrites, or mixed-valent manganites. Therefore, such DMS systems cannot be expected to be used in magnetic force related applications. On the other hand they should nd use as active layers in magneto-optic and spintronic device architectures through dopant-spin carrier-spin interactions. Early research efforts in this eld were focused on semiconductor matrices, especially GaAs (e.g., GaMnAs), and it took several years of intense efforts to develop appropriate materials manipulation strategies that led to robust and reproducible results that could be brought into the realm of technology development. This research has also led to a signicant enhancement in our understanding about the behavior of spins in semiconductor hosts, their exchange interactions and relaxation phenomena. With great advances in recent years in the synthesis/growth of high-quality functional metal oxide lms, nanomaterials, and interface systems, supporting a broad range of physical properties, there is now growing optimism about the possible emergence of a new frontier of oxide electronics.[20] Amongst the several different aspects being researched in this context, the explorations of doping effects have also acquired considerable signicance. As expected, one special class of doped metal oxides that has attracted considerable attention lately relates to magnetic dopants; the corresponding systems being broadly categorized as the DMS oxide (or DMSO) systems.

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3125

www.advmat.de www.MaterialsViews.com

REVIEW

Oxides differ in many ways from the conventional elemental and compound semiconductors, for example, in their structural forms and diversity, bonding characteristics, nature and types of defect states, special role of oxygen non-stoichiometry, anisotropies and their implications for properties, various types of ordering phenomena (specially in multi-component systems) and the broadest possible phase diagram of structure-stoichiometry-property relationships. Hence the subject of dopant effects in oxides promises exciting possibilities and new physics, not encountered or addressed in semiconductor physics. In this review we focus on the research effort aimed at using magnetic dopants to incorporate spin functionality (especially ferromagnetism /spin polarization) in nonmagnetic oxide systems. Most extensive research in this eld has been done on two wide bandgap functional n-type metal oxides, namely TiO2 and ZnO. As will be brought out, even in the case of such apparently simple oxide systems it has not yet been possible to reach a consensus in scientic community on many dopant distribution and defect related issues in spite of intense worldwide efforts. Although the scientic understanding generated by such efforts reects considerable promise in the application context, much more work is clearly needed before the full suggested potential of the DMSO systems is unleashed in the application domains of spintronics and magneto-optics. Research on the DMSO systems was triggered by the work of Matsumoto et al.[21] in 2000 on the observation of ferromagnetism in dilutely cobalt doped TiO2, though the rst case of wide bandgap DMSO was reported earlier in Mn:ZnO.[22] Since then intense research efforts have followed on these as well as other oxide systems with different dopants (and even defect doping) to realize robust ferromagnetic response in a uniformly doped oxide material. Although signicant new science and understanding have emerged through this process, the eld has not yet reached the same level of maturity and clarity about the attendant phenomena as the III-V semiconductor based DMS. The primarily hurdle has been the confusion about the uniformity vs clustering of dopants which appears to be sensitively linked to the growth methodology and conditions. Moreover, in view of the characteristic differences in the nature of lattice and carrier related properties between oxides and III-V semiconductors the mechanisms of exchange coupling in the two cases are proving to be different. On the positive side this has led to the development of interesting new theoretical models of exchange, but on the negative side the materials issues have made it difcult to compare such models with specic experimental results. In the following, we address with a fair degree of detail the problems related to novel magnetic effects in several DMSO systems. It is useful to mention from the outset that the eld is too vast and there have been a very large numbers of contributions from various groups around the world, making it almost impossible to cover and include all the publications. This review is therefore not intended to be fully exhaustive and all-inclusive. It is intended to bring out the spirit of the eld through examples chosen rather subjectively based on the literature with which the author has been particularly familiar. This does not at all imply that some work not directly discussed here is any less signicant.

Prof. S. B. Ogale received his Ph.D. from Pune University (India) in 1980. He worked as Physics Faculty of Pune University (19801996; Professor and Head of the Physics Department 19921995) and as a Senior Scientist at the University of Maryland, College Park (19962006). Presently he is a Senior Scientist at the National Chemical laboratory, and Adjunct Professor at IISER, Pune, India. He has worked as a Visiting Scientist in several Universities/Labs in the USA and Europe. His current research interests are spintronics, optoelectronics, and solar energy. He is an elected Fellow of the Indian Academy of Science.

2. Dening an Intrinsic Diluted Magnetic Semiconductor (DMS)


Before entering the main body of the review it is worth addressing a few issues that can dene the specics about some key points of discussion. The term diluted magnetic semiconductor (DMS) refers to a non-magnetic semiconductor wherein a homogeneous or uniform dilute (a few percent) concentration of a magnetic dopant (one with a net spin moment) is dispersed, and the resulting material exhibits magnetic order, especially ferromagnetism. Since one cannot normally get an ordered distribution of dopants, homogeneous or uniform essentially means a statistically uniform dopant distribution wherein the number distribution of near neighbor dopantdopant separation is not too broad. Since for dilute concentrations of magnetic dopants the mean separation between neighboring dopants is necessarily larger than that required for any conventional mechanism of ferromagnetic exchange between neighboring ions to be operative, the occurrence of ferromagnetism in DMS is termed as carrier mediated. Within the domain of carrier mediation different mechanisms are possible and have been suggested including itinerant carriers (RKKY), polarons, F-centers, etc. The strength of this interaction governs the value of the Curie temperature. Also, the effects of the interaction between the dopant magnetic moment and carrier spins give rise to the phenomena such as the anomalous Hall effect (AHE), optical magnetic circular dichroism (O-MCD), etc. Moreover, with the role of the carrier being central to such an exchange, any independent manipulation of carrier density such as by an external electric eld in an FET conguration or co-doping with another ion which can inuence the carrier density via valence control must have an inuence on the observed ferromagnetism. Although these can be regarded as the vital signs of the so called intrinsic or true DMS, caution needs to be exercised before arriving at the corresponding denitive conclusions because (as discussed later) the phenomena of AHE, O-MCD, and electric eld modulation can occur even in cases wherein the dopant

3126

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

distribution is not uniform and the dopants are clustered either partially or fully. When the dopants are clustered partially with precipitation of secondary magnetic phase(s), the situation is more complex because of mixed state effects. In such cases the corresponding magnitudes may be much weaker despite the fact that the vital signs of intrinsic DMS may be seen. Hence, occurrence of these phenomena alone is not enough to declare a material to be intrinsic DMS or DMSO. The strengths of the corresponding effects need to be judged vis--vis theoretical expectations. To be more denitive, the uniformity of dopants in the material needs to be directly established by nanoscale element specic scanning depth probes such as electron energy loss spectroscopy (EELS), secondary ion mass spectrometry, etc. Further the concentration of dopants needs to be ascertained as fully accounted for by such scanned data. This is because under certain growth conditions a large fraction of dopants could be siphoned to the surface, interface or grain boundary and may not be easily visualized in cross section scan probes in a specic localized volume along a specic line scan. For this reason it is also essential to get transmission electron microscopy images over different scaling lengths starting from very coarse (micrometers) to much ner (sub-nanometer regime). In many cases the HRTEM results show excellent data including structure, dopant distribution, etc. but a large chunk of the dopants could be localized elsewhere as clusters or secondary phases near the surface, interface or extended defect. Another matter of importance is the issue of probe depths. For instance, if surface-sensitive probes are used, one cannot conclude about the valence state of the full dopant compliment. The dopant may be in different valence states near the surface and in deeper regions. All these remarks and considerations emphasize the complexity of resolving matters regarding the desired atomistic state of the material especially when one is dealing with dilute dopant concentrations and weak magnetic signals. When clustering occurs either partially or fully, the analyses deserve even more careful considerations since these situations represent non-intrinsic or extrinsic DMS character. One could take a pragmatic view arguing that even if some clusters are present why worry about this if the spintronics properties are interesting and application-worthy? There may be a point in this argument as long as such complex extrinsic states of the material can be reproduced and sample-to-sample variations are in the acceptable zone for an application. This is generally not the case and hence the best situation is the one in which the growth scenario leads to statistically homogeneous dopant dispersion. If the magnetic dopant is like Mn or Cu, which do not have secondary ferromagnetic phases, partial clustering would just reduce the mean concentration of the uniformly dispersed dopant. However, if the dopant is like Fe or Co, some of the corresponding secondary phases are ferromagnetic or antiferromagnetic, and co-existence of such phases with the DMSO ferromagnetism emanating from partially uniform dopant distribution complicates the analysis. Thorough magnetic characterizations (inverse susceptibility, eld cooled and zero eld cooled magnetization in addition to temperature dependent hysteresis measurements) are then essential to elucidate the mechanisms. When the clusters are tiny (i.e., below a critical size that depends on crystal eld anisotropy energy parameters) one

encounters superparamagnetic relaxation. In this case even though all the spins are ferromagnetically aligned with reference to each other, the volume energy of pinning of the magnetization along any crystallographic direction becomes smaller than the thermal energy at a given temperature and the direction of the total magnetic moment of the ferromagnetic nanoparticle uctuates rapidly between different directions. This leads to an apparent paramagnetic response with a zero coercive eld. When the clusters become bigger than the critical size for superparamagnetic relaxation one gets into the regime of single domain particles for which the coercive eld is very high because of the absence of domain boundaries. All these considerations need to be factored into the analysis before one can conclude about a specic magnetic state of the sample and connect it to the micro-structural and chemical state. The complications outlined above have collectively contributed to some of the controversies in the experimental domain of the DMSO eld which are addressed in the review.

3. Diluted Magnetic Semiconducting Oxide (DMSO) Systems


As stated above, signicant work has been carried out over the past decade in search of intrinsic (i.e., not originating from extrinsic effects such as undesirable ferromagnetic clusters or secondary phases) diluted DMSO systems. The corresponding research has been quite controversial but has also led to several new insights that have red up the magnetism community to look for interesting research directions. In this pursuit some oxide systems have been prominently investigated because of the hope generated therein by early research as well as some application-worthy property feature they display. In the following we rst discuss the research on two materials systems that have been central in DMSO research, namely TiO2 and ZnO. We follow this with discussions on the reported observations of DMS ferromagnetism in other interesting oxide systems such as SnO2, Cu2O, La1xSrxTiO3+d, etc and the related materials issues. We then discuss the major rapidly developing subject of defect ferromagnetism, which has received quite a boost by the oxide DMS studies. We then conclude with a few remarks on possible directions for future research. A tabulated summary of most of the experimental research discussed in this review is given in the Supporting Information.

3.1. Magnetism in TiO2 with Dilute Doping of Transition Element Ion(s)


TiO2 is undoubtedly the most explored oxide based diluted magnetic semiconductor, because of the various well known special properties of this functional oxide[12] and also perhaps because of the rst report of DMSO ferromagnetism in this specic oxide by Matsumoto et al.[21] Briey, TiO2 is a wide bandgap metal oxide which has three basic polymorphs, namely rutile, anatase and brookite. The rutile and anatase phases, which have been explored in the context of DMSO studies, have indirect bandgaps of 3 and 3.2 eV, respectively, and are n-type conductors with shallow donor levels. Details

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3127

www.advmat.de www.MaterialsViews.com

REVIEW

of their physical properties are given in the review by Pearton et al.[12] Several Japanese groups, especially those of Kawasaki, Koinuma and co-workers have contributed a signicant body of interesting and insightful work to the DMSO eld. While their discovery[21] laid foundations of DMSO research, their subsequent research on TiO2 and ZnO systems has consistently addressed the central question to the eld: Are DMSO systems truly intrinsic? As expected, the same question has also been intensely addressed and debated by most other leading groups on the subject, as discussed later in the text. In the pioneering study that triggered tremendous interest and activity in the eld of DMSO, Matsumoto et al.[21] applied a combinatorial screening approach to the solid solubility of transition metals in titanium dioxide and examined their magnetic properties. They discovered transparent ferromagnetism in cobalt-doped anatase thin lms for cobalt concentration between 0 and 8%. Interestingly, magnetic domain structure (magnetic force microscopy or MFM image) was also observed in the lms, suggesting long range ferromagnetic ordering. Room temperature ferromagnetism with a magnetic moment of 0.32 B/Co was realized for 7% cobalt case. Such ferromagnetic lms were found to be conducting and exhibited a positive magneto-resistance (MR) of 60% at 2 K. The authors did discuss (and did not completely rule out) the possibility of granular precipitates of magnetic materials (e.g., Co metal clusters[23,24] as a possible cause of the observed ferromagnetism, though their XRD and TEM measurements did not show any apparent signatures of such metal granules.[24] The observed moment was suggested to be consistent with the low spin conguration of Co ions. Fukumura et al.[25] performed magneto-optical spectroscopy studies on the transparent ferromagnetic semiconductor anatase Co:TiO2 at room temperature. Not only was a large magneto-optic response observed from ultraviolet to visible range with ferromagnetic eld dependence, but the same was found to enhance with increasing Co content or carrier concentration. The MCD magnitude per unit thickness showed a huge peak around the absorption edge with a value of 10400 deg cm1 at 3.57 eV for a 10 mol% Co-doped case. Clearly, the coexistence of Co impurity and mobile carriers was thus shown to transform the band structure of host TiO2 to yield a ferromagnetic state. However, none of their lms exhibited anomalous Hall effect (AHE). Murakami et al.[26] examined the TiO2-CoO binary compositional spread in thin lm form grown on LaSrAlO4 (001) (LASO) single crystal substrate at a temperature gradient from 550 to 750 C for various properties. The oxygen ambient during deposition was kept at a pressure between 5 105 and 5 107 Torr. The samples with cobalt content of up to 50% showed c-axis oriented anatase peaks. For nearly 50% Co content the CoTiO3 peaks were noted to have maximum intensity and with increased Co content CoO peaks were seen to appear. The magnetization showed an increase with decreasing oxygen growth pressure. Magnetic domain structure was also observed in lms that were conducting. Existence of anatase phase and the conducting nature of the lm were thus shown to be indispensable for the emergence of ferromagnetism. Toyasaki et al.[27] reported on the anomalous Hall effect (AHE) in rutile phase Ti1xCoxO2 lms suggesting intrinsic

nature of ferromagnetism in this compound. Epitaxial thin lms were grown by laser molecular beam epitaxy (L-MBE) and the electron concentration (n) was controlled by varying the partial oxygen pressure during growth. Atomically controlled lm growth was conrmed by the intensity oscillation of reection high-energy electron diffraction (RHEED). The results indicated that AHE developed with increasing n and/or increasing x. The authors noted that in the classical model of AHE for ferromagnetic metals, xy is proportional to xx or xx2 (hence xy is nearly proportional to xx or xx2) as attributed to skew scattering or side jump, respectively. The scaling law for their samples was seen to fall between these behaviors, suggested to be a consequence of the semi-classical nature of the charge carrier dynamics due to low carrier mobility. It was further noted, however, that AHE scaled only by xx irrespective of the temperature, the dopant concentration x, or the oxygen deciency, bearing a single relationship extending over six orders of magnitude in AHE. This implied that the charge-carrier dynamics as related to AHE are not affected signicantly by carrier scattering by phonon, magnetic impurity or a defect, namely the oxygen deciency. Kawasaki, Koinuma and co-workers[28] also reported a systematic study of the structural, electronic and magneto-optic properties of the thin lms of anatase phase TiO2 doped with Co that were grown epitaxially on LaSrAlO4 (001) substrates using pulsed laser deposition (PLD). They controlled the oxygen pressure during growth to obtain insulating, semiconducting and metallic lms. No evidence of Co segregation was suggested to be found by different in situ and ex situ characterizations. Magneto-optic circular dichroism (MCD) spectra were also recorded at room temperature. A large Magneto-optical circular dichroism (MCD) peak was noted at a photon energy close to the bandgap of TiO2. The magnetic eld dependence of this signal was seen to match well with the magnetization. Also, the MCD intensity was observed to increase systematically with the increase of n or x. This once again strongly indicated that the interaction between the charge carriers and Co impurities is essential to realize ferromagnetism in this system. Subsequently, Toyosaki et al.[29] also performed MCD studies on Rutile Ti1xCoxO2d as a function of x and d values to explore the phase diagram for the appearance of ferromagnetism and to examine the connection between O(Optical)-MCD and AHE. Figure 1 compares the optical absorption and O-MCD data for 3% Co-doped TiO2. It is clear that the effect is negligible at low carrier concentration but is substantial at higher concentrations. Also, it correlates very well with the optical absorption data. Indeed for the high carrier concentration case the absorption as well as O-MCD results show similar blue shift. Moreover, the MCD signal at any photon wavelength changes systematically with magnetic eld. Importantly, a complete correlation is noted between the xx (conductivity)-x (dopant concentration) phase diagram for occurrence of ferromagnetism as derived from the MCD (photoexcited carriers) and AHE (dc driven carriers) measurements. These indicators together suggested carrier induced intrinsic mechanism of ferromagnetism in this case. Quilty et al.[30] performed X-ray photoemission spectroscopy measurements on the same Rutile case to reveal the electronic structure. No discernable signal corresponding to metallic cobalt was seen. The Co ions were found to be in the high-spin

3128

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

Figure 1. a) Absorption and b) MCD spectra at 300K for Ti0.97Co0.03O2 with different ne. Reprinted with permission from.[29] Copyright 2005, American Institute of Physics.

Co2+ conguration, consistent with the substitution for Ti. The valence band spectra showed evolution of a broad impurity band derived from the Co 3d orbitals located 2 eV below the Fermi level. The acceptor level of the Co impurity was found to be almost degenerate in energy with the conduction band minimum. The suggested strong hybridization between carriers in the Ti 3d t2g band and the t2g states of the high-spin Co2+ made a case for intrinsic ferromagnetism in this case. Toyosaki et al.[31] also fabricated rst Magnetic tunnel junctions (MTJs) employing the room temperature ferromagnetic semiconductor Ti1xCoxO2d and the ferromagnetic metal Fe0.1Co0.9 as electrodes, with an AlOx tunnel barrier. The tunneling properties were seen to depend systematically on the barrier thickness. The differential conductance and the tunneling magneto-resistance (TMR) were found to be signicantly asymmetric with respect to bias voltage at low temperature, which was attributed to the asymmetric junction structure and/ or the degraded interface of AlOx/Fe0.1Co0.9. Unfortunately only a limited amount of such device work has been reported in the DMSO eld thus far. Mamiya et al.[32] performed soft X-ray magnetic circular dichroism (XMCD) measurements on Rutile Co:TiO2 lms at the Co L2,3 edges at room temperature. Clear multiplet features characteristic of ferromagnetic Co2+ ions coordinated by O2 ions were seen, in sharp contrast to the featureless XMCD spectrum of Co metal or metallic clusters. The experimental data agreed well with a full atomic-multiplet calculation for the

Co2+ high-spin state in the D2h-symmetry crystal eld at the Ti site in Rutile TiO2, indicating that the ferromagnetism in this case arose from the Co2+ ions substituting the Ti4+ ions. The authors also estimated the orbital and spin moments of Co2+ from the experimental results and the corresponding values were found to be lower by a factor of 78. The discrepancy was attributed to the formation of magnetic dead layer. Ueno et al.[33] studied Anomalous Hall effect (AHE) in anatase Ti1xCoxO2d thin lms over the temperature range from 10 to 300 K. While the magnetic eld dependence of anomalous Hall resistance was found to be coincident with that of magnetization, it was seen to decrease at low temperature in spite of nearly temperature-independent magnetization. The Anomalous Hall conductivity was found to be proportional to the square of Hall mobility, suggesting that charge scattering strongly affected the AHE in this system. The anatase case was also shown to follow a scaling relationship to conductivity as AHE xx1.6, as was observed for the Rutile case suggesting identical mechanisms of AHE in the two polymorphs. Ueno et al.[34] also observed AHE in anatase Ti1xCoxO2d lms up to 600 K establishing the presence of spin polarized carriers in this system at high temperature. The anomalous Hall resistivity showed ferromagnetic hysteresis loop from 300 to 600 K without signicant change in magnitude, implying a Curie temperature higher than 600 K. Matsumura et al.[35] further examined the Ti site substitutionality of Co ions in Rutile and anatase TiO2 lms exhibiting ferromagnetism from the X-ray Bragg reection intensity measured around the K-absorption edge of Co. An anomaly due to the anomalous dispersion of the atomic scattering factor of Co was expected if the Co ions were substituting the Ti sites randomly. But none of the samples exhibited any anomaly, establishing that the Co ions were not exactly located at the Ti sites of TiO2. The absence of the anomaly was attributed to a signicant deformation of the local structure around Co due to the oxygen vacancy. Interestingly, when the same procedure was applied to the paramagnetic Zn1xCoxO case, direct evidence for Co substitution on the Zn sites was obtained. These results seemed to imply a signicant role of lattice deformation in DMSO ferromagnetism. In their relatively recent paper summarizing some of their work,[36] Kawasaki and co-workers have emphasized that their collective evidence points to a strong case of intrinsic ferromagnetism in the Co:TiO2 system. A striking piece of data presented in this paper (Figure 2) exhibits clear dependence of the saturation magnetization (MS) on both n, the carrier concentration, and x the dopant concentration. Particularly noteworthy is the decrease in MS (saturation magnetization) with increase in the carrier concentration for higher dopant concentration. The correlation between the AHE, O-MCD data and magnetization (lower left panel) is noteworthy. Moreover, the authors pointed out that the magnetic phase diagrams emanating from their AHE and O-MCD studies match and show that higher ne and x induce a ferromagnetic phase (lower right panel). Their studies further suggest that the local charge imbalance, the lattice distortion as well as the high-spin state of Co ions might be responsible for a large exchange coupling between localized spins and overlapping electron wave functions, leading to the observed high Curie temperatures.

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3129

REVIEW

www.advmat.de www.MaterialsViews.com

REVIEW

(a) x=0.03

41022 cm -3

2 1 (b) x=0.05
21020 cm -3 71021 cm -3

Magnetization (B /Co)

210 2

20

cm

-3

0 -1

2
0 410
19

cm

-3

1
-1

(c) x=0.10

21020 cm -3 41021 cm -3

0
710 -2 Hplane -3 -2 -1 0 1 300 K 2
18

cm

-3

-1 -2 -2

0H (T)
2 x=0.10 21021 cm -3 2 300 K 1 2kdeg. cm 1

-1

0H (T)
Electron density (cm3)
Magnetization (B/Co)
300 K Ferromagnetic

Hall resistivity ( cm)

1022 1021 1020 1019 1018 0

-MCD

1 Hplane 1

Paramagnetic

2 2

0H(T)

2 2

0.02

0.04

0.06

0.08

0.1

x in Ti 1-x Co x O2-

Figure 2. Magnetic-eld dependence of magnetization at 300K for Ti1xCoxO2 with different ne. (a) x = 0.03. Each data set is shifted vertically, (b) x = 0.05, (c) x = 0.10. Lower left panel shows the magnetic-eld dependence of Hall resistivity (line), magnetization (open circles) and MCD (lled circles) for Ti0.90Co0.10O2d. Reprinted with permission from.[36] Copyright 2008, New Journal of Physics.

Yamasaki et al.[37] have recently grown room-temperature ferromagnetic Co-doped TiO2 lms on glass substrates by the sputtering method. Nearly full-polarized magnetization, large magneto-optic effect and AHE were observed at room temperature. Importantly, the magneto-optic effect showed nearly fourfold enhancement in a one-dimensional magneto-photonic crystal (MPC) structure with a standard dielectric multilayer (SiO2/ TiO2), which was attributed to an increase in the electric eld of light in the intermediate magnetic layer sandwiched between dielectric layers. More such work exhibiting novel magnetooptic effects may help re up the interest in the DMSO eld. 3130

Chambers and co-workers have also done signicant amount of work on oxide based DMS systems, highlighting various microscopic details regarding the connection between the microstructural quality of a material in terms of clusters, precipitates, defects (point and extended) and the occurrence of ferromagnetism therein. In their initial work[38] Chambers and co-workers employed oxygen-plasma-assisted molecular-beam epitaxy (OPA-MBE) to grow CoxTi1xO2 anatase lms on SrTiO3 (001) substrates for x = 0.010.10, and measured their various properties. They found ferromagnetism at room temperature in the lms irrespective

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

of whether the lms were epitaxial or polycrystalline. They further found that the magnetic and structural properties depended critically on the Co distribution and varied signicantly with growth conditions. Co was shown to be substitutional in the anatase lattice and in the +2 formal oxidation state. The remanence was found to be much higher than that reported for PLDgrown lms. The Co moment was 1.26 0.12 B, much closer to the value expected for low-spin Co(II) than 1 B measured for PLD lms. In their further studies[39] performed on CoxTi1xO2 anatase lms grown on SrTiO3(001) and LaAlO3(001) substrates, they found that highest crystalline quality epitaxial lms with most uniform Co distribution resulted for the case of slow growth rate (0.01 nm/s) at 550600 C on LaAlO3(001). Growth at a higher rate (0.04 nm/s) led to secondary-phase Rutile nanocrystals to which Co was found to diffuse and segregate. Cobalt appeared to substitute for Ti and exhibited a +2 formal oxidation state. Both pure and Co-doped TiO2 lms either exhibited n-type semiconducting behavior (oxygen deciency), or were found to be insulating. In a subsequent related study[40] they showed that for OPA-MBE growth under certain specic conditions virtually all incident Co was seen to segregate to CoxTi1xO2x nanoscale clusters and the corresponding sample exhibited ferromagnetism. On the other hand, homogeneous lms of the same Co concentration were found to be non-ferromagnetic. These ndings showed that nanoscale CoxTi1xO2x anatase particles with large x are ferromagnetic. Since all bulk phases of the known TiCo binary oxides are either antiferromagnetic or low-temperature ferrimagnets, this result implied realization of an unexpected ferromagnetic material consisting of nanoscale Co-doped anatase with the Co mole fraction well in excess of that which produces ferromagnetism in large-area, homogeneous lms (x > 0.08). Chambers et al.[41] also employed Co K-shell near-edge and extended X-ray absorption ne structure analyses to study the charge state and local structure near Co in epitaxial anatase Co-doped lm grown on LaAlO3(001). They found that Co substitutes Ti in the lattice and is in the +2 oxidation state in the entire lm, creating an oxygen vacancy in the proximity for local charge balance. They did not nd any evidence for elemental Co or CoO, yet found ferromagnetism at and above room temperature, provided enough free electrons were present to make the material semiconducting. The moment was consistently found to be 1.11.2 B per Co, but the remanence was seen to vary inversely with the conductivity. Although these ndings are fairly consistent with true (intrinsic) DMS nature of the material, the authors did not place an unambiguous claim on the same. Kim et al.[42] also grew epitaxial Fe-doped TiO2 Rutile lms on Rutile TiO2 (110) substrates, and explored the resulting compositional, structural, morphological and magnetic properties. Clusters of mixed TiO2 Rutile and Fe3O4 were seen to form on the surface of an epitaxial Rutile lm during growth. Roomtemperature ferromagnetism was observed, and was attributed to the formation of secondary phase Fe3O4. In another study, the same group (Shutthanandan et al.[43]) synthesized ferromagnetic Co-doped Rutile TiO2 single crystals by high-temperature ion implantation and studied them by using various techniques. Cobalt was seen to be uniformly distributed to a depth of 300 nm with an average concentration

of 2 at%; except in the near-surface region where the concentration was found to be slightly higher (3 at%). Room temperature ferromagnetic behavior was observed with an effective saturation magnetization of 0.6 B/Co atom, with Co in a formal oxidation state of +2 throughout the implanted region. Based on the results of XANES, EXAFS, and XRD, it was found that 1/3 of the implanted Co was present as CoO secondary phase. CoO being antiferromagnetic in the bulk (ignoring the possibility that nano CoO may contribute to magnetization due to canted surface spins), only 2/3 of the Co atoms could be considered to be contributing to the observed ferromagnetism in this system. Droubay et al.[44] studied the case of Cr doped TiO2 for possible intrinsic DMS ferromagnetism using OPA-MBE. Chromium was found to be in +3 state throughout the lms by Cr K-shell X-ray absorption near-edge spectroscopy and no evidence for either elemental Cr or half-metallic CrO2 was found. The Cr distribution was found to be uniform in the doped region with Cr substituting Ti. Room-temperature ferromagnetism aligned in-plane was observed, with a saturation magnetization of 0.6 B/Cr atom for Cr concentration of 9%. The lms were found to be insulating as grown, which implied that the electrons associated with oxygen vacancies introduced for charge balance were localized. This further implied that a magnetic exchange other than carrier-mediated type could be operative. The authors suggested that the F-center model proposed by Coey[45] (discussed later in the theory section) could be consistent with their results, although other mechanisms could not be discounted. In this model, the F-center electrons are suggested to polarize the Cr spins yielding ferromagnetic coupling. A percolative ferromagnetic ordering as discussed by Kaminski and Das Sarma[46] could then lead to the observed bulk ferromagnetism. In their further work (Osterwalder et al.[47]) on highly ordered CrxTi1xO2 lms in rutile and anatase structures for Cr concentrations spanning a broad range from x = 0.02 to 0.16, they found from the X-ray photoelectron diffraction data that the Cr 2p level exhibited the same patterns and modulation amplitudes as those for Ti 2p, implying that a large fraction of the Cr atoms occupied the substitutional lattice sites in both structures. Cr was also shown to be in the Cr3+ state. The Cr-doped anatase lms showed in-plane ferromagnetic order with a saturation magnetization of 0.6 B/Cr for Cr concentration of 16%. No detectable formation of other Cr-containing phases was observed (even up to x 0.16), in stark contrast to the Co-doped anatase case, wherein most of the Co tended to segregate to a surface phase of Co-enriched CoxTi1xO2 epitaxial particles in registry with the underlying continuous lm with much smaller x. The Cr-doped Rutile lms were also insulating as the anatase lms but did not produce signicant ferromagnetic signals. These authors also brought out the characteristic differences between the anatase and Rutile cases in the context of Coeys F-center model.[45] Specically, they pointed out that there is a large difference in the orbital radius (r) of the F-center electron for the two phases (16 A in anatase, but only 2 A in Rutile), due to large difference in the static dielectric constants (k 31 for anatase and 100 for Rutile) and effective mass for conduction band electrons associated with donor states (m 1 m0 for anatase and 25 m0 for Rutile, m0 being the free

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3131

REVIEW

www.advmat.de www.MaterialsViews.com

REVIEW

electron rest mass). Clearly, in the anatase F-center states, multiple Cr dopants are included in each F-center volume, while in Rutile case they do not, explaining the non-observance of ferromagnetism in the latter. In a subsequent study Kaspar et al.[48] re-examined the mechanism of ferromagnetism in doped TiO2 anatase, using high quality epitaxial Cr:TiO2 as a model system. In contrast to the case of highly oriented but defective lms which exhibited ferromagnetism, structurally superior single crystal lms exhibited only negligible ferromagnetism. Similar results were also observed for the Co:TiO2 case. This study provided evidence that charge-compensating oxygen vacancies are not sufcient to activate ferromagnetism (as suggested by the F-center mechanism) and that presence of structural defects must have an important role. The role of structural defects was sensed in other works as well, such as the role of grain boundary defects in activating ferromagnetism by room-temperature aggregation of paramagnetic doped ZnO[49] and TiO2[50] nanocrystals under ambient conditions. The authors suggested that extended structural defects may give rise to charged point defects of a fundamentally different type, which could be associated with oxygen vacancies.[51] Continuing their work on the Cr-doped anatase TiO2 thin lms, Kaspar et al.[52] observed that the highly defected crystalline lms were ferromagnetic, but surprisingly without any free carrier spin polarization as revealed by Hall effect measurements. The lms grown at slow rate were nearly perfect crystalline but exhibited negligible ferromagnetism at all Cr concentrations. Vacuum annealing was used to generate additional oxygen defects and free carrier electrons, but this did not signicantly increase the ferromagnetic ordering in either the defected or perfectly crystalline lms. These results clearly contradicted with both the oxygen-vacancy induced free-carrier exchange and the F-center exchange mechanisms, once again emphasizing the primary role of extended structural defects in mediating the ferromagnetic ordering. Venkatesan, Ogale, and co-workers have also carried out considerable work on DMSO systems.[5363] They have elucidated the role of growth-condition-dependent dopant clustering and the occurrence of AHE even in the dopant-clustered case thereby questioning the validity of AHE as an unambiguous signature of intrinsic DMS (with R. L. Greene). They have also demonstrated electric-eld-induced changes in DMSO magnetism (with R. Ramesh) and dopant induced changes in the optical properties such as band edge, luminescence etc. (with H. D. Drew). They also discovered ferromagnetism in transitionelement-doped Cu2O and La0.5Sr0.5TiO3 systems. Separately, the Maryland theory group of S. Das Sarma has contributed new interesting models of ferromagnetism for the DMS systems. In their early study on TiO2 DMSO, Shinde et al.[53] examined thin lms of Co-doped anatase TiO2 grown by pulsed laser deposition at 700 C for dopant substitutionality, ferromagnetism, transport, magneto-transport, and optical properties. Their results showed limited (up to 2%) solubility of Co in the as-grown lms and formation of Co metal clusters at higher concentrations. The 7% Co doped sample exhibited a Curie temperature (TC) of over 1180 K and showed presence of 2050 nm Co clusters with small concentration of Co incorporated into the TiO2 matrix. High temperature annealing

caused by the rst magnetization measurement led to a TC of 650 K in the next measurement with almost full apparent dissolution of Co clusters and matrix incorporation of Co. Interestingly, this TC was found to be close to that of as-grown 1% doped TiO2 sample grown at a temperature of 700 C. Transport, magneto-transport, and optical studies also revealed interesting effects of the matrix incorporation of Co. These results gave a rst indication that the TC in Co:TiO2 intrinsic DMS may be as high as 650700 K. Guha et al.[54,55] performed Raman and photoluminescence (PL) studies on this series of samples. The Raman spectra conrmed their anatase phase. The photoluminescence spectra were characterized by a broad emission from self-trapped excitons at 2.3 eV (characteristic of anatase phase) at temperatures below 120 K. This peak showed a small blue-shift with increasing dopant concentration. The Co-doped samples also showed two spin-ip emission lines at 2.77 and 2.94 eV. Based on the published theoretical results, the authors concluded that the appearance of the observed 2.77 and 2.94 eV energy levels in the PL spectra well below the absorption edge of 3.2 eV for this system most probably arise from the cobalt eg energy levels. Indeed, the 170 meV difference between the two PL peaks agreed well with the energy separation between the spin-up and spin-down eg states of the LSDA calculations. To check the Ti-site substitutionality of doped cobalt ions Kulkarni et al.[56] performed ion channeling and 3.045 MeV He+ ion oxygen resonant backscattering experiments on the thin lms of Co:TiO2. They conrmed non-substitutionality of Co in lms deposited at 700 C, and their incorporation in the matrix by either annealing at a higher temperature of 875 C or deposition at this temperature. Higgins et al.[57] performed Hall effect measurements on Rutile phase Co:TiO2 thin lms and found n-type conduction over the entire range of Co concentration. They showed that lms with a low carrier concentration (10181019 cm3) yield a linear behavior in the Hall data, while those having a higher carrier concentration (10211022 cm3) display anomalous behavior near zero eld, the latter giving a clear evidence for an AHE in the heavily oxygen reduced samples. Indeed, the low-eld data showed a change in behavior at nearly the same point at which the magnetization of the sample saturated, which occurred at a eld signicantly lower than that for cobalt metal lms (H = 1.52 T). Also, the resistivity of each sample remained nearly the same for the two temperatures measured, and AHE remained relatively constant (in magnitude) as expected and suggested by the measurements. Although these observations seemed to suggest an intrinsic DMS character, the authors pointed out that the structural and chemical microstructures formed in samples prepared under highly reduced conditions could be quite complex, especially in view of the known formation of Magneli phases in the oxygen-reduced Ti-O system. This necessitated detailed studies on the samples to examine the relative proportion of dissolved and clustered cobalt to determine how the AHE could be interpreted in such cases. Thus, Shinde et al.[58] performed detailed magnetic and structural analysis of highly reduced Co-doped Rutile TiO2d lms which displayed anomalous Hall effect (AHE). The temperature and eld dependence of magnetization (Figure 3) clearly showed the collapse of coercive eld near and above 250 C and the collapse of all loops (inset (a)) onto a single curve above this temperature

3132

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

x=0.01
5K 50 K 100 K 150 K 200 K

15 M (emu) 0

380 K 350 K 300 K 250 K 200 K

-15 -1

Inset (a)

0 1 H/T (Oe/K) 300 K 350 K 20 10

250 K

Inset (d) 300 K

0 -10

380 K -1
400 HC (Oe) 300 200 100 0 0 100 200 300 400 0 T (K)

H (T) 1

-1500

H (Oe) -20 0 1500


Inset (c)
x=0.01 x=0.02

40 MR (emu) 30 20 10

Inset (b)
x=0.01 x=0.02

0 100 200 300 400 T (K)

Figure 3. M-H curves for Ti0.99Co0.01O2d lm measured at different temperatures. Inset (a) shows magnetization as a function of H/T for different temperatures. In insets (b) and (c) are plotted the temperature dependence of HC and MR for Ti0.99Co0.01O2d and Ti0.98Co0.02O2d lms. Inset (d) shows the magnetic eld dependence of Hall resistivity of the Ti0.99Co0.01O2d lm. Reprinted with permission from.[58] Copyright 2005, American Physical Society.

when plotted as M vs H/T. This established a clear case for superparamagnetic relaxation. Transmission electron microscopy data unequivocally established the presence of nanometersized superparamagnetic cobalt clusters of 810 nm size in the lms at the interface. Their observation of the co-occurrence of superparamagnetism and AHE raised signicant questions regarding the use of the AHE as a test of the intrinsic nature of ferromagnetism in DMS systems. It meant that although AHE is expected for intrinsic DMS, observation of AHE does not imply that one has intrinsic DMS. In contrast to AHE reported previously in samples with embedded magnetic clusters in a nonmagnetic matrix at a concentration near bulk percolation threshold,[59,60] the Co:TiO2d samples examined by Shinde et al.[58] had an average Co concentration of only 2%. However, due to the selective location of all superparamagnetic clusters at the interface, the average cluster density in the thin interface layer in the sample was clearly higher. The authors pointed out that even if one assumes all of the 2% cobalt atoms form

10 nm clusters near the interface, one nds the percentage of occupied area to be 7% with an inter-cluster (edgeedge) separation of 20 nm. This implied non-percolating cobalt metal inclusion. Therefore the authors invoked the argument adapted from the work of Shimshoni and Auerbach[61] on the observation of a quantized Hall effect in a non-percolating system to explain their data. As we point out later in the text, Shinde et al.[58] did not analyze the magnitude of the observed AHE. It turns out that although observation of AHE in itself may not be a clear signature of intrinsic (superparamagnetic cluster free) DMS, the magnitude of AHE and its comparison with theory can be. Simpson et al.[62] measured the optical properties of ferromagnetic Ti1xCoxO2d lms for low doping concentrations, 0 < x < 0.02, in the spectral range 0.2 < 5 eV. For well oxygenated lms (d 0) the optical conductivity was seen to be characterized by the absence of optical absorption below the onset of interband transitions at 3.6 eV and a blue-shift of the optical band edge was noted with increasing Co concentration. The absence of below bandgap absorption was clearly inconsistent with theoretical models which contained mid-gap magnetic impurity bands and suggested that strong on-site Coulomb interactions shifted the O band to Co-level optical transitions to energies above the gap. This further suggested the possibility of strong interaction effects on the optical transitions involving the Co ion. The possibility of an alloy being responsible for a shift within a rigid band picture (which is actually more appropriate for delocalized states) due to the average Ti/Co potentials could be refuted because the atomic potentials for Co being larger than those of Ti the Ti/Co band would be lower than the pure Ti bands in TiO2 resulting in a red shift, contrary to the observation. Since a localized picture was considered more appropriate for transition elements, the authors analyzed the processes operating on the optical transitions involving the Co levels within a localized picture and could conclude that O and Ti bands separate upon Co substitution. The large rate of separation ( = 5 eV) implied strong level repulsion that might occur for interstitial incorporation of Co. The authors pointed out that the observed large band-edge shift is especially interesting because it implied strong interactions required to provide the large exchange interaction and associated high ferromagnetic TC observed in this material. Another key test that has been consistently suggested for carrier-induced (intrinsic) ferromagnetism in DMS systems (but is unfortunately not pursued as much in DMSO systems) is the ability to manipulate the magnetism by electric eld induced modulation of carriers in the system. In this context, Zhao et al.[63] demonstrated a remarkable external electric eld induced reversible modulation of the room temperature magnetization in an epitaxial and insulating thin lm of dilutely cobaltdoped anatase TiO2 in a high quality epitaxial hetero-structure PbZr0.2Ti0.8O3/Co:TiO2/SrRuO3 grown on (001) LaAlO3. The top panel of Figure 4 shows the device structure and the epitaxial nature of the hetero-structure as reected by X-ray diffraction (which was also established by cross section HRTEM not shown here). The HRTEM data shown in the paper and recorded at different lateral locations did not show any signicant dopant clustering. It was argued that out of the intended 7% dopant concentration at least 4% was uniformly incorporated

REVIEW

xy (cm)

M (emu)

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3133

www.advmat.de www.MaterialsViews.com

REVIEW

LAO(100)

Pt

Pt

Pt

Pt

PbZr0.2Ti0.8O3 300nm Co:TiO2 60nm SrRuO3 75nm

LaAlO3

20

30 2 (deg.)

40

50

150

M (emu)

100 50 0 -50

70 35 0 -35 -70 -40 -20 0 20 40 Applied Voltage (V)

P (C/cm )

(a)

M (emu)

+30V -30V +30V -30V

100 95 90 85 +V -V +V -V Applied Voltage

-100

-150 -6000 -3000 50

3000 6000

M (emu)

25 0

+30V -30V +30V -30V


140 120 100 80

(b)

-25 -50 -750 -500 -250

Applied Voltage

+ - + -

250 500 750

H (Oe)
Figure 4. Top Panel shows the X-ray diffraction -2 scan of the deposited PZT/CTO/SRO/LAO multilayers. The inset is the schematic structure of the sample. Figure (a) and (b) show magnetic hysteresis loops of the CTO layer after several electric voltage poling on the PZT layer in a smaller and a larger H scale, respectively. The upper panel of the inset in (a) shows the ferroelectric hysteresis loops of the PZT layer, while the lower panel is the saturation magnetization of CTO as a function of the applied voltage on PZT. The inset in (b) is the coercive eld of CTO as a function of the applied voltage on PZT. Reprinted with permission from.[63] Copyright 2005, American Physical Society.

under the chosen conditions. Importantly, the modulation effect (lower panel of Figure 4) was observed at room temperature both on the saturation magnetization MS and coercive eld Hc. Note that SrRuO3 is not ferromagnetic at room temperature, hence the modulation emanates only from the DMSO layer. Its strength was almost 15% and the same could be established over several cycles. The authors argued that given the highly insulating nature of the samples, the itinerant electron based picture (RKKY-type scenario) discussed extensively in the context of GaMnAs magnetization could not apply. Therefore they invoked the other physical picture, namely the bound magnetic polaron (BMP) percolation picture[46] and the defect (F-center) state discussed by Coey et al.[45] to explain their observations. In this picture the F-center defects may very well be the bound magnetic polarons, which eventually form a percolating cluster producing global ferromagnetism. The authors indicated that the vacancy defects may themselves be susceptible to electric eld induced reorganization.[64] They argued that subjecting such insulating DMS system to strong external electric eld could lead to electric dipolar distortion of the magnetic polaron with attendant distortion of the shape of the corresponding wave function and eld driven redistribution effects. The corresponding changes in the local exchange energy and the global geometry of the percolation network could then inuence the magnetic moment and coercive eld of such a system. It is important to point out here that the electric eld modulation effect (Figure 4) discussed by Zhao et al.[63] is for an insulating anatase DMSO channel layer, while the occurrence of anomalous Hall effect (AHE) and superparamagnetism discussed by Shinde et al.[58] (Figure 3) is for a conducting Rutile lm formed under highly reduced conditions. Hence there is no conict between the two cases, since the sample states are entirely different. Unfortunately, both effects have not been demonstrated for a single sample state, which is highly desirable. Generally, demonstration of electric eld effect in conducting matrix requires ultrathin layer for an observable modulation of transport and in the DMSO case the corresponding magnetism is then very weak to examine the eld-induced changes therein. On the other hand in the insulating case, observation of AHE is difcult because of lack of carriers. Thus, alternative weak magnetism sensing schemes such as eld induced changes in optical MCD may have to be attempted for samples states which exhibit AHE. Since the concentration and distribution of oxygen vacancies in systems grown at low pressure appeared to be the cause of large variability and growth condition sensitivity of DMSO results, Zhang et al.[65] used another strategy of concurrent Nb doping to induce electrons into the conduction band due to the corresponding shallow level. It was shown that dilute niobium doping has a signicant effect on the ferromagnetism and microstructure of dilutely cobalt-doped anatase TiO2 lms. Epitaxial lms of anatase TiO2 with 3% Co, without and with 1% niobium doping were grown by PLD at 875 C at different oxygen pressures and examined for various properties. For growth at 105 Torr, niobium doping suppressed ferromagnetism, while it enhanced the same in lms grown at 104 Torr. Fu et al.[66] investigated the defect structure in these pure Codoped and (Co, Nb)-doped anatase TiO2 lms, both of which exhibited room temperature ferromagnetism. Dilute Nb doping

Intensity (a. u.)

PZT(001) SRO(100)

3134

HC (Oe)

TiO2(004)

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

PZT(002) SRO(200)

LAO(200)

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

was shown to signicantly (i.e. by 610 orders of magnitude depending on growth conditions) improve the conductivity and microstructure of the anatase TiO2 lm at the cost of a lower saturation magnetization (i.e. inverse correlation between lm structural perfection and ferromagnetism consistent with the results of Chambers et al.[38]). Z-contrast imaging and electronenergy-loss-spectroscopy study in the scanning transmission electron microscope showed cluster-free microstructure in the (Co, Nb)-doped thin lm although Co was found to be enriched into the surface forming a CoxTi1xyNbyO2d phase. In contrast, metallic Co, Rutile TiO2, and cobalt oxide nanoparticles combined with some crystallographic shear defect structures were observed in the case of pure Co-doped thin lm. More recently Ogale et al.[67] have performed detailed studies on epitaxial Rutile CoxTi1xO2 (x = 00.06) lms grown by PLD under different ambient atmospheres at a lower temperature of 400 C at which cobalt segregation observed for growth at high temperature of 700 C could possibly be avoided or at least dramatically reduced. They provided a combined window of HRTEM and EELS, X-ray absorption (XAS)/X-ray magnetic circular dichroism (XMCD), and magnetization measurements to elucidate the connection between microstructure and magnetism. Figure 5 is particularly interesting because it shows that under the chosen low temperature growth condition uniform incorporation of cobalt seems possible. But the paper also emphasized that one could be misled by this limited data since it is not possible to judge whether the entire amount of doped cobalt is included. In fact by using other surface layer sensitive techniques the authors established that some concentration of cobalt had segregated towards the surface. Indeed, their study brought out a mixed-state scenario of ferromagnetism involving intrinsic DMS (uniform dopant distribution at low dopant concentration) and coupled cluster magnetism, involving cobalt associations within the matrix at higher concentrations. They also showed that by matrix valence control during growth, it is possible to realize a uniform embedded cluster state and the related coupled cluster magnetism. They outlined an interesting comparison between the two cases of lms grown in vacuum and in Ar + H2 atmosphere for x 0.04. By analyzing the magnetization data within the molecular eld theory approximation, they inferred antiferromagnetic interaction between clusters in the sample deposited in Ar/H2. Very interestingly, the 1/ versus T curve for high-vacuum sample was seen to tilt upward at low temperatures indicating concurrent existence of ferromagnetic phase with Curie temperature lower than room temperature. The results suggested a competition between the intrinsic DMS ferromagnetic phase and antiferromagnetic coupling between clusters. This could also explain the higher magnetization in the high-vacuum grown sample than that in the Ar/H2-grown sample. The observed sensitivity of magnetization data to growth conditions even for low temperature (400 C) growth implies that the different microstates of the system are separated by only small energy scales. Ogale and co-workers (Bogle et al.[68]) have recently shown the benets of cobalt doping in TiO2 lms in the context of the resistive switching phenomenon. They showed that doping in an oxide matrix with a dopant having valence state other than the parent matrix ion can lead to a distributed network of oxygen vacancies that can help form controlled lamentary

TiO2

-cut Al 2O3
EELS

100 nm
Ti-L2,3 O-K Counts [a.u.]
9 8 7 6 5 4 3 2 1

Co L3 Co L2

450

500

550

600

650

700

750

800

850

Energy loss [eV]


Figure 5. a) High-resolution cross-section HAADF STEM image for a CoxTi1xO2 lm with x = 0.04 grown at 400 C in a vacuum greater than 1 106 Torr. b) EELS data recorded at various points across the lm crosssection. Reprinted with permission from.[67] Copyright 2005, American Chemical Society.

tracks in resistive switching, thereby improving the corresponding characteristics. This matter is however not related to the ferromagnetism aspect of this DMSO. It may be separately interesting to examine the effects of inclusion of such resistive switching layers in magnetic hetero-structures and then study the effect of external magnetic eld on the switching transport across the corresponding interfaces. Korean groups have also made signicant contributions to the DMSO eld. Kim et al.[69] investigated ferromagnetism in Co-doped anatase TiO2 thin lms using magnetic circular dichroism (MCD) at the Co L2,3 absorption edges. The magnetic moment was observed to be 0.1 B/Co, but the MCD spectral 3135

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advmat.de www.MaterialsViews.com

REVIEW

line shape was found to be nearly identical to that of Co metal (Co(0)) implying that the ferromagnetism originated from a small amount of clustered Co. With thermal treatments at 400 C, the MCD signal was seen to increase with an increase in the moment to 1.55 B/Co, close to 90% of the moment of Co metal. The cluster size in the annealed sample was observed to be 2060 nm. Their result once again emphasized the issue of cobalt inhomogeneity and clustering in this system. Noh, Kawai and co-workers[70] also examined anatase Ti0.96Co0.04O2 lms grown epitaxially on SrTiO3 (001) substrates by using PLD with in-situ reection high-energy electron diffraction (RHEED). The oxygen partial pressure during the growth was systematically varied. As the pressure decreased, the growth behavior was found to change from 2D layer-by-layer-like growth to 3D island-like growth which concurrently resulted in an increase in the saturation magnetization. These results along with TEM study showed the formation of cobalt clusters. They further investigated post-annealing effects on the magnetization of the PLD grown lms. Noh and co-workers[71,72] studied cobalt doped anatase lms made by cobalt ion implantation and found clustering effects as reected by the observation of superparamagnetism. Cobalt clustering effects have also been reported by Kim and co-workers[73] in chemical vapor deposited cobalt doped TiO2 lms and nano-belts. These works collectively brought out the possibilities of non-intrinsic effects in the case of this DMSO system. Krishnan and co-workers[74] employed RF magnetron sputtering followed by an annealing treatment to grow Co:TiO2 lms and showed that in their lms room temperature ferromagnetism coexisted with a dielectric state. TEM and X-ray absorption spectroscopy revealed a solid solution of Co in anatase phase, and Co was found to be in the +2 state substituting for Ti. A saturation moment of 1.1 B/Co was noted in their highly insulating lms, suggesting non-carried-mediated exchange mechanism. Hong et al.[75] have performed several studies on transition element doped TiO2 lms, and have addressed various issues including the possibilities of defect ferromagnetism. In their early study they substituted V, Cr, Fe, Co, and Ni partially for Ti in TiO2 and obtained ferromagnetism above room temperature in lms with good crystallinity. Vanadium doping was suggested to be the most promising case since it resulted in semiconducting lms with a giant magnetic moment. Interestingly, the V-doped sample showed a uniform magnetic force microscopy image (Figure 6) at room temperature that was distinctly different from the morphology image suggesting global ferromagnetic effect devoid of magnetic clusters. It may be noted that in their very rst study Matsumoto et al.[21] had also reported a scanning squid picture showing magnetic domains, but that result was at 3K. Based on their MFM data Hong et al.[75] suggested that this could be considered as strong enough evidence of intrinsic DMSO system. Gamelin and co-workers have done several interesting and insightful studies on DMS in metal oxides such as TiO2 and ZnO, the primary emphasis of their work being on the ZnO system. In contrast to most other research in the DMSO eld which has been focused on vapor deposited lms, their focus has been on DMSO nanomaterials and nanomaterial based lms. The general message of their work is the absence of

Figure 6. The topography (a) and the MFM (b) images taken at room temperature for the V:TiO2 lm grown on LaAlO3 at 650 C. The tip was magnetized perpendicular to the lm plane. Reprinted with permission from.[75] Copyright 2004, American Physical Society.

ferromagnetism (and occurrence of paramagnetism) in the colloidal system, with appearance of ferromagnetism only in nanoparticle aggregates or lms, suggesting inter-grain cooperative effects possibly through interface defects. Gamelin (Bryan et al.[50]) examined the most studied case of cobalt-doped anatase TiO2 but in the interesting form of nanocrystals and nanocrystal lms. By employing a number of techniques such as electronic absorption, magnetic circular dichroism (MCD), TEM, magnetic susceptibility, cobalt K-shell X-ray absorption spectroscopy (XAS), and extended X-ray absorption ne structure (EXAFS) measurements, they found that the nanocrystals were paramagnetic when isolated by surface-passivating ligands, weakly ferromagnetic (Ms 1.5 103 B/Co2+ at 300 K) when aggregated, and strongly ferromagnetic (Ms 1.9 B/Co2+ at 300 K) in spin-coated nanocrystalline lm form. XAS data revealed cobalt in the Co2+ oxidation state (Co2+ ions occupying the D2d anatase cation substitution sites and having a 4 4 E( T1g) low-symmetry ground state) strongly favoring existence of intrinsic ferromagnetism in cobalt-doped TiO2. Bryan et al[51] also studied colloidal Co2+ and Cr3+ doped TiO2 nanorods and nanocrystals. The nanorods in colloid form were found to be paramagnetic but showed room-temperature ferromagnetism when spin-coated aerobically into lms. The most important factor that activated ferromagnetism was found to be the creation of grain boundary defects, proposed to be oxygen vacancies at nanocrystal fusion interfaces. These defects could be passivated and the ferromagnetism destroyed by further aerobic annealing. These authors argued that the chemical composition of the nanocrystalline surfaces prior to aggregation gives further insight into the nature of interface defects. Oleic acid is known to coordinate to TiO2 nanorods by a chelating bidentate interaction of the -COO- group with a Ti4+-rich surface. Upon aggregation, such a metal-rich surface would present a deciency of lattice O2 ions, leading to a concentration of charged oxygen vacancies (Ovn) at the nanocrystalline fusion interfaces. The research discussed above leads one to conclude that transition element doped TiO2 is a complex DMSO system to understand and the DMS ferromagnetism therein depends rather signicantly on the growth methods and conditions as well as structural forms and conductivities. There is enough

3136

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

collective evidence to suggest that intrinsic DMS ferromagnetism can be stabilized in this system under controlled and specic growth conditions at least at low concentrations of < 23%. At high concentrations possibility of coexistence of intrinsic and extrinsic ferromagnetism exists, the concentration at which this would occur depends on the growth details. Also, the mechanisms of ferromagnetism in conducting lms and insulating lms are expected to be entirely different. Defects, both point and extended, inuence not only the occurrence of ferromagnetism, but the values of the saturation moment, coercive eld and the remanence as well. Secondary nanoscale phases can also contribute to magnetism even though they may not be ferromagnetic in the bulk. The location of such phases could generally be at the surface, the interface or an extended defect. Occurrence of anomalous Hall effect (AHE) or O-MCD alone may not represent strict tests of intrinsic carrier induced DMSO ferromagnetism, although they ought to be seen in case of the intrinsic systems. The strengths of these effects may yield better signatures of the connection between the microstate and the effect.

3.2. Magnetism in ZnO with Dilute Doping of Transition Element Ion(s)


ZnO is one of the most investigated functional metal oxides in various forms due to its unique set of physical (especially optical) properties and broad range of the corresponding application possibilities. It has a direct bandgap of 3.35 eV, shallow donor level (n-type conductivity, unless specially doped for p-type conduction) and a very high electron mobility of 200 cm2 V1 s1. Other property details are given in the review by Pearton et al.[12] Ueda et al.[76] examined for the rst time ferromagnetism in 3d-transition-metal-doped n-type ZnO lms grown on sapphire substrates using PLD technique, though Fukumura et al. had examined this system earlier[22] for solubility, electrical properties and magnetoresistance. Ueda et al.[76] found that while a few Zn1xCoxO lms showed ferromagnetic features, most others showed spin glass-like behavior. Also, the magnetic properties of these lms were seen to depend on the concentration of Co ions and carriers. The reproducibility of the method was however stated to be poor (<10%). Their Cr, Ni, or Mn doped ZnO lms did not show ferromagnetism. Ney et al.[77] studied the Zn1xCoxO system for element specic structural and magnetic information using synchrotron radiation. Co dopants were shown to exclusively occupy the Zn sites and purely paramagnetic behavior was seen for the isolated Co dopant atoms with a magnetic moment of 4.8 B. However, the total magnetization was seen to be less by 30%, revealing antiferromagnetic coupling of the Co-O-Co pairs. No sign of intrinsic ferromagnetism was noted for isolated or paired Co atoms. These authors further investigated the origin of ferromagnetism in epitaxial Co:ZnO lms which were seen to become weakly ferromagnetic after annealing in Zn vapor.[78] Conventional characterization techniques indicated no change after treatment. X-ray photoelectron spectroscopy depth proling however clearly showed the presence of Co(0) in the Zn-treated lms; the secondary phase being identied as

ferromagnetic CoZn. This work emphasized the importance of hidden secondary phase(s) in the context of all ferromagnetic Co:ZnO samples and other doped oxide DMS systems. In their early work Gamelin and cowrokers (Radovanovic et al[79]) reported on the synthesis and electronic absorption spectroscopy of highly crystalline and relatively monodispersed colloidal ZnO DMSO quantum dots (QDs) with Co2+ and Ni2+ ions as dopants. They employed ligand-eld electronic absorption spectroscopy to monitor dopant incorporation during nanocrystal growth and to conrm internal substitution. No magnetism data was reported, however. In a subsequent study from the same group involving alkaline-activated hydrolysis and condensation of zinc acetate solutions in dimethyl sulfoxide Schwartz et al.[80] showed that Co2+ and Ni2+ inhibit nucleation and growth of ZnO nanocrystals. In fact the dopants were shown to be quantitatively excluded from the critical nuclei but were incorporated nearly isotropically during subsequent nanocrystal growth. The Gibbs-Thompson relationship between lattice strain and crystal solubility was invoked to explain the decrease in nanocrystal size with doping. High-resolution low-temperature electronic absorption and magnetic circular dichroism (MCD) spectroscopy were used to address the issue of uniformity of dopant incorporation. Ferromagnetism with TC > 350 K was observed in the aggregated nanocrystals of Co2+:ZnO, suggesting the possibility of high temperature ferromagnetism in these systems. Their work also brought out that the coordination chemistry of the dopant ion strongly inuences both the nucleation and growth of nanocrystals, even at low dopant concentrations even in a highly compatible dopant/ host combination (e.g., Co2+ in ZnO) and can possibly inuence defect states and magnetism in such systems as well. They further demonstrated a ligand exchange processing method for removing dopant ions from the nanocrystal surfaces, ensuring exclusive internal doping in the nal colloids. Schwartz and Gamelin[81] also discovered that the 300 K ferromagnetic ordering in Co2+:ZnO system could be switched on and off with remarkable quantitative reproducibility by introducing and removing interstitial Zn (Zni). Figure 7 shows the reversible switching along with the modulation of other relevant physical parameters. These authors argued that incorporation of Zn interstitials into the paramagnetic lattice of Co:ZnO activates ferromagnetism by donating electrons, while the same are removed by oxidation killing the ferromagnetic ordering. They suggested double exchange mechanism of ferromagnetism involving electron delocalization amongst the substitutionally doped cobalt atoms. In yet another interesting work Kittilstved et al.[82] demonstrated manipulation of ferromagnetism in Mn2+:ZnO and Co2+:ZnO in predictable and reproducible ways using targeted p- and n-type chemical perturbations. As shown in Figure 8 a clear correlation was demonstrated between the nitrogen and high temperature ferromagnetism for Mn2+:ZnO and an inverse correlation for Co2+:ZnO, both as predicted by theoretical models based on charge carrier mediated ferromagnetism. These data are particularly important because they address the role of electrons Vs holes in the context of DMSO physics in the same sample via chemical tuning. As discussed elsewhere in the article some other studies have been performed separately on n-type and p-type samples, but these always leave open the

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3137

www.advmat.de www.MaterialsViews.com

REVIEW

Figure 7. (a) Reversible 300 K ferromagnetic ordering in Co2+:ZnO achieved by introduction and removal of interstitial Zn; (b) Quantitative reversibility of the 300 K 4A2 4T1(P) ligand-eld absorption intensity and 10 K Co2+ paramagnetism (two curves following squares); and the 300 K ferromagnetic saturation moment [MS(300 K)], and 300 K conductivity (two curve following open circles), all for the same lm; c) Schematic illustration of the effect of Zn interstitial on magnetic ordering. Reprinted with permission from.[81]

questions about the differences in the states of the samples grown under different conditions. Liu et al.[83] performed photocurrent and photoconductivity measurements on nanocrystalline Co2+:ZnO electrodes in combination with absorption and magnetic circular dichroism

Figure 8. Magnetization at 300 K vs eld loops for (a) 0.20% Mn2+:ZnO and (b) 3.5% Co2+:ZnO lms prepared from Ncapped ( , ) or O-capped ( , ) colloids. Insets: statistics for 300 K ferromagnetic saturation moments obtained from 10 independent lms for each case. Reprinted with permission from.[82] Copyright 2005, American Physical Society.

(MCD) spectroscopic measurements to elucidate the mechanism of photo-induced carrier generation therein. They observed two broad Co2+ charge transfer (CT) bands extending throughout the visible energy range; the lower one assigned as the Co2+ to the conduction band excitation and the higher energy CT band assigned to the valence band to Co2+ excitation. Theoretical models of magnetism in the DMSO systems have suggested that ferromagnetic ordering in these cases is stabilized by exchange interactions between the magnetic dopants and additional free or defect-bound charge carriers. These authors thus argued that the corresponding exchange interactions may be parameterized as conguration interactions involving CT excited states. In another study Liu et al.[84] introduced conduction band (CB) electrons into colloidal semiconductor nanocrystals and examined manifestations of TM2+ to e(CB) interactions by EPR spectroscopy. This work brought out new and novel opportunities of spin manipulation in such systems. Kittilstved et al.[82] compared and contrasted high-temperature magnetic ordering in Mn2+ and Co2+ doped ZnO diluted magnetic semiconductors to explore as to why manganese is different from cobalt. A clear difference was noted in carrier polarity requirements for ferromagnetism in Mn2+:ZnO (p type) and Co2+:ZnO (n type) systems, and was attributed to the different charge-transfer electronic structures in the two cases. In another work aimed at exploring the issues related to spinoidal decomposition, the same group (White et al.[85]) studied a series of free-standing Zn1xCoxO nanocrystal samples with 0.0 < x < 1.0. With increasing x they observed decrease in the size of the unit cell, inhomogeneous broadening of the Co2+ d-d absorption bands at intermediate x, increase in Dq and decrease in B, consistent with the changing structural parameters as described by the ligand-eld theory. Increase in x was also seen to redistribute absorption and MCD intensity from the ZnO band-to-band feature to a Co2+ centered band, which was eventually seen to evolve into the CT-type bandgap absorption of wurtzite-CoO. The magnetization per Co2+ was also seen to decrease with increasing x due to antiferromagnetic coupling, with super-statistical Co2+ clustering at small x. None of the compositions studied in this work showed superparamagnetic or ferromagnetic response implying that spinodal decomposition alone cannot be the origin of the ferromagnetism in Zn1xCoxO. It also ruled out the possibility of ferromagnetic signals emanating from uncompensated surface spins in highly Co2+-enriched Zn1xCoxO nanostructures. The authors invoked theoretical studies that suggested possibility of high ordering temperatures even with isotropic dopant distributions, given the right combination of specic dopant-defect pair concentrations and additional n-type charge carriers.[86] In these calculations the role of vacancy defects is to bring the Co2+ d levels in resonance with shallow donor defects, as indicated also in other DFT calculations.[87,88] These authors also pointed to the broad body of phenomenological evidence suggesting that ferromagnetism in Zn1xCoxO and several other related oxides may be linked to the presence of extended structural defects. Towards this end they suggested that grain boundary effects are a possibility that is experimentally most substantiated but least understood theoretically. They argued that the localization of ferromagnetism along grain boundaries may explain

3138

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

the paradoxical combination of high ordering temperatures and low saturation moments observed in many oxide DMSOs and could also provide an explanation for the puzzling disconnect between magnetic and magneto-transport properties of some Zn1xCoxO lms. Rao and co-workers[89] synthesized Mn- and Co-doped ZnO samples with 6% and 2% dopant concentrations by low-temperature (400500 C) decomposition of acetate solid solutions. The samples were characterized by various methods to ensure the dopant substitutionality. All such samples failed to show ferromagnetism. Instead, their magnetic properties were best described by a CurieWeiss type behavior. The authors emphasized the need for addition of carriers by co-doping with other cations or defects induced carrier incorporation for realization of ferromagnetism in this system. Shifting gears to vapor-deposited ZnO-based thin lm DMSO work, to which J. M. D Coey and his group at Trinity College Dublin have made signicant and interesting contributions, M. Venkatesan et al.[90,91] reported observation of roomtemperature ferromagnetism in (110) oriented ZnO lms made from targets containing 5 at % of Sc, Ti, V, Fe, Co, or Ni. However, no ferromagnetism was seen in similar lms doped with Cr, Mn, or Cu ions. Figure 9 shows that from Sc to Zn the observed variation is rather systematic. Large moments of 1.9 B and 0.5 B per dopant atom were seen for Co- and V-containing oxides, respectively, with signicant anisotropy. Study of the concentration dependence of the magnetic moment revealed a giant moment of about 6B/Co in the low concentration limit. A moment of 0.3 B per Sc was also seen, which was a surprise because both Sc3+ and Zn2+ are non-magnetic ions. No obvious correlation was seen between the conductivity and cation valence (Ti, Cr, and Fe cases are insulating, and the others are conducting) or conductivity and magnetism. The authors suggested that the unusual and huge anisotropy of

Figure 9. Magnetic moment of (Zn1xMx)O lms, M = Sc; Ti;; Cu; Zn, measured at room temperature. Solid circles are for the eld applied perpendicular to the lm plane and open circles are for the eld applied in the plane of the lm. The moment is expressed as B/M. The trend measured at 5 K is similar. Reprinted with permission from.[90] Copyright 2004, American Physical Society.

the magnetization of the oriented lms and the large moment per cobalt ion at low concentrations may probably be related. They further argued that the giant moments that were rst reported for SnO2 doped with Co,[92] and later conrmed for Mn, Fe, and Co doping[93] may require another source of magnetism in the dilute limit. They suggested that two-electron defects such as Ft centers (an adjacent pair of F centers) for which a low-lying triplet state could be stabilized by exchange with cobalt or V0 centers (cation vacancies) may be good candidates in this respect. They pointed out that since ZnO lms themselves are not ferromagnetic, the defects in Co-doped ZnO must be somehow switched on by Co, which suggested Ft centers. They further pointed out that the anisotropy may arise from an orbital moment associated with the molecular orbital(s) around the oriented magnetic point defects. Following their previous work, Coey et al.[90,91] further addressed the issue of the large anisotropy of magnetism in their transition element doped high-quality (110) lms of ZnO prepared on R-cut sapphire. They could rule out the possibility of secondary phases. Their observations and analyses led them to conclude that the observed anisotropy does not have its primary origin in crystaleld effects on the dopant ions. They suggested an entirely new type of ferromagnetism having its origin in lattice defects at the lmsubstrate interface. The inference regarding the evidence of an interfacial origin was drawn from the absence of any systematic concentration or thickness dependence of the moment. Krishnamurthy et al.[93] performed site-selective and elementsensitive XAS studies on Co:ZnO system in the vicinity of the Co L2,3 edge, the oxygen K edge, and at the Zn L3 edge. Ferromagnetic Co:ZnO lms showed additional components in the O K edge XAS spectrum that were absent in undoped lms, which could possibly be attributed to both hybridization with unoccupied Co 3d states and lattice defects such as oxygen vacancies. The shift in the onset of the oxygen K edge XAS of 0.5 eV was proposed to be due to the presence of oxygen vacancies, a view reinforced when considered in conjunction with optical measurements that exhibited a bandgap narrowing of only 0.1 eV. All these observations were suggested to be consistent with a polaron percolation model suggested by Coey.[45] M. Venkatesan et al.[94] also studied the effects of Al codoping on the ferromagnetism and other physical properties of 5% Co-doped ZnO using nominally the same experimental procedure as in their earlier work.[90,91] They found a bandedge shift, which varied with carrier concentration as n2/3. The doped lms, which contained coherent Co clusters of 48 nm size, exhibited a ferromagnetic moment of 0.31.0 B/Co. The magnetism was found to be progressively destroyed by Al doping concurrently with a reduction in Co-cluster formation. It was concluded that these materials could not be regarded as intrinsic dilute magnetic semiconductors. Kaspar et al.[95] also explored epitaxial thin lms of cobaltdoped ZnO (Co:ZnO) deposited by pulsed laser deposition (PLD) on C-plane and R-plane sapphire (Al2O3). The lms showed high structural quality, and the X-ray absorption spectroscopy (XANES and EXAFS) conrmed well-ordered Co substitution for Zn without secondary phases. Deposition, post-deposition processing and Al co-doping conditions were varied to achieve a wide range of n-type conductivities (104105 -cm). No signicant room temperature ferromagnetism was seen in any of

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3139

www.advmat.de www.MaterialsViews.com

REVIEW
3140

these lms. This conrmed the earlier observations of Chambers and co-workers[96] that itinerant conduction band electrons alone are not sufcient to induce ferromagnetism in Co:ZnO. The lack of ferromagnetism was attributed to the possibility of lack of lowangle grain boundaries. Khare et al.[97] examined 2% Co-doped ZnO epitaxial lms grown on sapphire for the systematic of ferromagnetism under different annealing treatments in oxygen atmosphere and Zn vapor, and showed that the ferromagnetism is not linked to conductivity, but rather to a small concentration of Zni interstitial defect. This, in their view, would then support the bound polaron percolation model of ferromagnetism.[45] Jayakumar et al.[98] showed that high-temperature ferromagnetism can be activated in Co- and Li-doped ZnO samples by surface treatment and that this is an intrinsic property of a defect-stabilized state that cannot be attributed to secondary phases. Radonovic et al.[79] observed ferromagnetism with TC > 350 K in the diluted magnetic semiconductor Ni2+:ZnO synthesized from solution. Whereas colloidal Ni2+:ZnO nanocrystals were paramagnetic, their aggregation gave rise to ferromagnetism which was attributed to the increase in domain volumes and the generation of new lattice defects Figure 10. Schematic description of the split-batch experiment demonstrating activation of 2+ C upon aggregation. The temperature depend- high-T100]ferromagnetism in Mn :ZnO by amine decomposition. Reprinted with permission from.[ Copyright 2005, American Chemical Society. ence of hysteresis (magnetization, coercive eld) was found to be unusual and was discussed in terms of temperature dependent exchange interacat 300 K, the lms prepared with N-capped colloids showed a tion involving paramagnetic Ni2+ ions (spin-orbit ground-state strong magnetic hysteresis with MS 1.22 B/Mn2+ and a coersplitting of paramagnetic Ni2+ ions exchange coupled to the fercive eld of 65 Oe. Even upon several repetitions N-capped colromagnetic domains). loids yielded ferromagnetic lms, whereas O-capped colloids Norberg et al.[99] also examined colloidal Mn2+:ZnO quantum yielded paramagnetic lms. Similar results were seen for caldots and their room-temperature ferromagnetic nanocrystalcined powders, establishing that the magnetism of Mn2+:ZnO line thin lms. They removed Mn2+ ions from the surfaces of depends critically on factors other than the transition metal as-prepared nanocrystals yielding high-quality internally doped dopants themselves, explaining the variability of results in the Mn2+:ZnO colloids. Robust ferromagnetism was observed in eld. spin-coated thin lms of such nanocrystals, with saturation Very recently Droubay et al.[101] have studied epitaxial lms of moments as large as 1.35 B/Mn2+ at 300 K. They speculated Mn2+ doped ZnO grown by pulsed laser deposition on -Al2O3 that calcination of amines may have introduced p-type nitrogen (0001) using targets created from Mn2+ doped ZnO nanopardefects into the ZnO lattice. To test this hypothesis, Kittilstved ticles. Mn (II) was found to substitute for Zn (II) in the wuret al.[100] performed a split-batch experiment wherein one batch tzite ZnO. A paramagnetic dichroic component was noted from of high-quality 0.20% Mn2+:ZnO nanocrystals was split into two Mn and no magnetic component was observed from either O equal parts. Half of these were capped with trioctylphosphine or Zn. The dichroism data revealed that Mn2+ distribution in oxide (TOPO, O-capped), and the other half with trioctylamine the ZnO lattice is not stochastic. Mn2+ was seen to have a ten(N-capped) under identical conditions. Colloidal suspensions of dency to substitute with higher effective local concentrations such nanocrystals in toluene were spin-coated on fused-silica than anticipated for stochastic doping. The authors conjectured substrates and were calcined aerobically at 500 C for 2 min per the possibility that such non-stochastic doping could be advancoat (about 20 coats, 1 micrometer thickness). Figure 10 shows tageous in stabilizing a ferromagnetic ground state for p-type 300 K magnetization loops collected for the two lms. RemarkMn2+:ZnO lms. ably, the samples prepared with O-capped and N-capped nanocIn the context of the Mn:ZnO story an interesting report rystals showed distinctly different results. While little magnetiappeared in the literature by Sharma et al.,[102] which created zation was observed in lms made with O-capped nanocrystals considerable interest. They reported the rst observation of

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

ferromagnetism above room temperature in dilute (<4 at%) Mn-doped ZnO with average magnetic moment of 0.16 B/ Mn. This sample was obtained by uniform sintering of a mixture of (<4%) MnO2 and ZnO at 500 C for several hours. They could realize room-temperature ferromagnetic ordering in bulk pellets, in thick transparent lms, and powders of the same material. They emphasized that while the previously reported high-temperature (T > 700 C) processing methods led to clustering with absence of room temperature ferromagnetism in such samples, their low temperature processed samples were free of clusters (uniform Mn distribution in ZnO matrix) and still ferromagnetic at room temperature. Ogale and co-workers[103] followed this discovery with more detailed work on the low-temperature-processed Mn-Zn-O system using suitably designed bulk and thin-lm samples. They demonstrated that ferromagnetism in this system does not originate from carrier-induced interaction between separated Mn atoms in ZnO as claimed by Sharma et al.,[102] but rather from a new metastable phase. They further showed that this ferromagnetism persists up to 980 K, and further heating transforms the metastable phase and kills the ferromagnetism. By examining the interface diffusion and reaction between thin-lm bilayers of Mn and Zn oxides, they showed that a uniform solution of Mn in ZnO does not form under low-temperature processing. Instead, a metastable ferromagnetic phase develops by Zn diffusion into Mn oxide, strongly suggesting that the observed ferromagnetic phase is oxygenvacancy-stabilized Mn2xZnxO3. This conclusion was completely borne out by the subsequent work of Garcia et al.[104] (and also other works by different groups[105,106]), who suggested the possibility of double exchange at the interface as the possible origin of ferromagnetism. During the past few years another interesting system, namely Cu-Zn-O, is being researched and discussed in the literature intensely. Although some early works (e.g., M. Venkatesan et al.[90]) did not nd ferromagnetism in dilutely Cu-doped ZnO lms, Buchholz et al.[107] examined a series of Cu-doped ZnO lms grown by pulsed-laser ablation under different conditions and found that the ones exhibiting n-type behavior were nonmagnetic while those showing p-type behavior were ferromagnetic with a Curie temperature above 350 K. The magnetic moment per Cu atom was seen to decrease as the Cu concentration increased. Antiferromagnetic interaction contribution of a growing number of neighboring pairs of Cu ions was suggested to be responsible for this decrease. Chakraborty et al.[108] also conrmed ferromagnetism in Cu-Zn-O lms grown by PLD, but did not discuss carrier type and concentration issue. Contrary to the observation of Buchholz et al.[107] of occurrence of ferromagnetism only in the p-type case, Hou et al.[109] found ferromagnetism in n-type Cu-Zn-O lms as well. Tiwari et al.[110] supported this view that p-type carriers are not necessary for ferromagnetism in this system. Xu et al.[111] also observed ferromagnetism in lms grown by PLD in nitrogen atmosphere at 570 C, but not in lms grown at 655 C which were of better structural quality. Thus they attributed ferromagnetism to defect states in ZnO. Interestingly, when Keavney et al.[112] probed the spin asymmetries in the Cu 3d, O 2p, and Zn 3d states individually in a soft X-ray absorption spectroscopy (XAS) and MCD studies, they found a paramagnetic component originating

REVIEW

from the Cu 3d, and no magnetic signal in the O or Zn. They also did not nd any dichroic signal consistent with ferromagnetism originating from any of these states. Sudakar et al.[113] reported ferromagnetism in reactive magnetron co-sputtered Cu-Zn-O lms and found ferromagnetism therein as well. In an interesting suggestion based on their HRTEM observations they attributed the observed ferromagnetism to the CuO planar nanophase inclusions. Ferromagnetism was also reported by Wu et al.[114] in Cu doped ZnO nanorods and the ferromagnetism was attributed to defect states (oxygen vacancies). In a fairly detailed investigation Ma et al.[115] reported the local structures of a series of Cu-doped zinc oxide lms grown by PLD using polarization-dependent X-ray-absorption spectroscopy. Their results showed that the lms where Cu exists solely in a clustered form are not ferromagnetic. They argued based on unique characteristics observed in substitutionally Cu-doped zinc oxide ferromagnetic lms that the observed enhanced covalence of the CuZn-O bond is a prerequisite for the spin alignments. More recently Ran et al.[116] have observed ferromagnetism in 2% Cu doped ZnO lms grown by helicon magnetron sputtering, but the moment per Cu was found to be lower (0.15 B). Once again defect mechanism was emphasized. Owens[117] has reported observation of ferromagnetism in sintered CuO and ZnO nanoparticles with ordering temperature of 520 K. Most recently, Straumal et al.[118] have analyzed several experimental publications on ZnO-based DMSO systems focusing on the connection between the ratio of grain boundary area to grain volume. Interestingly it was found that ferromagnetism is observed only when this ratio happens to be above a certain threshold value. Their observation of reproducible room temperature ferromagnetism and proportionality of magnetization to lm thickness makes a strong case for defect ferromagnetism related to the grain boundaries in this system. The various observations by Gamelin and co-workers discussed earlier appear to be broadly consistent with this conclusion. In summary, although the work on transition-element-doped ZnO has posed some issues and questions about the intrinsic nature of this DMS system, there is ample magnetic, magnetooptic, and photoelectron spectroscopy evidence to suggest that this system, when grown under appropriate conditions, could be an intrinsic DMSO with fairly uniform substitutional dopant incorporation. The possibility of realizing ferromagnetism in nanoparticle ZnO-based DMSO lms and their reversible tunability are truly exciting propositions. Similarly the possibilities of new metastable ferromagnetic phases in TM-Zn-O systems are encouraging.

3.3. Magnetism in Other Metal Oxide Systems (SnO2, Cu2O, La1xSrxTiO3+d) with Dilute Doping of Transition Element Ion
Although the primary focus of the DMSO research has been on TiO2 and ZnO, several interesting works have appeared in the literature on the search for ferromagnetism in other oxide systems doped with dilute concentration of transition element impurities. These studies have led to newer questions as well as insights into the possible exchange mechanisms of ferromagnetism

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3141

www.advmat.de www.MaterialsViews.com

in systems with dilute dispersions of magnetic ions. In this section we summarize some such studies. The Maryland group (Kale et al.[119]) examined for the rst time the occurrence of DMS ferromagnetism in (p-type) Cu2O. Thin lms of 5% Co doped Cu2O were grown on single-crystal (001) MgO substrates by PLD without and with 0.5% co-doping with Al, V, or Zn. All the lms showed phase-pure character under the chosen optimum growth conditions. Spin-glass-like behavior was observed in Co-doped lms without co-doping, while room temperature ferromagnetism was found only for the case of Co:Cu2O lms co-doped with Al. The fact that the room temperature ferromagnetic signal was not seen for a low resistivity Zn co-doped Co:Cu2O sample but was seen for a relatively resistive Al co-doped Co:Cu2O sample implied that the mechanism of ferromagnetism in this system is most possibly not carrier induced. The possibility of pure Co metal clusters was considered unlikely based on the rather low value (0.44 B/ Co) of the observed moment per Co (Co metal has 1.67 B/Co and Co nanoclusters have 2.16 B/Co).[106] Interestingly, Zn and V are 3d transition elements with orbitals that are compatible with those of Cu and Co which are also 3d elements. Zn has a xed valence state of +2 whereas V can have a mixed valence. Neither of these co-dopants, however, induce a ferromagnetic state in Co-doped Cu2O. On the other hand, Al, which not only has the smallest ionic radius amongst the co-dopants but also s and p as the outermost orbitals (and therefore no orbital compatibility with the 3d elements) does induce ferromagnetism. The authors argued therefore that an orbital defect state may be responsible for ferromagnetism. Zunger and co-workers[120] studied the case of Co:Cu2O theoretically and suggested a new mechanism which would not require partially lled gap states. This was in clear contrast to the prevalent models for diluted magnetic semiconductors systems (such as p-d exchange or double-exchange) which did rely on the presence of such states. In their new model ferromagnetic coupling was shown to arise from the occupation of the previously unoccupied levels when two transition metal impurities form a close pair. Within this model they have not only explained the ferromagnetic coupling between Co impurities in Cu2O, but have also shown the tendency of the system towards Co clustering. The ferromagnetism is thus suggested not to be carrier mediated as observed experimentally. Ogale, Venkatesan, and co-workers[121] discovered above room temperature ferromagnetism in pulsed laser deposited lms of Co-doped Ti-based oxide perovskite (La0.5Sr0.5TiO3). This system showed characteristics of an intrinsic diluted magnetic semiconductor at low concentrations (<2%), but was seen to develop inhomogeneity at higher cobalt concentrations. Depending on the oxygen pressure during growth the lm properties ranged from opaque metallic to transparent semiconducting and yet these were ferromagnetic. Kulkarni et al.[56] performed oxygen resonant backscattering studies on these La0.5Sr0.5Ti0.93Co0.07O3d lms deposited over a range of oxygen pressures and found a systematic variation of oxygen content in the lm, which correlated well with the associated change in the lm resistivity. The magnetic and optical properties of the relatively resistive La0.5Sr0.5Ti0.985Co0.015O3d lm grown at 102 Torr showed that lm grown under such high pressure qualied as a transparent DMSO system. The Curie temperature (TC) for

such lms was close to 550 K, about 100 K higher than that of the relatively conducting lm grown at 104 Torr. The decrease in magnetization to zero value was found to be gradual for the resistive lm suggesting that the mechanism in this case could be percolative.[122] Qiao et al.[123] examined the temperature dependence of the structural, electrical transport and magnetic properties of La0.5Sr0.5Ti0.93Co0.07O3+d thin lms above room temperature and found that the structure of sample changes at high temperatures. Correspondingly, the resistivity of the samples was found to increase dramatically with the disappearance of ferromagnetism. This transition was shown to be related to the oxygen absorption induced structural distortion, which resulted in the localization of carriers. This work clearly showed that carriermediated coupling between magnetic ions is essential for ferromagnetism in this system. Zhang et al.[124] have reported a systematic study of the magnetic properties and Hall effect on pulsed laser deposited 5% Co: (La,Sr)TiO3 thin lms, especially grown at high substrate temperature. This system was found to be superparamagnetic in nature as evidenced by several protocols of magnetic measurements. Yet, the prole of the measured Hall resistivity vs magnetic eld was found to be identical to the magnetic hysteresis loops. This once again highlighted the limitations of AHE as a tool to test the intrinsic nature of ferromagnetism in DMS systems, supporting a previous report by the Maryland group.[53] However, it was noted that the magnitude of the AHE signal observed in this extrinsic DMSO case was much lower (by a few orders of magnitude) than that found in the intrinsic longrange ferromagnetically ordered DMS. This raised the interesting possibility of using the magnitude of AHE rather than simply its occurrence, as a criterion for elucidating intrinsic vs extrinsic DMS. In an extrinsic DMS system wherein the superparamagnetic clusters are well separated (dilute doping) and do not couple to each other, only polarize the carriers in the nearby region and the corresponding AHE strength is far weaker. However, in the intrinsic DMS, especially for the carrier mediated type, the carriers strongly couple to the local magnetic ions which are homogenously distributed. Therefore, the population of carriers contributing to AHE in the intrinsic DMS is much larger than that in the extrinsic case. Fert and co-workers[125] reported observation of tunneling magnetoresistance (TMR) in Co-doped LaSrTiO3 (Co-LSTO) and demonstrated the existence of signicant spin polarization in this system. The TMR experiments were performed on magnetic tunnel junctions involving Co-LSTO and Co electrodes. They could positively exclude the presence of Co clusters in the Co-LSTO layer from an extensive structural analysis combined with HRTEM and Auger electron spectroscopy. They could thus conclude that the ferromagnetism and high spin polarization in this case are intrinsic properties of this DMSO highlighting its possible signicance for spintronics. This is once again one of the very few examples of the use of DMSO as an active layer in a device conguration, concurrently addressing the most important property of relevance to spintronics, namely the spin polarization. Clearly more such studies are needed for the eld to consolidate and grow in the right direction. Zhang et al.[126] explored the possibility of realizing DMS ferromagnetism in Nb:SrTiO3, which is an n-type oxide conductor.

REVIEW
3142

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

They grew thin lms of (0.5%, 1%) Nb:SrTiO3 dilutely doped with (2 at%) magnetic transition elements (Cr, Co, Fe, Mn) and studied them by various techniques which established high crystalline quality of the lms with no impurity phase(s) and highly uniform dopant distribution. Although the lm conductivity was seen to improve dramatically by Nb doping, no ferromagnetism was found in any of the samples over the temperature range of 365 down to 5 K. This case presented a contrast to another titanium oxide namely cobalt doped (La,Sr)TiO3. Based on the analyses and comparison of the carrier density and the spin density in the two systems the authors argued that these results are contrary to the expectations based on carrier mediated RKKY type physical picture. This prompted them to suggest that a new mechanism may have to be sought to explain the origin or absence of magnetism in these oxide based conductive systems. Ogale et al.[92] examined the possibility of achieving ferromagnetism in dilutely cobalt doped SnO2 lms grown by PLD on R-plane sapphire, which led to high quality single crystalline Rutile phase sample. Ferromagnetism was observed in such lms for cobalt concentrations <3% examined in the work. The lms were found to be semiconducting and optically highly transparent. Interestingly, the lms of Sn0.95Co0.05O2d not only exhibited ferromagnetism with a Curie temperature close to 650 K, but also a giant magnetic moment of 7.50.5 B/Co. The moment was seen to drop rapidly with increase in the Co content. Giant moments have been observed earlier in transition metal doped palladium and alkali metal systems,[127133] and these have been a subject of interesting scientic debate for many years. The giant moment in the Co:SnO2 case was noted to be much larger than the value of 1.67 B/Co for cobalt metal, that for small Co clusters (2.1 B/Co), or that for any of the standard Co oxides wherein the orbital moment is quenched. The authors conjectured the possibilities that the atoms surrounding the cobalt atoms may have acquired a moment through electronic effects, or the orbital moment of cobalt remains unquenched for some peculiar reason. The authors further noted that argon annealing of a Sn0.95Co0.05O2d lm led to a drop in the moment, and caused concomitant changes in the transport and structural features. This indicated that the state supporting the giant moment was possibly a highly metastable state. Subsequent work has shown that it is not trivial to stabilize such a state. Several other groups have also reported observation of giant magnetic moments in their samples, but questions regarding the origin of such moments, their stability, sensitivity to growth conditions and potential extrinsic contributions have remained mostly controversial as discussed below. Coey and co-workers[45] studied thin lms of Sn0.95Fe0.05O2 grown by PLD and found that these are transparent ferromagnets with Curie temperature of 610 K and spontaneous magnetization of 2.2 A m2 kg1. Iron was found to be in high-spin Fe3+ state but only 23% of the iron was noted to have ordered magnetically. The moment per ordered iron ion, 1.8 B, was found to be greater than that for any simple iron oxide with superexchange interactions. Ferromagnetic coupling of ferric ions via oxygen vacancy F-center (resembling Kasuyas bound magnetic polaron) was proposed to explain the high Curie temperature. Such oxygen vacancy should arise naturally to ensure charge

neutrality when a trivalent ion is substituted for Sn4+. The radius of the F-center electron orbital is of the order a0, where a0 is the Bohr radius and is the dielectric constant of SnO2 (13). The overlap of the corresponding electron wave-function with the d-orbitals of neighboring Fe ions was suggested to be responsible for the ferromagnetic exchange. Fitzgerald et al.[134] examined and found occurrence of room temperature ferromagnetism in Rutile phase (Sn1xMx)O2 (M = Mn, Fe, Co, x = 0.05) ceramic samples. Room temperature saturation magnetization values of 0.11 or 0.95 B per Mn or Fe atom were observed with Curie temperatures of 340 and 360 K, respectively. In the case of Fe doping about 85% of the iron was found to be in a magnetically ordered high spin Fe3+ state. Fitzgerald et al.[135] did further detailed work on ferromagnetism in dilutely transition element doped SnO2 lms grown by PLD on R-cut sapphire substrates. Ferromagnetism at room temperature was found with Cr, Mn, Fe, Co, or Ni, but not with other 3d cations. Once again the extrapolated Curie temperatures were generally in excess of 500 K. Intriguingly, the moment of the lms was found to be roughly independent of doping level up to about 15 at %, with a value per unit substrate area of 200 100 B nm2. Highest (giant) values of moment were found for iron and the trend in all cases was from higher moment at low concentration to lower values at higher concentrations. Also, the magnetization was found to be highly anisotropic. The samples were found to be degenerate n-type semiconductors with a Hall mobility of 100 cm2 V1 s1. Surprisingly, no AHE or magnetoresistance (MR) was observed at room temperature. The authors explained the variation of magnitude of the magnetic moment with increasing dopant concentration in terms of the distribution of magnetic impurities in the Rutile lattice. Isolated impurities possessed the full spin-only moment, but nearest-neighbor pairs were suggested to be coupled antiferromagnetically, making no net contribution to the magnetization. Higher associations contributed differently according to this picture, but led to a progressive net reduction in magnetization. The authors emphasized that the observed large moments at low concentrations are inexplicable if the only source of magnetism in doped SnO2 is the transition metal dopant. As possible explanation for such behavior the authors suggested the presence of magnetic centers in the doped SnO2 lms in the vicinity of the substrate/lm interface which are not identied with the 3d cations. The suggestion that the interface played a role was supported by their observation that a ferromagnetic moment was observed when V-doped SnO2 lms were deposited on some specic substrates (e.g., LaAlO3) but not others (e.g., Al2O3). They inferred the thickness of the interface region wherein the magnetic centers are located to be 10 nm. Since the moment is rather large and may not be accountable entirely by the interface layer, and since undoped SnO2 lms showed no ferromagnetism, these authors suggested an interesting possibility that the defect centers have a nonmagnetic ground state, but that a low-lying magnetic excited state is stabilized by exchange coupling to a 3d dopant ion. Absence of AHE indicated that there was negligible 3d density of states at EF. Hong et al. have also examined Ni-doped SnO2 and Cr-doped ZnO lms fabricated under various conditions.[136] They found that the structural defects and/or oxygen vacancies inuence the magnetism in these systems signicantly. They found that

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3143

REVIEW

www.advmat.de www.MaterialsViews.com

REVIEW

in some cases elimination of defects might contribute to a signicant reduction of the ferromagnetic ordering. Filling up of oxygen vacancies was also noted to degrade the magnetic moment. Interestingly, the authors suggested that some specic defect types could also help maintain its ferromagnetic ordering. The Gamelin group has also worked on the doped SnO2 system but in the nanoparticle form. Archer et al.[137] studied colloidal Ni2+-doped SnO2 nanocrystals in the context of occurrence of magnetism. The Ni2+ dopants were found to occupy pseudo-octahedral Sn4+ cation sites of Rutile SnO2 without local charge compensation. Robust ferromagnetism with 300 K saturation moment Ms 0.8 B/Ni2+ was found in spin-coated lms only, suggesting a key role of interfacial fusion defects in magnetism. Reversibility of the paramagnetic-ferromagnetic phase transition was also observed. The phase transition was seen to reverse under air but not under N2, suggesting the microscopic identity of the activating defect to be interfacial oxygen vacancies (VO2,). Although relatively less investigated, the SnO2 case appears to be somewhat different in its details as compared to the TiO2 and ZnO cases. The issue of giant moments has captured signicant attention in this specic context, but the origin of the same has been suggested to be related primarily to the defects states introduced in this system under certain growth conditions and interactions of the defects with the magnetic dopants. More decisive work is clearly needed to be performed on this oxide. Hong et al.[138] have studied Ni-doped In2O3 thin lms grown on sapphire and MgO substrates by laser ablation under different conditions. They presented evidence of room temperature ferromagnetism in single phase crystalline lms with Ni substituting for Indium atoms with a uniform dopant distribution over the whole lm thickness. They showed that lms grown at a low temperature of 550 C had the same Ni concentration as in the bulk target, and had best crystallinity and the largest magnetic moment 0.7B/Ni.

4. Defect Ferromagnetism and Ferromagnetism Without Magnetic Impurities or Constituents in Metal Oxides
The high sensitivity of the observed ferromagnetism in most oxide based DMS systems to the growth/synthesis conditions, issues of inhomogeneity of samples and non-reproducibility of results from different laboratories (once again representing high sensitivity to growth conditions) and the difculties in relating the experimental results to known models of ferromagnetism have led scientists to conclude that in these DMSO systems various defect states (point and extended) must play a major role. Moreover there are a number of examples wherein no magnetic impurity or element is involved but ferromagnetism is encountered under specic synthesis conditions. This has led to the re-emergence of a new research frontier of defect ferromagnetism. It has also given an impetus to theoretical modeling and rst principles calculations of electronic states of defect solids. While the issue of the role of defects in ferromagnetism has been discussed in the past literature, the one which makes

defects as the primary origin or source of magnetic moment or the mechanism of exchange is rather new. In the following paragraphs, we discuss this emerging scenario briey. The suggestions by Coey about the possible role of oxygen vacancy F-center defects in the mechanisms of ferromagnetism in certain insulating DMSO lms and the works of Sawatzky,[139] Zunger[140] and others brought to focus the central role of defect states in DMS ferromagnetism. However, the discussion was triggered further by the rather unusual result reported by Coey and co-workers[141] of an unexpected ferromagnetism in undoped HfO2 lms with a Curie temperature exceeding 500 K and a magnetic moment of about 0.15 B per HfO2 formula unit. Since magnetic order in an insulator is understood to require the cation to have partially lled shells of d or f electrons, this discovery challenged our basic understanding in this regard, because neither Hf4+ nor O2 are magnetic ions, and the d and f shells of the Hf4+ ion are either empty or full. Moreover, the magnetization was found to be remarkably anisotropic, being up to three times greater when the magnetic eld is applied perpendicular to the lm plane than when it is parallel to it. The coercive eld for the lm at 5 K was found to be about 5 millitesla (mT). While the results of HfO2 could be reproduced by the authors in lms of different thicknesses prepared at different oxygen pressures and on different substrates, no ferromagnetism was found in similarly prepared lms of zinc or tin oxides. The authors suggested that by allowing the impurity band to mix with the empty 5d states of hafnium and transfer a fraction of an electron for each vacancy, the 5d states would polarize the impurity band and provide the required ferromagnetic coupling. The anisotropy suggested a large orbital contribution to the 5d moment. Coey et al. followed up their initial report on unexpected ferromagnetism in undoped hafnia with further work.[91] They found that thin lms of HfO2 produced by PLD on sapphire, yttria-stabilized zirconia, or silicon substrates were ferromagnetic at room temperature with little hysteresis and fairly strong anisotropy. The extrapolated Curie temperatures were far in excess of 400 K. It was also found that the moment did not scale with lm thickness, but in terms of substrate area it was 150400 B nm2. Interestingly, Pure HfO2 powder was also seen to develop a weak magnetic moment on heating in vacuum, which got eliminated on annealing in oxygen. These data strongly suggested lattice defects in the interface layers as the likely source of the magnetism. Following the work of Coey et al., Abraham et al.[142] addressed the issue of ferromagnetism in undoped HfO2 lms and in the process established the limits of magnetism in thin, electronic grade, hafnium oxide, and hafnium silicate lms grown onto silicon by chemical vapor deposition (CVD) and atomic layer deposition (ALD). They did not nd any ferromagnetism in these samples to the limit of sensitivity of their measurement. Importantly, they brought out that contamination by handling with stainless-steel tweezers could lead to a measurable magnetic signal. Moreover they showed that the magnetic properties associated with such contamination (including the magnitude of moment, magnetization eld dependence, and anisotropy) were similar to those attributed to ferromagnetic HfO2 by Coey et al.[143] This revelation brought out a very important consideration which has perhaps been central to

3144

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

the confusions regarding non-concurrence of results from different laboratories in the DMSO eld. This is not to say that all of the observations and reports of growth condition sensitivity are necessarily inaccurate. As discussed, the case of defect ferromagnetism is signicantly strong as a fact of reality than attributable simply to contamination. However, the fact that contamination can lead to ferromagnetic effects that also have a systematic which could apparently reect true or intrinsic ferromagnetic effects is rather dangerous and needs researchers to take further precaution in reporting their research results. This consideration also makes one ponder over the reports of large or giant moments where the inverse linear concentration dependence of the moment could be equally well explained by a roughly constant extrinsic moment added to the measurement through contamination and then one dividing this value by carrier concentration assuming that it all arose from the intrinsic ferromagnetism. Since the degree of contamination added to different samples and in samples made in different laboratories would clearly be signicantly different, it seems improbable that all such data and systematic may originate in contamination alone. There seems to be a fair case for growth condition dependent defect related ferromagnetism. In contrast to manganite, magnetite or ferrite physics to which the oxide magnetism community has been glued for decades, the actual measurement signals in the case of DMS and DMSO systems are very weak and therefore eliminating any small possibility of contamination becomes all the more important. One case of such contamination that was noted immediately by some laboratories is the use of stainless steel blade to remove the samples after a high temperature growth. Even if stainless steel is non-magnetic, the debris that may come off represents a transformed material or a compound that can be ferromagnetic. This prompted researchers to keep away from steel blades and tweezers while handling DMSO samples. There can be several other sources of extrinsic magnetic signals and irreproducible DMSO results. These may emanate from the materials and processes used in experiments as well as the measurement protocols and procedures employed. For instance, the work of Salzer et al.[144] brought out that the substrates themselves can be a source of magnetic contamination. These authors found that all the sapphire substrates they used showed ferromagnetism and the same got reduced substantially after surface cleaning treatments suggesting adhering surface contamination. They conjectured that transition element impurities could be the source of ferromagnetism, although no clear correlation between the impurity concentration and saturation moment could be established. Apparently, such impurities seemed to contribute mainly to the measured paramagnetic signal. Nevertheless, their work emphasized caution in this respect in view of the weak signal case one is faced with. Ney et al.[145] have analyzed the artifacts that may emanate in a SQUID measurement which is invariably used for small signal magnetic measurements. They showed that observation of magnetic hysteresis is a necessary but not sufcient criterion to prove ferromagnetism in samples which give small signal. They brought out the typical signatures of cleaving edge contamination of the sample or misaligned magnetic eld. They further demonstrated the high degree of sensitivity of anisotropy measurements in case of samples with weak magnetic

response causing artifacts from the sample holder and sample geometry. They also discussed the issue of residual hysteresis in the context of trustworthy vs false signal. They concluded that for signals below 4 107 emu artifacts are possible and the corresponding claims of ferromagnetism become questionable unless complementary techniques are used. It is important to emphasize further that things like silver paste used to mount the substrate can also give false ferromagnetic contributions if they are not selected and applied judiciously. If one uses silver pastes that may have impurities, they can get hardened and ferromagnetic compounds can form after processing the paste at high temperature during lm growth. One must also consider contamination coming from dies used for making pellets, especially if pre-ablation or surface cleaning is not performed carefully before lm growth. Finally, the plastic sieves, holders, Teon tape, etc. that may be used during SQUID measurements need to be analyzed for possible magnetic signals before sample loading. With such precautions it is hoped that the newer breed of papers may have more concurrence from different laboratories and hopefully a good reproducibility of results could be realized. Rao et al.[146] (Maryland group) performed several experiments to search for ferromagnetism in undoped and cobaltdoped high-k dielectric HfO2 lms. Contrary to Coeys result, their study did not show ferromagnetism in any undoped HfO2 lms over a broad range of growth conditions. On the other hand, they did observe room temperature ferromagnetism in dilutely Co-doped HfO2 lms, but the corresponding origin was judged to be extrinsic due to a Co rich surface layer, at least for the regime of growth conditions explored. The cobalt layer on the surface was inferred from SIMS measurement which showed that Co signal is strong on the surface and decreases quickly within the rst 50 nm. Following the theoretical proposal by Sawatzky et al.,[139] Zunger[140] had concluded from his theoretical work that for CaO the estimated vacancy concentration required for obtaining ferromagnetism was 1021 cm3, or about 5% Ca vacancies, which exceeded by 3 orders of magnitude the computed equilibrium concentration of vacancies. Tirosh and Markowich[147] noted that vacancies in HfO2 may have a similar interaction radius and with a higher percolation threshold than that for the CaO lattice a higher vacancy concentration would be needed to obtain ferromagnetism in HfO2. They argued that far-fromequilibrium preparation methods may be needed to produce crystals with the high concentration of the defects required. They implemented this strategy by synthesizing HfO2 nanorods by a modied synthesis procedure which enabled them to achieve and control high defect density. Under such high density condition they did observe ferromagnetism as conjectured by Zunger.[140] Interestingly, they also observed a reduced bandgap of 4 eV in the ferromagnetic sample consistent with the theoretical prediction by Pemmaraju and Sanvito.[86] Recently, Zhang et al.[148] (Maryland Group) observed magnetic effect in another functional non-magnetic material, Nbdoped anatase TiO2 when grown under specic conditions, through electrical measurements. One advantage of the electrical measurement of the magnetic effect (Kondo scattering) is that it reveals key information on the coupling between the itinerant carriers and localized magnetic moments. As a result, the

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3145

www.advmat.de www.MaterialsViews.com

REVIEW
Figure 11. a) A typical X-ray diffraction pattern of Ti1xNbxO2 thin lms grown on LaAlO3 substrates; LAO denotes the substrate peaks and A denotes the anatase peaks. b) Sample geometry used in electronic-transport measurements. H is the applied magnetic eld, which is along the z-axis for Hall measurements. In order to avoid a MR contribution, H was swept from negative to positive values and the Hall resistance is given by: rH = (rxy (H) rxy(-H))/2. For measurements of the angular dependence of the MR, the magnetic eld was rotated in the xz plane, with being the angle between the z-axis and H. Temperature dependence of c) the resistivity, , and d) the Hall constant of Ti1xNbxO2 thin lms (x = 0.01, 0.03, 0.05, 0.07). The inset shows the carrier concentrations derived from the Hall constants (1 kOe = 0.1 T). Reprinted with permission from.[148]

electrical results exclude any contribution from unintentional metallic contamination of the lms during handling or from magnetic impurities in the substrate. Figure 11 shows the X-ray diffraction data, the geometry for electrical measurements, as well as the resistivity and Hall coefcient data for different niobium concentrations. The upturn in resistivity below a certain temperature and its ln T dependences are clear. Also, nearly temperature independent RH implies that Niobium donors are degenerate and that the upturn in resistivity should only be attributed to temperature dependence of scattering processes. Through the measurements of magnetic eld dependence of resistivity and the corresponding anisotropy the authors proved that Kondo scattering is the primary source of scattering with only a small contribution of weak localization. Zhang et al.[148] provided evidence for the cation-vacancy origin of the magnetism from X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS) measurements, supported by rst-principles calculations. They observed that there could be two possible origins of the local magnetic moments: specic cation valence states or other defects states. If some portion of substituted niobium is in the +4 valence state or Ti is in the +3 valence state, then the unpaired d-electron in Nb4+ or Ti3+ may act as a localized spin. XPS measurements however refuted this possibility, since it showed Nb in 5+ state and Ti in 4+ state only. The authors then performed XAS measurements, which revealed systematic increase in the possible contribution originating from Ti 3d holes or oxygen interstitials with increase in oxygen pressure. First principles calculations showed that no magnetic moment is produced by neutral O-vacancy (NbTi+VO) or interstitial O (NbTi+Oi) in Nb-doped TiO2. Only VTi defects were found to act as magnetic centers. Figure 12 shows the XAS

data and the suggested origin of the local magnetic moment by cation vacancy. Pan et al.[149] reported another case of ferromagnetism in a system with non-magnetic elements, namely carbon doped ZnO. They predicted ferromagnetism in this case by rst principles calculations and observed the same experimentally in PLD lms. Their work suggested that the ferromagnetism emanates from Zn-C system in ZnO environment. Hong et al.[150] reported observation of room temperature ferromagnetism in undoped TiO2, HfO2, and In2O3 thin lms grown under specic conditions. They found that in TiO2 and HfO2 the ferromagnetism is basically in-plane, since application of magnetic eld perpendicular to the lm plane led to paramagnetism for TiO2 lms and a mixed state of diamagnetism and a small component of paramagnetism for HfO2 lms. In the case of In2O3 lms, no anisotropy was found for lms on MgO, however for lms on Al2O3 ferromagnetism was found to be in the perpendicular conguration. These authors also found a strong thickness dependence of the magnetic moment in these undoped oxide thin lms. Hong et al.[151] have also reported room temperature ferromagnetism in laser deposited undoped SnO2 thin lms grown on LaAlO3 substrates Experiments on Mn-doped SnO2 lms revealed that Mn doping did not play any key role in the ferromagnetism in this system. Rao and co-workers[152] reported an interesting and intriguing observation of room-temperature ferromagnetism in nanoparticles (730 nm diameter) of several nonmagnetic oxides such as CeO2, Al2O3, ZnO, In2O3, and SnO2. They observed that the saturated moments in CeO2 and Al2O3 case were comparable to those observed in transition-metal-doped wide bandgap

3146

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

implications of such surface-related ferromagnetism to carriermediated DMS or DMSO ferromagnetism. Finally, it is important to point out that the recent work of Brinkman et al.[154] on the observation of ferromagnetism at the interface of non-magnetic materials (SrTiO3/LaAlO3), which cannot exactly be qualied under defect ferromagnetism, has added to the richness of physics at structural and compositional (valence) discontinuities in solids that remains to be unfolded in different systems. Some of the ingredients of related physics may be applicable to the analyses of defect ferromagnetism and surface ferromagnetism as well. Clearly the subject of defect ferromagnetism has just begun to be unfolded for metal oxide systems. A combined experimental and theoretical effort is necessary to render precise insights into this exciting but complex problem. It will be a separate effort of course to see how one could stabilize the defect states in specic systems to exploit them in spintronics or magneto-optic applications.

5. Models and Theoretical Work


Given the intriguing nature of results on several DMSO systems it is not surprising that the subject has triggered the curiosity of theorists and several theorists have contributed to the development of new ideas, models and schemes to explain the experimental results and to predict new possibilities. Some of these studies are devoted to the rst principles (density functional theory, DFT) calculations and others to the development of new models of exchange and global ferromagnetic order. In the context of the most investigated Co:TiO2 DMSO system Sullivan and Erwin[155] employed rst-principles (density functional theory) microscopic calculations of the formation energy, electrical activity and the magnetic moment of Co dopants and different type of native defects in anatase TiO2. Using these inputs and equilibrium thermodynamics they found that under O-poor growth conditions that are usually employed during growth a substantial fraction of Co dopants occupies interstitial sites as donors, which lead to n-type behavior due to incomplete compensation by substitutional Co acceptors. All of the cobalt (substitutional and interstitial) was suggested to be in Co2+ state, consistent with experimental results. They also found that oxygen vacancies do not play a signicant role in Co doped anatase TiO2. Geng and Kim[156] reported on the interplay between the local structure and magnetism in anatase Co:TiO2 based on ab initio density functional theory (DFT) calculations. Their calculated formation energy of a pair of substitutional Co ions indicated that they have a tendency to cluster without noticeable effect on the low spin state of Co, a condition distinctly different from a cobalt metal (Co(0)) cluster. Further, their calculations showed that interstitial Co is strongly attracted to a substitutional Co; in fact, more strongly to a substitutional Co pair. They also found that interstitial cobalt is in low spin state and destroys the spin polarization of the surrounding substitutional cobalt. Janisch and Spaldin[157] performed rst-principles DFT study within the local spin density approximation (LSDA) of the Co-doped TiO2 system and calculated the magnetic ordering and electronic properties as a function of the Co concentration, and the distribution of Co dopants and oxygen vacancies. They

Figure 12. a) Schematic showing how a cation vacancy leads to a local magnetic moment; left: crystalline structure of anatase TiO2; right: top view of a TiO plane with and without a Ti vacancy. The big spheres represent Ti and O ions, respectively, while the small spheres with arrows represent electrons provided by Ti and O atoms, respectively. The arrows represent spins, the big dashed circles represent Ti vacancies, the small dashed circles with arrows represent holes, and the circular periphery encompassing spins is a guide for the eye showing the formation of the net magnetic moment. b) Ti L-edge XAS spectra of an undoped TiO2 lm grown at 104 Torr, and Ti0.95Nb0.05O2 lms grown at 105, 104, and 2 104 Torr. The inset in (b) shows the concentration of magnetic moment as a function of O partial pressure for the Ti0.95Nb0.05O2 lms. Reprinted with permission from.[148]

semiconducting oxides. Bulk samples obtained by sintering such nanoparticles at high temperatures in air or oxygen however were noted to make the material diamagnetic. They attributed the ferromagnetism to the oxygen vacancies at the surfaces of nanoparticles, and suggested this ferromagnetism may be a universal feature of metal oxide nanoparticles. Subsequently, Rao and co-workers[153] also observed room-temperature ferromagnetism in undoped GaN and CdS nanoparticles with an average diameter in the range 1025 nm synthesized by simple routes. It would be interesting to examine the connection and

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3147

www.advmat.de www.MaterialsViews.com

REVIEW

found that Co atoms prefer to substitute on neighboring sites of the Ti lattice and, importantly, this leads to a ferromagnetic super-exchange based on the well established GoodenoughKanamori-Anderson rules. Moreover, it was emphasized that these local interactions being ferromagnetic they do not compete with the long-range carrier-mediated interactions that are also ferromagnetic, thereby leading to a robust ferromagnetic state. Weng et al.[158] examined the Co-doped anatase TiO2 system by rst-principles calculations in pseudopotential plane-wave formalism within the local-spin density approximation, supplemented by the full-potential linear augmented plane-wave method. Their calculated results for substitutional Co for Ti showed Co in its low spin state, consistent with experimental results. Interestingly, the oxygen vacancy was seen to enhance the ferromagnetism, having a larger effect on the electronic structure and optical properties than the Co-doping concentration alone. Weng et al.[159] also performed ab initio calculations of the absorption and optical MCD spectra for the Co-doped anatase and Rutile TiO2 systems. The positive MCD signals around the absorption edge and critical points obtained in their calculated spectra implied that the p-d exchange interaction between the O 2p and Co 3d electrons should be ferromagnetic, once again supporting the case for intrinsic ferromagnetism in these systems. Murugan et al.[160] studied the electronic and magnetic properties of Dual-impurity-doped (M1 and M2, from M1 = Cr, Mn, Fe, Co, and Ni, and M2 = Mo, W, and Re) TiO2 (Rutile) using rst-principles calculations within the generalized gradient approximation. Their results brought out the overlapping of the highest occupied impurity states with the bottom of the conduction band of the host system and the halfmetallic nature in (Fe,W) dual doped TiO2. It was further noted that these impurities preferentially substituted on the adjacent Ti sites and formed stable magnetic complexes with oxygen atoms. Badaeva et al.[161] performed linear response time-dependent hybrid density functional theoretical analysis of Co2+- and Mn2+doped ZnO quantum dots (QDs) with sizes up to 300 atoms (1.8 nm diameter), and investigated QD size effects on the absorption spectra, focusing on charge-transfer (CT or photoionization) excited states. CT transitions involving excitation of dopant d-electrons to the ZnO conduction band were found to occur lowest in energy, with additional CT transitions corresponding to the promotion of ZnO valence band electrons to the dopant d-orbitals appearing at higher energies, consistent with the experimental results. The CT energies were also found to depend on the QD diameter. In addition to CT transitions, the Co2+-doped ZnO QDs also exhibited characteristic d-d excitations whose experimental energies were reproduced well and were not found to depend on the size of the QD. Spaldin[162] performed a computational study of ZnO in the presence of Co and Mn substitutional impurities to identify potential ferromagnetic ground states within the (Zn,Co)O or (Zn,Mn)O systems, and found that robust ferromagnetism is not realized by substitution of Co or Mn on the Zn site, unless additional carriers (holes) are also incorporated. Computationally, Spaldin simulated p-doping by imposing a high concentration of Zn vacancies, though this is clearly experimentally unfeasible. Spaldin therefore proposed and computationally

showed that doping with Cu may be a feasible way of making ZnO p-type that could support ferromagnetism with dilute transition element doping. In her computation, when two separated Co ions were included in a 32 atom unit cell with a single Cu ion placed as far as possible from the Co ions, she obtained a ferromagnetic ground state more than 80 meV lower in energy than the corresponding antiferromagnetic state. In a recent paper Dietl[163] addressed the issue of the experimentally observed disagreement between photoemission and optical ndings in magnetically doped GaN and ZnO. He argued that in these materials the conditions for the strong coupling between the band hole and the localized spins are met. Such strong coupling between valence-band holes and localized spins leads to a midgap ZhangRice-like state, a sign reversal of the apparent p-d exchange integral, and an increase of the bandgap with the magnetic ion concentration. His model accounts for the signicantly reduced magnitude of the p-d exchange integral in (Zn,Co)O and (Zn,Mn)O. The positive sign of this integral, as implied by the model, points to inverted sub-band ordering in ZnO. Zunger[164] analyzed a number of dopant related theoretical studies and laid out seven practical doping principles for oxide matrices. He suggested that the enthalpy of forming anion (cation) vacancies decreases under host cation-rich (anionrich) conditions. He further suggested that cation-substituting dopants should be more soluble under host-cation poor (hostanion rich) growth conditions. There have been some theoretical studies which have offered new ideas and scenarios of magnetism in these DMS systems without adhering to the specics of any particular system. Das Sarma and co-workers[165] developed theory of diluted magnetic semiconductor (DMS) ferromagnetism in quasi-two dimensions (relevant to ultrathin lms) and found that long range ferromagnetic order can be stabilized by even a small amount of anisotropy. They showed that nite temperature effects in carriermediated effective interaction between impurity moments introduce a strong dependence on the density of carriers. They also found that the 2D DMS TC is strongly suppressed compared with the corresponding 3D DMS TC. In the 3D case which they also investigated, they found that impurity correlations have only small effects on TC with the neutrally correlated random disorder producing the nominally highest TC. Interestingly, they further found that when approached from the high temperature paramagnetic side the ferromagnetic order sets in through a random magnetic clustering phenomenon consistent with the percolative transition scenario. Dagotto and co-workers[166] performed generic Monte Carlo study of a lattice spin-fermion model for diluted magnetic semiconductors numerically, with an objective to establish a phase diagram by improving upon the previously used meaneld approximations. They obtained Curie temperatures (TC) by varying the dopant spin and hole densities, and importantly the impurity-hole exchange J in units of the hopping t. Interestingly the optimal values were found in the intermediate regime between those for the itinerant and localized carriers. They found a clustered state above the Curie temperature with suppressed ferromagnetism at intermediate and large J/t. Nayak et al.[167] have studied the case of Co doped ZnO (Zn1xCoxO) diluted magnetic semiconductors from

3148

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

rst-principles and Monte Carlo simulations. By calculating the exchange interactions using the KorringaKohnRostoker (KKR) Greens function method, they used the corresponding exchange coupling constants in a classical Heisenberg model to examine the magnetic phase transitions by the Monte Carlo technique. They also addressed some specic aspects related to overestimation of Zn 3d and O 2p hybridization and underestimation of the bandgap with respect to DFT calculations within the generalized gradient approximation (GGA). Their results showed that no magnetism is sustained by defect free substitution of Co for Zn in ZnO at low (<20%) concentration (which is already too high in the DMS context). This is in clear contrast to several experimental and even some theoretical reports. Ferromagnetism, with Curie temperature below room temperature, was found at intermediate Co concentrations (20% to 65%). Sanyal et al.[168] have examined the chemical and magnetic interactions in Co doped ZnO by ab initio density functional calculations and have brought out a strong tendency of clustering between Co atoms. Importantly they have also established that clustering of magnetic atoms in DMS affects the electronic structure as well. They have further shown that the Curie temperature for a homogeneous sample is low, whereas allowing for an inhomogeneous growth results in very high Curie temperatures. Once again these results contradict some other theoretical works and also raise questions about the high Curie temperatures observed and attributed to uniform cobalt distributions in some experiments. In the context of defect ferromagnetism, Sawatzky and coworkers[139] suggested that large-bandgap nonmagnetic materials, such as CaO, could exhibit local magnetic moments associated with Ca vacancies that, at concentrations as low as 3%, could transform the nonmagnetic insulators into ferromagnetic (FM) half-metals. Zunger and co-workers[140] showed, however, that magnetic percolation of the FM interactions of such cation vacancies on a face centered-cubic (fcc) lattice requires a minimum of 4.9% vacancies. Their totalenergy calculations for CaO showed that, because of the high energy of vacancy formation, one cannot obtain more than 0.003% vacancies under thermal equilibrium at room temeprature. Zunger and co-workers[169] also studied the intriguing case of the possibility of ferromagnetism in undoped HfO2. They found that not all Hf vacancies have a local magnetic moment. They estimated that for equilibrium growth scenario the concentration of Hf vacancies with magnetic moment at room temperature needed for collective (percolative) ferromagnetism is eight orders of magnitude higher than what is feasible, once again suggesting that the only possibility that one could hope to bridge such a vast difference would be via use of highly non-equilibrium material growth. As stated earlier, the work of Tirosh et al.[147] has borne out this suggestion in the case of HfO2 nanorods grown under non-equilibrium conditions. Separately, using density functional theory, Pemmaraju and Sanvito[86] showed that the isolated cation-vacancy sites in HfO2 lead to the formation of high-spin defect states that are ferromagnetically coupled with a short-range magnetic interaction. In order to explain high Curie temperatures (much above 300 K) and rather large moments (approaching or even exceeding spin only moments at low concentrations) observed

in some high k dielectrics (generally n-type oxides that may be partially compensated) with degenerate or thermally activated semi-conductivity, Coey et al.[170] came up with a model for ferromagnetic exchange mediated by shallow donor electrons that form bound magnetic polarons which overlap to create a spinsplit impurity band (schematic shown in Figure 13, top panel). They showed that in this framework the Curie temperature in the mean-eld approximation should vary as (x)1/2, where x and are the concentrations of magnetic cations and donors, respectively. They further showed that only when empty minority-spin or majority-spin d states lie at the Fermi level in the impurity band, high Curie temperature could occur. The magnetic phase diagram (Figure 13, lower panel) based on this model could include diverse states such as semiconducting and metallic ferromagnetism, cluster paramagnetism, spin glass and canted antiferromagnetism. Coey et al.[170] argued that an electron associated with a particular defect will be conned to a hydrogenic orbital of radius rH = (m/m)a0, where is the high-frequency dielectric constant, m is the electron mass, m is the effective mass of the donor electrons, and a0 is the Bohr radius. As the donor concentration increases, the 1s orbitals would overlap to form an impurity band. At low concentrations, the electrons may remain localized because of the inuence of correlations and potential uctuations in a narrow band, but above a critical donor concentration the impurity band states would become delocalized, and metallic conduction would set in. With increase in the density of defects, the hydrogenic orbitals associated with the randomly positioned defects should overlap. The polaron percolation threshold and the cation percolation threshold would then be the two landmarks on the magnetic phase diagram. The authors also provided an estimate for the Curie temperature within this picture. Their model implied that high Curie temperatures require hybridization and charge transfer from a donor-derived impurity band to unoccupied 3d states at the Fermi level. Coey[171] also analyzed the unusual magnetic effects contradicting the accepted paradigms of localized or band magnetism encountered in the case of four ferromagnetic systems, namely irradiated graphite, non-stoichiometric CaB6, thin lms of HfO2, and doped nonmagnetic oxides. He noted that all of these have small ferromagnetic moments and Curie temperatures well above room temperature despite the absence of atoms with partially lled d or f shells. He identied that the common feature in all these cases was the presence of lattice or bond defects, capable of giving rise to an impurity band. He offered insights into different ways in which this impurity band may become spin-polarized. In one of his subsequent papers, Coey[172] analyzed the diverse, unusual and unexpected aspects of DMSO physics. He suggested that a rst move towards out of box thinking was to abandon the idea that the moment necessarily resided on the magnetic cation, implying that some extra moment may possibly lie elsewhere in the system including defects induced by the dopant or present otherwise. He pointed out that a defect such as a cation vacancy, or an aliovalent substitution such as Fe2+ for Ti4+ in TiO2 could introduce an electronic defect on the oxygen in proximity for charge neutrality, creating an O2p5 anion. The oxygen holes are strongly correlated, and

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3149

www.advmat.de www.MaterialsViews.com

REVIEW

Isolated polaron

Antiferromagnetic pair

Isolated ion

Overlapping polarons

Spin glass Canted antiferromagnet

Ferromagnetic metal

Ferromagnetic insulator 3p Clusters Antiferromagnet

Para 0 0.2 xp 0.4 x

Figure 13. a) Representation of magnetic polarons. A donor electron in its hydrogenic orbit couples with its spin anti-parallel to impurities with a 3d shell that is half-full or more than half-full. Cation sites are represented by small circles. Oxygen is not shown; the unoccupied oxygen sites are represented by squares. b) The magnetic phase diagram for dilute ferromagnetic semiconductors. The electrons are localized in the shaded area. xp and p are the cation and donor polaron percolation thresholds, respectively. is the ratio of the radius of the hydrogenic donor orbital to the Bohr radius. Reprinted with permission from.[170] Copyright 2005, Nature Publishing Group.

they can form extended molecular orbits around the defect site, which couple ferromagnetically. These orbitals may be extended in space, thereby reducing the percolation threshold, and the Exchange between these nearly-lled degenerate states should be ferromagnetic. He further pointed out that there is enough indirect experimental evidence that the strange ferromagnetism is related to defects, and the dilute doping of the oxides with magnetic cations could be something of a smokescreen in this context. He pointed to their peculiar experimental observation that in Fe-doped TiO2 case iron was found to be paramagnetic, but the lm to be ferromagnetic. Coey et al.[173] have also proposed a new model for ferromagnetism associated with defects in the bulk or at the surface of nanoparticles in an attempt to explain the so called phantom ferromagnetism observed in the diverse set of samples encompassing undoped and doped lms, and functionalized nanomaterials. The model is based on the identication that a narrow local density of states Ns(E) is associated with the defects, but the Fermi level will not normally coincide with a peak in it. If however a local charge reservoir (e.g., a dopant cation coexisting in two different charge states) or a chargetransfer complex at the surface, is provided then it may be possible for electron transfer to raise the Fermi level to a peak in the local density of states, leading to Stoner splitting of Ns(E). Spontaneous Stoner ferromagnetism can thus arise in percolating defect-rich regions. Finally, in the context of the Cu:ZnO case of recent interest, Ye et al.[174] have performed rst-principles simulations and have predicted that this should be a half-metallic dilute magnetic semiconductor with the ferromagnetism emerging from the hybridization of the Cu 3d and O 2p bands. Interestingly however, they did not nd any close relation between the ferromagnetism and the hole density. They found that oxygen vacancies tended to destroy the ferromagnetism. L. Huang et al.[175] also noted a half-metallic ground state in this system. They too found that n-type defects tended to destroy ferromagnetism while p-type defects enhanced the ferromagnetic stability, apparently consistent with the report of Buchholz et al.[107] D. Huang et al.[176] also addressed the problem from rst principles and concluded that p-type ZnO:Cu could have ferromagnetic properties, but n-type ZnO:Cu would not. They found that the neutral substitutive Cu defects are favorable to clustering, maintaining ferromagnetic ordering. Lathiotakis et al.[177] studied an interesting case of Co and Cu co-doping in ZnO and suggested that the Cu ions enhance ferromagnetism in Zn(Co,Cu)O by acting as superexchange

3150

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

mediators while causing a remote delocalization through the hybridization of the Cu d3z2r2 spin-majority states with the O2p states. Recently, Ferhat et al.[178] have performed rst-principles calculations based on the density functional theory on Cu-doped ZnO and have found a net magnetic moment of 0.57 B for Cu at a composition of 6.25%. They also conrmed an appreciable bandgap reduction in ZnO in agreement with recent experiments. The ferromagnetism and bandgap narrowing are identied to be due principally to the strong p-d mixing of O and Cu. The above discussion clearly shows that even on theoretical grounds there are no clear agreements between works by different groups on the connection between dopant types and their distribution, defects, and ferromagnetism. Thus, the eld remains equally controversial on the theoretical front as on the experimental side. On the positive side, the controversy has brought the subject of defect ferromagnetism into the scientic focus. While the importance of defects can easily be acknowledged, resolution and clarity are needed on the precise nature of interaction and coupling between the dopants and defects, and their specicity in the context of particular materials systems.

6. Conclusions and Outlook


The research on diluted magnetic semiconducting oxide (DMSO) systems has witnessed various developments, involving twists and turns, starting with the initial excitement leading to intensive research efforts in many laboratories around the world, yielding interesting results but also raising new questions and concerns related to reproducibility issues, non-intrinsic effects, and the possible role of defects. This scenario has also attracted theoretical attention for the creation of new models of ferromagnetism. In contrast to the conventional structurally and chemically homogeneous single-phase ferromagnetic systems involving exchange between neighboring atoms with orbital overlaps, the DMS system intrinsically places a signicant burden on our ability to control dopant distribution, disorder and clustering effects. Unfortunately, the connection between ferromagnetism and the atomistic/electronic state of a material system under such a dilute doping scenario is too sensitive to allow easy comparisons between works of different groups using different growth techniques and parameter spaces. This is precisely the reason that there has been confusion about some issues that has not yet been fully resolved. The current view of the scientic community towards this eld is therefore one of cautious optimism. Based on the analyses of the series of reported observations and the pertinent discussions in the published literature one could summarize the current state of affairs as follows: 1. At low magnetic dopant concentrations (<23%) and in thin lm samples grown at low temperatures (<400 C) the reports of ferromagnetism broadly point to the desirable intrinsic DMSO scenario. At higher dopant concentrations and higher growth temperatures generally dopant clustering has been noted either within the matrix itself or at the surface, interface or extended defects. 2. Anomalous Hall-effect (AHE), optical magnetic circular dichroism (O-MCD) and electric eld modulation of

magnetism (electric eld effect studies), which are generally regarded as the cornerstones of intrinsic DMS FM have been reported in different systems. Although AHE and O-MCD have been reported in the same sample, AHE and eld effect have not been reported in the same sample. The conductivity requirement of an AHE measurement forbids use of thick layers in an electric eld effect conguration, and for thin layers the magnetic measurements become difcult (and sometimes questionable) due to weak signal strength. It may be useful to employ optical (Kerr) measurement of magnetization of thin layers in a eld effect conguration to accomplish such a composite measurement. Another option could be to explore DMSO systems wherein higher concentrations of dopants can be uniformly incorporated (e.g., as suggested for the Co:ZnO case) for eld effect and AHE measurements. In such cases the magnetization measurements may be possible even for thinner layers due to higher effective magnetic moment and the conductivity may be moderately lower but within the measurable regime due to dopant scattering. Also, although AHE can be observed even in the sample with a clustered state (superparamagnetism), which does not represent an intrinsic DMS, its magnitude is much smaller than that expected for the case of uniform dopant incorporation. Thus theoretical studies are needed to predict as to what is expected in terms of the magnitude of AHE for the two microstates (uniform or clustered doping) against which the experimental observations can be tested. Similarly theoretical inputs can be expected to be valuable for both the O-MCD and eld effect studies. Most theoretical studies thus far have focused on the occurrence of the phenomenon, but not on the connection between the state of the material and the specic observation probe outcomes. 3. There have been only a limited number of studies on p-type DMSO systems, possibly because of a limited number of p-type oxide systems available and the difculties encountered in converting natural n-type (shallow oxygen vacancy level) oxides such as ZnO to p-type conduction by doping. More work along these lines on p-type materials (including some ternary systems) may be interesting. Research on ntype ternary oxides is also lacking. 4. There have not been many studies of spin polarization in DMSO systems. Since this is the key property that will control spintronics applications and will also reect upon the spin worthiness of the material irrespective of its homogeneous or inhomogeneous status as a material system, the corresponding studies using synchrotron methods, scanning tunneling spectroscopy techniques as well as any other methods are highly desirable. Also, the studies of magnetic force microscopy, which reect cooperative magnetic effects, have been rather limited and scattered. The problem is the weak magnetic signal, which has also been the problem that has led to confusions regarding the intrinsic Vs extrinsic origin of magnetization. Once again it may be useful to perform these studies on systems wherein higher dopant concentration can be incorporated substitutionally and uniformly. 5. There is a need for more detailed atomistic characterizations on dopant distributions and their chemical states. More extensive use of techniques such as high resolution TEM combined with element specic sensitive detection in

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3151

www.advmat.de www.MaterialsViews.com

6.

7.

8.

9.

10.

scanning mode such as by electron energy loss spectroscopy (EELS) is highly desirable. Also, attention needs to be paid to the HRTEM and TEM observations on various length scales, because even in the cases of dopant clustered samples, local TEM can look deceptively good. Combined use of various synchrotron based techniques can also bring much greater clarity on the issues of importance to the eld.[179180] There have been hardly any sustained studies on femtosecond laser induced non-equilibrium spin population in DMSO materials and the related spin diffusion/relaxation effects. Such work has been followed in the case of GaMnAs type III-V DMS systems and has yielded interesting and denitive results. Work along these lines on DMSO systems will help bring new insights to the research debate since it directly addresses the issues of signicance to spintronics applications. DMSO researchers have over the years devoted considerable attention and energy addressing and solving materials problems, but this has also led to very little or limited research efforts on the device side involving DMSO as an active layer in a multilayer spin device system. Using materials and growth conditions that bear the signature of intrinsic DMS character much more research on devices is required to be carried out because this can throw light on the nature of the carrier spin system. Defect states introduced by growth methods and parameters, or those introduced for local charge balance due to valence differences between the dopant and the host have inuenced the analysis of magnetization results because they can either contribute to the magnetization themselves or through their role in the exchange mechanism. This calls for more detailed atomistic characterizations and modeling of the defects states. The defect-related considerations have led to the re-emergence of the eld of defect ferromagnetism. Although of great academic interest, it is not clear whether defect ferromagnetic states could be stabilized under device processing conditions and could actually be utilized. In contrast the case of interface ferromagnetism (which is a sort of defect ferromagnetism, the interface being regarded as the defect) is of great interest because the corresponding state of the material can be growth controlled and stabilized. Research on nanomaterials has revealed that in most cases the dilutely magnetic-impurity-doped oxide (or even semiconductor) nanoparticles themselves are paramagnetic, but their agglomerates or assembled and sintered lms exhibit ferromagnetism. Capping strategies for nanoparticle agglomerates/lms have been shown to enable an engineered and reversible ferromagnetic response. In a recent publication Gamelin and co-workers[181] have reported a very interesting result on light-induced spontaneous magnetization of Mn2+ spins in colloidal Mn-doped CdSe nanocrystals. Such effects cannot only facilitate the manipulation of spin effects for spintronics applications, but could also help to elucidate DMS mechanisms as well. Research along these lines on favorable DMSO systems could be very interesting. In a different context, C. N. R. Rao and co-workers[16] have suggested that the surface-conned universal ferromagnetism observed by them in most inorganic

nanoparticles[151152] could be used in many interesting ways. They have shown that the classic ferroelectric BaTiO3 can be rendered multiferroic by this approach.[16] It will be interesting to explore these possibilities further. 11. Although most of the results discussed in this review relate to cation doping and attendant ferromagnetic effects, considerable efforts are being carried out on anion doping in functional metal oxides such as TiO2 and ZnO in the context of photocatalysis and other optoelectronic effects.[182184] It would be interesting for the DMSO community to examine certain aspects of these studies and possibly exploit them in manipulating and controlling DMSO ferromagnetism. It is clear that the DMSO eld needs a fresh approach to solving the central issues of uniformity and reproducibility, with scientists being able to nd some common ground for broad agreements. Also, some denitive steps by leading groups towards the realization of spintronics device functions in DMSO materials would go a long way in putting this eld on rmer footing.

REVIEW

Supporting Information
Supporting Information is available online from Wiley InterScience or from the author.

Acknowledgements
It is a pleasure to thank Dr. Esther Levy and editors at Advanced Materials for their invitation to write this review. I thank them also for their patience and a couple of deadline extensions. It is a great pleasure to thank Professor T. Venkatesan and members of his former group at the University of Maryland, college park, especially Dr. Sanjay Shinde, Dr. Darshan Kundaliya, Dr. Sankar Dhar, Prof. M.S.R. Rao, Prof. Yongang Zhao, Dr. Ram Janay Chaudhary, Dr. Sangeeta Kale, Dr. Shixiong Zhang and others for creating a stimulating research environment where most of the DMSO work was performed. I would also like to thank Prof. J.M.D. Coey, Prof. M. Kawasaki, Dr. Fukumura, Prof. Daniel Gamelin and Prof. N. H. Hong for allowing the use of some of the gures in their published papers and the publishers of the corresponding papers for granting permissions for the same. Thanks are also due to Prof. Richard Greene, Prof. Dennis Drew, Prof. R. Ramesh and their students for fruitful collaborations and to Prof. S. Das Sarma and Prof. A. J. Millis for providing theoretical insights. The author would like to acknowledge the NSF MRSEC program (project Director Prof. Ellen Williams) and DARPA (project Directors Prof. T. Venkatesan, Prof. R. Ramesh) at UMD for support to the work performed at Maryland and the BRNS CRP spintronics program, CSIR and DST, Govt. of India for support to the work done at NCL, Pune, India. The author is thankful to Dr. S. Sivaram, Director, National Chemical Laboratory, Prof. K. N. Ganesh, Director, IISER, Pune, and Dr. S. Pal, Head Physical Chemistry, NCL, for their support. Finally, it is a pleasure to thank the members of my present research group at NCL, Pune, especially Vivek Dhas, Subas Muduli, Tushar Jagadale, Shruti Agarkar, Onkar Game, Meenal Deo, and Vishal Thakare for their help in preparing this article. Finally, I want to thank all the reviewers for their constructive and helpful suggestions. Received: November 14, 2009 Revised: March 6, 2010 Published online: June 9, 2010

3152

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

[1] S. A Wolf, D. D Awschalom, R. A Buhrman, J. M. Daughton, S. V. Molnar, M. L. Roukes, A. Y. Chtchelkanova, D.M. Treger, Science 2001, 294, 1488. [2] S. D. Sarma, American Scientist 2001, 89, 516. [3] I. Zutic, J. Fabian, S. D. Sarma, Reviews of Modern Physics 2004, 76, 323. [4] D. D. Awschalom, M. E. Flatt, Nat. Phys. 2007, 3, 153. [5] C. Felser, G. H. Fecher, B. Balke, Angew Chemie 2007, 46, 668. [6] J. K. Furdyna, J. Appl. Phys. 1988, 64, R29. [7] H. Ohno, Science 1998, 281, 951. [8] T. Dietl, H. Ohno, F. Matsukura, J. Cibert, D. Ferrand, Science 2000, 287, 1020. [9] T. Dietl, Semocond. Sci. Technol. 2002, 17, 377. [10] T. Dietl, Nat. Mater. 2003, 2, 646. [11] W. Prellier, A. Fouchet, B. Mercey, J. Phys: Condens. Matter 2003, 15, R1583. [12] S. J. Pearton, W. H. Heo, M. Ivill, D. P. Norton, T. Steiner, Semicond. Sci. Technol 2004, 19, R59. [13] J. D. Bryan, D. R. Gamelin, Prog. Inorg. Chem. 2005, 54, 47. [14] A. H. MacDonald, P. Schiffer, N. Samarth, Nat. Mater. 2005, 4, 195. [15] S. Kuroda, N. Nishizawa, K. Takita, M. Mitome, Y. Bando, K. Osuch, T. Dietl, Nat. Mater. 2007, 6, 440. [16] A. Sundaresan, C. N. R. Rao, Nano Today 2009, 4, 96. [17] F. Pan, C. Song, X. J. Liu, Y. C. Yang, F. Zeng, Materials Science and Engineering 2008, R 62, 1. [18] K. Potzger, S. Zhou, Phys. Status Solidi B 2009, 246, 1147. [19] R. Janisch, P. Gopal, N. A Spaldin, J. Phys.: Condens. Matter 2005, 17, R657. [20] Thin Films and Heterostructures for oxide electronics (Ed: S. B. Ogale), Springer Verlag, Berlin, 2005. [21] Y. Matsumoto, Science, 2001, 291, 854. [22] T. Fukumura, Z. Jin, A. Ohtomo, H. Koinuma, M. Kawasaki, Appl. Phys. Lett. 1999, 75, 3366. [23] J. Q. Xiao, J. S. Jiang, C. L. Chien, Phy. Rev. Lett. 1992, 68, 3749. [24] S. Mitani, H. Fujimori, S. Ohnuma, J. Magn. Magn. Mater. 1997, 165, 141. [25] T. Fukumura, Y. Yamada, K. Tamura, K. Nakajima, T. Aoyama, A. Tsukazaki, M. Sumiya, S. Fuke, Y. Segawa, T. Chikyow, T. Hasegawa, H. Koinuma, M. Kawasaki, Jpn. J. Appl. Phys 2003, 42, L105. [26] M. Murakami, Y. Matsumoto, M. Nagano, T. Hasegawa, M. Kawasaki, H. Koinuma, Appl. Surf. Sci. 2004, 223, 245. [27] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T. Hasegawa, H. Koinuma, M. Kawasaki, Nat Mater 2004, 3, 221. [28] Y. Yamada, H. Toyosaki, A. Tsukazaki, T. Fukumura, K. Tamura, Y. Segawa, K. Nakajima, T. Aoyama, T. Chikyow, T. Hasegawa, H. Koinuma, M. Kawasaki, J. Appl. Phys. 2004, 96, 5097. [29] H. Toyosaki, T. Fukumura, Y. Yamada, M. Kawasaki, Appl. Phys. Lett. 2005, 86, 182503. [30] J. W. Quilty, A. Shibata, J.-Y. Son, K. Takubo, T. Mizokawa, H. Toyosaki, T. Fukumura, M. Kawasaki, Phys. Rev. Lett. 2006, 96, 027202. [31] H. Toyosaki, T. Fukumura, K. Ueno, M. Nakano, M. Kawasaki, J. Appl. Phys. 2006, 99, 08M102. [32] K. Mamiya and T. Koide A. Fujimori H. Tokano H. Manaka A. Tanaka H. Toyosaki, T. Fukumura, M. Kawasaki, Appl. Phys. Lett. 2006, 89, 062506. [33] K. Ueno, T. Fukumura, H. Toyosaki, and M. Nakano M. Kawasaki, Appl. Phys. Lett. 2007, 90, 072103. [34] K. Ueno, T. Fukumura, H. Toyosaki, M. Nakano, T. Yamasaki, Y. Yamada, M. Kawasaki, J. Appl. Phys. 2008, 103, 07D114. [35] T. Matsumura, D. Okuyama, S. Niioka, H. Ishida, T. Satoh, Y. Murakami, H. Toyoski, Y. Yamada, T. Fukumura, M. Kawasaki, Phys. Rev. B. 2007, 76, 115320. [36] T. Fukumura, H. Toyosaki, K. Ueno, M. Nakano, M. Kawasaki, New J. of Phys 2008, 10, 055018.

[37] T. Yamasaki, T. Fukumura, Y. Yamada, M. Nakano, K. Ueno, T. Makino, M. Kawasaki, Appl. Phys. Lett. 2009, 94, 102515. [38] S. A. Chambers, S. Thevuthasan R. F. C. Farrow, R. F. Marks, J. U. Thiele, L. Folks, M. G. Samant, A. J. Kellock, N. Ruzycki, D. L. Ederer, U. Diebold, Appl. Phys. Lett. 2001, 79, 3467. [39] S. A. Chambers, C. M. Wang, S. Thevuthasan, T. Droubay, D. E. McCready, A. S. Lea, V. Shutthanandan, C. F. Windisch Jr, Thin Solid Films 2002, 418, 197. [40] S. A. Chambers, T. Droubay, C. M. Wang, and A. S. Lea R. F. C. Farrow, L. Folks, V. Deline, S. Anders, Appl. Phys. Lett. 2003, 82, 1257. [41] S. A. Chambers, S. M. Heald, T. Droubay, Phys. Rev. B 2003, 67, 100401. [42] Y. J. Kim, S. Thevuthasan, T. Droubay, A. S. Lea, C. M. Wang, V. Shutthanandan, S. A. Chambers, R. P. Sears, B. Taylor, B. Sinkovic, Appl. Phys. Lett. 2004, 84, 3531. [43] V. Shutthanandan, S. Thevuthasan, S. M. Heald, T. Droubay, M. H. Engelhard, T. C. Kaspar, D. E. McCready, L. Saraf, S. A. Chambers, B. S. Mun, N. Hamdan, P. Nachimuthu, B. Taylor, R. P. Sears, B. Sinkovic, Appl. Phys. Lett. 2004, 84, 4466. [44] T. Droubay, S. M. Heald, V. Shutthanandan, S. Thevuthasan, S. A. Chambers, J. Osterwalder, J. Appl. Phys. 2005, 97, 046103. [45] J. M. D. Coey, A. P. Douvalis, C. B. Fitzgerald, M. Venkatesan, Appl. Phys. Lett. 2004, 84, 1332. [46] S. D. Sarma, E. H. Hwang, A. Kaminski, Phys. Rev. B 2003, 67, 155201. [47] J. Osterwalder, T. Droubay, T. Kaspar, J. Williams, C. M. Wang, S. A. Chambers, Thin Solid Films 2005, 484, 289. [48] T. C. Kaspar, S. M. Heald, C. M. Wang, J. D. Bryan, T. Droubay, V. Shutthanandan, S. Thevuthasan, D. E. McCready, A. J. Kellock, D. R. Gamelin, S. A. Chambers, Phys. Rev. Lett. 2005, 95, 217203. [49] P. V. Radovanovic, D. R. Gamelin, Phys. Rev. Lett. 2003, 91, 157202. [50] J. D. Bryan, S. M. Heald, S. A. Chambers, D. R. Gamelin, J. Am. Chem. Soc. 2004, 126, 11640. [51] J. D. Bryan, S. A. Santangelo, S. C. Keveren, D. R. Gamelin, J. Am. Chem. Soc. 2005, 127, 15568. [52] T. C. Kaspar, T. Droubay, V. Shutthanandan, S. M. Heald, C. M. Wang, D. E. McCready, S. Thevuthasan, J. D. Bryan, D. R. Gamelin, A. J. Kellock, M. F. Toney, X. Hong, C. H. Ahn, S. A. Chambers, Phy. Rev. B 2006, 73, 155327. [53] S. R. Shinde, S. B. Ogale, S. D. Sarma, J. R. Simpson, H. D. Drew, S. E. Loand, C. Lanci, J. P. Buban, N. D. Browning, V. N. Kulkarni, J. Higgins, R. P. Sharma, R. L. Greene, T. Venkatesan, Phys. Rev. B 2003, 67, 115211. [54] S. Guha, K. Ghosh, J. G. Keeth, S. B. Ogale, S. R. Shinde, J. R. Simpson, H. D. Drew, T. Venkatesan, Appl. Phys. Lett. 2003, 83, 3296. [55] S. Guha, K. Ghosh, J. G. Keeth, S. B. Ogale, S. R. Shinde, J. R. Simpson, H. D. Drew, T. Venkatesan, Appl. Phys. Lett. 2004, 84, 1613. [56] V. N. Kulkarni, S. R. Shinde, Y. G. Zhao, R. J. Choudhary, S. B. Ogale, R. L. Greene, T. Venkatesan, Nucl. Instrum. and Meth. for Phys. Res. B. 2004, 21920, 902. [57] J. S. Higgins, S. R. Shinde, S. B. Ogale, T. Venkatesan, R. L. Greene, Phys. Rev. B 2004, 69, 073201. [58] S. R. Shinde, S. B. Ogale, J. S. Higgins, H. Zheng, A. J. Millis, V. N. Kulkarni, R. Ramesh, R. L. Greene, T. Venkatesan, Phys. Rev. Lett.. 2004, 92, 166601. [59] B. A. Aronzon, A. B. Granovski, D. Yu. Kovalev, E. Z. Meilikhov, V. V. Rylkov, M. V. Sedova, J. Exp. Theor. Phys. 2000, 71, 687. [60] B. Raquet, M. Goiran, N. Ngre, J. Lotin, B. Aronzon, V. Rylkov, E. Meilikhov, Phy. Rev. B 2000, 62, 17144. [61] E. Shimshoni, A. Auerbach, Phy. Rev. B 1997, 55, 9817.

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3153

www.advmat.de www.MaterialsViews.com

REVIEW

[62] J. R. Simpson, H. D. Drew, S. R. Shinde, R. J. Choudhary, S. B. Ogale, T. Venkatesan, Phys. Rev. B 2004, 69, 193205. [63] T. Zhao, S. R. Shinde, S. B. Ogale, H. Zheng, T. Venkatesan, R. Ramesh, S. Das Sarma, Phys. Rev. Lett. 2005, 94, 126601. [64] N. Chandrasekhar, O. T. Valls, A. M. Goldman, Phy. Rev. Lett. 1993, 71, 1079. [65] S. X. Zhang, S. B. Ogale, L. F. Fu, S. Dhar, D. C. Kundaliya, W. Ramadan, N. D. Browning, T. Venkatesan, Appl. Phys. Lett. 2006, 88, 012513. [66] L. F. Fu, N. D. Browning, S. X. Zhang, S. B. Ogale, D. C. Kundaliya, T. Venkatesan, J. Appl. Phys. 2006, 100, 123910. [67] S. B. Ogale, D. Kundaliya, S. Mehraeen, L.-f. Fu, S. Zhang, A. Lussier, J. Dvorak, N. Browning, Y. Idzerda, T. Venkatesan, Chem. Mater. 2008, 20, 1344. [68] K. A. Bogle, M. N. Bachhav, M. S. Deo, N. Valanoor, S. B. Ogale, Appl. Phy. Lett. 2009, 1, 95. [69] J.-Y. Kim, J.-H. Park, B.-G. Park, H.-J. Noh, S.-J. Oh, J. S. Yang, D.-H. Kim, S. D. Bu, T.-W. Noh, H.-J. Lin, H.-H. Hsieh, C. T. Chen, Phy. Rev. Lett. 2003, 90, 0174011. [70] D. H. Kim, J. S. Yang, K. W. Lee, S. D. Bu, T. W. Noh, S.-J. Oh, Y.-W. Kim, J.-S. Chung, H. Tanaka, H. Y. Lee, T. Kawai, Appl. Phys. Lett. 2002, 81, 2421. [71] D. H. Kim, J. S. Yang, Y. S. Kim, T. W. Noh, S. D. Bu, S.-I. Baik, Y.-W. Kim, Y. D. Park, S. J. Pearton, J.-Y. Kim, J.-H. Park, H.-J. Lin, C. T. Chen, Phy. Rev. B 2005, 71, 014440. [72] D. H. Kim, J. S. Yang, Y. S. Kim, D.-W. Kim, T. W. Noh, S. D. Bu, Y.-W. Kim, Y. D. Park, S. J. Pearton, Y. Jo, J.-G. Park, Appl. Phys. Lett. 2003, 83, 22. [73] S. H. Kang, H. Nguyen, T. Quynh, S. G. Yoon, E. T. Kim, Z. Lee, App. Phys. Lett. 2007, 90, 102504. [74] K. A. Grifn, A. B. Pakhomov, C. M. Wang, S. M. Heald, K. M. Krishnan, Phys. Rev. Lett. 2005, 94, 157204. [75] N. H. Hong, J. Sakai, W. Prellier, A. Hassini, A. Ruyter, F. Gervais, Phys. Rev. B. 2004, 70, 195204. [76] K. Ueda, H. Tabata, T. Kawai, Appl. Phys. Lett. 2001, 79, 988. [77] A. Ney, K. Ollefs, S. Ye, T. Kammermeier, V. Ney, T. C. Kaspar, S. A. Chambers, F. Wilhelm, A. Rogalev, Phys. Rev. Lett. 2008, 100, 157201. [78] T. C. Kaspar, T. Droubay, S. M. Heald, M. H. Engelhard, P. Nachimuthu, S. A. Chambers, Phys. Rev. B 2008, 77, 201303. [79] P. V. Radovanovic, N. S. Norberg, K. E. McNally, D. R. Gamelin, J. Am. Chem. Soc. 2002, 124, 15192. [80] D. A. Schwartz, N. S. Norberg, Q. P. Nguyen, J. M. Parker, D. R. Gamelin, J. Am. Chem. Soc. 2003, 125, 13205. [81] D. A. Schwartz, D. R. Gamelin, Adv. Mater. 2004, 16, 2115. [82] K. R. Kittilstved, N. S. Norberg, D. R. Gamelin, Phys. Rev. Lett. 2005, 94, 147209. [83] W. K. Liu, G. M. Salley, D. R. Gamelin, J. Phys. Chem. B 2005, 109, 14486. [84] W. K. Liu, K. M. Whitaker, K. R. Kittilstved, D. R. Gamelin, J. Am. Chem. Soc. 2006, 128, 3910. [85] M. A. White, S. T. Ochsenbein, D. R. Gamelin, Chem. Mater. 2008, 20, 7107. [86] C. D. Pemmaraju, S. Sanvito, Phy. Rev. Lett. 2005, 94, 217205. [87] S. Lany, H. Raebiger, A. Zunger, Phy. Rev. B 2008, 77, 241201. [88] A. Walsh, J. L. F. DaSilva, S-H. Wei, Phy. Rev. Lett 2008, 100, 256401. [89] C. N. R. Rao, F. L. Deepak, J. Mater. Chem. 2005, 15, 573. [90] M. Venkatesan, C. B. Fitzgerald, J. G. Lunney, J. M. D. Coey, Phy. Rev. Lett. 2004, 93, 177206. [91] C. B. Fitzgeral, M. Venkatesan, J. G. Lunney, L. S. Dorneles, J. M. D. Coey, Appl. Sur. Sci, 2005, 247, 493. [92] S. B. Ogale, R. J. Choudhary, J. P. Buban, S. E. Loand, S. R. Shinde, S. N. Kale, V. N. Kulkarni, J. Higgins, C. Lanci, J. R. Simpson, N. D. Browning, S. D. Sarma, H. D. Drew, R. L. Greene, T. Venkatesan, Phys. Rev. Lett. 2003, 91, 077205.

[93] S. Krishnamurthy, C. McGuinness, L. S. Dorneles, M. Venkatesan, J. M. D. Coey, J. G. Lunney, C. H. Pattersonb, K. E. Smith, T. Learmonth, P.-A. Glans, T. Schmittc, J.-H. Guo, J. Appl. Phys. 2006, 99, 08M111. [94] M. Venkatesan, P. Stamenov, L. S. Dorneles, R. D. Gunning, B. Bernoux, J. M. D. Coey, Appl. Phys. Lett. 2007, 90, 242508. [95] T. C. Kaspar, T. Droubay, S. M. Heald, P. Nachimuthu, C. M. Wang, V. Shutthanandan, C. A. Johnson, D. R. Gamelin, S. A. Chambers, New J. of Phys 2008, 10, 055010. [96] K. R. Kittilstved, D. A. Schwartz, A. C. Tuan, S. M. Heald, S. A. Chambers, D. R. Gamelin, Phy. Rev. Lett. 2006, 97, 037203. [97] N. Khare, M. J. Kappers, M. Wei, M. G. Blamire, J. L. MacManusDriscoll, Adv. Mater. 2006, 18, 1449. [98] O. D. Jayakumar, I. K. Gopalakrishnan, S. K. Kulshreshtha, Adv. Mater. 2006, 18, 1857. [99] N. S. Norberg, K. R. Kittilstved, J. E. Amonette, R. K. Kukkadapu, D. A. Schwartz, D. R. Gamelin, J. Am. Chem. Soc. 2004, 126, 9387. [100] K. R. Kittilstved and D. R. Gamelin, J. Am. Chem. Soc. 2005, 127, 5292 [101] T. C. Droubay, D. J. Keavney, T. C. Kaspar, S. M. Heald, C. M. Wang, C. A. Johnson, K. M. Whitaker D. R. Gamelin, S. A. Chambers, Phy. Rev. B 2009, 79, 155203. [102] P. Sharma, A. Gupta, K.V. Rao, F. J. Owens, R. Sharma, R. Ahuja, J. M. O. Guillen, B. Johansson, G. A. Gehring, Nat. Mater. 2003, 2, 673. [103] D. C. Kundaliya, S. B. Ogale, S. E. Loand, S. Dhar, C. J. Metting, S. R. Shinde, Nat. Mater 2004, 3, 709. [104] M. A. Garca, M. L. Ruiz-Gonzalez, A. Quesada, J. L. Costa-Kramer, J. F. Fernandez, S. J. Khatib, A. Wennberg, A. C. Caballero, M. S. Martn-Gonzalez, M. Villegas, F. Briones, J. M. Gonzalez-Calbet, A. Hernando, Phy. Rev. Lett 2005, 94, 217206. [105] J. P. Bucher, D. C. Douglass, L. A. Bloomeld, Phy. Rev. Lett. 1991, 66, 3052. [106] H. J. Fan, C. Liu, M. Liao, Chem. Phys. Lett. 1997, 273, 353. [107] D. B. Buchholz, R. P. H. Chang, J. H. Song, J. B. Ketterson, Appl. Phys. Lett. 2005, 87, 082504. [108] D. Chakraborti, J. Narayan, J. T. Prater, Appl. Phys. Lett 2007, 90, 062504. [109] D. L. Hou, X. J. Ye, H. J. Meng, H. J. Zhou, X. L. Li, C. M. Zhen, G. D. Tang, Appl. Phys. Lett 2007, 90, 142502. [110] A. Tiwari, M. Snure, D. Kumar, J. T. Abiade, Appl. Phys. Lett 2008, 92, 062509. [111] Q. Xu, H. Schmidt, S. Zhou, K. Potzger, M. Helm, H. Hochmuth, M. Lorenz, A. Setzer, P. Esquinazi, C. Meinecke, M. Grundmann, Appl. Phys. Lett. 2008, 92, 082508. [112] D. J. Keavney, D. B. Buchholz, Q. Ma, R. P. H. Chang, Appl. Phys. Lett. 2007, 91, 012501. [113] C. Sudakar, J. S. Thakur, G. Lawes, R. Naik, V. M. Naik, Phy. Rev. B 2007, 75, 054423. [114] Z. F. Wu, X. M. Wu, L. J. Zhuge, X. M. Chen, X. F. Wang, Appl. Phys. Lett. 2008, 93, 023103. [115] Q. Ma, D. B. Buchholz, R. P. H. Chang, Phy. Rev. B 2008, 78, 214429. [116] F-Y. Ran, M. Tanemura, Y. Hayashia, T. Hihara, Journal of Crystal Growth 2009, 311, 4270. [117] F. J. Owens, J. Magn. Magn. Mater. 2009, 321, 3734. [118] B. B. Straumal, A. A. Mazilkin, S. G. Protasova, A. A. Myatiev, P. B. Straumal, G. Schtz, P. A. van Aken, E. Goering, B. Baretzky, Phys. Rev. B. 2009, 79, 205206. [119] S. N. Kale, S. B. Ogale, S. R. Shinde, R. L. Greene, T. Venkatesan, Appl. Phys. Lett. 2003, 82, 2100. [120] H. Raebiger, S. Lany, A. Zunger, Phy. Rev. Lett. 2007, 99, 167203. [121] Y. G. Zhao, S. R. Shinde, S. B. Ogale, J. Higgins, R. J. Choudhary, V. N. Kulkarni, R. L. Greene, T. Venkatesan, S. Loand, C. Lanci,

3154

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2010, 22, 31253155

www.advmat.de www.MaterialsViews.com

REVIEW

[122] [123]

[124]

[125]

[126]

[127] [128] [129] [130] [131] [132] [133] [134] [135]

[136] [137] [138] [139] [140] [141] [142] [143] [144] [145] [146]

[147] [148]

J. Buban, N. Browning, S. D. Sarma, A. J. Millis, Appl. Phys. Lett. 2003, 83, 2199. A. Kaminski, S. D. Sarma, Phy. Rev. Lett. 2002, 88, 2472021. P. T. Qiao, Z. H. Zhao, Y. G. Zhao, S. P. Zhang, W. Y. Zhang, S. B. Ogale, S. R. Shinde, T. Venkatesan, S. Loand, C. Lanci, Thin Solid Films 2004, 468, 8. S. X. Zhang, W. Yu, S. B. Ogale, S. R. Shinde, D. C. Kundaliya, W.-K. Tse, S. Y. Young, J. S. Higgins, L. G. Salamanca-Riba, M. Herrera, L. F. Fu, N. D. Browning, R. L. Greene, T. Venkatesan, Phys. Rev. B 2007, 76, 085323. G. Herranz, R. Ranchal, M. Bibes, H. Jaffre, E. Jacquet, J.-L. Maurice, K. Bouzehouane, F. Wyczisk, E. Tafra, M. Basletic, A. Hamzic, C. Colliex, J.-P. Contour, A. Barthelemy, A. Fert, Phy. Rev. Lett. 2006, 96, 027207. S. X. Zhang, S. B. Ogale, D. C. Kundaliya, L. F. Fu, N. D. Browning, S. Dhar, W. Ramadan, J. S. Higgins, R. L. Greene, T. Venkatesan, Appl. Phys. Lett. 2006, 89, 012501. P. Gambardella, S. S. Dhesi, S. Gardonio, C. Grazioli, P. Ohresser, C. Carbone, Phy. Rev. Lett. 2002, 88, 047202. G. Bergmann, M. Hossain, Phy. Rev. Lett. 2001, 86, 2138. S. K. Kwon, B. I. Min, Phy. Rev. Lett. 2000, 84, 3970. H. Beckmann, G. Bergmann, Phy. Rev. Lett. 1999, 83, 2417. G. Bergmann, Phy. Rev. B 1981, 23, 3805. J. Flouquet, O. Taurian, J. Sanchez, M. Chapellier, J. L. Tholence, Phy. Rev. Lett. 1977, 38, 81. R. N. Bozorth, P. A. Wolff, D. D. Davis, V. B. Compton, J. H. Wernick, Phy. Rev. 1961, 122, 1157. C. B. Fitzgerald, M. Venkatesan, A. P. Douvalis, S. Huber, J. M. D. Coey, T. Bakas, J. Appl. Phys. 2004, 95, 7390. C. B. Fitzgerald, M. Venkatesan, L. S. Dorneles, R. Gunning, P. Stamenov, J. M. D. Coey, P. A. Stampe, R. J. Kennedy, E. C. Moreira, U. S. Sias, Phys. Rev. B 2006, 74, 115307. N. H. Hong, J. Sakai, N. T. Huong, V. Briz, Appl. Phys. Lett. 2005, 87, 102505. P. I. Archer, P. V. Radovanovic, S. M. Heald, D. R. Gamelin, J. Am. Chem. Soc. 2005, 127, 14479. N. H. Hong, J. Sakai, N. T. Huong, N. Poirot, A. Ruyter, Phys. Rev. B. 2005, 72, 045336. I. S. Elmov, S. Yunoki, G. A. Sawatzky, Phys. Rev. Lett. 2002, 89, 216403. J. Osorio- Guilln, S. Lany, S. Barabash, A. Zunger, Phy. Rev. Lett. 2006, 96, 107203. M. Venkatesan, C. B. Fitzgerald, J. M. D. Coey, Nature 2004, 430, 630. D. W. Abraham, M. M. Frank, S. Guha, Appl. Phys. Lett. 2005, 87, 252502. J. M. D. Coey, M. Venkatesan, P. Stamenov, C. B. Fitzgerald, L. S. Dorneles, Phy. Rev. B 2005, 72, 024450. R. Salzer, D. Spemann, P. Esquinazi, R. Hohne, A. Setzer, K. Schindler, H. Schmidt, T. Butz, J. Magn. Magn. Mater. 2007, 317, 53. Ney, T. Kammermeier, V. Ney, K. Ollefs, S. Ye, J. Magn. Magn. Mater. 2008, 320, 3341. M. S. R. Rao, D. C. Kundaliya, S. B. Ogale, L. F. Fu, S. J. Welz, N. D. Browning, V. Zaitsev, B. Varughese, C. A. Cardoso, A. Curtin, S. Dhar, S. R. Shinde, T. Venkatesan, S. E. Loand, S. A. Schwartz, Appl. Phys. Lett. 2006, 88, 142505. E. Tirosh, G. Markovich, Adv. Mater. 2007, 19, 2608. S. Zhang, S. B. Ogale, W. Yu, X. Gao, T. Liu, S. Ghosh, G. P. Das, A. T. S. Wee, R. L. Greene, T. Venkatesan, Adv. Mater. 2009, 21, 2282.

[149] H. Pan, J. B. Yi, L. Shen, R. Q. Wu, J. H. Yang, J. Y. Lin, Y. P. Feng, J. Ding, L. H. Van, J. H. Yin, Phys. Rev. Lett. 2007, 99, 127201. [150] N. H. Hong, J. Sakai, N. Poirot, V. Briz, Phys. Rev. B. 2006, 73, 132404. [151] N. H. Hong, N. Poirot, J. Sakai, Phys. Rev. B. 2008, 77, 033205. [152] A. Sundaresan, R. Bhargavi, N. Rangarajan, U. Siddesh, C. N. R. Rao, Phys. Rev. B 2006, 74, 161306. [153] C. Madhu, A. Sundaresan, C. N. R. Rao, Phys. Rev. B 2008, 77, 201306. [154] A. Brinkman, M. Huijben, M. V. Zalk, J. Huijben, U. Zeitler, J. C. Maan, W. G. Van Der Wiel, G. Rijnders, D. H. A. Blank, H. Hilgenkamp, Nat. Mater. 2007, 6, 493. [155] J. M. Sullivan, S. C. Erwin, Phy. Rev. B 2003, 67, 144415. [156] W. T. Geng, K. S. Kim, Phys. Rev. B 2003, 68, 125203. [157] R. Janisch, N. A. Spaldin, Phys. Rev. B 2006, 73, 035201. [158] H. Weng, X. Yang, J. Dong, H. Mizuseki, M. Kawasaki, Y. Kawazoe, Phys. Rev. B 2004, 69, 125219. [159] H. Weng, J. Dong, T. Fukumura, M. Kawasaki, Y. Kawazoe, Phy. Rev. B 2006, 73, 121201. [160] P. Murugan, R. V. Belosludov, H. Mizuseki, T. Nishimatsu, T. Fukumura, M. Kawasaki, Y. Kawazoe, J. Appl. Phys. 2006, 99, 08M105. [161] E. Badaeva, C. M. Isborn, Y. Feng, S. T. Ochsenbein, D. R. Gamelin, X. Li, J. Phys. Chem. C, 2009, 113, 8710. [162] N. A. Spaldin, Phys. Rev. B 2004, 69, 125201. [163] T. Dietl, Phys. Rev. B 2008, 77, 085208. [164] A. Zunger, Appl. Phy. Lett. 2003, 83, 57. [165] D. J. Priour, Jr, E. H. Hwang, S. D. Sarma, Phy. Rev. Lett. 2005, 95, 037201. [166] G. Alvarez, M. Mayr, E. Dagotto, Phy. Rev. Lett. 2002, 89, 277202. [167] S. K. Nayak, M. Ogura, A. Hucht, H. Akai, P. Entel, J. Phys.: Condens. Matter 2009, 21, 064238. [168] B. Sanyal, R. Knut, O. Grns, D. M. Iusan, O. Karis, O. Eriksson, J. Appl. Phys. 2008, 103, 07D131. [169] J. Osorio-Guilln, S. Lany, S. V. Barabash, A. Zunger, Phy. Rev. B 2007, 75, 184421. [170] J. M. D. Coey, M. Venkatesan, C. B. Fitzgerald, Nat. Mater. 2005, 4, 173. [171] J. M. D. Coey, Solid State Sci. 2005, 7, 660. [172] J. M. D. Coey, Curr. Opin. Solid State Mater. Sci. 2006, 10, 83. [173] J. M. D. Coey, K. Wongsaprom, J. Alaria, M. Venkatesan, J. Phys. D: Appl. Phys. 2008, 41, 134012. [174] Lin-Hui Ye, A. J. Freeman, B. Delley, Phy. Rev. B 2006, 73, 033203. [175] L. M. Huang, A. L. Rosa, R. Ahuja, Phy. Rev. B 2006, 74, 075206. [176] D. Huang, Y. Zhao, D. Chen, Y. Shao, Appl. Phys. Lett. 2008, 92, 182509. [177] N. N. Lathiotakis, A. N. Andriotis, M. Menon, Phy. Rev. B 2008, 78, 193311. [178] M. Ferhat, A. Zaoui, R. Ahuja, Appl. Phys. Lett 2009, 94, 142502. [179] A. Ney, M. Opel, T. C. Kaspar, V. Ney, S. Ye, K. Ollefs, T. Kammermeier, S. Bauer, K.-W. Nielsen, S. T. B. Goennenwein, M. H. Engelhard, S. Zhou, K. Potzger, J. Simon, W. Mader, New J. Phys. 2010, 12, 013020. [180] A. Ney, T. Kammermeier, K. Ollefs, S. Ye, V. Ney, T. C. Kaspar, S. A. Chambers, F. Wilhelm and A. Rogalev, Phys. Rev. B 2010, 81, 054420. [181] R. Beaulac, L. Schneider, P. I. Archer, G. Bacher, D. R. Gamelin, Science 2010, 325, 973. [182] X. Qiu, C. Burda, Chem. Phys. 2007, 339, 1. [183] T. L. Thompson, J. T. Yates Jr, Chem. Rev. 2006, 106, 4428. [184] X. Chen, S. S. Mao, Chem. Rev. 2007, 107, 2891.

Adv. Mater. 2010, 22, 31253155

2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

3155

Вам также может понравиться