Refine
Has Fulltext
- yes (7)
Is part of the Bibliography
- yes (7)
Document Type
- Doctoral Thesis (7)
Keywords
- Spintronics (7) (remove)
Institute
In pursuit of a novel generation of devices, exploration of spin properties of the particles is needed. Spintronics is a modern field in physics which exploits spin properties to be used in addition to the charge degree of freedom. Since the conductivity mismatch problem presents a fundamental obstacle for electrical spin injection from a ferromagnetic metal into a diffusive semiconductor [SFM+00], other means for injecting spin-polarized carriers must be used. With a tunnel contact, it is possible to achieve a highly spin-polarized room-temperature tunnel injection [JWS+05]. We used a novel approach and applied magnetic RTDs for spin manipulation. In this work, properties of all-II-VI magnetic resonant tunneling diodes (RTDs), as applied to spintronics, were reported. Growth conditions were optimized to increase the peak-to-valley ratio, and the design of the RTDs was optimized for observation of spin related transport effects. When an external magnetic field was applied, spin manipulation became possible. Selforganized CdSe quantum structures were grown and investigated using optical means. After embedding them into a (Zn,Be)Se tunneling barrier, the properties were assessed by the resonant tunneling.
Diese Arbeit enthält Untersuchungen von Magnetowiderstandseffekten in (Ga,Mn)As basierten ferromagnetischen Halbleiterdioden. Die Resultate wurden in den folgenden Artikeln veröffentlicht: [1] C. Rüster, T. Borzenko, C. Gould, G. Schmidt, L.W. Molenkamp, X. Liu, T.J.Wojtowicz, J.K. Furdyna, Z.G. Yu and M. Flatt´e, Very Large Magnetoresistance in Lateral Ferromagnetic (Ga,Mn)As Wires with Nanoconstrictions, Physical Review Letters 91, 216602 (2003). [2] C. Gould, C. Rüster, T. Jungwirth, E. Girgis, G.M. Schott, R. Giraud, K. Brunner, G. Schmidt and L.W. Molenkamp, Tunneling Anisotropic Magnetoresistance: A Spin-Valve-Like Tunnel Magnetoresistance Using a Single Magnetic Layer, Physical Review Letters 93, 117203 (2004). [3] C. Rüster, C. Gould, T. Jungwirth, J. Sinova, G.M. Schott, R. Giraud, K. Brunner, G. Schmidt and L.W. Molenkamp, Very Large Tunneling Anisotropic Magnetoresistance of a (Ga,Mn)As/GaAs/(Ga,Mn)As Stack, Physical Review Letters 94, 027203 (2005). [4] C. Rüster and C. Gould, T. Jungwirth, E. Girgis, G.M. Schott, R. Giraud, K. Brunner, G. Schmidt and L.W. Molenkamp, Tunneling anisotropic magnetoresistance: Creating a spin-valve-like signal using a single ferromagnetic semiconductor layer, Journal of Applied Physics 97, 10C506 (2005).
Since the late 20th century, spintroncis has become a very active field of research [ŽFS04]. The prospect of spin based information technology, featuring strongly decreased energy consumption and possibly quantum-computation capabilities, has fueled this interest. Standard materials, like bulk gallium arsenide (GaAs), have experienced new attention in this context by exhibiting extraordinarily long lifetimes for nonequilibrium spin information, which is an important requirement for efficient spin based information storage and transfer. Another important factor is the lengthscale over which spin information can be transported in a given material and the role of external influences. Both aspects have been studied experimentally with innovative optical methods since the late 1990s by the groups of D. D. AWSHALOM and S. A. CROOKER et al. [KA99, CS05, CFL+05]. Although the pioneering experimental approaches presented by these authors led to a variety of insights into spin propagation, some questions were raised as well. Most prominently, the classical Einstein relation, which connects the mobility and diffusivity of a given particle species, seemed to be violated for electron spins in a bulk semiconductor. In essence, nonequilibrium spins appeared to move (diffuse) faster than the electrons that actually carry the spin. However, this contradiction was masked by the fact, that the material of interest was n-type GaAs with a doping concentration directly at the transition between metallic and insulating behavior (MIT). In this regime, the electron mobility is difficult to determine experimentally. Consequently, it was not a priori obvious that the spin diffusion rates determined by the newly introduced optical methods were in contradiction with established electrical transport data.
However, in an attempt to extend the available data of optical spin microscopy, another issue surfaced, concerning the mathematical drift-diffusion model that has been commonly used to evaluate lateral spin density measurements. Upon close investigation, this model appears to have a limited range of applicability, due to systematic discrepancies with the experimental data (chapter 4). These deviations are noticeable in original publications as well, and it is shown in the present work that they originate from the local heating of electrons in the process of optical spin pumping. Based on insights gained during the second half of the 20th century, it is recapitulated why conduction electrons are easily overheated at cryogenic temperatures. The main reason is the poor thermal coupling between electrons and the crystal lattice (chapter 3). Experiments in the present work showed that a significant thermal gradient exists in the conduction band under local optical excitation of electron–hole pairs. This information was used to develop a better mathematical model of spin diffusion, which allowed to derive the diffusivity of the undisturbed system, due to an effective consideration of electron overheating. In this way, spin diffusivities of n-GaAs were obtained as a function of temperature and doping density in the most interesting regime of the metal–insulator-transition.
The experiments presented in this work were performed on a series of n-type bulk GaAs samples, which comprised the transition between metallic conductivity and electrical insulation at low temperatures. Local electron temperature gradients were measured by a hyperspectral photoluminescence imaging technique with subsequent evaluation of the electron–acceptor (e,A$^0$) line shape. The local density of nonequilibrium conduction electron spins was deduced from scanning magneto-optic Kerr effect microscopy. Numerical evaluations were performed using the finite elements method in combination with a least-squares fitting procedure.
Chapter 1 provides an introduction to historical and recent research in the field of spintronics, as far as it is relevant for the understanding of the present work. Chapter 2 summarizes related physical concepts and experimental methods. Here, the main topics are semiconductor optics, relaxation of hot conduction electrons, and the dynamics of nonequilibrium electron spins in semiconductors.
Chapter 3 discusses optical heating effects due to local laser excitation of electron–hole pairs. Experimental evaluations of the acceptor-bound-exciton triplet lines led to the conclusion that the crystal lattice is usually not overheated even at high excitation densities. Here, the heat is efficiently dissipated to the bath, due to the good thermal conductivity of the lattice. Furthermore, the heating of the lattice is inherently limited by the weak heat transfer from the electron system, which on the other hand is also the reason why conduction electrons are easily overheated at temperatures below ≈ 30 K. Spatio-spectral imaging of the electron–acceptor-luminescence line shape allowed to trace the thermal gradient within the conduction band under focused laser excitation. A heat-diffusion model was formulated, which reproduces the experimental electron-temperature trend nicely for low-doped GaAs samples of n- and p-type. For high-doped n-type GaAs samples, it could be shown that the lateral electron-temperature profile is well approximated by a Gaussian. This facilitated easy integration of hot electron influence into the mathematical model of spin diffusion.
Chapter 4 deals with magneto-optical imaging of optically induced nonequilibrium conduction-electron spins in n-GaAs close to the MIT. First, the spectral dependence of the magneto-optic Kerr effect was examined in the vicinity of the fundamental band gap. Despite the marked differences among the investigated samples, the spectral shape of the Kerr rotation could be described in terms of a simple Lorentz-oscillator model in all cases. Based on this model, the linearity of the Kerr effect with respect to a nonequilibrium spin polarization is demonstrated, which is decisively important for further quantitative evaluations.
Furthermore, chapter 4 presents an experimental survey of spin relaxation in n-GaAs at the MIT. Here, the dependence of the spin relaxation time on bath temperature and doping density was deduced from Hanle-MOKE measurements. While all observed trends agree with established literature, the presented results extend the current portfolio by adding a coherent set of data.
Finally, diffusion of optically generated nonequilibrium conduction-electron spins was investigated by scanning MOKE microscopy. First, it is demonstrated that the standard diffusion model is inapplicable for data evaluation in certain situations. A systematic survey of the residual deviations between this model and the experimental data revealed that this situation unfortunately persisted in published works. Moreover, the temperature trend of the residual deviations suggests a close connection to the local overheating of conduction electrons. Consequently, a modified diffusion model was developed and evaluated, in order to compensate for the optical heating effect. From this model, much more reliable results were obtained, as compared to the standard diffusion model. Therefore, it was shown conclusively that the commonly reported anomalously large spin diffusivities were at least in parts caused by overheated conduction electrons.
In addition to these new insights some experimental and technological enhancements were realized in the course of this work. First, the optical resolution of scanning MOKE microscopy was improved by implementing a novel scanning mechanism, which allows the application of a larger aperture objective than in the usual scheme. Secondly, imaging photoluminescence spectroscopy was employed for spatially resolved electron-temperature measurements. Here, two different implementations were developed: One for lattice-temperature measurements by acceptor–bound exciton luminescence and a second for conduction-electron temperature measurements via the analysis of the electron–acceptor luminescence line shape.
It is shown in the present work that the originally stated anomalously high spin diffusivities were caused to a large extent by unwanted optical heating of the electron system. Although an efficient method was found to compensate for the influence of electron heating, it became also evident that the classical Einstein relation was nonetheless violated under the given experimental conditions. In this case however, it could be shown that this discrepancy did not originate from an experimental artifact, but was instead a manifestation of the fermionic nature of conduction electrons.
Ferromagnetic semiconductors (FS) promise the integration of magnetic memory functionalities and semiconductor information processing into the same material system. The prototypical FS (Ga,Mn)As has become the focus of semiconductor spintronics research over the past years. The spin-orbit mediated coupling of magnetic and semiconductor properties in this material gives rise to many novel transport-related phenomena which can be harnessed for device applications. In this thesis we address challenges faced in the development of an all-semiconductor memory architecture. A starting point for information storage in FS is the knowledge of their detailed magnetic anisotropy. The first part of this thesis concentrates on the investigation of the magnetization behaviour in compressively strained (Ga,Mn)As by electrical means. The angle between current and magnetization is monitored in magnetoresistance(MR) measurements along many in-plane directions using the Anisotropic MR(AMR) or Planar Hall effect(PHE). It is shown, that a full angular set of such measurements displayed in a color coded resistance polar plot can be used to identify and quantitatively determine the symmetry components of the magnetic anisotropy of (Ga,Mn)As at 4 K. We compile such "anisotropy fingerprints" for many (Ga,Mn)As layers from Wuerzburg and other laboratories and find the presence of three symmetry terms in all layers. The biaxial anisotropy term with easy axes along the [100] and [010] crystal direction dominates the magnetic behaviour. An additional uniaxial term with an anisotropy constant of ~10% of the biaxial one has its easy axis along either of the two <110> directions. A second contribution of uniaxial symmetry with easy axis along one of the biaxial easy axes has a strength of only ~1% of the biaxial anisotropy and is therefore barely visible in standard SQUID measurements. An all-electrical writing scheme would be desirable for commercialization. We report on a current assisted magnetization manipulation experiment in a lateral (Ga,Mn)As nanodevice at 4 K (far below Tc). Reading out the large resistance signal from DW that are confined in nanoconstrictions, we demonstrate the current assisted magnetization switching of a small central island through a hole mediated spin transfer from the adjacent leads. One possible non-perturbative read-out scheme for FS memory devices could be the recently discovered Tunneling Anisotropic MagnetoResistance (TAMR) effect. Here we clarify the origin of the large amplification of the TAMR amplitude in a device with an epitaxial GaAs tunnel barrier at low temperatures. We prove with the help of density of states spectroscopy that a thin (Ga,Mn)As injector layer undergoes a metal insulator transition upon a change of the magnetization direction in the layer plane. The two states can be distinguished by their typical power law behaviour in the measured conductance vs voltage tunneling spectra. While all hereto demonstrated (Ga,Mn)As devices inherited their anisotropic magnetic properties from their parent FS layer, more sophisticated FS architectures will require locally defined FS elements of different magnetic anisotropy on the same wafer. We show that shape anisotropy is not applicable in FS because of their low volume magnetization. We present a method to lithographically engineer the magnetic anisotropy of (Ga,Mn)As by submicron patterning. Anisotropic strain relaxation in submicron bar structures (nanobars) and the related deformation of the crystal lattice introduce a new uniaxial anisotropy term in the energy equation. We demonstrate by both SQUID and transport investigations that this lithographically induced uniaxial anisotropy overwrites the intrinsic biaxial anisotropy at all temperatures up to Tc. The final section of the thesis combines all the above into a novel device scheme. We use anisotropy engineering to fabricate two orthogonal, magnetically uniaxial, nanobars which are electrically connected through a constriction. We find that the constriction resistance depends on the relative orientation of the nanobar magnetizations, which can be written by an in-plane magnetic field. This effect can be explained with the AMR effect in connection with the field line patterns in the respective states. The device offers a novel non-volatile information storage scheme and a corresponding non-perturbative read-out method. The read out signal is shown to increase drastically in samples with partly depleted constriction region. This could be shown to originate in a magnetization direction driven metal insulator transition of the material in the constriction region.
This thesis reports a successful fabrication and characterisation of ferromagnetic/superconductor junction (F/S) on graphene. The thesis preposes a fabrication method to produce F/S junctions on graphene which make use of ALD grown Al2O3 as the tunnel barrier for the ferromagnetic contacts. Measurements done on F/G/S/G/F suggests that by injecting spin polarised current into the superconductor, a spin imbalance is created in the quasiparticle density of states of the superconductor which then diffuses through the graphene channel. The observed characteristic curves are similar to the ones which are already reported on metallic ferromagnet/superconductor junctions where the spin imbalance is created using Zeeman splitting. Further measurements also show that the curves loose their characteristic shapes when the temperature is increased above the critical temperature (Tc) or when the external magnetic field is higher then the critical field (Hc) of the superconducting contact. But to prove conclusively and doubtlessly the existence of spin imbalance in ferromagnet/superconductor junctions on graphene, more devices have to be made and characterised preferably in a dilution refrigerator.
Die Zielsetzung dieser Arbeit war die elektrische Spininjektion in Halbleiter zu erforschen und Methoden zu deren Realisation zu entwickeln. Hierzu wurden in dieser Arbeit III-V und II-VI Halbleiterheterostrukturen mit Hilfe von Photolumineszenz-, Elektrolumineszenz- und Anregungsspektroskopie untersucht. Die Messungen wurden bei Temperaturen im Bereich von 1.6 K bis 50 K durchgeführt und es wurden Magnetfelder bis zu 9 T verwendet. Die elektrische Spininjektion in einen nicht magnetischen Halbleiter wurde zum ersten mal in dieser Arbeit nachgewiesen. Hierzu wurden zwei neuartige Konzepte verwendet und miteinander verbunden. Zum einen wurde die Detektion von spinpolarisierten Strömen mit Hilfe von optischen Übergängen durchgeführt. Zum anderen wurde in dieser Arbeit erstmals ein semimagnetischer II-VI Halbleiter als spinpolarisierender Kontakt verwendet. Durch die optische Detektion wurden die bisherigen Magnetowiderstandsmessungen zur Bestimmung der Spininjektion abgelöst und durch die Verwendung von semimagnetischen Halbleitern wurde eine neue Klasse von Materialien für die Anwendung in spinselektiven Halbleiterheterostrukturen gefunden. Für den optischen Detektor der Elektronenpolarisation wurde eine GaAs/(Al, Ga)As Leuchtdiode (Spin-LED) verwendet, in die über das p-dotierte Substrat unpolarisierte Löcher und über den n-dotierten semimagnetischen Halbleiter spinpolarisierte Elektronen injiziert wurden. Das durch die Rekombination der Ladungsträger aus der LED emittierte Licht wurde in Oberflächenemission detektiert. Aufgrund der Auswahlregeln für optische Übergänge in Halbleitern mit Zinkblendestruktur ist es möglich, anhand der zirkularen Polarisation der Elektrolumineszenz, die Polarisation der injizierten Elektronen anzugeben. Abhängig vom externen Magnetfeld wurde die zirkulare Polarisation der Lichtemission von Spin-LEDs analysiert. Diese Polarisation erreichte schon bei geringen externen Magnetfeldern von z.B. 0.5 T sehr hohe Werte von bis zu 50 %. Im Vergleich dazu ist die intrinsische Polarisation von GaAs/(Al, Ga)As Heterostrukturen mit bis zu 5 % sehr gering. An den Spin-LEDs wurden Photolumineszenzmessungen zu der Bestimmung der intrinsischen Polarisation durchgeführt und zusätzlich wurde die Elektrolumineszenz von GaAs LEDs ohne manganhaltigen Kontakt analysiert. Mit Hilfe dieser Referenzmessungen konnten Seiteneffekte, die z.B. durch die magneto-optisch aktive manganhaltige Schicht in den Spin-LEDs verursacht werden können, ausgeschlossen werden. Insgesamt war es möglich die elektrische Spininjektion in Halbleiter eindeutig nachzuweisen.
Spin-Orbit Torques and Galvanomagnetic Effects Generated by the 3D Topological Insulator HgTe
(2021)
Nature shows us only the tail of the lion. But I have no doubt that the lion belongs with it even if he cannot reveal himself all at once. Albert Einstein
In my dissertation, I addressed the question of whether the 3D topological insulator mercury telluride (3D TI HgTe) is a suitable material for spintronics applications. This question was addressed by investigating the SOTs generated by the 3D TI HgTe in an adjacent ferromagnet (Permalloy) by using the ferromagnetic resonance technique (SOT-FMR).
In the first part of the dissertation, the reader was introduced to the mathematical description of the SOTs of a hybrid system consisting of a topological insulator (TI) and a ferromagnet (FM). Furthermore, the sample preparation and the measurement setup for the SOT-FMR measurements were discussed. Our SOT-FMR measurements showed that at low temperatures (T = 4.2 K) the out-of-plane component of the torque is dominant. At room temperature, both in-plane and out-of-plane components of the torque could be observed. From the symmetry of the mixing voltage (Figs. 3.14 and 3.15) we could conclude that the 3D TI HgTe may be efficient for the generation of spin torques in the permalloy [1]. The investigations reported here showed that the SOT efficiencies generated by the 3D TI HgTe are comparable with other existent topological insulators (see Fig. 3.17). We also discussed in detail the parasitic effects (such as thermovoltages) that can contribute to the correct interpretation of the spin torque efficiencies.
Although the results reported here provide several indications that the 3D TI HgTe might be efficient in exerting spin-torques in adjacent ferromagnets [2], the reader was repeatedly made aware that parasitic effects might contaminate the correct writing and reading of the information in the ferromagnet. These effects should be taken into consideration when interpreting results in the published literature claiming high spin-orbit torque efficiencies [2–4]. The drawbacks of the SOT-FMR measurement method led to a further development of our measurement concept, in which the ferromagnet on top of the 3D TI HgTe was replaced by a
spin-valve structure. In contrast with our measurements, in this measurement setup, the current flowing through the HgTe is known and changes in the spin-valve resistance can be read via the GMR effect.
Moreover, the SOT-FMR experiments required the application of an in-plane magnetic field up to 300 mT to define the magnetization direction in the ferromagnet. Motivated by this fact, we investigated the influence of an in-plane magnetic field in the magnetoresistance of the 3D TI HgTe. The surprising results of these measurements are described in the second part of the dissertation. Although the TI studied here is non-magnetic, its transversal MR (Rxy) showed an oscillating behavior that depended on the angle between the in-plane magnetic field and the electrical current. This effect is a typical property of ferromagnetic materials and is called planar Hall effect (PHE) [5, 6]. Moreover, it was also shown that the PHE amplitude (Rxy) and the longitudinal resistance (Rxx) oscillate as a function of the in-plane magnetic field amplitude for a wide range of carrier densities of the topological insulator.
The PHE was already described in another TI material (Bi2−xSbxTe3) [7]. The authors suggested as a possible mechanism the scattering of the electron off impurities that are polarized by an in-plane magnetic field. We critically discussed this and other theoretical proposed mechanisms existent in the literature [8, 9].
In this thesis, we attempted to explain the origin of the PHE in the 3D TI HgTe by anisotropies in the band structure of this material. The k.p calculations based on 6-orbitals were able to demonstrate that an interplay between Rashba, Dresselhaus, and in-plane magnetic field deforms the Fermi contours of the camel back band of the 3D TI HgTe, which could lead to anisotropies in its conductivity. However, the magnetic fields needed to experimentally observe this effect are as
high as 40 T, i.e., one order of magnitude higher than reported in our experiments. Additionally, calculations of the DoS to assess if there is a difference in the states for Bin parallel and Bin perpendicular to the current were, so far, inconclusive. Moreover, the complicated dependence of Rashba in the p-conducting
regime of HgTe [10] makes it not straightforward the inclusion of this term in the band structure calculations.
Despite the extensive efforts to understand the origin of the galvanomagnetic effects in the 3D TI HgTe, we could not determine a clear mechanism for the origin of the PHE and the MR oscillations studied in this thesis. However, our work clarifies and excludes a few mechanisms reported in the literature as the origin of these effects in the 3D TI HgTe. The major challenge, which still needs to be overcome, is to find a model that simultaneously explains the PHE, the gate dependence, and the oscillations in the magnetoresistance of the 3D TI HgTe as a function of the in-plane magnetic field.
To conclude, the author would like to express her hope to have brought the reader closer to the complexity of the questions addressed in this thesis and to have initiated them into the art of properly conducting electrical transport measurements on topological insulators with in-plane magnetic fields.