Refine
Has Fulltext
- yes (3) (remove)
Is part of the Bibliography
- yes (3)
Year of publication
- 2020 (3) (remove)
Document Type
- Doctoral Thesis (3)
Keywords
- Molekularstrahlepitaxie (3) (remove)
Institute
The subject of this thesis is the fabrication and characterization of magnetic topological
insulator layers of (V,Bi,Sb)\(_2\)Te\(_3\) exhibiting the quantum anomalous Hall
effect. A major task was the experimental realization of the quantum anomalous
Hall effect, which is only observed in layers with very specific structural,
electronic and magnetic properties. These properties and their influence on the
quantum anomalous Hall effect are analyzed in detail.
First, the optimal conditions for the growth of pure Bi\(_2\)Te\(_3\) and Sb\(_2\)Te\(_3\) crystal
layers and the resulting structural quality are studied. The crystalline quality of
Bi\(_2\)Te\(_3\) improves significantly at higher growth temperatures resulting in a small
mosaicity-tilt and reduced twinning defects. The optimal growth temperature is
determined as 260\(^{\circ}\)C, low enough to avoid desorption while maintaining a high
crystalline quality.
The crystalline quality of Sb\(_2\)Te\(_3\) is less dependent on the growth temperature.
Temperatures below 230\(^{\circ}\)C are necessary to avoid significant material desorption,
though. Especially for the nucleation on Si(111)-H, a low sticking coefficient is
observed preventing the coalescence of islands into a homogeneous layer.
The influence of the substrate type, miscut and annealing sequence on the growth
of Bi\(_2\)Te\(_3\) layers is investigated. The alignment of the layer changes depending on
the miscut angle and annealing sequence: Typically, layer planes align parallel to
the Si(111) planes. This can enhance the twin suppression due to transfer of the
stacking order from the substrate to the layer at step edges, but results in a step
bunched layer morphology. For specific substrate preparations, however, the layer
planes are observed to align parallel to the surface plane. This alignment avoids
displacement at the step edges, which would cause anti-phase domains. This results
in narrow Bragg peaks in XRD rocking curve scans due to long-range order in
the absence of anti-phase domains. Furthermore, the use of rough Fe:InP(111):B
substrates leads to a strong reduction of twinning defects and a significantly reduced
mosaicity-twist due to the smaller lattice mismatch.
Next, the magnetically doped mixed compound V\(_z\)(Bi\(_{1−x}\)Sb\(_x\))\(_{2−z}\)Te\(_3\) is studied in
order to realize the quantum anomalous Hall effect. The addition of V and Bi to
Sb\(_2\)Te\(_3\) leads to efficient nucleation on the Si(111)-H surface and a closed, homogeneous
layer. Magneto-transport measurements of layers reveal a finite anomalous
Hall resistivity significantly below the von Klitzing constant. The observation of
the quantum anomalous Hall effect requires the complete suppression of parasitic
bulklike conduction due to defect induced carriers. This can be achieved by optimizing
the thickness, composition and growth conditions of the layers.
The growth temperature is observed to strongly influence the structural quality.
Elevated temperatures result in bigger islands, improved crystallographic orientation
and reduced twinning. On the other hand, desorption of primarily Sb is
observed, affecting the thickness, composition and reproducibility of the layers.
At 190\(^{\circ}\)C, desorption is avoided enabling precise control of layer thickness and
composition of the quaternary compound while maintaining a high structural
quality.
It is especially important to optimize the Bi/Sb ratio in the (V,Bi,Sb)\(_2\)Te\(_3\) layers,
since by alloying n-type Bi\(_2\)Te\(_3\) and p-type Sb\(_2\)Te\(_3\) charge neutrality is achieved at
a specific mixing ratio. This is necessary to shift the Fermi level into the magnetic
exchange gap and fully suppress the bulk conduction. The Sb content x furthermore
influences the in-plane lattice constant a significantly. This is utilized to
accurately determine x even for thin films below 10 nm thickness required for the
quantum anomalous Hall effect. Furthermore, x strongly influences the surface
morphology: with increasing x the island size decreases and the RMS roughness
increases by up to a factor of 4 between x = 0 and x = 1.
A series of samples with x varied between 0.56-0.95 is grown, while carefully
maintaining a constant thickness of 9 nm and a doping concentration of 2 at.% V.
Magneto-transport measurements reveal the charge neutral point around x = 0.86
at 4.2 K. The maximum of the anomalous Hall resistivity of 0.44 h/e\(^2\) is observed
at x = 0.77 close to charge neutrality. Reducing the measurement temperature
to 50 mK significantly increases the anomalous Hall resistivity. Several samples
in a narrow range of x between 0.76-0.79 show the quantum anomalous Hall effect
with the Hall resistivity reaching the von Klitzing constant and a vanishing
longitudinal resistivity. Having realized the quantum anomalous Hall effect as the
first group in Europe, this breakthrough enabled us to study the electronic and
magnetic properties of the samples in close collaborations with other groups.
In collaboration with the Physikalisch-Technische Bundesanstalt high-precision
measurements were conducted with detailed error analysis yielding a relative de-
viation from the von Klitzing constant of (0.17 \(\pm\) 0.25) * 10\(^{−6}\). This is published
as the smallest, most precise value at that time, proving the high quality of the
provided samples. This result paves the way for the application of magnetic topological
insulators as zero-field resistance standards.
Non-local magneto-transport measurements were conducted at 15 mK in close
collaboration with the transport group in EP3. The results prove that transport
happens through chiral edge channels. The detailed analysis of small anomalies in
transport measurements reveals instabilities in the magnetic phase even at 15 mK.
Their time dependent nature indicates the presence of superparamagnetic contributions
in the nominally ferromagnetic phase.
Next, the influence of the capping layer and the substrate type on structural properties
and the impact on the quantum anomalous Hall effect is investigated. To
this end, a layer was grown on a semi-insulating Fe:InP(111)B substrate using the
previously optimized growth conditions. The crystalline quality is improved significantly
with the mosaicity twist reduced from 5.4\(^{\circ}\) to 1.0\(^{\circ}\). Furthermore, a layer
without protective capping layer was grown on Si and studied after providing sufficient
time for degradation. The uncapped layer on Si shows perfect quantization,
while the layer on InP deviates by about 5%. This may be caused by the higher
crystalline quality, but variations in e.g. Sb content cannot be ruled out as the
cause. Overall, the quantum anomalous Hall effect seems robust against changes
in substrate and capping layer with only little deviations.
Furthermore, the dependence of the quantum anomalous Hall effect on the thickness
of the layers is investigated. Between 5-8 nm thickness the material typically
transitions from a 2D topological insulator with hybridized top and bottom surface
states to a 3D topological insulator. A set of samples with 6 nm, 8 nm, and
9 nm thickness exhibits the quantum anomalous Hall effect, while 5 nm and 15 nm
thick layers show significant bulk contributions. The analysis of the longitudinal
and Hall conductivity during the reversal of magnetization reveals distinct differences
between different thicknesses. The 6 nm thick layer shows scaling consistent
with the integer quantum Hall effect, while the 9 nm thick layer shows scaling expected
for the topological surface states of a 3D topological insulator. The unique
scaling of the 9 nm thick layer is of particular interest as it may be a result of
axion electrodynamics in a 3D topological insulator.
Subsequently, the influence of V doping on the structural and magnetic properties
of the host material is studied systematically. Similarly to Bi alloying, increased
V doping seems to flatten the layer surface significantly. With increasing V content,
Te bonding partners are observed to increase simultaneously in a 2:3 ratio
as expected for V incorporation on group-V sites. The linear contraction of the
in-plane and out-of-plane lattice constants with increasing V doping is quantitatively
consistent with the incorporation of V\(^{3+}\) ions, possibly mixed with V\(^{4+}\)
ions, at the group-V sites. This is consistent with SQUID measurements showing
a magnetization of 1.3 \(\mu_B\) per V ion.
Finally, magnetically doped topological insulator heterostructures are fabricated
and studied in magneto-transport. Trilayer heterostructures with a non-magnetic
(Bi,Sb)\(_2\)Te\(_3\) layer sandwiched between two magnetically doped layers are predicted
to host the axion insulator state if the two magnetic layers are decoupled and in
antiparallel configuration. Magneto-transport measurements of such a trilayer heterostructure
with 7 nm undoped (Bi,Sb)\(_2\)Te\(_3\) between 2 nm thick layers doped with
1.5 at.% V exhibit a zero Hall plateau representing an insulating state. Similar results
in the literature were interpreted as axion insulator state, but in the absence
of a measurement showing the antiparallel magnetic orientation other explanations
for the insulating state cannot be ruled out.
Furthermore, heterostructures including a 2 nm thin, highly V doped layer region
show an anomalous Hall effect of opposite sign compared to previous samples. A
dependency on the thickness and position of the doped layer region is observed,
which indicates that scattering at the interfaces causes contributions to the anomalous
Hall effect of opposite sign compared to bulk scattering effects.
Many interesting phenomena in quantum anomalous Hall insulators as well as axion
insulators are still not unambiguously observed. This includes Majorana bound
states in quantum anomalous Hall insulator/superconductor hybrid systems and
the topological magneto-electric effect in axion insulators. The limited observation
temperature of the quantum anomalous Hall effect of below 1 K could be increased
in 3D topological insulator/magnetic insulator heterostructures which utilize the
magnetic proximity effect.
The main achievement of this thesis is the reproducible growth and characterization
of (V,Bi,Sb)2Te3 layers exhibiting the quantum anomalous Hall effect. The
detailed study of the structural requirements of the quantum anomalous Hall effect
and the observation of the unique axionic scaling behavior in 3D magnetic
topological insulator layers leads to a better understanding of the nature of this
new quantum state. The high-precision measurements of the quantum anomalous
Hall effect reporting the smallest deviation from the von Klitzing constant
are an important step towards the realization of a zero-field quantum resistance
standard.
The subject of this thesis is the growth of Hg\(_{1-x}\)Cd\(_2\)Te layers via molecular beam epitaxy (MBE).
This material system gives rise to a number of extraordinary physical phenomena related to its electronic band structure and therefore is of fundamental interest in research.
The main results can be divided into three main areas, the implementation of a temperature measurement system based on band edge thermometry (BET), improvements of CdTe virtual substrate growth and the investigation of Hg\(_{1-x}\)Cd\(_2\)Te for different compositions.
Im Rahmen dieser Arbeit wurden GaInP/GaAs/GaInNAs 3J-Mehrfachsolarzellen in einem MBE/MOVPE-Hybridprozess hergestellt und untersucht. Der verwendete Hybridprozess, bei dem nur die GaInNAs-Teilsolarzelle mittels MBE hergestellt wird, kombiniert diese beiden Technologien und setzt sie entsprechend ihrer jeweiligen Vorteile ein. Die gezeigten Ergebnisse bestätigen grundsätzlich die Machbarkeit des Hybridprozesses, denn eine Degradation des mittels MBE hergestellten GaInNAs-Materials durch die Atmosphäre im MOVPE-Reaktor konnte nicht festgestellt werden. Dieses Resultat wurde von im Hybridprozess hergestellten 3J-Mehrfachsolarzellen, die GaInNAs-Teilsolarzellen enthalten, bekräftigt. Die offene Klemmspannung einer gezeigten Solarzelle erreichte bereits 2,59 V (AM1.5d) bzw. 2,48 V (AM0) und liegt damit jeweils über einer als Referenz hergestellten 2J-Mehrfachsolarzelle ohne GaInNAs. Die mittlere interne Quanteneffizienz der enthaltenen GaInNAs-Teilsolarzelle liegt bei 79 %. Die Berechnungen auf Grundlage dieser Effizienz unter Beleuchtung mit AM1.5d und unter Beleuchtung mit AM0 zeigten, dass nicht die enthaltene GaInNAs-Teilsolarzelle Strom limitierend wirkt, sondern die mittels MOVPE gewachsene GaInP-Teilsolarzelle. Die experimentell bestimmte Kurzschlussstromdichte der hergestellten Mehrfachsolarzelle ist wegen dieser Limitierung etwas geringer als die der 2J-Referenzsolarzelle. Der MOVPE-Überwachsvorgang bietet zwar noch weiteres Verbesserungspotential, aber es ist naheliegend, dass der Anwachsvorgang auf dem MBE-Material soweit optimiert werden kann, dass die aufgewachsenen GaInP- und GaAs-Schichten frei von Degradation bleiben. Damit bietet der Hybridprozess perspektivisch das Potential günstigere Produktionskosten in der Epitaxie von Mehrfachsolarzellen mit verdünnten Nitriden zu erreichen als es ausschließlich mittels MBE möglich ist.
Im Vorfeld zur Herstellung der 3J-Mehrfachsolarzellen wurden umfassende Optimierungsarbeiten des MBE-Prozesses zur Herstellung der GaInNAs-Teilsolarzelle durchgeführt. So wurde insbesondere festgestellt, dass das As/III-Verhältnis während dem Wachstum einen entscheidenden Einfluss auf die elektrisch aktive Dotierung des GaInNAs-Materials besitzt. Die elektrisch aktive Dotierung wiederum beeinflusst sehr stark die Ausdehnung der Raumladungszone in den als p-i-n-Struktur hergestellten GaInNAs-Solarzellen und hat damit einen direkten Einfluss auf deren Stromerzeugung. In der Tendenz zeigte sich eine Zunahme der Stromerzeugung der GaInNAs-Teilsolarzellen bei einer gleichzeitigen Abnahme ihrer offenen Klemmspannung, sobald das As/III-Verhältnis während des Wachstums reduziert wurde. Durch eine sehr exakte Kalibration des As/III-Verhältnisses konnte ein bestmöglicher Kompromiss zwischen offener Klemmspannung und Stromerzeugung gefunden werden. Eine gezeigte GaInNAs-Einfachsolarzelle erreichte eine mittlere interne Quanteneffizienz von 88 % und eine offene Klemmspannung von 341 mV (AM1.5d) bzw. 351 mV (AM0). Berechnungen auf Grundlage der Quanteneffizienz ergaben, dass diese Solarzelle integriert in eine 3J-Mehrfachsolarzelle unter dem Beleuchtungsspektrum AM1.5g eine Stromdichte von 14,2 mA/cm^2 und unter AM0 von 17,6 mA/cm^2 erzeugen würde. Diese Stromdichten sind so hoch, dass diese GaInNAs-Solarzelle die Stromproduktion der GaInP- und GaAs-Teilsolarzellen in einer gängigen Mehrfachsolarzelle erreicht und keine Ladungsträgerverluste auftreten würden. Aufgrund ihrer höheren offenen Klemmspannung gegenüber einer Ge-Teilsolarzelle bietet diese GaInNAs-Teilsolarzelle das Potential die Effizienz der Mehrfachsolarzelle zu steigern. Messungen der Dotierkonzentration in der GaInNAs-Schicht dieser Solarzelle ergaben extrem geringe Werte im Bereich von 1x10^14 1/cm^3 bis 1x10^15 1/cm^3 (p-Leitung). In Ergänzung zu den Optimierungen des As/III-Verhältnisses konnte gezeigt werden, dass sich ein Übergang von p- zu n-Leitung im GaInNAs mit der Verringerung des As/III-Verhältnisses erzeugen lässt. Nahe des Übergangsbereiches wurden sehr geringe Dotierungen erreicht, die sich durch eine hohe Stromproduktion aufgrund der Ausbildung einer extrem breiten Verarmungszone gezeigt haben. Durch eine reduzierte offene Klemmspannung der bei relativ geringen As/III-Verhältnissen hergestellten Solarzellen mit n-leitendem GaInNAs konnte auf das Vorhandensein von elektrisch aktiven Defekten geschlossen werden. Generell konnten die gemessenen elektrisch aktiven Dotierkonzentrationen im Bereich von üblicherweise 10^16 1/cm^3 mit hoher Wahrscheinlichkeit auf elektrisch aktive Kristalldefekte im GaInNAs zurückgeführt werden. Eine Kontamination des Materials mit Kohlenstoffatomen in dieser Größenordnung wurde ausgeschlossen.