Nothing of this would be possible without support of my supervisor Alberta Bonanni, who gave me the opportunity and inspiration to do this work. I am thankful to her for giving me a guidance and sharing her experience to find solutions for everyday life challenges. She is an admirable and great leader, who managed to establish a friendly atmosphere and warm relations in a group of talented scientists – special people with different cultural background that are always mutually helpful and supportive. No words can describe my gratitude to her. My special thanks to Bogdan Faina, for his support and being always there for me, when I needed help, no matter if it was with a setup in lab, or with any situation even outside work. I could always rely on him and he would do everything in to order to help. Many thanks to Aitana Tarazaga Mart´ın-Luengo, my former officemate. From her PhD experience I learned a lot, which helped me to overcome many challenges during my time. I am thankful to her for advices to sort out any concern that I had, whether concerning work, private, or simply tiny and even sometimes meaningless. I could always share it with her and she would guide me to find a solution. I am extremely thankful to Rajdeep Adhikari for his support and help, especially when it concerns a scientific topic. He would always answer my questions better than any book could explain. Besides showing an example of a great scientist, he is the one who gave me the mindset that we have to remain children in our souls in order to stay young forever. My gratitude to Giulia Capuzzo, my closest former work-companion. We shared long hours in lab fabricating time demanding and challenging heterostructures, and with a mutual support we managed to succeed. Thanks also for giving me a push to register for my first Linz half marathon. My thanks to Margherita
Group III-nitrides are nowadays used in many different areas due to their unique properties. By utilizing ternary compounds of AlN, GaN, and InN, the band gap can be varied over a large range, enabling to fabricate optoelectronic devices operating over the entire visible spectrum, as well as in the infrared and ultraviolet spectral range. Especially in the blue and ultraviolet spectral range, nitrides are without any alternative for fabricating devices like LEDs or lasers operating with high efficiency. Only progress in the development of group III-nitrides made the production of white LEDs possible, which are used in almost all lighting applications today. Although the band gap of ternaries containing In can be small enough to cover the infrared spectral range utilizing interband transitions, the efficiency of these structures would be poor due to several difficulties associated with the incorporation of In [ 1 – 3 ]. However, intersubband transitions within the conduction band of GaN-based quantum wells feature energies suited for infrared applications. Conceivable devices are for example quantum cascade lasers operating at high powers or at high frequencies [ 4 , 5 ], or fast infrared detectors [ 6 ].
Despite their great success, group-IIInitrideheterostructures suffer from a major drawback which is the occurrence of large stress leading to straining and relaxation of the material. This is unwanted because defects are formed which deteriorate the crystal quality. Also the material surface can roughen or the material can crack. The large amount of stress in nitride epitaxy originates from large differences in lattice parameters between the binary compounds GaN, AlN and InN . The lattice mismatch between GaN and AlN is 2.4 % , which is very large compared to GaAs and AlAs (-0.14 % ). This causes relaxation processes which affect material properties like the band gap energy . Especially ternary alloys as AlGaN, InGaN and AlInN pose a challenge to process control because stress, strain and process parameter fluctuations can cause the composition to vary. If relaxation starts in ternary alloys and the lattice parameter is changed gradually, the energy barrier for the incorporation of certain atomic species into the solid phase can be lowered or increased. This causes the composition of ternary alloys to change during the relaxation process and is termed as compositional pulling , . Binary materials as GaN or AlN are much more stable and robust regarding their epitaxial processes. Compositional pulling cannot occur for these materials. To combine the robustness
This work is dedicated to ll gaps in the knowledge about the self-assembled growth of GaN NWs and group III-nitride NWHs by PAMBE. Additionally, a PCG process that allows the fabrication of GaN NW templates with a diameter ≤ 100 nm is introduced. Both growth techniques are used to fabricate GaN NW templates for the subsqeuent overgrowth with group III-nitride NWHs. The manipulation of the internal electric elds in AlN/GaN nanodisc superlattices (NDSLs) was studied by following two complementary approaches. Free carrier screening was investigated in Ge doped GaN NDs while the recently proposed Internal Field Guarded Active Region Design (IFGARD)  was applied to nominally undoped GaN NDs embedded in AlN barriers. The inuence of the atom source operating parameters on the structural and optical properties of GaN NWs and In x Ga 1−x N/GaN
Single layer thin films of YBCO and LCMO and heterostructures as well as superlattices of both materials were deposited single crystalline (110) oriented NGO. The investigations on an other substrate than STO have been performed to check whether the results found on STO are intrinsic or substrate related. The lattice mismatch between YBCO and (110) oriented NGO is less than 0.3 % in contrast to STO with 1.3 %. Nevertheless, the strain induced by STO will not affect the physical properties of the thin films as shown by Driza et al. . Malik et al.  reported about YBCO/LCMO superlattices ablated on NGO (110) and LSAT (001). They used these substrates because the lattices parameters are nearly matching YBCO and LCMO and they are not undergoing any strctural transition below 300 K in contrast to STO. Malik et al. discovered that an additional annealing step is required to achieve a complete oxidation of the superlattices. So they annealed their samples for 12 hours at 485 °C in oxygen in a separate furnace. They correlate the necessity of an additional annealing to the crystalline perfection of their samples. Since the diffusion of oxygen in YBCO is very low and the diffusion coefficients differ between the ab-plane (D ab (760 K) ≈ 10 -15 m²/s) and the c-axis (D c (760 K) ≈ 10 -19 m²/s) as reported
Due to strong spontaneous and piezoelectric polarization fields in AlGaN-based het- erostructures, a MQW band structure is formed that exhibits tilted valence bands and conduction bands within the quantum wells and barrier layers . As described by the quantum-confined Stark effect (QCSE) , such strong polarization fields lead to reduced B coefficients due to a reduced overlap of electron and hole wavefunctions as well as a red shift of the emission wavelength, all of which can be reduced by employing very thin quantum wells . Further, electrical confinement and QCSE can be influ- enced by the barrier height – the band gap offset between AlGaN barriers and AlGaN quantum wells. As AlGaN-based MQWs can be categorized as type I semiconductor heterostructures with straddling gap, the band gap offset is divided into fractions for conduction band and valence band, e. g. with 70/30 ratio . The optical polarization of the MQW emission is determined by conduction and valence bands involved in the radiative recombination and can be controlled by adjusting the Al mole fraction in quantum wells and barriers, quantum well thickness, or strain state in order to obtain higher transverse-electric (TE) or transverse-magnetic (TM) emission polarization . An electron blocking layer (EBL) is required in order to provide high carrier injection efficiency (CIE). The main purpose is the prevention of electron leakage into the p-side layers such as the AlGaN:Mg SPSL, as parasitic recombination channels with holes in the p-layers would reduce the efficiency of LEDs. At the same time an EBL needs to facilitate efficient hole injection into the MQW. The performance of an EBL is affected by the respective band alignment to the last barrier of the MQW as well as to the AlGaN:Mg SPSL, mainly influenced by band gap, doping or thickness of the EBL. The AlGaN:Mg short-period superlattice (SPSL) is the p-type layer of the LED heterostructure, responsible for efficient vertical transport of holes from the p-contact to the EBL and MQW layers. Therefore, the AlGaN:Mg SPSL needs to provide high p-type conductivity. In order to obtain high LEE, strong UV absorption has to be avoided, e. g. by using UV transparent AlGaN:Mg SPSLs with wide band gap in combination with UV reflective contacts. However, achieving both high conductivity and high UV transparency at the same time appears outstandingly challenging for Mg-doped AlGaN alloys (section 4.3 and section 4.4).
observed depending on material heights. However, no rim zone damage like white layer, foamy or porous structure, cracks or microstructure modifications occurred. Compared to steel, the main cut produced a lower roughness and trim cuts a higher roughness at almost the same magnitude of discharge energy. However, cutting rates of silicon nitride are higher. Trim cuts are no longer necessary to remove a white layer or foamy structure, so the number of trim cuts could be reduced. Furthermore, an explanatory approach was shown to describe the achieved topography.
A special case of the deep defect is the so called DX-center. which has been studied for Si in AlGaAs alloys in most detail . In literature, different models have been proposed, describing its electronic properties. In the following we will discuss the proposals by Lang et al. [56,57]. Their major observations were (i) a defect with a very small capture cross section at low temperatures (ii) persistent photo-conductivity (PPC) and (iii) very large differences between optical and thermal ionization energies. They attributed these properties to a metastable defect, which occurs in three charge states (+/0/-) within the bandgap. While the (+/0) states are shallow donor configurations, the negatively charged (acceptor like) state refers to the stable ground state. Thus, the neutral d (0) state
For PL experiments, the samples were mounted in a cold-finger cryostat with the temperature controlled from T ¼ 7 K to RT. PL was excited with a frequency-doubled argon laser (k ¼ 244 nm) and collected into a Jobin- Yvon HR460 monochromator equipped with an ultraviolet- enhanced charge-coupled device (CCD) camera. The diame- ter of the excitation spot on the sample was about 100 lm. The excitation power was kept around 100 lW, low enough to avoid screening of the internal electric field. The low- temperature (T ¼ 7 K) PL spectra of the samples under study are presented in Fig. 5 . In the case of the GaN/AlN QWs, the spectral structure of the emission is due to monolayer thick- ness fluctuations in the QWs, as described elsewhere. 33 In the rest of the structures (GaN/AlN QDs and InGaN/GaN QDs and QWs), the broader linewidth makes it possible to observe the superimposition of a Fabry-Perot interference pattern related to the total nitride thickness.
In order to realize a light–emitting nanostructure with an active material of desired dimen- sionality and emission energy, a heterostructure of the binary (or ternary, . . . ) III–nitride alloys is synthesized. If the active–material inclusion therein has a different lattice constant than the matrix, the lattice is deformed at the heterointerfaces, with a strain distribution that minimizes the total strain energy. Notably, a difference in lattice constants can be crucial to achieving the desired growth mode, e. g. in the strain–induced self–assembly (Stranski– Krastanov growth) of QDs. Since pyroelectric crystals are always also piezoelectric, the strain deformation entails a strain–induced dipole (Fig. 2.1 c), where the relation between strain state and piezoelectric field P pz (r) is given in terms of the electro–mechanical tensor, containing the material–dependent piezoelectric coefficients. Particularly, the piezoelectric coefficients of the III–nitrides are about an order of magnitude larger than those found in arsenide or phosphide material systems [32, 33]. While the orientation of the spon- taneous polarization is always along the [000¯ 1] direction, the piezoelectric component can be either parallel or anti–parallel, depending on the strain state. Specifically, the strain–induced polarization is parallel to the spontaneous polarization in case of biaxial tensile strain and anti–parallel for biaxial compressive strain (cf. Fig. 2.1 c) . The total internal polarization P (r) = P sp (r) + P pz (r) is then given simply as the sum of pyroelectric and piezoelectric contributions, with an associated polarization–charge density ρ(r) = −∇ · P (r). For instance, the abrupt change of the polarization at an InGaN/GaN heterointerface leads to a quasi–two–dimensional sheet charge σ of up to ∼ 10 14 cm −2 (in units of the elementary charge) . Depending on the sign of this bound charge density and the position of the Fermi level, a two–dimensional electron (or hole) gas will form, that can partially compensate the bound charges.
The dependencies that follow from the analysis, described above will be called intensity composition relationship in the following. Each intensity composition relationship depicted in Fig. 1 a–c shows an in- creasing, nearly linear behavior of the Voronoi intensity with increasing number of substitute atoms. This is because a lighter element is sub- stituted with a heavier one and thus increasing the mean atomic number Z of the column. The slope of the plot depends on the atomic number of the substitute atom as well as the difference to the replaced atom, since heavier elements have a higher influence on the STEM HAADF intensity. To visualize this effect more clearly, a second x-axis has been plotted in Fig. 1 a–c. There, the plotted intensities were nor- malized to the surrounding base material, i.e. GaAs for sample I, GaP for sample II and Si for sample III. The limits of the axis reflect the slope of the plot. This means that the intensity difference between pure GaAs and InAs ( Fig. 1 a) or pure GaP and GaAs ( Fig. 1 b) is smaller than the difference between pure Si and pure Ge ( Fig. 1 c). Comparing sample I and II, it is noticeable that the points of pure GaAs do not match in Voronoi intensity. This is due to the different inner detector angles (63 mrad for sample I and 74 mrad for sample II), which affects the intensity strongly. The inner detector angles of sample I and III were 63 mrad and 68 mrad and thus are more comparable. The second axis in
16 nm is caused by the low refractive index contrast in group III nitrides.
Experimental results from several groups are summarized in the paper from Carlin et al. . Compared to these data, a maximum reflectivity around 75 % with 10 pairs is ranked acceptable for nitride-based DBR. The reflectivity data from our sample are even slightly higher than the given data of AlGaN/GaN-DBR but this can be attributed to the very small stopband wavelength close to the absorption edge of GaN for the SL-DBR sample. This increases the refractive index contrast. The graph in red (in fig. 4.22) shows the reflectivity spectrum of a buffer structure (sapphire substrate, AlN, GaN) comparable to the one in the SL-DBR sample. The reflectivity of the buffer structure shows no stopband peak but also periodic oscillations between 400 and 800 nm. By comparison of the two graphs, it becomes clear that the modulation observed for the SL-DBR on the long-wavelength side of the stopband peak originates from the buffer structure. Between 550 and 800 nm, the reflectivity graphs of SL- DBR (black) and buffer structure (red) look quite identical and only the oscillation frequency of the buffer structure is higher. This could be related to differences in layer thicknesses. Between 400 and 550 nm, the spectra for SL-DBR and buffer structure look more different. This can be attributed to a convolution of DBR reflectivity ripples from the adjacent stopband with the thickness-related oscillation from the buffer for the SL-DBR structure. The node at approximately 530 nm is also related to the layer thicknesses in the buffer stack. A simulational analysis of the reflectivity will be given at the end of this chapter.
Semiconducting oxides and nitrides are, due to their large band gap energies, the two most interesting groups to employ in optoelectronic devices operating in the visible and UV spectral range. There are to mention Zinc Oxide (ZnO) and Gallium Oxide (Ga2O3) on the oxide side and Gallium Nitride (GaN) together with Indium Nitride (InN) and Aluminum Nitride (AlN) for the nitrides. These semiconducting materials combine unique properties on their crystallography and growth mechanisms, as well as on their optical, electrical and magnetic properties. Hence it is not surprising that with these materials it was possible to build novel displays, light emitters, data storages, bio- and environmental-sensors and energy generating- or saving-devices. For any device application one has to solve problems related to the growth mechanisms of the materials. Defect characterization of the materials is a necessity, since relevant physical properties are affected by intrinsic and extrinsic defects. There are various characterization tools ranging from the electrical- or optical- and magnetic methods to microscopy’s such as electron- or atomic force microscopy which give information on the structural- or surface-properties. The choice which one suits best to achieve the given purpose depends on the specific information one needs.
ﬁelds at the aluminum and gold junctions become equivalent so that both junctions determine simultaneously the transport of carriers.
This phenomenon will not occur for symmetrically shaped energy selective contacts assuming an equivalent contribution of both junctions. Furthermore, this behavior is predicted for two Schottky barriers at the interfaces. The electric ﬁeld of the Schottky-Barrier needs to point in the same direction of the ﬁeld of the P-N deple- tion layer. Rajasekharan et al. have applied a diﬀerent fabrication procedure for a double SB device [ 242 ]. They have directly formed Schottky-barriers with erbium (low work function) and palladium (high work function) and backgated the channel (three terminal device). The backgating ensures an additional band bending so that the ﬁeld of the depletion layers at the metal silicon interfaces points in the same direction. A minimum of the current swing was observed for -20 V but Rajasekha- ran et al. have not paid attention to this particular behavior. In principal, this phenomenon oﬀers a promising approach to examine metal-nitride-contact forma- tion of asymmetric design. Once a metal contact is well characterized, the second one can be studied by the inverse subthreshold slope. Also, the ultra-thin device layer of 20 nm SOI thickness reveals excellent subthreshold swings, which are close to the limit by thermionic emission. In addition, the subthreshold swing provides a simple determination of the ideality factor [ 243 ]. Focusing on the minimum S min
6.2.4 demonstrates the large impact of a small contaminated area on the speciﬁc resistance indicating a silicon surface related issue. All these experiments point to the depletion region of the Schottky-Mott contact but an explicit identiﬁcation of this cause is not presented in this thesis. An analysis of the lift-oﬀ steps is necessary to clarify the circumstance of any residual materials or contaminants causing a rise of the contact resistance. Here, a contact resistivity study by transmission line method is an appropriate approach to analyze in detail the junction resistance [ 238 ]. The results of the vanadium based nitride MOSFET indicate that a clean and pure metal evaporation fabrication process is crucial for a low subthreshold swing. For instance, the grain size of the evaporated material determines the speciﬁc resistance [ 239 ]. SEM images of in-house evaporated aluminum and vanadium and externally deposited aluminum layer show diﬀerent grain sizes. While vanadium exhibits an excellent smoothness and no small grain formation (see Fig. 6.5 a), the aluminum layers have much smaller grain sizes (see Fig. 6.5 b and c). Especially the in-house evaporated aluminum has even smaller fragments, which is conﬁrmed by the huge surface roughness. This might be an indication for various work functions and a detailed study of the evaporated metal should take this into account.
Normaldruck bereits bei ca. 800 °C stattfindet, unterdrückt werden kann. Es gelang somit durch Anwendung von Hochdruck, die maximale Synthesetemperatur ungefähr zu verdoppeln. Dies hat deutlich verbesserte Kristallisationsbedingungen zu Folge. Das Hauptproblem bei konventionellen Synthesen von Nitridophosphaten, nämlich ungünstige Kristallisationsbedingungen infolge limitierter Synthesetemperatur, konnte somit beseitigt und infolgedessen die Zahl der bekannten kristallographisch eindeutig charakterisierten ternären Phosphor(V)-nitride von neun auf achtzehn verdoppelt werden. Insbesondere wurde die Zahl der bekannten hochkondensierten Nitridophosphate (molares Verhältnis P : N > 1 : 2) von drei auf neun erhöht sowie eine neue Modifikation des binären Phosphor(V)-nitrids P 3 N 5 erhalten. Daher kann das im Rahmen der Dissertation
significantly different. 10 The charge transfer processes involved in contact type and core/shell type semiconductor systems are shown in Figure II-2. In contact type semiconductor system the two particles are in contact with each other and both holes and electrons are accessible for selective oxidation and reduction process on different particle surfaces. On the other hand, in core/shell type heterostructures the electrons get injected into the energy levels of the core semiconductor (if it has a conduction band potential which is lower than that of the shell). Therefore, only one charge carrier is accessible at the surface in a core/shell type semiconductor system, thus making selective charge transfer possible at the semiconductor electrolyte interface. The other charge carrier (here electron) gets trapped within the inner semiconductor particle and is not readily accessible for the reduction reaction. However, in the coupled heterostructure composite nanoparticles, no electric field is necessary, as the charge separation is achieved by the tunnelling of electrons. 11 It has been established that the