Refine
Document Type
- Doctoral thesis (3)
Has Fulltext
- yes (3)
Is part of the Bibliography
- no (3)
Language
- English (3)
Keywords
- Röntgenspektroskopie (3) (remove)
Institute
The chemical and electronic structure of chalcopyrite absorbers with different bulk band gap energies, Egbulk, [i.e., low-gap Cu(In,Ga)Se2 (CIGSe, Egbulk ~ 1.2 eV) and wide-gap CuInS2 (CIS, Egbulk ~ 1.5 eV)] and of buffer/absorber heterointerfaces based on these materials are studied with soft and hard x-ray spectroscopy techniques. Mechanisms that benefit (limit) the performance of low(wide)-gap chalcopyrite-based solar cells are identified. This knowledge is used to develop surface tailoring treatments to optimize buffer/absorber heterointerfaces based on wide-gap chalcopyrites and improve the performance of their solar cells.
Photoemission spectroscopy (PES) characterization of the two absorbers (i.e., CIGSe and CIS) reveal compositional-depth profiles. The changes detected in CIGSe include: a near surface Ga-depletion, a strongly Cu-poor surface and a strong presence of surface Na that (likely) occupies Cu vacancies. A similar Cu-deficiency is found in CIS. The depth-composition changes result in significant widening of the band gap at the surface, Egsurf, (i.e., CIGSe, Egsurf: 1.70 ± 0.2 eV and CIS, Egsurf: 1.88 ± 0.2 eV) as evident by ultraviolet photoelectron spectroscopy (UPS) and inverse photoemission spectroscopy (IPES) measurements. Differences in the interaction of the CIGSe and CIS surfaces with deposited buffer materials are identified. PES and modified Auger parameter studies reveal strong intermixing at the CdS/CIGSe and ZnS/CIGSe heterointerfaces. S L2,3 x-ray emission spectroscopy (XES) measurements of CIGSe substrates submitted to CdS chemical bath deposition (CBD-CdS) treatments show the formation of In2S3 and defect-rich/nanostructured CdS at the interface, compounds with higher band gap values than the measured Egsurf for CIGSe. S L2,3 XES spectra of CIGSe substrates submitted to CBD-ZnS treatments reveal the formation of (Zn,In)(S,Se)2 chemical analogs at the interface. PES and XES measurement series show that the CdS/CIS heterointerface is more abrupt, with no detected interface chemical species. Direct measurement of the band alignment of these heterointerfaces reveals: an ideal conduction band offset (CBO) configuration for CdS/CIGSe (i.e., CBO: +0.11 ± 0.25 eV), a spike CBO configuration for ZnS/CIGSe (i.e., CBO: +1.06 ± 0.4 eV), and a highly unfavorable cliff CBO configuration for CdS/CIS (i.e., CBO: -0.42 ± 0.25 eV). The performance of solar cell devices based on these heterointerfaces is correlated to their CBO configuration.
Two surface tailoring approaches intended to correct the CBO configuration of the CdS/CIS heterointerface are presented. One method is based on rapid thermal processing (RTP) selenization treatments of CIS absorbers, aiming to exchange Se for S in treated samples. The idea behind this approach is to modify the surface of a wide-gap chalcopyrite so that it forms a more favorable heterointerface with CdS, such as heterointerfaces within low-gap chalcopyrite devices. X-ray fluorescence analysis and PES measurements of RTP-treated CIS samples show a greater treatment effect at the surface of the sample compared to the bulk (i.e., surface [Se]/[S+Se] range: 0.23 ± 0.05 to 0.83 ± 0.05, compared to bulk [Se]/[S+Se] range: 0.01 ± 0.03 to 0.24 ± 0.03). Tuning of the Cu:In:(S+Se) surface composition from a Cu-poor 1:3:5 to a 1:1:2 stoichiometry is observed in RTP-treated CIS samples with lower to higher surface Se contents, respectively. UPS measurements show a shift in valence band maximum toward the Fermi level in samples with higher surface Se content (i.e., -0.88 ± 0.1 to -0.51 ± 0.1 eV), as expected for a reduction in Egsurf due to exchange of Se for S. Ultraviolet-visible spectrophotometry reveals a reduction in the optical band gap of samples with greater Se incorporation (i.e., from 1.47 ± 0.05 to 1.08 ± 0.05 eV), allowing for a working window for optimization purposes.
The second tailoring method involves surface functionalization of CIS absorbers with dipole-charge-inducing self-assembled monolayers (SAM) of benzoic acid derivatives and thiol molecules. The introduction of dipole charges between a heterointerface can tune the relative alignment of the electronic bands composing its electronic structure; thus, use of a suitable dipole-inducing SAM could correct the CBO misalignment in the CdS/CIS heterointerface. UPS measurements of the secondary electron cut-off region of CIS samples treated with a selected set of SAMs show a work function modulation of CIS (i.e., 4.4 ± 0.2 eV - 5.2 ± 0.2 eV). Small gains in solar cell parameters of solar cells based on SAM-modified heterointerfaces are measured.
An overview of the performance of chalcopyrite(kesterite)-based solar cells in relation to the electronic properties of their corresponding buffer/absorber heterointerface suggests that optimization approaches extending beyond the buffer/absorber heterointerface may be needed for further performance gains in wide-gap chalcopyrite-based solar cell devices.
Current memory technologies, such as DRAM, SRAM, and NAND Flash, which are approaching very difficult issues related to the continuous scaling to and beyond the 16 nm generation, has led research over the past two decades to the discovery of several new memory technologies. In recent years, new emerging nonvolatile memories (NVMs), such as phase-change random access memory (PCRAM), ferroelectric random access memory (FRAM), magnetic random access memory (MRAM), and resistive random access memory (RRAM), have been intensively studied. Among these candidates, RRAM is a very promising and worldwide studied candidate for alternative NVM and a high potential successor for Flash in terms of energy consumption (write current in the μA range compared to mA) and simplicity of process integration.
A fully CMOS compatible TiN/Ti/HfO2/TiN RRAM module was successfully integrated with a select transistor (1T1R memory) in IHP’s technology. Nonetheless, reliability and insufficient understanding of the resistive switching mechanism are the two main issues limiting this memory technology development for e.g. wireless sensor network (WSN) applications. The still unclear atomic-scale mechanism of HfO2-based resistive switches and the identification of the material changes within the insulator must be addressed to suggest a knowledge-based improvement of device performance. In this frame, the Ti/HfO2 interface is thoroughly investigated in this Thesis by complementary materials science techniques.
First, the investigation of the as-deposited Ti/HfO2/TiN cells revealed that: (1) the Ti layer scavenges oxygen atoms stronger from amorphous (a-HfO2) than from monoclinic (m-HfO2) HfO2 films; (2) not only oxygen vacancies but also other impurities in the atomic vapor deposited (AVD) a-HfO2 film, such like nitrogen and carbon (probably resulting from the used AVD precursor chemistry) are present in the HfO2 insulator.
Next, the electrical characterization of Ti/AVD a-HfO2/TiN cells (with voltage applied to the Ti top electrode while TiN bottom electrode was grounded) revealed a clockwise bipolar resistive switching behavior after an electroforming process at positive voltage polarity. Besides, the chemical and electronic changes observed by hard X-ray photoelectron spectroscopy (HAXPES), indicate the creation of n-type dopants in the a-HfO2 film during the electroforming process, probably related to the formation of positively charged oxygen vacancies in a-HfO2 by the electrochemically induced Ti/a-HfO2 interface oxidation.
In order to directly compare electrical with electronic and chemical changes of one and the same RRAM cell, an in-operando HAXPES technique was developed. These unique studies have revealed the following characteristics of the Ti/AVD a-HfO2/TiN cells: (1) the as-deposited cells are able to switch at low electrical power; (2) However, this resistive switching is not stable and an electroforming process with a slightly increased power is required to stabilize the switching event; (3) Electrical changes correlated with HAXPES results and literature indicate that (i) the forming/set electrical power defines the oxygen vacancies concentration in the a-HfO2 and thus the stability of the resistive switching properties and (ii) the stable resistive switching can be described by a push-pull model of oxygen vacancies migration under the influence of an electrical field; (4) Besides, carbon segregation at the Ti/a-HfO2 interface – while increasing the electrical power or cycling the device – shows that the defects physics is not limited only to oxygen vacancies; other defects may thus contribute under electrical stress to the resistive switching phenomenon and need to be included in theoretical models to correctly describe the switching characteristics.
Finally, according to the presented HAXPES results, the Ti/AVD a-HfO2/TiN RRAM cells are classified to the valence change mechanism. The resistive switching mechanism is attributed to the creation and rupture of oxygen vacancies-based conducting filaments and the Ti/HfO2 interface oxidation is of central importance for the defect balance of the RRAM cell. Most importantly, a reduction of carbon content in the AVD-deposited HfO2 improved the reliability of these memory cells.
The properties of Si thin films for solar cells, the interaction with different substrates and the influence of dopants are examined with synchrotron based x-ray spectroscopy – primarily x-ray emission spectroscopy (XES) and hard x-ray photoelectron spectroscopy (HAXPES). The films are studied as-deposited (i.e., amorphous, a-Si) and after conversion into polycrystalline (poly-Si) employing solid phase crystallization (SPC). Si L2,3 XES spectra of thin-film Si samples can be described by a superposition of a-Si and monocrystalline Si-wafer (c-Si) reference spectra. According to a quantification based on that superposition principle, none of the investigated samples are completely crystallized – a measurable a-Si component always remains (5-20 %) regardless of deposition and treatment conditions. Based on additional results from electron back scattering diffraction different models are developed which may explain this finding. According to these models, the remnant a-Si component can be attributed to amorphous/disordered material at the grain boundaries. Using one of these models, the thickness of this grain-surrounding material s could be approximated to be (1.5 ± 0.5) nm. Further investigations of the SPC process reveal a faster crystallization for boron-doped samples, and a slower crystallization for phosphorous-doped samples, when compared to the crystallization of undoped a Si:H thin films. The peculiarities of B K XES spectra (and observed changes upon SPC) indicate that boron could act as a nucleation center promoting crystallization. Si L2,3 XES spectra of a-Si:H and P-doped poly-Si exhibit spectral features above the valence band maximum at 100 eV that could be attributed to a-Si defect states and n+-dopant states, respectively. The SPC crystallization velocity of Si thin films on ZnO:Al/glass is found to be faster than that on SiNx/glass substrate. Multiple indications for oxidization at the poly-Si/ZnO:Al interface are found based on our Si L2,3 XES analysis. Spatially resolved x-ray photoelectron spectroscopy data support this and even suggest the formation of sub-oxides or zinc silicate as an interface species. The electronic structure of the buried a-SiOx:H(B)/ZnO:Al and µc-Si:H(B)/ZnO:Al interfaces are unraveled with “depth resolved” hard x-ray photoelectron spectroscopy. A surface band bending limited to the very surface of the silicon layers is found. The valence band maxima for the Si cover layers and the ZnO:Al TCO are determined and interface induced band bending for both interfaces are derived. At the a-SiOx:H(B)/ZnO:Al interface a tunnel barrier of (0.22 ± 0.31) eV and at µc-Si:H(B)/ZnO:Al interface a tunnel barrier of (-0.08 ± 0.31) eV is determined. This explains a previously empirically found solar cell efficiency increase produced by introducing a µc-Si:H(B) buffer layer between an a-Si p-i-n cell and the ZnO:Al/glass substrate.