Theses 2018

Entangled Photons Recombination from Broadband Parametric Down-Conversion

Conclusions: In this thesis, a study on the dependence of the peak-to-background ratio in the SFG spectrum on the number of frequency modes in PDC has been carried out. Particularly, a theoretical analysis for the optical elements to be utilized in the experimental setup has been performed. For example, the most suitable nonlinear crystal to generate the broadest PDC spectral bandwidth with the higher parametric gain. As a result of this analysis, BBO crystal with 5mm length has been chosen. Also the optimal material of the prisms has been chosen with the most practical geometry of the dispersion compensator to compensate for all the positive GVD in PCD.

Moreover, an experimental setup has been built to study the dependence of the peak-to-background ration on the GVD compensation and the number of frequency modes. The most important results are summarized as follows:

  • An efficient pump reconstruction has been observed. The highest measured peak-to-background ration is 20 +/-1.0 for pump power 200 mW using a high resolution monochromator. In addition, the spectral bandwith of the peak coincides with the pump bandwidth.
  • The peak-to-background ration demonstrates a high dependence on the time delay, which is caused by the GVD, between signal and idler beams. The peak-to-background ratio increases when the GVD is compensated and it reaches its maximum with zero GVD. Positive and negative GVD have the same effect on the peak-to-background ratio.
  • The thesis demonstrates the dependence of the measured peak-to-background ratio on the number of the frequency modes in PDC. The peak-to-background ratio increases with increasing the number of the frequency modes in the PDC, similarly to the Fedorov ratio, which is commonly used operational measure of entanglement. Such dependence reveals a new method for the PDC entanglement characterization.

Finally, the experimental setup can be further improved using spatial light modulator (SLM) with spectral-physe holograms to compensate for the higher order dispersion or to modulate the phases of the signal and idler and study their effect on the reconstructed pump spectrum and the peak-to-background ratio.

On the Investigation of the Reducction of Phase Singularities in Speckles Interferometry

Summary: In this thesis, the following conclusions are confirmed:

  1. The number of phase singularities reduction was achieved due to the incoherent averaging of the speckle fields using an extended light source in Michelson interferometer. Those phase singularities are undefined phase points that join together creating phase edges, preventing successful unwrapping for the required phase map. The reduction mechanism shows an interesting complexity. To investigate that complexity, a hybrid setup between Michelson and Mach-Zehnder is built in order to separate the reference arm from the object arm that allows enlarging the speckles size without affecting any required reference points.
  2. The experiment has confirmed the simulation results, that there are two main effects of the reduction mechanism:
  • The speckle get smaller proportional of the adding of different incoherent point sources, which in effect reduces the number of averaged phase edges and phase singularities.
  • The speckles structure get more ordered, that seems to raise the probability of two phase edges to be at the same location in the phase maps before and after deformation measurements, which as a result, reduces the phase singularities drastically at the attained phase deformation measurements. 

Summary: The obtained results from chapter 5 are summarised in Table 14, 15 and Figure 27. The results are sorted according to the used machine learning technique and data processing method, where “aup” is the abbreviation for the areas under the peak, and “ratio aup” stands for the ration of the areas under the peak.

The best sensitivity result for the nerve tissue classification is obtained by the one-dimensional CNN model, tje sensitivity is 91.9%. Muscle discrimination is relatively well done by all the models and data processing techniques. The best sensitivity for fat discrimination is obtained by LDA with the PCA model.

According to the results, despite the fact that LDA is the only linear algorithm used, it shows the best on average performance whose F1-score is equal to 95.3%. The deep learning methods on certain validation sets show high F1-scores. The highest F1-score is achieved by the DNN method with PCA data pre-processing on the 5th validation set.

All classification techniques demonstrate lower nerve and fat tissue classification accuracies than muscle. Low differentiation performance of fat and nerve are also mentioned in the literature.

The table also demonstrate that the LDA and DNN with PCA based algorithms show better results than the other data processing techniques. Meanwhile, 2D and 1D CNNs perform better with ‘ratio of the area under the peak’ data processing. CNN with PCA did not perform well, because two-dimensional spatial information is lost during eigenvalue decomposition in PCA. Unfortunately, despite fine parameters tuning, the DNN algorithm with ration aup could not produce any sensible results, therefor the results are not include into the table.

Polarization Tailored Raman Spectroscopy

Conclusion and outlook: In this thesis, we successfully demonstrated a novel measurement scheme, namely polarization tailored Raman spectroscopy (PTRS). We used a tightly focused radially polarized beam, and a tightly focused azimuthally polarized beam to excite phonons within a GaN pillar of sub-wavelength dimensions.

Previously, z-sensitive Raman spectra had only been accessible by rotating the substrate when working with linearly polarized excitation beams. The strong longitudinal electric field component, present at the optical axis in the focal field distributions of the tightly focused radially polarized beam, enabled us to measure those spectra in a reflection configuration under normal incidence, relative to the substrate. The pillar under investigation exhibited Raman spectra, predominantly resulting from the longitudinal electric field component when placed on the optical axis of the tightly focused radially polarized beam. The pillar’s diameter was sufficiently small (d::::: 166nm), to yield a relatively weak contribution ofthe lateral field components to the Raman spectra. Moreover, we observed a strong dependence ofthe Raman spectra on the lateral position of the GaN pillar relative to the optical axis for both excitation beams used in this thesis. The presented measurements can be considered a successful proof of principle.

Strang features, which are in compliance with the theoretical predictions, could be observed in the changes of the relative Raman intensities for lateral displacements. Raman spectra were recorded at different lateral positions along a line, as well as in the form of a 2D-scan for the tightly focused radially polarized beam. The Raman line scan conducted with a tightly focused azimuthally polarized beam shows strong influence of a magnetic multipole mode in the particle under investigation. Additional Raman peaks have been observed that may be explained by quasi-TO phonons and surface phonons upon further evaluation.

A tentative fitting scheme for the extraction of the Raman tensor elements of the structure under investigation was proposed. The method is based on a least squares minimization of the difference of the calculated Raman intensity ratios resulting from FDTD simulations and the measured Raman intensity ratios. In the present thesis this method yielded diverging results, because too many parameters were unknown. In combination with thorough characterization of both the exciting field distribution and the geometry ofthe particle under investigation, this evaluation technique could reveal basic material properties ofthe investigated structure.

Possible follow up experiments could include measurements with further excitation beams. Tightly focusing a Hermite-Gaussian TEM 10 mode would lead to interesting field distributions for PTRS. In these focal field distributions, there is a node line along which no lateral electric field is defined. On this line,the z-component exhibits a maximum on the optical axis, similar to the field distribution in the focal plane of the tightly focused radially polarized beam [11].

Examining further GaN pillar samples ofvarying sizes could reveal more details about the lower Raman shifts that were obtained within this thesis. The technique could also be applied to pillars lying on the substrate, by either choosing sufficiently small pillars or suitable excitation beams. Rotating the polarization of linearly polarized light could already yield valuable insight in the phonon modes in a lying pillar. One could also aim to selectively excite particular modes in the structures, similar to reference [12] .

The effects on the Raman spectra resulting from selective excitation by tailoring the excitation fields to the resonances supported by the nanospecimen could become another topic for future examinations.

The selective excitation of higher order multi pole modes in a lying pillar could for example reveal the lattice properties ofthe pillar at different positions along its axis.

Moreover, the interaction of crystal phonons with light carrying optical angular momentum could be further investigated. The application of PTRS could be especially interesting in combination with specimens exhibiting Raman optical activity [38] . In this case, the transverse angular momentum carried by the tightly focused radially polarized beam could be used to observe the effects in more convenient measurement configurations, just like the longitudinal electric field enabled us to do in this thesis.

Furthermore, the polarization tailored Raman spectroscopy technique demonstrated in this thesis could be extended by combining it with other enhancement approaches for Raman spectroscopy. Roy et al. used a radially polarized beam for the excitation of a plasmonic tip to enhance Raman spectra [91].

Similarly, one could place a plasmonic nanostructure, i.e. a gold split ring resonator with well known properties, close to the structure under investigation and control the near fields around the Raman specimen by selectively exciting different modes in the plasmonic structure. The selective excitation of dark modes by tailored light could also be considered within the approach of plasmonic nanostructure enhanced Raman spectroscopy.

Finally, incorporating a spatial and angular polarization analysis of Raman signals into the measurement scheme may reveal new insight in the characteristic properties of the Raman effect.

Conclusons and outlook: The aim of the master project was the detection of quadrature squeezing of a broadband optical state. So far, standard techniques to extract the quadrature information of an optical state rely on detectors that are too slow to resolve the quadrature oscillation, if the state has a broadband spectrum. The detectors are limited to GHz bandwidth, but optical states can provide spectra with bandwidth of 150 THz. Nonlinear interferometry can be used to close the gap between the bandwidth limitation of detectors and the available bandwidth in light fields.
In this project, twin-beams were generated via high-gain parametric down-conversion in an aperiodically poled LiNbO3 crystal. A technique is used where such a crystal (source) is  followed by another crystal (analyzer), forming a nonlinear interferometer. Efficient detection of quadrature squeezing is achieved by unbalancing source and analyzer gain values. The experimental work with nonlinear interferometer reveals both potential and limitations of this technique. The main conclusions of the presented experiments are:

  • Observation of squeezing in broadband configuration. Two nonlinear interferometer configurations with aperiodically poled crystals as analyzers were tested. Squeezing was
    observed in both configurations. However, the detected interference spectra do not coincide with the theoretical predictions in both cases. The disagreement between measurement and theory is subject of ongoing investigations.
  • Observation of squeezing in narrowband configuration. In a third nonlinear interferometer configuration a periodically poled crystal acts as the analyzer and the unbalancing
    of source and analyzer gain is achieved. Squeezing of 2.2 dB is observed in this configuration and the period of the fringe agrees well with the theory. The disadvantage of the
    configuration is the narrow bandwidth of the analyser that allows one to detect only one interference fringe.

The results leave open the influence of the temporal walk-off on the fringe period. Unbalancing of the source and the analyzer gain values with an aperiodically poled crystal as analyzer should be implemented next. Two suggestions on how the unbalancing of the gain values can be achieved are:

  • A tighter focusing of the pump beam into the nonlinear interferometer results in a decreased Rayleigh length. Moving the analyzer into the beam waist and the source out of the waist could be sufficient to unbalance the gain values.
  • The parametric down-conversion process only takes place if the pump polarization is correct. For a further attempt, one could try to achieve different pump polarizations in the source and in the analyzer to tailor the pump power in each crystal.

In this Thesis, the detection of quadrature squeezing is demonstrated with unbalanced nonlinear interferometry. Due to the inherent unbalancing of source and analyser gain values, the
validity of the unbalanced gain technique is verified in the last described nonlinear interferometer configuration, but only for a narrowband analyzer spectrum. In the case of an aperiodically poled analyzer crystal squeezing is also observed, but the interference spectra do not agree with the theoretical predictions.

Thermal and Mass Diffusivities of Linear Alcohols Containing Dissovled Gases by Dynamic Light Scattering

Abstract: In the present work, simultaneous measurements of thermal diffusivities and Fick diffusion coefficients in the liquid phase of binary mixtures of primary alcohols containing dissovled gases were performed by dynamic light scattering. The total average expanded uncertainties (k=2) were 14.2 for thermal and 6.8 for Fick diffusivities. By the investigation of 1-hexanol or 1-decanol with H2, He, N2 or CO at infinite dilution of the gas between 303 K and 423 K, the dependences of thermal and Fick diffusivities on temperature and on solute as well as solvent were studied. The thermal diffusivity was found to decrease with increasing temperature and agrees for all solutes with values calculated from literature for the pure solvents. The Fick diffusion coefficient shows Arrhenius-like behaviour with temperature and is considerably larger for the systems with light gases compared to the same systems with N2 or CO. It was observed that the Fick diffusion coefficient is higher for the He than for H2, despite its larger molar mass, while no considerable difference between N2 and CO was found over a wide temperature range. Thermal diffusivities are larger in 1-decanol than in 1-hexanol, while the opposite is the case for the Fick diffusion coefficient. In the binary mixture of 1-hexanol with dissolved CO2, thermal and Fick diffusivities were measured over a wide range of concentrations at different temperatures between 303 K and 353 K. The Fick diffusion coefficient shows a strongly non-ideal behaviour, which is expressed by a maximum around 20 mol% of CO2, followed by a pronounced decrease with increasing concentration of CO2. Due to the experimental limitations, a minimum expected between 75 and 95 mol% of CO2 could not be resolved. The method of polarization difference Raman spectroscopy was applied to experimentally determine the mixture concentration. The determination of CO2 molar fractions has an estimated uncertainty (k=2) of 12% and the comparison with literature data yields an average absolute relative deviation of 3.8%.

Investigation of SRAFs in EUV Lithography for Alternative Absorber Materials

Optical lithography is one step in the process for fabricating integrated circuits. This technique transfers patterns from a photomask onto a photoresist atop a semiconductor wafer. Extreme Ultraviolet Lithography is an emerging technology set to overtake Deep Ultraviolet Lithography as the dominant commercial method for this task. The reduced wavelength of light used in EUV allows printing of much smaller features with fewer and less complicated processing steps than its predecessor technology. Unfortunately, the 13.5 nm wavelength of light requires the use of a reflective system, rather than a transmissive one, which gives rise to a number of new effects unique to EUV. As such, many of the resolution enhancement techniques developed for DUV lithography must be re-evaluated in these new systems. This work investigates sub-resolution assist features (SRAF) in EUV lithography in the form of line-space-patterns for four absorber materials: Ta, to represent current commercial masks; RuTe, a low-absorption material with good refractive index contrast; Ni, a high-absorption material; and AttPSM, an idealized attenuated PSM material with 6% reflectivity and 180° phase shift. These investigations consider a high-NA=0.55 system with anamorphic reduction of 4 in x and 8 in y with a 20% central obscuration. Optimization methods are used to find the optimal SRAF sizing, position, and main feature bias for a three

– process window overlap of a 20 nm dense mask with 40 nm and 60 nm semi-dense masks with a target CD of 10 nm. SRAFs were found to be around 6 nm in width with slight asymmetric placement. Known phase effects induced by SRAFs are evaluated through diffraction analysis and followed by a Zernike analysis to observe the impact on image formation. Just as in DUV, SRAFs in EUV lithography redistribute energy of the diffraction orders to better represent a dense structure on the mask. The presence of phase shifts arising from SRAFs is confirmed and shown to be stronger for the less-absorbent materials. The Zernike analysis is limited in accuracy due to the small pitches of the investigated masks, but shows considerable influence arising from SRAF size and placement on aberrations present in the system. The source of these effects is believed to arise predominately from the absorber component of the mask. An evaluation of separate mask components through the use of a “hybrid” mask model simulation is performed to evaluate this. Additional investigations on illumination evaluate the effect of source area on the best focus, normalized image log slope, and telecentricity error with small variations in SRAF width, as well as a broader observation on illumination direction by mapping individual source points across the pupil. Two investigations beyond the classical analysis look into variable SRAF depth, rather than width, and the potential of CD-retargeting as a technique to replace SRAFs in EUV Lithography.

All Solid-State Spectral Broadenin and Temporal Compression of Femtosecond Pulses from a Ytterbium Doped Thin-disk Laser

Conclusions: In this thesis a recently proposed method for spectral broadening in bulk material is further investigated with focus on pulse compression after broadening stage. The laser source is a Ytterbium doped thin-disk laser with 35J.LJ pulse energy, repetition rate of 100 KHz and initial pulse duration of about 300 fs . Along with many available approaches to spectral broadening such as gas-filled hollow-core fibers, the multi-plat e glass setup is a cheap and robust solution for broadening. This approach allows us to exploit the air gaps between the plates to avoid self-focusing and beam collapse inside the material. In order to choose the best candidate, fused silica and BK7 plates were examined with more emphasis on the amount of broadening and durability. The broadening stage is followed by a prism compressor and the beam diagnostic consist ing of a Frequency Resolved Optical Gating (FROG) setup and an Optical Spectrum Analyzer (OSA). With the compressor a pulse duration of 52 fs can be achieved, which is close to the transform limit of 42 fs. We also performed numerical simulations in order to investigate the dynamics of such a nonlinear system.

Abstract: As laser beam melting technology is getting adopted for serial production of parts it is necessary to ensure the quality of the finished parts. Coaxial melt pool monitoring is been considered as a tool in the scope of the thesis, experiments to study directional, positional and layer height influcence on melt pool signal has been discussed. Also in the second half, support structure design and scan parameter influence on roughness along transition zuone and impact of remelting in reducing it has been discussed.

Weak Measurement Enhanced Localization of Nanoparticles Using Structured Light

Abstract: In this work we elaborate on a position sensing technique based on far field polarization measurements. In particular, we apply the weak measurement ansatz to the light scattered by a single nano-antenna. We excite gold nanoparticles resonantly using a polarization tailored focused light beam. The dipole moment of the nanoparticle depends on the position of the particle within the focal field. To determine the dipole moment, we investigate the scattered light in the far field. Moreover, to be able to detect minor changes of the dipole moment we project the polarization onto a Cartesian basis and mix TE- and TM-components of the scattered light in a suited ratio. This postselection is similar to that used in weak measurements. To proof the concept, we investigate different ratios of TE- and TM- components and compare our experimental results with a theoretical model.

Application of Deep Learning Algorithms for Lithographic Mask Characterization

Abstract: Optical Lithography is a technique used to transfer patterns from a given photomask to a photoresist on top of a semiconductor wafer. One of the key challenges in lithographic printing is the appearance of defects on the photomask. Printable defects affect the lithographic process by causing errors in both the phase and magnitude of the light and of the sizes and location of the printed features. Since it is not yet possible to produce defect-free masks, methods to inspect and repair mask defects play a significant role. This master thesis proposes and investigates the application of Convolutional Neural Networks (CNNs) to characterize and classify defects on lithographic masks. CNN as one of the algorithms in deep learning have achieved good results in image classification for other problems in the past.

The simulation software Dr.LiTHO of Fraunhofer IISB is used to simulate aerial images of defect-free masks and of masks with different types and locations of defects. Specifically we compute images of regular arrays of 38 nm and 25 nm wide squares to be imaged with typical settings of EUV lithography (l = 13.5 nm, NA = 0.33). Only absorber defects on the mask are studied. We simulated 5 types of defects (extrusion, intrusion, oversize, undersize and center spot).

Depending on the position of the defects, extrusion and intrusion are further classified in to 8 types which results in a total of 19 classes for the classification. The final architecture of the CNN contains 5 convolutional layers (conv. layers), where most of them are followed by a max-pooling layer. A mixed size of  filters is used for the conv. layers (3 3) and (5 5). The convolution stride is fixed to 1 pixel. The spatial padding of conv. layer input is such that the spatial resolution is preserved after convolution, i.e. the padding is 1 pixel for all conv. layers. Two separate networks are trained for detection of the defect types and location. Another algorithm is used to combine the results from the two networks.

An accuracy of 99.9% on the training set and 99.3% on the test set is achieved for detection of the defect type. The network trained for location detection results in 98.7% training accuracy and 98.0% for the test set. The performance of the models is also measured with other performance measures like confusion matrix, precision and recall. The robustness of the models is studied by systematically removing images of certain defect sizes from the training set. Moreover, we investigate the relation between defect sizes and the accuracy of the models. The defect detection model can predict the types of defects with a size of 5 nm or 0.12l NA , which is well below the classical resolution limit. The location detection model shows a slightly different behavior for the different dataset of images. It predicts location of a defect above the size of 4 nm for 25 nm features and of 6 nm for 38 nm features.

A Polyvinyl Chloride Plastisol (PVCp) based optofluidic Lab-on-a-chip Device to Mimic the Optical and Accoustic Properties of Vascularized Human Tissue

Conclusions: During the research work, the PVCp material was chosen to be the matrix material of the optofluidic chip due to its intrinsic characteristics such as insoluble in water and stable for a long time. Although it also forms the albinism when immersing in water, this phenomenon can be healed in 3 hours. An optofluidic chip composed of PVCp sub- and superstrate with the defined thickness and a hollow microfluidic structure embedded is able to achieve the similar morphology characteristics of tissues especially the human cutis including the two layered and the microvasculature structure. With the application of the SIMDOS diaphragm pump assigned and the SDOCT imaging, the vasculature geometry and micro-circulation of the fabricated chip are reconstructed properly as well as the lamination between layers are also investigated.

To achieve the similar optical properties of human cutaneous tissues, different amounts of Titanium Oxide powder and ink was inserted in the PVCp matrix and tune their absorption coefficient μα and reduced scattering coefficient μς which cover the properties of tissues mostly in the visible and infrared region.

Moreover, by inserting different concentrations of the softener in the PVCp matrix, its similar Young’s modules E and the acoustic velocity with that of soft tissues was almost achieved and the results here indicate that the experimental values of Young’s modulus almost match with that of soft tissue and the tendency of acoustic speed variated with the softener concentration can validate its dependency on the mechanical elasticity.

Addition informations