Single-photon emitters are at the heart of quantum optics and photonic quantum-information technologies. Identifying and characterizing such quantum emitters requires sophisticated experimental physics. In this experimental master end project, we aim to design and build an optical setup that allows for the characterization of single photon sources. More specifically, we aim to build a setup that allows for photoluminescence (PL) spectroscopy and second-order autocorrelation (g(2)) measurements of single-photon emission (SPE).
The initial design of the optical setup is based on that of a micro-photoluminescence setup and optimized for single-photon emission from indium arsenide phosphide (InAsP) quantum dots (QDs). This design therefore includes a continuous-flow liquid-helium cryostat, and a high resolution monochromator that can be used to extract the emission line of a single QD transition.
The final version of the experimental setup is demonstrated to be able to perform high resolution PL spectroscopy measurements. g(2) Measurements are performed on room temperature hexagonal Boron Nitride (hBN) SPE using a free-space Hanbury-Brown and Twiss (HBT) interferometer containing two geiger-mode avalanche photodiodes (APDs) indicating pronounced photon antibunching. Due to malfunctioning of the cryostat, however, measurements have exclusively been performed at room temperature.

As demand for chips increases and critical dimension keeps shrinking, the inspection of wafer becomes one of the critical challenges in the high volume production of chips. Coherent Fourier scatterometry (CFS) is a scanning bright field technique that is capable to detect nanoparticles on Si surfaces. The optical readout of CFS consists of a two pixel split detector that gives a voltage signal based on the intensity difference between the two halves. While an area of the wafer is raster scanned, the flat surface with no particles gives a zero voltage and on the particle an asymmetrical (non-zero) signal will be detected.

In this thesis, we demonstrate the use of synthetic optical holography (SOH) as a new method to improve the sensitivity of CFS. By adding a reference mirror with a piezo stage to the setup, we can interfere the scattered field with a plane reference wave. And by moving the mirror in steps, we can change the phase of the reference wave for every line of the raster scan, such that the 2D scan represents a digital off-axis hologram. Applying the standard digital holography reconstruction process, we retrieve a signal that equals the complex far field difference scaled by the reference field amplitude.

Overall, we see consistent signal-to-noise ratio (SNR) improvement over the conventional CFS. For the detection of a polystyrene latex (PSL) particle with a diameter of 60 nm (λ/10) on a silicon wafer, this new implementation leads to a SNR of 10 dB, which is about 4 dB over the filtered conventional CFS signal. For measurements of a dust particle using a very low amount of incident power (0.0014 mW on wafer), a SNR gain of more than 6 dB is achieved compared to filtered conventional CFS, due to the attenuation of low frequency electronic noises. Therefore, the implementation of SOH improves the sensitivity for detecting small nanoparticles and allows low power applications, such as biological imaging.

Superconducting nanowire single-photon detectors (SNSPDs) are characterized by their quantum limited ability to accurately detect single photons, with low jitter, high detection efficiency and low dark count rate. To achieve this, the detector is cooled to 2-3K, bringing the device in a superconductive state, and is then biased with a direct current (DC) close to its critical current. When a photon impinges the detector, the depairing of Cooper-pairs by the photon leads to local destruction of the superconductivity. The growth of this non-superconducting area, first across and then along the nanowire, leads to the development of a measurable resistance and hence the production of detection pulses. Increasing system detection efficiency (SDE) of detectors has been a long-term goal in the community. Recently ultrahigh efficiency detectors (SDE>98%) have been demonstrated. It has also been shown that the wavelengths dependence of SDE, typically defined by a quarter wavelength cavity, is modulated by fiber-detector airgap (Fabry–Pérot). Measuring such modulations and finding the optimal operation wavelength manually is a time consuming and tedious process. In this thesis a setup for automatic measurement of SDE versus wavelength was developed and benchmarked. For the tested detector, efficiencies were found ranging between 22% and 95% in the wavelengths range between 1260nm and 1650nm. For the optical circuit using Single Mode (SM) fibers only, the automated SDE measurements were unreliable due to shifts in polarisation during measurements. Using PM fibers led to efficiencies very similar to the values measured manually.

This thesis discusses the application of deep learning to Coherent Fourier
Scatterometry data in order to quickly and reliably detect nanoparticles on
surfaces. An introduction to deep learning is followed by a review of the
experimental setup and used software. After that, results are presented of
classification accuracy tests on various datasets containing images obtained
from scatterometry scans. We show that a relatively simple convolutional
neural network can achieve an accuracy as high as 98% on a 200 image test
set. We compare this to the accuracy of a non-deep learning, clustering
based classification algorithm and conclude that deep learning is a more
suitable method for particle classification. Then, three methods of open set
recognition are applied. We show that it is possible to reject 80% of a fooling
dataset at the cost of rejecting 10% of the normal data. Finally, the results
are discussed and placed in the context of future work on this subject.

Single-photon detection is extremely important for a number of different applications such as quantum cryptography, CMOS testing and even biomedical research. Most of the applications of single-photon detector require high efficiency combined with high time resolution, high count rates and low dark counts. Superconducting nanowire single-photon detectors has emerged provides this combination unlike any other available single-photon detectors. Some of the applications of superconducting nanowire single-photon detectors (SNSPD's) require larger area SNSPD's without affecting it's performance. For this purpose, multi-pixel SNSPD's would provide a solution. To this end, prototype 4- and 16-pixel detectors have been characterised in terms of a number of performance parameters such as critical current, count rate behaviour and timing jitter. Aside from this, simulations have been performed on detection statistics and relative illumination of the individual pixels of a multi-pixel SNSPD. Results show that for 4-pixel detectors, all desired aspects have been achieved, but not yet combined in one detector. The 16-pixel detectors showed excellent critical currents and count rate behaviour but, most likely due to the fact that they were prototype chips with little to no protection, had several not functioning pixels.

For wavefront reconstruction, often the combination of a Shack-Hartmann sensor and a reconstruction method utilizing the Cartesian derivatives of Zernike polynomials (the least-squares method) is used, which is known to introduce cross-talk. In Janssen (2014) a new wavefront reconstruction algorithm is introduced to be used with the Shack-Hartmann sensor that does not present cross-talk. To our knowledge, this method has never been demonstrated. In the current research, Janssen's method and the conventional least-squares method are compared on a modified Michelson interferometer setup with a spatial light modulator to first remove the aberrations of the complete system, and subsequently introduce specific aberrations for which the Zernike coefficients are reconstructed using both reconstruction methods. It is found that both methods work equally well with optimal fitting powers. When fitting fewer than the optimal amount of fitting powers, it was found that Janssen's method accurately recovered Zernike coeficient amn when n + 1 powers are fit, while the conventional method achieves the same accuracy only at n + 2 fitting powers, especially when aberrations introducing cross-talk are present. When more than the optimal amount Zernike powers are used, it is seen that Janssen's method presents aliasing at lower fitting powers. At the optimal fitting power, the RMS landscape for both methods is shown to be similar, and therefore they are equally susceptible to errors due to mis- estimation of the center position and radius of the beam on the ShackHartmann
sensor. Lastly, it is shown that theoretically, Janssen's method is at least 3.5 times slower than the conventional least-squares method due to the use of complex valued Zernike polynomials.

The goal of the research is to find out whether it is possible to measure a wavefront qualitatively and quantitatively with Janssen’s recently developed analytical algorithm. The spot displacements given by the wavefront sensor are used to relate the derivatives of the wavefront with respect to x and y with the Zernike coefficients of the wavefront itself. The results are compared with some conventional and commercially used methods: an iterative integration method of the SHS, which have been implemented in a Matlab script. Moreover, some of them are compared with the reconstructed wavefront obtained through interferometry. Janssen's method turns out to be a reliable and fast (regarding computation time) wavefront reconstruction method to qualify which aberrations are present.