Silicon integrated devices for quantum photonics in the telecom band

Quantum information can provide many advantages by exploiting quantum mechanics properties. The qubit is the fundamental unit of information, and its superposition and entanglement properties are responsible of its advantage over classical information. Different systems are studied to implement the qubit, but photons seem the best candidates for developing quantum communication protocols. For realizing quantum applications in photonics, nonclassical states of light (single-photon or entangled-photon states) have to be generated. The first realization of such sources made use of bulk optical components, but the need of building scalable, compact and integration-compatible implementations has recently grown. So, the silicon platform has become one of the leading technologies in photonics for its compatibility to the CMOS fabrication process. The goal of this thesis is to study silicon integrated nonlinear optical components and characterize their ability of emitting photons through four-wave mixing (FWM). In particular, two main problems are addressed: the assessment of the efficacy of integrated filters on spectrally clean the generated photons and the lack of laser emission in silicon. Then, we also focus on a topic related to quantum computing, that describes a method to emulate quantum gates by using classical waves. First, we study the case of BWs, constructed by periodically shrinking the width of a waveguide. Usually FWM emission is 9-10 orders of magnitude smaller than the intensity of the pump and the emitted photons are symmetrically located around it. So, the generated photons have to be spectrally cleaned from the pump with 100-dB rejection filters. All the studied solutions are made of several microns of silicon waveguide and can thus generate unwanted photon pairs that could pollute the quantum state at the output. We, then, perform a stimulated FWM experiment on a BW by keeping the signal’s frequency fixed and varying the pump’s frequency in order to probe the FWM process across the BW stopband. By calculating the generation rate that would be obtained by spontaneous FWM, we find that, at the bottom of the stopband, it would be of few Hz. The spontaneous generation rate is thus at least five order of magnitudes smaller than that of the most efficient silicon sources, thus ruling out the possibility of the introduction of spurious photon pairs by integrated filters. Second, we propose a way of bypassing the need of an external laser acting as optical pump in silicon chips, by designing a fiber-loop cavity with the silicon integrated source of light (ring resonator) inside of it and an external amplifier to provide gain. One ring resonance is chosen as pump frequency and laser emission is achieved with sufficient power to observe FWM emission. First, we employ a low-Q ring and stimulated FWM is observed. Then a JSD measurement is performed, suggesting strong correlations between the emitted photons. Second, a high-Q ring is introduced and spontaneous FWM is observed. A coincidence measurement then shows that signal and idler are emitted at the same time. Finally, we show that all quantum computing protocols can be realized by propagating a single particle in a linear network as long as one is interested only in detection probabilities at individual outputs. We prove this point by implementing the BV algorithm in a three-qubit configuration in an analog electronic circuit and show that any quantum information protocol can be realized in a classical-wave network at the price of an exponential increase of the employed resources with the number of qubits. However, in the BV algorithm the qubits are never entangled. So, we show that the BV algorithm can be emulated in an electronic circuit where the number of resources scales linearly with the number of qubits, showing that the strength of quantum information lies in entanglement.