Abstract
The development of high-speed demanding data systems and the growing use of wide-bandwidth requiring applications nowadays make the data traffic more and more severe. One promising avenue to solve this problem is optical signal processing, taking advantage of its large time-bandwidth capacities and the possibility to process signals at the line rate of fiber optic communications. Meanwhile, the improvement in photonic integrated circuits (PICs) and materials with high nonlinearities make it possible to deploy optical signal processing in modern optics. Nonlinear optics is one of the most successful outcomes of modern optics. When light with sufficient intensity is injected into a medium, it responds to the optical intensity in a nonlinear way, where light components with new frequencies are generated and coupled with each other. As a result, various nonlinear optical phenomena arise, forming the domain of nonlinear optics. The significance of nonlinear optics lies in the advanced optical processing that it enables and has yielded numerous applications, including wavelength conversion, optical parametric amplification, frequency comb generation, and supercontinuum generation. It’s therefore highly useful for enabling the implementation of a wide range of all-optical processes on a chip. In nonlinear optics, material platforms with a large nonlinear figure of merit (??????????=??2(??TPA·??)/) and CMOS-compatibility are highly sought after, for the benefits of enhanced efficiency of nonlinear optical processes, low-cost fabrication, and easy integration. vi In this thesis, nonlinear integrated optical devices are designed and implemented on the Ultra-Silicon-Rich Nitride (USRN: Si7N3) platform, which combines the advantages of both Si and Si3N4 while minimizing their drawbacks. Compared with Si3N4, USRN has a larger linear refractive index (3.1), a higher Kerr nonlinearity (2.8×10-13 cm2/W), and doesn’t suffer from TPA losses. Therefore, it’s an ideal platform for nonlinear optical applications, such as supercontinuum generation (SCG) and high optical parametric gains. Based on the USRN platform, we first propose a new way for laser spectral manipulation in a 2-stage device, consisting of a 3-mm-long cladding-modulated Bragg grating (CMBG) and a concatenated 7-mm-long channel waveguide. By operating at frequencies close to the grating stopband edge, we access strong grating-induced dispersion, while maintaining low losses and high transmissivity. We further exploit the red-shift in the Bragg grating stopband due to the thermo-optic effect to achieve tunable dispersion, leading to varying degrees of spectral broadening. We observe an increase in the bandwidth of the output pulse spectrum from 69 nm to 106 nm as temperature decreases from 70 ? to 25 ?. This approach provides a new avenue to achieve on-chip laser spectral tuning without loss in pulse energy. Then we demonstrate the SCG with high coherence in a USRN waveguide using 500 fs, 17 pJ pulses. The generated supercontinuum spans from 1006 nm to 2240 nm at the –40 dB level and is characterized to possess an average |g12| exceeding 0.90 across the wavelength range from 1260 nm to 1700 nm. Compared with other on-chip SCG, this work relaxes the pulse width requirement from tens of femtoseconds to 500 fs, while requiring low pulse energies. vii Lastly, we study enhanced SCG in a 2-stage USRN device in this thesis. The spectral enhancement is strongly dependent on the Bragg-soliton effect temporal compression occurring in the CMBG stage which is linked to both device and input pulse parameters. Through numerical simulation, we obtain the optimal device design, including a CMBG length of 0.5 mm, a waveguide width of 600 nm, a waveguide height of 300 nm, a waveguide length of 6 mm, together with the optimal pulse conditions, including a pulse width of 1.9 ps and a pulse peak power of 4.4 W. We fabricate the 2-stage device with the optimal design and perform the spectral broadening using pulses with the optimal conditions. A supercontinuum with a bandwidth of 610 nm at –30dB level is experimentally demonstrated in the 2-stage USRN device, representing a 5× enhancement compared to that in a reference waveguide. viii