Ferroelectric materials offer a broad range of application-relevant properties, including spontaneous polarization switchable by electric field. Archetypical representatives of this class of materials are perovskites, currently used in applications ranging from sensors and actuators to information storage in nonvolatile memories. The latter application poses significant technological challenges because perovskites cannot be easily integrated in CMOS fabrication. The two main issues are a high annealing budget required to obtain functional films and the instability of ferroelectricity at the nanoscale that prevents aggressive scaling. In this context, the discovery a decade ago of ferroelectricity in hafnia (HfO2) thin films was groundbreaking: they are readily employed in CMOS processing and were shown to retain their properties down to 5 nm and lower, overcoming the main limitations of perovskites and making them ideal candidates for low-power ferroelectric devices. Research then focused on understanding the physics underlying polarization response of these materials while developing reliable fabrication process for smooth integration in logic and memory devices. This thesis contributes to CMOS-compatible processing of ferroelectric hafnia and addresses intriguing and partly controversial aspects of its switching properties. The majority of HfO2 films are grown by Atomic Layer Deposition (ALD), a mature but slow and not so flexible technique. The layers are amorphous and need high thermal budget annealing to become ferroelectric. This work proposes an alternative Pulsed Laser Deposition (PLD) fabrication route for ferroelectric hafnia thin films with conventional TiN electrodes and very low thermal budget, well within the requirements for CMOS compatibility. The superior flexibility of PLD is exploited to obtain already partially crystallized films that only need low thermal budget annealing to become ferroelectric. Besides the development of a CMOS-compatible fabrication route, detailed investigation of ferroelectricity is performed by off-resonance Piezoresponse Force Microscopy (PFM). Optimization of this technique allowed probing extremely weak response, with vertical resolution below 0.1 pm. This is accomplished without having to rely on resonance amplification, thus avoiding possible related artifacts. The same technique enabled to obtain reliable PFM through a top electrode and dielectric layer. Ferroelectricity in undoped HfO2 is explored, giving insight into phase evolution at nanometer scale during the wake-up process. A study on ultrathin HZO demonstrated the potential of such films for future generations of memory devices with mV-range operation. Comparing HZO films of different thickness, an anomalous scaling of coercive field is noted: a new nucleation model is proposed to interpret this, with important practical implications. Finally, the developed methodology is employed in the analysis of ferroelectric/dielectric bilayers used for Negative Capacitance (NC) applications, offering an alternative look to the debate on intrinsic or nucleation-driven switching. In this thesis, technological challenges of this new class of ferroelectrics are taken as starting point to propose an alternative fabrication method and explore the physics behind ferroelectricity in thin films. The results reveal the fascinating complexity of hafnia ferroelectrics and the enormous potential that they hold for next-generation applications.