Modern communication devices demand challenging specifications in terms of miniaturization, performance, power consumption and cost. Every new generation of radio frequency integrated circuits (RF-ICs) offer better functionality at reduced size, power consumption and cost per device and per integrated function. Passive devices (resistors, inductors, capacitors, antennas and transmission lines) represent an important part of the cost and size of RF circuits. These components have not evolved at the same level of the transistor devices, especially because their performance is strongly degenerated when they scale down in size. The low resistivity silicon used to build the transistors also imposes prohibitive levels of RF losses to these passive devices. Radio frequency microelectromechanical systems (RF MEMS) are enabling technologies capable to bring significant improvement in the electrical performances and expressive size and cost reduction of these functions, with unparallel introduction of new functionalities, unimaginable to attain when using bulky, externally connected discrete components. High quality factor (Q) inductors are amongst ones of the most needed components in RF circuits and at the same time ones that are most affected by thin metallization and substrate related losses, demanding considerable research effort. This thesis presents a contribution toward the development of thick metal fabrication technologies, covering also the design, modeling and characterization of high quality factor and high self-resonant frequency (SRF) RF MEMS passive devices, with a special emphasis on spiral inductors. A new approach using damascene-like interconnect fabrication steps associated to low κ dielectrics (polyimide), highly-conductive thick copper electroplating, chemical mechanical polishing (CMP) and tailored substrate properties delivered quality factors in excess of 40 and self resonant frequencies in excess of 10 GHz, performances in the current state-of-the-art for integrated spiral inductors built on top of silicon wafers. Furthermore, the developed process steps are compatible with back-end processing used to fabricate modern IC interconnects and have a low thermal budget (< 250 °C), what makes it a good choice to build above-IC passives without degenerating the performance of passivated RF-CMOS circuits. Deep reactive ion etching (DRIE) of quartz substrates was also studied for the fabrication of spiral inductors, offering excellent RF performances (Q exceeding 40 and SRF exceeding 7 GHz). A new doubly-functional quartz packaging concept for RF MEMS devices was developed. This technique process both sides of the packaging wafer: the top is used to embed high quality factor copper inductors while the bottom is thermo-mechanically bonded to another RF MEMS wafer, offering a semi-hermetic SU-8 epoxy-based seal. The bonding process was optimized for high yield, to be compatible with SF6-plasma-released MEMS and to present low level of RF losses. Band pass filters for the GSM (1.8 GHz) and WLAN (5.2 GHz) standards were fabricated and characterized by RF measurements and full wave electromagnetic simulations. Although further development is need in order to predict the frequency response accurately, insertion losses as low as 1.2 dB were demonstrated, levels that cannot be usually attained using on-chip passives. Systematic analysis, RF measurements, electromagnetic simulations and equivalent circuit extraction were used to model the behavior of the fabricated devices and establish a methodology to deliver optimum performances for a given technological profile and specified performance targets (quality factor, inductance and frequency bandwidth). A simple yet accurate physics-based analytical model for spiral inductors was developed and proved to be accurate in terms of loss estimation for thick metal layers. This model is capable to accurately describe the frequency-dependent behavior of the device below its first resonant frequency over a large device design space. The model was validated by both measurements and full wave electromagnetic simulations and is well suited to perform numeric optimization of designs. The proposed models were also systematized in a Matlab® toolbox.