Rubidium | EPFL Graph Search

Rubidium is the chemical element with the symbol Rb and atomic number 37. It is a very soft, whitish-grey solid in the alkali metal group, similar to potassium and caesium. Rubidium is the first alkali metal in the group to have a density higher than water. On Earth, natural rubidium comprises two isotopes: 72% is a stable isotope 85Rb, and 28% is slightly radioactive 87Rb, with a half-life of 48.8 billion years—more than three times as long as the estimated age of the universe. German chemists Robert Bunsen and Gustav Kirchhoff discovered rubidium in 1861 by the newly developed technique, flame spectroscopy. The name comes from the Latin word rubidus, meaning deep red, the color of its emission spectrum. Rubidium's compounds have various chemical and electronic applications. Rubidium metal is easily vaporized and has a convenient spectral absorption range, making it a frequent target for laser manipulation of atoms. Rubidium is not a known nutrient for any living organisms. However,

Low Temperature Hermetic Wafer Bonding and Microfabrication of Wall-Coated Alkali Vapor Cells for Chip-Scale Atomic Clocks

Rahel Strässle

The increasing demand for highly precise, yet compact and low power timing devices pushes the research in the field of chip scale atomic clocks (CSACs). At the heart of a CSAC, there is a vapor cell containing a few micrograms of alkali metal, such as rubidium or cesium. Hermetic sealing of this cell is essential not just to avoid the strong reaction of the alkali metal to oxygen and hydrogen, but also to maintain a stable operation atmosphere that ensures minimized clock shifts. Since alkali atoms loose their spin polarization when colliding with the cell walls, a buffer gas is commonly added in alkali cells to reduce the mean free path. An alternative approach to increase the polarization lifetime is to use antirelaxation wall-coatings, which have been studied extensively for larger glass-blown alkali vapor cells but not yet for microfabricated cells. This is mainly because anodic bonding, which is the standard process for sealing microfabricated alkali vapor cells, degrades the known antirelaxation coatings with its elevated temperatures and the electric field strength. In this work, the problem of implementing the wall-coatings in microfabricated cells is addressed. In a first step, a hermetic and low temperature thin-film indium-bonding technique was developed, which is compatible with the most common antirelaxation coatings and meets the requirements to fabricate alkali vapor cells. The technique is based on the thermocompression of indium and can be applied on wafer level with temperatures ranging from room temperature to temperatures below indium’s melting point at 156 ◦ C. It yields a leak rate below 1.5 × 10−13 mbar l/s and a tensile strength between 10 and 25 MPa, depending on the substrates and the bonding parameters. Wafer-level bonding at such low temperatures with the reported level of hermeticity is unique in the field of MEMS packaging. This novel bonding technique was then successfully applied to microfabricate alkali vapor cells at the lowest temperature alkali cells have been sealed before. The gas contamination in an evacuated cell was lower than 1 mbar and the suitability of indium-bonded cells for atomic clock applications was demonstrated with a buffer-gas filled cell by recording the double-resonance signal. The obtained clock signal linewidth was 2.02 kHz, only slightly higher than the theoretically predicted intrinsic linewidth for this cell design. After validation of the bonding technique for alkali vapor cells, a wall-coating for the cell walls was integrated in the microfabrication process. The bulk and surface properties of selected antirelaxation wall-coatings materials, such as octadecyltrichlorosilane (OTS), paraffin, an alkene mix, and Parylene, were investigated for compatibility with alkali cells and microfabrication. The gathered knowledge led to the fabrication of the first microfabricated rubidium vapor cell with an OTS coating that shows antirelaxation properties.

EPFL2013

AIN Thin Film Structures for High-Q Resonators

Alvaro Artieda

The development of wireless and mobile communication in the last decade led to the elaboration of new components based on thin films with continually increasing performance and reduced fabrication cost. A fundamental element for mobile communication is the passband filter defining the total bandwidth, the so-called RF filter. Thanks to outstanding characteristics, thin film bulk acoustic wave resonators (TFBAR's) based on piezoelectric AlN thin films have conquered parts of this market against surface acoustic wave devices based on piezoelectric crystals. AlN is today the most favoured material for such these devices, thanks to its excellent electro-acoustic properties. This thesis explores paths for a new type of application in oscillators for time and frequency control in frequency range of several GHz. In portable applications, this domain is largely dominated by quartz crystals. Their resonance frequency – typically around 33 kHz – is multiplied to reach higher frequency domains. For such applications, the main issues are high quality factors and temperature stability of the resonators. This exploration implied two research directions, the first focused on AlN thin film property-process relationships for films grown on amorphous substrates, and the second on the property-design relations of such new devices. High quality factors and temperature stability are key issues in oscillator applications. We proposed to apply symmetric composite resonators (composite TFBAR) including two amorphous SiO2 thin films sandwiching high quality piezoelectric thin films of AlN. This required the growth of AlN on amorphous and insulating surfaces. We thus studied the growth of AlN on such substrates to evaluate the impact on film microstructure and morphology, mechanical stress, quality of orientation, piezoelectric constant d33,f , and polarization uniformity. Substrate roughness and substrate bias power (voltage) were identified as crucial growth parameters. In parallel, we also optimized Mo-electrode thin films for optimal AlN/Mo thin film structures. This was necessary to derive properties from pure AlN TFBAR's, and to produce HBAR's for comparison with our composite BAR's. The impact of the general device design, comprising active area surface, apodization and lateral edge design, was investigated concerning the Q-factor, kt2 and parasitic behaviour of devices. The important role of excitation of parasitic resonances was highlighted. We could show that also the microstructure of AlN plays a key role in their excitation and propagation. AlN exhibits a polar structure with one polar axis (c-axis). The growth process has to provide mechanisms to align the polar axis of all grains. No reorientation is possible by external means. The stability of the piezoelectric AlN layer with applied electric field guarantees a linear response for sensors and actuators. The high stiffness, combined with low density leads to a high longitudinal acoustic velocity of AlN, thus enabling the use of very high application frequencies. Almost all the reports so far have concentrated on depositing AlN on metallic layers, such as Mo, or Pt. The ability to grow high quality AlN thin film on amorphous substrates increases the possibility of developing new devices and applications. The orientation and stress properties, as well as the piezoelectric constant and the polarization of polycrystalline AlN thin films deposited on amorphous substrates, were investigated. Thin sputtered amorphous silicon layers of defined nanoscale roughness were assessed for their ability to manipulate the mechanical stress in AlN thin films that need to be grown on thermal oxide. With a roughness variation from 1 to 11 Å in the Si film, a stress variation from -700 to +200 MPa was observed in the AlN film, which shows the high impact of substrate roughness on the stress in AlN films. The rocking curve width of the 1.2 µm AlN deposited on the Si layers showed a pronounced, linear increase as a function of Si layer roughness. Low density grain boundaries appeared, and the overall stress changed to positive values. A study of the piezoelectricity and the grain polarity of AlN thin films grown on amorphous SiO2 thin films has been done. d33,f of 5.0 pm/V were achieved at low substrate roughness and low mechanical AlN film stress. This value is comparable to values reached on oriented Pt. Increasing substrate roughness and stress leads to a deterioration of d33,f which is correlated with a higher density of opposite polarity grains detected by piezoresponse force microscopy (PFM). Extrapolating to 100% uniform polarity, a maximum d33,f value of 6.1 pm/V is derived. High-Q, bulk acoustic wave composite resonators based on a symmetric layer sequence of SiO2-AlN-SiO2 sandwiched between electrodes have been developed. The SiO2 film thickness was varied while the piezoelectric AlN film had a constant thickness of 1.2 µm. Experimental results of quality factors (Q) and coupling coefficients (kt2) are in agreement with finite element calculations. Q-factors of about 2000 were observed for the first harmonic of the 310 nm oxide devices. Third harmonic resonances were very intense for devices with 770 nm oxide double layers. The temperature drift reveals the impact of the SiO2 layers, which is more pronounced on the first harmonic, reducing the TCF to about 0 ppm/K. Hybrid harmonic bulk acoustic resonators (HBAR) based on an AlN/Si structure were fabricated. In order to increase the coupling kt2 of the simple device, an additional film of AlN was deposited on the top of the HBAR to optimize the excitation energy localization in the stack. The gain of the coupling was more than a factor of 2. A general study of the TFBAR performance is given in this work. The performance of circular, polygonal and elliptical electrodes were compared according to the area. The figure of merit Q · kt2 is higher in the case of elliptical design electrodes in comparison with others. Due to its high acoustic impedance and low electrical resistivity molybdenum is the best electrode material used in this work. The use of the lateral edge design on symmetric SiO2-AlN-SiO2 composite resonators slightly reduces the strength of spurious resonances of the third harmonic, but decreases the Q-factor of the resonances. This composite design has the advantage of attenuating the spurious resonances. In order to study the impact of the microstructure of AlN films on electroacoustic properties of TFBARs, simple Mo-AlN-Mo devices with different values of AlN in-plane stress (from -500 MPa to +500 MPa) were developed. These devices showed a clear correlation between the stress and the intensity of the spurious resonances, devices with tensile stressed AlN films reached Q-factors of about 1800. The impact of the microstructure on the transmission of in-plane acoustic waves is responsible for this phenomena. Voided grain boundaries, that are correlated with tensile films, reduce this transmission. These devices showed that the use of amorphous SiO2 thin films in the stack of resonators increased the Q-factor and the figure of merit Q · f to (∼8.0 ·1012 Hz), which is close to the ultimate value of Quartz (∼1.6·1013 Hz). The reached value compares also well with a novel improved HBAR structure reaching a Q · f of (∼2.7·1012 Hz). The composite resonator in addition exhibits a much higher coupling coefficient, which facilitates the implementation of the electronics. The admittance curves of the composite resonators could be well simulated by combining finite element modelling with simple circuits including parasitic elements. The derived materials quality factors of SiO2 and for the Al/AlN/Pt stack amounted to 50'000 and 3'000, respectively. Finally, composite resonators working at 3.4 GHz, corresponding to the half ground state hyperfine separation of rubidium atoms, were developed. The Q-factor was equal to 2300 and the TCF equal to 1.5 ppm/K. A figure of merit Q · f of ∼0.8 ·1013 Hz and a kt2 of 1% were reached. This device will be implemented in an oscillator as planned to realize an atomic clock control circuit.

EPFL2010

Bose-Einstein condensation in semiconductors: myth or reality?

Bose-Einstein condensation, predicted for a gas of non interacting bosons in 1924 by Einstein, has been demonstrated for the first time in 1995 in a dilute gas of rubidium atoms at temperatures below 10(-6) K. In this work, it is shown that Bose-Einstein condensation can be achieved at around 15-20 K in a solid state system by using microcavity polaritons, which are composite bosons of mass ten billion times lighter than that of rubidium atoms. [DOI: 10.2971/jeos.2008.08023]