The development work included analytical and Monte Carlo modeling to optimize the gas shower heads. A good agreement between the two approaches was achieved. For instance one could predict the impact of a substrate tilt on impinging flux distribution. The new reactor was designed, and manufactured according to the created CAD drawings. The assembled machine was delivered during the 12th project month.

Taking into account that the amplitude and the phase of the reference wave keep constant during the SNOM image acquisition, the amplitude and the phase of the optical near-field is extracted by a lock-in detection at the Δf frequency. This heterodyne interferometric detection (fig. 2) is also an elegant solution to measure the very weak near-field signal collected by the SNOM probe, since it increases the signal to noise ratio.

LN and LT wafers were prepared by SAES and delivered to partners. All the wafers are double side polished and fully characterized in terms of trace elements (plasma based mass spectroscopy), crystalline quality (X-ray transmission topography), and optical absorption from 200 to 6000 nm wavelength. The first choice precursors were the metal-organic liquids of lithium tert-butoxide (Li-O-t-Bu) and niobium ethoxide (Nb(OEt)5). These were synthesized and characterized by SAFC. The characterization before delivery included NMR, mass spectrometry, and thermo-gravimetric analysis to assess volatility as a function of temperature. In the following, the CBE group of EPFL and ABCD determined thermal properties of precursors (vapor pressure as a function of temperature, chemical decomposition), tested the precursor vaporization as a function of temperature, and the resulting pressure in the pre-chamber system, which governs the gas flow to the substrate.
The PLD partners NIL and the PLD group of EPFL concentrated on the following materials systems:
• LiNbO3 (LN)
• Tungsten bronze K6Li4Nb10O30 (KLN)
• Tungsten Bronze SrxBa5–xNb10O30 (SBN)
• Perovskite (K1/2-xNa1/2-xLi2x)(Nb1-yTay)O3 (KNLNT)
Most of the films were characterized with regard to structural properties, including techniques such as X-ray diffraction; secondary electron microscopy (SEM) and transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), energy dispersive analysis of X-ray emission (EDAX). At some of the films functional test were performed like the evaluation of dielectric constant, piezoelectric activity, and piezo-sensitive AFM (atomic force microscopy), and the optical index of refraction.

From the modeling of flux distribution, a min-max uniformity of +/- 0.5 % was calculated across a 6" wafer. The measured distribution showed a very satisfying variation of +/- 2 % - along one diameter even +/- 0.5 % - for a deposition at 560 °C. Possibly, temperature uniformity, tightness of gas shower head, and precision of wafer alignment must be improved to achieve the predicted values.
The showerhead was also tested for creating defined gradients in the thickness distribution of TiO₂ by selectively closing segments. The obtained gradients coincided in a satisfactory manner with the calculated impingement distributions, thus proving that the programmable shower head is useful for combinatorial research.

The high-resolution TEM images at the interface indicate the presence of an intermediate layer, most likely full of dislocations to relax the stress. This can be explained by the large lattice misfit of -7.5% (RT).

The 170 nm thick film showed a RMS roughness of 5.5 nm. Such small surface roughness is a key to achieving low-loss optical waveguides. Raman spectroscopy proved to be a fast measurement technique for identifying the crystalline lithium niobate films close to stoichiometric composition. In future work, the thickness needs to be increased to around 400 nm in order to allow for wave guiding at 1.5 µm wavelength.

A side result of this research was the study on solid solutions of HfO2 and Nb2O3. The variability of the relatively high refractive index in this system might be of interest for optical devices based on graded refractive index materials.

Conductive transparent oxide thin films of Al doped ZnO were deposited in top of SBN/STON structures. The optical properties were evaluated by means of ellipsometry as a function of the applied DC electric field. The obtained loops show that the films exhibit ferroelectric properties in as much it can be partially poled with negative voltages (top electrode). Most importantly a tunable refractive index is obtained with a Δn range of 0.12.

Initial depositions proved to have rather high surface roughness however (see fig. 10), so attempts were made to optimise the microstructure to the levels observed when MgO (100) substrates were used. This was in response to difficulties encountered by our CNRS partners when undertaking SNOM analysis of our films. The reason for the increased surface roughness was attributed to the unstable MgO (111) surface, which is prone to both hydroxylating in air and reconstructing under the conditions necessary to deposit LiNbO3. This had a subsequent effect on the epitaxial quality of any film that was deposited.

Increasing the Mn-doping to 2 mol% lead to a suppression in the relative permittivity and a further reduction in loss at both positive and negative fields, when compared to undoped and 1 mol% doped films.

Polarisation hysteresis loops were also obtained for 0-2% Mn-doped NKLNTS films. Once again, the reduction in leakage current for doped films manifested itself in the observed results. Without the aid of Mn it was not possible to achieve saturation. The addition of dopants made possible the application of larger electric fields and consequently the shape of the hysteresis loops began to resemble that of a genuine ferroelectric. Piezoresponse force microscopy (PFM) was used to probe the local piezoresponse from a region of a 1 mol% Mn-doped film. The predominantly bright contrast in the phase image indicates the film was out-of-plane self-polarised.

The next dopant added to the base NKLNS composition was CaTiO3. Existing literature for KNN ceramics doped with CaTiO3 suggested that it was beneficial in stabilising piezoelectric properties across a wider temperature range. The effect of CaTiO3 doping on thin films was therefore felt worthy of investigation. The dielectric constant was found to fall with increasing CaTiO3 content. Losses for both the 1.5 mol% and 3 mol% films were reduced in comparison to the undoped film; however the losses under positive fields were greater than under negative fields suggesting the presence of p-type conduction, as previously discussed. This was supported by the shape of the polarisation hysteresis loops acquired from the doped samples. It was impossible to apply a strong enough field to reach saturation, as was the case with undoped films.

The transmission properties of this novel device were assessed by optical near-field measurements, compared to the transmission spectra of the same PhC drilled into bulk LN and calculated by a two dimensional finite-difference time domain method. We were able to show a strong improvement in the transmission properties of the LN PhC by etching the pores into a thin layer rather than into a thick LN wafer, due to the almost vertical sidewall of the holes within the film, and putting the bottom of the holes outside the wave guiding layer.

This result is of great interest for the development of very efficient active devices based on LN PhCs. Moreover, the required tunable voltage will be lower in a thin film than for a bulk device.

Whispering gallery mode resonators
Whispering gallery mode (WGM) resonators are photonic components, which are attracting a lot of interest as these devices can store optical power over long time intervals. They can form very narrow band optical spectral filters and if doped with laser active ions such as rare earths can form laser sources with interesting spectral characteristics. Because the resonances in such structures are very sensitive to environmental factors WGM resonators can form excellent bio-sensors. A method for the fabrication of micron-size WGM resonators in micro-structured lithium niobate has been developed within the framework of the project. The method involves domain engineering and micro-structuring of the surface of the crystal by inhibition of poling and chemical etching followed by surface tension reshaping at high temperatures, which smoothens the features (see example in fig. 17). Besides interesting geometries obtained through transformation driven by surface tension, it is also possible to tailor designed geometries such as hexagonal shapes. Resonances in a 200 µm wide hexagon could be observed.

The bottom electrode was patterned by dry etching, using a photoresist mask, and chlorine chemistry in a dry etcher. The PZT (Pb(Zr0.30, Ti0.70)O3 in this case) was deposited onto the bottom electrode by a sol-gel process. Standard conditions were used to fabricate {100} oriented films. A PbTiO3 seed layer was used to achieve high {001}-texture. A photolithographic lift-off process was used to pattern the top electrode of Pt. Another photolithography was then performed in order to define the etching holes in the PZT film. The PZT was dry etched using fluorine chemistry. The last step was the wet etching of the MgO substrate using 10 vol.% H3PO4 solution at 80 °C. Unfortunately we were unable to liberate the BAW membranes completely, meaning that the TFBAR lost a lot of the acoustic energy into the substrate. The admittance of the TFBAR reveals a broad resonance. The coupling coefficient kt2 reached a value of 23 %, however the quality factor is low due to the acoustic emission into the substrate.

All the niobate films studied so far for piezoelectric applications show large difficulties in poling. The reason is the large leakage in combination with large coercive fields. Both might be connected, in as much as defects can be responsible for increased leakage as well as for pinning ferroelectric domains. A process flow for TFBAR’s on small MgO substrates was developed. However, we still have a problem in liberating the membrane completely from the substrate by surface micromachining with a wet etchant. The admittance could be measured and showed very large coupling constants in case of PZT thin films. The study shows how much more easy lead containing ferroelectrics (usually containing titanates) can be processed as compared to the alkali niobates.

The advancement in micromachining and shaping of surface structures of LN crystals shows that this approach allows a quicker progress towards real applications. Possibly one can combine such techniques with thin films (e.g. amorphous silica and metals) to realize more complex structures. The combination of single crystal LN and LT with thin films of LT and LN would be of high interest. It was found that epitaxial homoepitaxy is not trivial as the surface of the single crystal substrate becomes unstable in the vacuum at higher temperatures. Again, CBE techniques might be more successful for such growth because of the lower deposition temperature. All the mentioned effects of roughness, cracks, leakage and poling problems are of course also detrimental for ultrasound applications in the GHz frequency range. It appears that CVD techniques might also be superior in this case. Furthermore, the project has lead to an improvement of scanning near field microscope (SNOM) techniques. The heterodyne SNOM proved to be very useful to characterize photonic structures of thin films at the nano and micron level, enabling us to demonstrate a better defined stop band edge in a thin film waveguide of LN/MgO.