One of the two reconfiguration techniques studied in this project is based on the use of chalcogenide glasses. The latters are part of the Phase Change Material (PCM) family and can switch from a high resistivity amorphous state to a low resistivity crystalline one under a heating process. Chalcogenide glasses are well known since a long time in different applications as rewritable memory for instance. They are now studied for reconfiguration in microwave applications. The main interests in this context are the large resistivity switching ratio that could be expected (4 to 6 orders) and the switching time (from few or few tens of nanoseconds especially with an optical control). The two main compositions of chalcogenide glasses studied in this context are the Germanium Telluride (GeTe) and, to the lesser extent, the GST (Ge2Sb2Te5).
The DATERAC project is the first one focusing on the chalcogenide glasses for microwave applications in Rennes. IETR lab has so opened a collaboration with the Glasses and Ceramics team of the ISCR lab. This team is internationally well known for its numerous works about chalcogenide glasses. As all beginning collaboration about technological aspects, the first steps consist of establishing the manufacturing process and characterizing the deposited films for the targeted applications. These are the main objectives of this first WP.
This WP is decomposed into 2 sub-tasks corresponding to these 2 main objectives:
- The development of the thin film deposition process on silicon substrate
- The characterization of chalcogenide glasses for microwave applications.
One should note that the choice of silicon substrate is a direct consequence of our final target which will consist in associating 2 reconfiguration methods on the same system. As the second reconfiguration method, developed in WP3, is based on the integration of localized doping areas on silicon substrate, the choice of this latter has become obvious.
Development of the thin film deposition process on silicon substrate
An exploratory action made in 2019-2020, also with the support of CominLabs, validated the fact that our partner ISCR is able to deposit chalcogenide glasses on silicon substrate. So, one of the first task of this project is to develop a complete deposition and shaping process of the chalcogenide glasses on this substrate.
The chalcogenide glasses chosen at the beginning of the project is GST (Ge2Sb2Te5), directly available from our ISCR partner. We firstly focus on the thickness of the deposition. Our objective was initially to have a glass thickness of 1 um in order to be close to the one of the aluminium one (2 um) used as metal for the design of the passive parts of the reocnfigurable circuits. First tests showed that such a thickness can be done easily.
These tests was also used to analyse the real glasses composition finally obtained. The initial objective was 22% of Germanium (Ge), 22% of Antimony (Sb) and 56% of Tellurium (Te). As well using commercial or homemade sputtering targets, the EDS (Energy-Dispersive Spectroscopy) analyses present globally a final composition of 25-27% of Ge and Sb and around 45 to 50% of Te according to the samples. Moreover, these compositions can varied along the wafer as presented in Figure 1.1.
Figure 1.1 – Example of EDS analysis of a Ge2Sb2Te5 deposition on Si wafer (2 inches).
The second part of this task focus on the shaping of the glasses deposition. Because ISCR Glass & Ceramics team is not used to work in microwave applications, their sizing means were not really adapted to our needs. First tests were so made using different homemade masks. The most interesting results were obtained using thick film of Kapton. The latter was engraved using laser and then used as mask to limit the deposition on the targeted areas. Such a method gives a good resolution and the final sizings are close to the targeting ones. An example is presented in Figure 1.2. The final length is 4.8 mm instead of an expected one of 5 mm and the width is 180 to 190 um instead of 200 um. Nevertheless, there seems to be a gradient on the edges and the ends are rounded.
Figure 1.2 – Example of chalcogenide glasses deposition using homemade Kapton mask.
Nevertheless, the arrival of a team of the FOTON institute in the project has given us an access to NanoRennes microtechnology platform. The chalcogenide glasses deposition could then be done by a lift-off process offering us a better resolution. First deposition using this process will be done at the end of March 2022.
Characterization of chalcogenide glasses for microwave applications
First characterizations were made on the samples discussed above. They concern the transition from the amorphous state to the crystalline one. The crystallization of the chalcogenide glasses is very fast and is obtained for a temperature between 150 to 180°C. The state switch was obtained using a ventilated oven in order to avoid oxydation troubles. The use of an oven does not allow us to switch back the chalcogenide glasses to their initial amorphous state because this operation needs to be able to decrease extremely quickly the temperature of the samples. It is possible only using lasers. In the crystalline state, some conductivity characterizations were also made. Nevertheless, results are not reliable at that time because of measurement system faults and impact of the substrate.
New characterizations were so made at FOTON institute on the new deposition samples. Based on thin layers of Ge2Sb2Te5 deposited by ISCR by means of RF magnetron sputtering, and processed by FOTON Institute to realize patterns (see insert in Figure 1.3), a more than 5 decades variation of GST film resistivities have been measured as function of the annealing temperature (Figure 1.3, red dots), in agreement with the work presented in the open literature. Very recently, we even succeeded in getting this structural transition on such patterns combining laser illumination and sweeping as shown on Figure 1.3 (blue stars) as function of laser power.
During these laser illumination tests, several configurations and characterizations were made. First, we crystallized rectangular samples using different spot shapes (circular or elliptical). A pulse laser Q-switched from CNI was used (λ=355 nm, pulse width = 7 ns (freq=7 kHz), <P>=0,12 W). The GST crystallization were obtained on 60×1500 um2 line with a 60 um-diameter circular spot and on 15×1500 um2 line with a 15×600 um2 elliptical spot by manual movement (50 um/s). Of course, the crystallization of larger surfaces and its reproducibility will need the automation of the laser movement. The assembly of a motorized bench is currently under study. The laser fluence was also studied in order to determine the crystallization threshold (> 15 mJ/cm2) and the burnt one (70 mJ/cm2). The useful fluence band is so between 20 to 60 mJ/cm2.
