**2.2 GeTe-based coplanar waveguide (CPW) radio frequency (RF) switches**

#### *2.2.1 Fabrication process for GeTe RF switches*

For this fabrication process, oxide and nitride layers were used as isolation layers. Some substrates had oxide layer, and some have nitride layers. Silicon substrate was thermally oxidized to make silicon dioxide (300 nm thickness) film. On the other hand, plasma-enhanced chemical vapor deposition (PECVD) was used to make silicon nitride (100 nm thickness) film. On top of these isolation layers, GeTe strips for signal line were deposited. GeTe layers were fabricated using RF sputtering method (parameters were 300 W RF power, 10 mTorr pressure, 20.1 sccm Ar gas flow for ignition, and 2 min time). The prepared GeTe films were in amorphous phase because of room temperature deposition. GeTe strips had length of 30 μm, width of 10 μm, and thickness of 100 nm. After that, bimetallic Ti/Au ground lines and bimetallic Ti/Au pads on both sides of GeTe strips were deposited using E-beam evaporation method and lift-off method. Ten-nanometer-thick Ti works as an adhesion layer for 100 nm gold layer, and it is essential to get better electrical contact. **Figure 3** shows the fabricated GeTe CPW RF switching devices [2].

### *2.2.2 Performance analysis of GeTe RF switches*

Before doing RF switching testing, thermal characterization was done. Thermal characterizations were done using thermal chuck where devices were directly placed on the chuck. **Figure 4** shows the thermal characterization results, and it is found that all devices show resistance change of ~six orders of magnitude. **Figure 4a** and **b** shows nonvolatile phase transitions. These devices were fabricated using oxide and nitride isolation layer, and GeTe film resistance never came back to its original values. This is an indication of permanent crystallization. **Figure 4c** and **d** shows volatile phase transitions. These devices were directly fabricated on substrates (c-Si and Al2O3 substrates, respectively) without any isolation layer, and they came back to their original values. Also, **Figure 4c** does not show ideal I-V behavior because of higher conductivity of c-Si substrate [2].

This heat-induced phase transition can be explained by the phenomena of heat transfer via conduction through multiple layers of materials. For GeTe on

#### **Figure 3.**

*(a) Microscopic image of a device with GeTe strip; (b) a smaller device on sapphire substrate for I-V measurement; and (c) a CPW RF switch having signal and ground lines [2].*

*Germanium Telluride: A Chalcogenide Phase Change Material with Many Possibilities DOI: http://dx.doi.org/10.5772/intechopen.108461*

**Figure 4.**

*Semi-logarithmic resistance versus temperature curves of GeTe: (a) on SiO2/c-Si substrate; (b) on Si3N4/c-Si substrate; (c) on c-Si substrate; and (d) on Al2O3 substrate (LINEAR scale curves are in the insets) [2].*

SiO2/c-Si and GeTe on Si3N4/c-Si, there were two layers of materials in between GeTe layer and thermal chuck. For GeTe on c-Si and GeTe on Al2O3, there was only one layer of material in between GeTe layer and thermal chuck. Heat vertically moves upward direction from thermal chuck. Small crystallites are formed due to this upward flow from substrate to GeTe via dielectric layer, and these crystallites are responsible for the GeTe resistance change. Additionally, GeTe on Al2O3 shows volatile transition following a different path. This is because of the difference in thermal expansion coefficients between Al2O3 and GeTe (5.3 and 0.56, respectively), which may result breaking of the conductive crystallites throughout the cooling process [2].

To perform S-parameters analysis of GeTe-based RF switches, S11 and S21 were investigated in frequency range of 10 MHz-20 GHz. It was found that when frequency increased, S11 went down from −11 dB to −16 dB in crystalline phase. In crystalline phase, reflectivity increases because of incorporation of small crystallites due to frequency increase. In amorphous phase, full return loss was found as S11 remained constant according to frequency change. These are shown in **Figure 5**. S21 was found constant in crystalline phase, and the value was −3 dB. In amorphous phase, S21 had different values in different frequency range. In 10–100 MHz range, it was −55 dB and remained constant. But when frequency increases from 100 MHz to 20 GHz, S21 gradually increased, and the value went up to −28 dB. This might be because of frequency induced crystallization in amorphous phase. The signal line resistance was found as 75 Ω and 3.5 MΩ in ON and OFF states, respectively. For making highperformance RF switch, it was suggested that S21 losses can be reduced by reducing the width of GeTe section which can eventually reduce ON state resistance from 75 Ω to 50 Ω [2].

#### **Figure 5.**

*Analysis of S parameters for GeTe-based RF switches. In this analysis, 31.6% and 10% of full signal strength are equivalent to −10 dB and −20 dB [2].*

#### **Figure 6.**

*GeTe gold split-ring resonators (SRRs): (a) GeTe SRRs; (b) GeTe-in-gap SRRs.*
