**7. Fabricated RF MEMS switches**

Due to better flexibility for large systems and wide band applications, metal to metal contact switch was chosen over the capacitive switch. The CPW centre conductor was 60µm wide, 20µm gap and 210µm ground widths which resulted in characteristic impedance of 50Ω. The beam was suspended 2.5µm from the substrate. The ground planes around the beam were suspended as it provided easy access to beam and electrode when being used in biasing systems. The switches were fabricated using the developed six mask all metal process. A dimple was used at the bottom of front centre tip of the cantilever beam to reduce the stress sensitivity of the beam. Front tip contact area was small as compared to conventional cantilever beams because of following reasons. First, small contact points would reduce the metal-to-metal stiction and would increase the contact pressure. Secondly, it gave better isolation.

Figure 14 displays the proposed arrow beam design. The length of the beam is 120µm and the width of the beam is 60µm. The beam has been curved inside from the front with a front tip 20µm in width.

Fig. 14. SEM of fabricated arrow beam cantilever based RF MEMS switch

Figures 15 and 16 show the two proposed cantilever beam designs. In figure 15, the beam labeled Design-1 has three supported cantilever bars which behave like three springs while moving the beam during the actuation. The length of each cantilever bar is 20µm and the width is 10µm. The gap between each cantilever bar is 15µm and provides symmetry to the beam structure. All three cantilever bars are connected with an anchor which is 20µm in length and 60µm wide. The supported cantilever bars are then connected with a beam of length 100µm and width of 60µm.

Due to better flexibility for large systems and wide band applications, metal to metal contact switch was chosen over the capacitive switch. The CPW centre conductor was 60µm wide, 20µm gap and 210µm ground widths which resulted in characteristic impedance of 50Ω. The beam was suspended 2.5µm from the substrate. The ground planes around the beam were suspended as it provided easy access to beam and electrode when being used in biasing systems. The switches were fabricated using the developed six mask all metal process. A dimple was used at the bottom of front centre tip of the cantilever beam to reduce the stress sensitivity of the beam. Front tip contact area was small as compared to conventional cantilever beams because of following reasons. First, small contact points would reduce the metal-to-metal stiction and would increase the contact pressure. Secondly, it gave better

Figure 14 displays the proposed arrow beam design. The length of the beam is 120µm and the width of the beam is 60µm. The beam has been curved inside from the front with a front

Fig. 14. SEM of fabricated arrow beam cantilever based RF MEMS switch

length 100µm and width of 60µm.

Figures 15 and 16 show the two proposed cantilever beam designs. In figure 15, the beam labeled Design-1 has three supported cantilever bars which behave like three springs while moving the beam during the actuation. The length of each cantilever bar is 20µm and the width is 10µm. The gap between each cantilever bar is 15µm and provides symmetry to the beam structure. All three cantilever bars are connected with an anchor which is 20µm in length and 60µm wide. The supported cantilever bars are then connected with a beam of

**7. Fabricated RF MEMS switches** 

isolation.

tip 20µm in width.

Fig. 15. SEM of fabricated RF MEMS switch

Figure 16 shows the beam labeled Design-2, with three supported cantilever bars and an extended cantilever at the front. The dimensions of the three supported cantilever beams are the same as that for Design-1, with the centre 60µm×60µm and the extended cantilever at the front 40µm×20µm.

Fig. 16. SEM of fabricated RF MEMS switch

Plasma Based Dry Release of MEMS Devices 285

Figure 18 shows another design of switch which has a standard cantilever with dimensions of 80µm×60µm at the rear and an extended cantilever at the front with 40µm×20µm. Some metal particles can be seen on front portion of the cantilever which came after testing of the

The measurement setup for actuation voltage and RF performance was employed using two test configurations, i.e, preliminary screening and RF characterization. Preliminary screening was made using Cascade Microtech 10000 probe station with Tungsten needle connected to a Sony Tektronix 370 Programmable Curve Tracer. No RF performance was analyzed at this stage. The curve tracer was programmed to 0-100V DC signal with a step of 2V increment. The switches actuated at 19V and 23V. At this stage, the contact confirmation

For RF performance, a two port on wafer measurement of the RF MEMS switches was performed using HP-8510 vector network analyzer (VNA) from 0-40 GHz. A Cascade Microtech 10000 probe station was used. RF probing was done using Cascade Microtech GSG RF probes with a pitch of 100µm. The SOLT (short-open-line-through) method was

An HP 4140 DC voltage source was used to actuate the switch during RF characterization. The actuation voltage for the RF MEMS switches was applied between the cantilever beam and the lower actuation pad. Two Picosecond Pulse Labs 5590 DC blocks were connected

A two port on wafer measurement of the RF MEMS switches have been performed from 0- 40 GHz. When our switches were unactuated and beam was in up position, switches were in OFF state.When switches were actuated and the beam was pulled down, switches were in

In order to determine the RF performance of the switch the insertion loss, return loss and isolation of the switches were measured. Isolation of the switch was measured when signal was in OFF state. Figure 19 and 20 illustrate the measured S-parameters for Design-1 and Design-2 respectively. As shown, Design-1 had an isolation of 28dB at 20GHz and better than 23dB at 40GHz. For Design-2, the isolation of the switch was 30dB at 20GHz and better

The return loss and insertion loss of the switches were measured when signal passed through the ON state. Design-1 has a return loss better than 22dB at 20GHz and 19dB at 40GHz, for Design-2 it was better than 20dB at 20GHz and 18dB at 40GHz. This reveals

Insertion loss for Design-1 was 0.75dB at 20GHz and 1.15dB at 40GHz where as insertion loss for the Design-2 was 0.8dB at 20GHz and 1.3dB at 40GHz. Higher insertion loss was

was made between two surfaces while measuring the contact resistance.

used for the calibration of the system before each test sequence.

between VNA and RF cables connected with RF probes.

device while dragging the probe for contact.

**8. Experimental results** 

**8.1 Electrical performance** 

ON state.

**8.1.1 Isolation** 

than 28dB at 40GHz.

**8.1.2 Insertion loss and return loss** 

good impedance matching to 50Ω of our RF MEMS designs.

Fig. 17. SEM of fabricated RF MEMS switch

Figure 17 shows the RF MEMS switch which has the same dimensions of beam as explained in figure 16, instead of three supporting bars, has two supporting bars with a single cross bar link intended to increase the strength of the two low spring constant supporting bars.

Fig. 18. SEM of fabricated novel RF MEMS switch

Figure 18 shows another design of switch which has a standard cantilever with dimensions of 80µm×60µm at the rear and an extended cantilever at the front with 40µm×20µm. Some metal particles can be seen on front portion of the cantilever which came after testing of the device while dragging the probe for contact.

### **8. Experimental results**

284 Microelectromechanical Systems and Devices

Figure 17 shows the RF MEMS switch which has the same dimensions of beam as explained in figure 16, instead of three supporting bars, has two supporting bars with a single cross bar link intended to increase the strength of the two low spring constant

Fig. 17. SEM of fabricated RF MEMS switch

Fig. 18. SEM of fabricated novel RF MEMS switch

supporting bars.

The measurement setup for actuation voltage and RF performance was employed using two test configurations, i.e, preliminary screening and RF characterization. Preliminary screening was made using Cascade Microtech 10000 probe station with Tungsten needle connected to a Sony Tektronix 370 Programmable Curve Tracer. No RF performance was analyzed at this stage. The curve tracer was programmed to 0-100V DC signal with a step of 2V increment. The switches actuated at 19V and 23V. At this stage, the contact confirmation was made between two surfaces while measuring the contact resistance.

For RF performance, a two port on wafer measurement of the RF MEMS switches was performed using HP-8510 vector network analyzer (VNA) from 0-40 GHz. A Cascade Microtech 10000 probe station was used. RF probing was done using Cascade Microtech GSG RF probes with a pitch of 100µm. The SOLT (short-open-line-through) method was used for the calibration of the system before each test sequence.

An HP 4140 DC voltage source was used to actuate the switch during RF characterization. The actuation voltage for the RF MEMS switches was applied between the cantilever beam and the lower actuation pad. Two Picosecond Pulse Labs 5590 DC blocks were connected between VNA and RF cables connected with RF probes.

#### **8.1 Electrical performance**

A two port on wafer measurement of the RF MEMS switches have been performed from 0- 40 GHz. When our switches were unactuated and beam was in up position, switches were in OFF state.When switches were actuated and the beam was pulled down, switches were in ON state.

#### **8.1.1 Isolation**

In order to determine the RF performance of the switch the insertion loss, return loss and isolation of the switches were measured. Isolation of the switch was measured when signal was in OFF state. Figure 19 and 20 illustrate the measured S-parameters for Design-1 and Design-2 respectively. As shown, Design-1 had an isolation of 28dB at 20GHz and better than 23dB at 40GHz. For Design-2, the isolation of the switch was 30dB at 20GHz and better than 28dB at 40GHz.

#### **8.1.2 Insertion loss and return loss**

The return loss and insertion loss of the switches were measured when signal passed through the ON state. Design-1 has a return loss better than 22dB at 20GHz and 19dB at 40GHz, for Design-2 it was better than 20dB at 20GHz and 18dB at 40GHz. This reveals good impedance matching to 50Ω of our RF MEMS designs.

Insertion loss for Design-1 was 0.75dB at 20GHz and 1.15dB at 40GHz where as insertion loss for the Design-2 was 0.8dB at 20GHz and 1.3dB at 40GHz. Higher insertion loss was

Plasma Based Dry Release of MEMS Devices 287

attributed to following reasons. First, the higher contact resistance was achieved which was due to high surface roughness of the metal surface and smaller contact area. The surface roughness value is 18nm for dimple and 22nm for signal line contact area which showed

(a) (b)

(c) (d)

(e) (f)

Secondly, dimple surface might have an uneven surface contact with signal line contact area. To validate this observation, a simulation test was conducted to see the dimple movement in different stages of the actuation. Figure 21 showed the movement of the cantilever beam with dimple in different stages when actuation bias was applied. The dimple made a perfect smooth contact with lower contact surface as shown in figure 21(c)

Fig. 21. Simulated view of dimple contact during different stages of actuation

that surfaces of both contact points are quite rough.

Fig. 19. Measured S-parameters of the switch using Design-1

Fig. 20. Measured S-parameters of the switch using Design-2

Fig. 19. Measured S-parameters of the switch using Design-1

Fig. 20. Measured S-parameters of the switch using Design-2

attributed to following reasons. First, the higher contact resistance was achieved which was due to high surface roughness of the metal surface and smaller contact area. The surface roughness value is 18nm for dimple and 22nm for signal line contact area which showed that surfaces of both contact points are quite rough.

(c) (d)

(e) (f)

Secondly, dimple surface might have an uneven surface contact with signal line contact area. To validate this observation, a simulation test was conducted to see the dimple movement in different stages of the actuation. Figure 21 showed the movement of the cantilever beam with dimple in different stages when actuation bias was applied. The dimple made a perfect smooth contact with lower contact surface as shown in figure 21(c)

Plasma Based Dry Release of MEMS Devices 289

whole research The helpful advice by Dr Eric Gauja during the fabrication is greatly

Abe, T.; Messner, W. C. & Reed, M. L. (1995). Effective Methods to Prevent Stiction during

Alley, R. L.; Cuan, G. J.; Howe, R. T. & Momvopoulos, K. (1992). The Effect of Release Etch

Campbell, S. A. (1996). *The Science and Engineering of Microelectronic Fabrication*, Oxford

Chan, K. Y.; Daneshmand, M.; Mansour, R. R. & Ramer, R. (2007). Novel Beam Design for

Forsen, E.; Davis, Z. J.; Dong, M.; Nilsson, S. G.; Montelius, L. & Boisen, A. (2004). Dry

Harsh, K. F.; Zhang, W.; Bright, V. M. & Lee, Y. C., (1999). Flip Chip Assembly for Si Based

Israelachvili, J. N. (1991). *Intermolecular and Surface Forces,* Academic Press Ltd, ISBN 0-12-

Madou, M. J. (2002). *Fundamentals of Microfabrication,* pp. 276-277, CRC Press, Florida, ISBN

Mastrangelo, C. H. (2000). Suppression of Stiction in MEMS, *Proceedings of Material Research* 

Mastrangelo, C. H. & Hsu, C. H. (1993). Mechanical Stability and Adhesion of

Orpana, M. & Korhonen, A. O. (1991). Control of Residual Stress of Polysilicon Thin Films

87942-585-7, 24-27 June 1991, San Francisco, CA USA

*Microelectromechanical Systems*, Vol. 2, No. 1, pp. 33-43, ISSN 1057-7157 Mastrangelo, C. H. & Hsu, C. H. (1992). A Simple Experimental Technique for Measurement

Microstructures under Capillary Forces – Part I : Basic Theory, *Journal of* 

of the Work of Adhesion of Microstructures, *Proceedings of IEEE 5th Technical Digest on Solid State Sensors and Actuators Workshop*, pp. 2008-2012, ISBN 0-7803-0456-X, SC,

by Heavy Doping in Surface Micromachining, *Proceedings of International Conference on Solid State Sensors and Actuators Transducers 91,* Vol. 23, pp. 957-960, ISBN 0-

74, pp. 487-490, doi: 10.1016/j.mee.2004.03.022, ISSN 0167-9317

University Press, ISBN 0-19-510508-7, New York USA

Post Release Etch Processing, *Proceedings of International Conference on Micro Electro Mechanical Systems (IEEE MEMS 95)*, pp. 94-99, ISBN 0-7803-2503-6, 29th Jan – 2nd

Processing on Surface Microstructure Stiction, *Proceedings of IEEE 5th Technical Digest on Solid State Sensors and Actuators Workshop*, pp. 202-207, ISBN 0-7803-0456-

Compact RF MEMS Series Switches, *Proceedings of Asia Pacific Microwave Conference (APMC 2007)*, pp. 229-232, ISBN 978-1-4244-0749-4, 11-14 Dec 2007, Bangkok

Release of Suspended Nanostructures, *Journal of Microelectronic Engineering,* No. 73-

RF MEMS, *Proceedings of 12th International Conference on Micro Electro Mechanical Systems (IEEE MEMS 99)*, pp. 273-278, ISBN 0-7803-5194-0, 17-21 Jan 1999, Orlando,

appreciated.

**11. References** 

Feb 1995. Netherland

Thailand

FL USA

USA

375181-0, London UK

0-8493-0826-7, USA

*Society Symposium*, Vol. 605, pp. 105-116

X, 22-25 June 1992, SC, USA

but at this point full boundary conditions were not enforced. When boundary condition were fully enforced and beam was placed in the hold down position the complete surface of dimple was not in contact and front surface of dimple has lifted up as shown in figure 21(e). This phenominon lead to higher insertion loss.

#### **8.1.3 Actuation voltage**

The measured actuation voltage of the Design-1 is 19V and Design-2 is 23V. A number of release holes can be observed in the fabricated switches. The effect on electrostatic force due to release holes had already been rationalized with inclusion of 40% of fringing field effect during simulation of spring constant of the beam designs (Rebeiz, 2003).
