Design of low actuation voltage RF MEMS switch

Low-loss microwave microelectromechanical systems (MEMS) shunt switches are reported that utilize highly compliant serpentine spring folded suspensions together with large area capacitive actuators to achieve low actuation voltages while maintaining
of 4
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  1 Design of Low Actuation Voltage RF MEMS Switch Sergio P. Pacheco 1 , Linda P. B. Katehi 1 , and Clark T.-C. Nguyen 21 Radiation Laboratory and  2 Center for MicrosystemsDepartment of Electrical Engineering and Computer ScienceUniversity of MichiganAnn Arbor, Michigan 48109-2122 USA A BSTRACT Low-loss microwave microelectromechanical systems(MEMS) shunt switches are reported that utilize highly com-pliant serpentine spring folded suspensions together withlarge area capacitive actuators to achieve low actuation volt-ages while maintaining sufficient off-state isolation. The RFMEMS switches were fabricated via a surface micromachin-ing process using PI2545 Polyimide 1 as the sacrificial layer.The switch structure was composed of electroplated nickeland the serpentine folded suspensions had varying number of meanders from 1 to 5. DC measurements indicate actuationvoltages as low as 9 V with an on-to-off capacitance ratio of 48. Power handling measurement results showed no ”self-biasing” or failure of the MEMS switches for power levelsup to 6.6W. RF measurements demonstrate an isolation of -26 dB at 40 GHz.I. I NTRODUCTION The advent of bulk and surface micromachining tech-niques during the 1970s enabled the emergence of micro-electromechanical systems (MEMS). Micromechanical mi-crowave switches were firstdemonstrated by Petersen ascan-tilever beams using electrostatic actuation [1]. The advan-tage of MEMS switches over their solid state counterpartssuch as FETs or PIN diodes is their extremely low series re-sistance and lowdrive power requirements. Inaddition, sinceMEMS switches do not contain a semiconductor junction,they exhibit negligible intermodulation distortion.Recently, shunt microwave switches have been developedin the K/Ka band [2], [3], [4]. These devices are primar- ily designed for low-loss applications that do not requirefast rates such as airborne and/or deep space communica-tion. The switches are usually electrostatic in nature andcommonly driven by bias voltages in the 30 - 50 V range.The primary goal of the research presented in this paper is todevelop electrostatic shunt K/Ka-band microwave switchesthat rival the performance of the above designs while achiev-ing very low pull-in voltages. This is achieved by designingswitch structures with extremely compliant folded suspen- 1 DuPont Electronic Materials, P. O. Box 80334, Wilmington, DE. sion springs and large electrostatic actuation area.II. M ECHANICAL AND  RF D ESIGN Figure 1 shows a Scanning Electron Micrograph (SEM) of a released MEMS RF switch with serpentine folded suspen-sions containing four meanders. Meanders (N=4)Capacitive PadActuation PadActuation PadLyLxLcLswcwsg0gap between switch and FGCPW linetthicknessof switchwcond Fig. 1. SEM of fabricated MEMS RF switch.TABLE IP HYSICAL  D IMENSIONS OF  MEMS  MICROWAVE SWITCHSHOWN IN  F IG . 1 L s  250 µ m  L x  250 µ m L c  20 µ m  L y  250 µ mt 2 µ m  w cond  60 µ mw 5 µ m  g 0  3 µ mN 4  K  z  0.521 N/mmass 3.23x10 − 9 kg  V    pi  1.94 VTable I lists the physical dimensions of the MEMS mi-crowave switch shown in Fig. 1.The mechanical design consists of folded suspensions of serpentine format attached to square plates that provide theelectrostatic area of actuation. Attached to the actuationplates is the center plate that provides the parallel capaci-tance at the center conductor of the finite ground coplanarwaveguide (FGCPW) line. The entire structure is anchored 0-7803-5687-X/00/$10.00 (c) 2000 IEEE  to the substrate at the ends of the folded suspensions. At themoment of actuation, the necessary pull-in voltage is appliedbetween the ground plane of the FGCPWline and the switch.Once the switch clamps down, the high capacitance presentat the center conductor provides a virtual short at RF.The formula for actuation or pull-in voltage is given by: V    pi  =   8 K  z g 3 o 27 Aǫ o (1)where  K  z  is the spring constant in the z-direction,  g o  is theinitial gap between the switch and the bottom electrode,  A isthe area of the actuation pads and  ǫ o  is the permittivity of air.In order to lower the pull-in voltage of the structure, threedifferent routes can be pursued: (1) increasing the area of ac-tuation, (2) diminishing the gap between the switch and bot-tom electrode, and (3) designing a structure with low springconstant. In the first case, the area can only be increased byso much before compactness becomes a prevailing issue. Inthe second case, the return loss associated with the RF signalrestricts the value of the gap. The third route is the one withthe most flexibility, since the design of the springs does notconsiderably impact the size, weight, and/or RFperformanceof the circuit.The design of the folded suspension is of crucial impor-tance in realizing MEMS switches with low actuation volt-age. The folded suspension was chosen due to its ability toprovide very low values of spring constant in a compact areaas well as providing high cross-axis sensitivity between ver-tical and lateral dimensions. The spring constant in the z-direction,  k z  for one of the suspensions is given by [5], [6]: k z  = Ew   tL c  3 1 +  L s L c  L s L c  2 + 12  1+ ν  1+ ( wt  ) 2   (2)where  E   and  ν   are the Young’s modulus and Poisson’s ratiofor the metal. The total spring constant is the sum of all foursuspensions attached to the structure: K  z  = 4 k z N   (3)where N is the number of meanders in the suspension (Fig-ure 1 contains four meanders). The spring constant decreaseslinearly with successive addition of meanders to the foldedsuspension.FiveMEMSswitch designs werefabricated and measured.All designs were identical except for containing differentnumber of meanders in their folded suspensions. MEMSswitches with one, two, three, four and five meanders werestudied.III. F ABRICATION  P ROCESS A five mask batch process is used to fabricate the MEMSswitching circuits. The samples are composed of 400  µ mhigh-resistivity silicon. Figure 2 shows the process flow usedto fabricate the MEMS RF switches. (a)(b)(c)(d)(e)Si SubstrateSi SubstrateSi SubstrateSi SubstrateSi SubstrateCircuit MetalSilicon NitridePolyimideElectroplated NickelReleased Structure Fig. 2. Process flow for MEMS RF switch. The process is as follows: (a) 500/7500˚A of Ti/Au is de-posited and the circuit metal layer is defined via a liftoff process; (b) 1000˚A of plasma enhanced chemical vapor de-position (PECVD) silicon nitride (Si 3 N 4 ) is patterned overthe locations of switch actuation; (c) a sacrificial layer 3 µ mthick of polyimide DuPont PI2545 is spun cast, soft baked,and patterned for anchor points; (d) 2 µ m of nickel is electro-plated to define the switch structure; (e) sacrificial etching of the polyimide layer and supercritical CO 2  drying and releaseof the switch structure.IV. R ESULTS AND  D ISCUSSION Actuation voltage measurements were performed using aHP 4275A Multi-Frequency LCR meter with an internal biasoption.Table II lists the predicted and measured values for pull-in voltages for each given suspension design. As expected,the actuation voltage dropped with increasing number of me-anders. However, the values are considerably higher thanthe calculated design values. This increase can be tracked toseveral factors. First, a slight overplating of the beam struc- 0-7803-5687-X/00/$10.00 (c) 2000 IEEE  TABLE IIM EASURED ACTUATION VOLTAGES FOR  MEMS  SWITCH Number of meanders  V    pi  - Design  V    pi  - Measured1 3.90 V 35 V2 2.75 V 28 V3 2.24 V 20 V4 1.94 V 15 V5 1.74 V 9 Vture to 2.6 µ m instead of the design value of 2 µ m accountsfor an increase of 1.48 times in the actuation voltage. Sec-ond, the structures exhibit a certain amount of warping in thefolded suspensions due to vertical stress gradients across thethickness of the beams. This causes the actuation pads to beapproximately 5 µ m above the ground planes upon release,thus increasing the pull-in voltage by an additional factor of 2.15. Third, the remaining increase is due to the high intrin-sic residual axial tensile stress developed in the nickel filmduring the fabrication process (in the order of 150 MPa).Figure 3 shows the actuation of the circuits and the respec-tive on and off capacitances. The on and off capacitances Bias Voltage [V]010203040    C  a  p  a  c   i   t  a  n  c  e   [   f   F   ] 05001000150020002500N=5N=4N=3N=2N=1 Fig. 3. Plot of measured pull-in voltage vs. capacitance for vary-ing number of spring meanders. were measured at 47 fF and 2.25 pF respectively, resulting inan on-to-off ratio of 48.The power handling capabilities of the RF MEMS weremeasured at X-Band. Fig. 4 is a diagram of the power mea-surement setup. The power source was an X-band (8-12GHz) TWT amplifier and the variable attenuator was used tovary the output power from 1 mW up to the maximum levelof 6.6 W. The power coming out of the MEMS switch wascoupled to a 20 dB attenuator and fed into the power meter.The power handling capabilities of the RF MEMSswitches can be limited either by the current density on the 83624Synthesized   SweeperVariableAttenuatorTWTRF inRF out 0 50 dB DUT10 dB CouplerPowerMeter20 dBAttenuatorLoadIsolator   Fig. 4. X-band power measurement set-up transmission line causing excessive heating or by the actu-ation of the switches due to the average RF voltage on theFGCPW line (denoted as ”self biasing”). Since the electro-static force acting on the switch can derive from either a pos-itive or negative voltage, the average voltage level of the rec-tified sine wave due to the RF power on the FGCPW line isattracting the switch.The predicted average voltage on the FGCPW line forthe maximum power level of 6.6 W is 8.56V, which is notenough to cause any of the switch designs to actuate. Actu-ally, since the RF power is delivered to the center conductorof the FGCPW line, the area of electrostatic attraction is ap-proximately eight times smaller and a pull-in voltage of 26 Vwould be needed to actuate a MEMS switch with five mean-ders. That pull-in voltage corresponds to a RF power level of 66.6 W. As predicted, the MEMS switch did not ”self-bias”at any moment. Fig. 5 shows a plot of the power readings forthe cases of the switch off and on.   101520253035404550−10−50510152025303540      P  o  w  e  r   O  u   t   [   d   B  m   ]   Variable Attenuator [dB]   Switch Off   Switch On Fig. 5. Measured power levels for switch off (-) and switch on (x) Switches were actuated using bias voltages from 10-40V at power levels from 1mW to 5.5W. There was no ob-served catastrophic failure of the switches and/or dielectricfilm. The dielectric strength for a PECVD Si 3 N 4  thin filmis 5x10 6 V/cm [7], corresponding to a breakdown voltage of  0-7803-5687-X/00/$10.00 (c) 2000 IEEE  50 V for a 1000˚A thick film. There was a slight bending of the folded suspensions at power levels above 4.5 W with theswitch on. The suspensions reverted back to their srcinalposition once the power was decreased or turned off. Thiswarping was due to heating of the circuit as the power levelsincreased.The RF response of the system was measured using a8510C Vector Network Analyzer, Alessi Probe Station, andGGB Picoprobe 150 micron pitch coplanar probes. A TRLcalibration software, MULTICAL, developed at NIST [8],[9] is used to deembed the effects of the probe tips and feed-lines from the measurement response thus extending the ref-erence plane up to the MEMS switch under test. The plot inFig. 6 shows the response of the FGCPW line with a switchin the off position. Notice the increase of the return loss at Frequency [GHz]10203040    S  c  a   t   t  e  r   i  n  g   P  a  r  a  m  e   t  e  r  s   [   d   B   ] -50-40-30-20-100S11S21 Fig. 6. Plot of measured RF response of FGCPW line with switchin off position. higher frequencies due to the capacitance introduced by theswitch. In addition, the difference in insertion loss as com-pared to a FGCPW line without a switch is was measured tobe of only 0.16 dB at 40 GHz.The RF characteristics with the switch on is shown inFig. 7 and demonstrates an isolation of approximately -26 dBat 40 GHz. The isolation can be further improved by decreas-ing the thickness of the dielectric between the switch andcenter conductor, thus increasing the on capacitance. How-ever, care must be taken in order to make sure that the thinfilm thickness is sufficient not to cause dielectric breakdownand current flow.V. C ONCLUSIONS MEMS RF switches with actuation voltages as low as9Vand excellent RFcharacteristics have been demonstrated.Measurements show an isolation of better than -15 dB above15 GHz. These MEMS switches are suitable for wireless Frequency [GHz]10203040    S  c  a   t   t  e  r   i  n  g   P  a  r  a  m  e   t  e  r  s   [   d   B   ] -30-25-20-15-10-50S21S11 Fig. 7. Plot of measured RF response of FGCPW line with switchin off position. and/or space systems where low power consumption is es-sential.VI. A CKNOWLEDGMENTS The authors at the University of Michigan gratefully ac-knowledge the support of this research by the SOAC/JPLunder the CISM Project and the Department of Defense Re-search and Engineering (DDR&E) Multidisciplinary Univer-sity Research Initiative (MURI) on “Low Power Electronics”managed by the Army Research Office (ARO) under GrantDAAH04–96–1–0001.R EFERENCES [1] K. E. Petersen, “Micromechanical membrane switches on silicon,”  IBM Journal of Research and Development  , vol. 23, pp. 376–385,July 1971.[2] C. Goldsmith, T. H. Lin, B. Powers, W. R. Wu, and B. Norvell,“Micromechanical membrane switches for microwave applications,”in  IEEE MTT-S International Microwave Symposium Proceedings ,pp. 91–94, 1995.[3] J. J. Yao and M. F. Chang, “A surface micromachined miniatureswitch for telecommunications applications with signal frequenciesfrom dc up to 4 GHz,” in  The 8th International Conference on Solid-State Sensors and Actuators Digest  , pp. 384–387, 1995.[4] C. Goldsmith, J.Randall, S.Eshelman, and T. H. Lin, “Characteristicsof micromachined switches at microwave frequencies,” in  IEEE MTT-S International Microwave Symposium Proceedings , pp. 1141–1144,1996.[5] E. P. Popov,  Introduction to Mechanics of Solids . Prentice-Hall, 1968.[6] J. E. Shigley and L. D. Mitchell,  Mechanical Engineering Design .McGraw-Hill, 4 th ed., 1983.[7] M. Madou,  Fundamentals of Microfabrication . New York: CRCPress, 1997.[8] R. B. Marks and D. F. Williams, “Program multical,” tech. rep., NIST,August 1995. Rev. 1.[9] R. B. Marks, “A multiline method of network analyzer calibration,”  IEEE Transactions on Microwave Theory and Techniques , vol. 39,pp. 1205–1215, July 1991. 0-7803-5687-X/00/$10.00 (c) 2000 IEEE


Apr 28, 2018

Arvore Generosa

Apr 28, 2018
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!