PIERS ONLINE, VOL. 4, NO. 4, 2008 433

RF MEMS Extended Tuning Range Varactor and Varactor BasedTrue Time Delay Line Design

Yaping Liang1, C. W. Domier2, and N. C. Luhmann, Jr.2

1Department of Electronics and Information Engineering, Hangzhou Dianzi University, China2Department of Applied Science, University of California, Davis, USA

Abstract— MEMS varactors are one of the important passive MEMS devices. Their applica-tions include use in VCOs, tunable impedance matching networks, tunable filters, phase shifters,and true time delay lines. The shunt capacitive structure has been employed in most of theconventional MEMS varactor designs because of its simplicity. However, the capacitance ratio ofthis conventional shunt capacitive MEMS varactor is limited to 1.5 because of the MEMS Pull-In effect, which happens when the deflection between the MEMS top and bottom metal platesincrease beyond 1/3 of the airgap between the two metal plates. At that time, the top metalplate will quickly snap down. This effect is the major limitation in MEMS varactor designs andcan cause nonlinearity and mechanically instability. In order to eliminate this Pull-In effect, theauthor employed the so-called MEMS extended tuning range structure. This structure utilizesa variable height top metal beam with separate actuation parts. The airgap between the centerpart of the top beam and the bottom plate has been designed to be less than 1/3 of the airgapbetween the top beam and the bottom actuation pads. When DC bias is applied to the actuationparts, the entire top beam will move down together. Consequently, before the Pull-In effecthappens at the actuation parts, the center part has already traveled through its entire tuningrange, which means that the capacitive ratio of this kind of MEMS varactor can go to infinity.

A fabrication process employing a GaAs substrate has been designed based on surface micro-machining technology. The maximum capacitance ratio of the designed MEMS extended tuningrange varactor is 5.39 with a Cmax value of 167 fF. Based on this MEMS varactor design, aKa-band MEMS varactor based distributed true time delay line has been designed. This dis-tributed true time delay line includes a high impedance CPW transmission line with 70 Ω un-loaded impedance at 28 GHz and eight MEMS extended tuning range varactors based on thevaractor design periodically loaded on the CPW line. The testing results show that a 56 phasedelay variation has been achieved at 28 GHz. The measured insertion loss at 28 GHz is −1.07 dBat the up-state and −2.36 dB at the down-state. The measured return losses, S11 and S22, areboth below −15 dB at 28GHz and below −10 dB over the entire tested frequency range of 5GHzto 40 GHz.

1. INTRODUCTION

MEMS varactors are one of the important passive MEMS devices. They have considerable ad-vantages compared with other semiconductor devices, including low loss, very high Q at mm-wavefrequencies, high power handling capability, low power consumption, and high IIP3. The RFMEMS varactor can be employed in a phase shifter or true time delay line design to replace theGaAs Schottky varactor diode for low-loss, broadband, and high frequency applications in mod-ern communication, automotive and defense applications. It can also be used in low loss tunablecircuits including matching networks, tunable filters, and low noise oscillators.

2. RF MEMS EXTENDED TUNING RANGE VARACTOR

Conventional RF-MEMS varactors usually employ a shunt parallel plate capacitor whose capac-itance is determined by the spacing between a fixed bottom plate and a movable suspended topplate. Electrostatic actuation occurs when an electrostatic force is created by applying a DC voltagebetween the capacitor plates, thereby displacing the movable plate toward the fixed plate. How-ever, this shunt capacitance MEMS varactor structure suffers from the so-called Pull-In effect [1].It happens when the displacement between the two metal plates exceeds 1/3 of the entire air-gap. At that moment, the electrostatic attraction force loses balance with the mechanical restoringforce and that causes the two metal plates to quickly snap into contact. The Pull-In effect is themajor limitation in MEMS varactor designs. It will cause nonlinearity and mechanical instability

PIERS ONLINE, VOL. 4, NO. 4, 2008 434

Figure 1: Model of a MEMS extended tuning range varactor structure.

of the MEMS varactors. In order to avoid the snap down, the designed capacitance ratio of theconventional MEMS capacitive varactor is usually set to 1.2 to 1.5 [2].

In order to eliminate this Pull-In effect, one approach is to employ the so-called MEMS extendedtuning range structure [3]. This structure, as shown in Figure 1, utilizes a variable height top metalbeam E1 with separate actuation parts E3. The airgap between the center part of the top beamE1 and the bottom plate E2 has been designed to be less than 1/3 of the airgap between the topbeam E1 and the bottom actuation pads E3. When DC bias is applied to the actuation parts, theentire top beam E1 will move down together. Consequently, before the Pull-In effect happens atthe actuation parts, the center part has already traveled through its entire tuning range, whichmeans that the capacitive ratio of this kind of MEMS varactor can theoretically approach infinity.

A MEMS extended tuning range varactor has been designed at 28 GHz on a GaAs substrate byusing the Ansoft HFSS and Agilent ADS simulation tools. Figure 2 shows the designed five-maskfabrication process. The most important and difficult step in building this extended tuning range

Figure 2: Fabrication process steps.

PIERS ONLINE, VOL. 4, NO. 4, 2008 435

structure is to form the variable height top metal beam E1. Here, it has been realized by spinningtwo layers of photoresist continuously with different masks to pattern, see Figure 2(c) and (d). Thefirst step (Figure 2(a)) is to evaporate 0.7µm of gold to form the signal lines and actuation padsusing a gold lift-off process. The second step (Figure 2(b)) is to use PECVD to deposit 3000 A ofSi3N4 and to use dry Reactive Ion Etching (RIE) to form the dielectric layer between the bottomand top metal beams. The third step (Figure 2(c)) is to spin 1µm thick photoresist and patternthe anchor points. The fourth step (Figure 2(d)) is to spin another 2µm thick photoresist layerand pattern the anchor points and the center lower beam E2. The fifth step (Figure 2(e)) is toelectroplate 2 µm of gold and use photolithography to form the upper beam E1. The final step(Figure 2(f)) is to use a dry etch to remove the sacrificial layer and release the whole structure.

Top beam

CPW transmission line

Figure 3: SEM picture of a MEMS extended tuningrange varactor.

Figure 4: On-wafer C-V testing results.

Figure 3 shows an SEM picture of one of the fabricated MEMS varactors. On-wafer mea-surements by using an HP 4279A C-V meter have been employed and the results show that themaximum capacitance ratio is 5.39 with a Cmax value of 167 fF (see Figure 4).

Figure 5: SEM picture of a MEMS varactor basedtrue time delay line.

sLline

sCline

CVaractor

s

Figure 6: Equivalent circuit of unit section LC lad-der network.

3. RF MEMS VARACTOR BASED TRUE TIME DELAY LINE

RF MEMS varactor based true time delay line technology employs a distributed LC ladder structureby parallel loading the MEMS varactors on high impedance coplanar waveguide (CPW) transmis-sion lines. Figure 5 shows an SEM picture of a portion of one of the fabricated MEMS extendedtuning range varactor based true time delay lines. The unit section equivalent circuit of the dis-tributed LC ladder network is shown in Figure 6. When the operation frequency is far below the

PIERS ONLINE, VOL. 4, NO. 4, 2008 436

Bragg cutoff frequency of the LC ladder network, the group velocity remains essentially constantas the frequency is varied [4].

This MEMS varactor based true time delay line comprises 8 MEMS extended tuning rangevaractors loaded on a 70 Ω CPW transmission line operated at 28 GHz. The on-wafer testingresults show that the insertion loss at 28 GHz is −2.36 dB in the down-state; the return loss, S11

and S22, are both below −15 dB at 28 GHz (see Figure 7). The measurement phase delay is 56 at28GHz (see Figure 8).

Retu

rn L

oss (

dB

)

S21

S11

S22

Freqency (GHz)

Figure 7: Measured down-state S-parameters of theMEMS varactor based true time delay line.

In

sert

ion L

oss (

dB

)Freqency (GHz)

Figure 8: S21 phase delay.

4. CONCLUSIONS

A novel RF MEMS extended tuning range varactor structure has been employed to eliminate thePull-In effect of the conventional MEMS varactor designs. On-wafer measurement results showthat the maximum capacitance ratio is 5.39 for the extended tuning range MEMS varactors. A28GHz proof-of-principle MEMS varactor based true time delay line design employed the MEMSextended tuning range varactor structure. The maximum phase delay is 56 with a usable rangeextending from 5 to 40GHz over which the line has demonstrated both low insertion loss and highreturn loss.

ACKNOWLEDGMENT

The authors are grateful to Miao Lu, Xiaodong Hu, and Yongjun Yan, of the Hebei SemiconductorResearch Institute in China, for fabricating our MEMS varactors and delay lines. The authorswould also like to thank Mehmet Ozgur and Michael Huff of the MEMS and NanotechnologyExchange, funded by DARPA, for fabricating additional MEMS devices and circuits. This workwas supported in part by the U.S. Department of Defense under Grant No. NBCH1050014 and bythe U.S. Department of Energy under Grant No. DE-FG02-99ER54531.

REFERENCES

1. Senturia, S. D., Microsystem Design, Kluwer Academic Publishers, Boston, MA, 2001.2. Rebeiz, G. M., RF MEMS Theory, Design, and Technology, John Wiley & Sons, Inc., Hoboken,

New Jersey, 2003.3. Zou, J., C. Liu, et al., “Development of a wide tuning range MEMS tunable capacitor for

wireless communication system,” Technical Digest of International Electron Devices Meeting,403–406, 2000.

4. Hsia, R. P., “Nonlinear transmission lines and applications,” PhD Dissertation, UC Davis,1997.