when discontinuous sections are close. So, instead of using thosemethods, Ld and Cd in Table 1 were calculated by measuring theimage impedance. Then, the phase velocity of a discontinuoustransmission line is given by

vp �1

��Lc � Li � Ld��Cc � Ci � Cd�.

So, from Table 1 it is obvious that the discontinuous transmissionline is much shorter than the conventional transmission line for thesame electrical wavelength.

4. CIRCUIT LAYOUT AND MEASUREMENT RESULTS

Figure 4 shows the layout of the compact reflection-mode phaseshifter employing slow-wave microstrip lines. We used MA-COMMA46H120 varactor diodes, whose capacitance range is from 0.2to 1.1 pF. The circuit was made on Rogers RO3010 circuit board.Compared to conventional microwave circuits, the areas of the3-dB coupler and the reflection loads in this figure are reduced by45% and 35%, respectively. Figure 5 shows the measurementresults of insertion loss and the relative phase shift between 1.9 and2.1 GHz. Insertion loss is less than 3.5 dB for �300° phase shiftthroughout the measured frequencies, and phase shift range is 435°at the center frequency. The insertion loss in this structure mainly

arises from the series resistance of the varactors and the resistanceof the silver epoxy. The admittance range of the varactor diode isfrom 0.0025 to 0.0138 S at 2 GHz. These values result in anapproximately 360° phase shift (see Fig. 2). However, parasiticparallel capacitance between a device and the ground plane shouldbe added to the varactor capacitance range. This moves the admit-tance range to the right side of Figure 2 and increases the phase-shift range to be greater than the calculated one. The effect ofparasitic capacitance depends on where the original admittance islocated in this graph.

5. CONCLUSION

We have developed a reflection-mode phase shifter that exhibitslarge phase variation. Without using inductors, we can realize�300° phase shift with �3.5-dB insertion loss from 1.9 to 2.1GHz. The input and the output match was excellent due to the useof a 3-dB hybrid coupler. We also described a size-reductiontechnique that employs slow-wave microstrip lines. The sizes ofthe 3-dB coupler and the reflection loads were reduced by 45% and35%, respectively.

REFERENCES

1. J.I. Upshur and B.D. Geller, Low-loss 360° X-band analog phaseshifter, IEEE MTT-S Int Microwave Symp Dig 1 (1990), 487–490.

2. M. Kumar, R.J. Menna, and H.C. Huang, Broadband active phaseshifter using dual-gate MESFET, IEEE Trans Microwave Theory Tech29 (1981), 1098–1102.

3. J. Martel, R.R. Boix, and M. Horno, Equivalent circuits for MISmicrostrip discontinuities, IEEE Microwave Guided Wave Lett 3(1993), 408–410.

4. D.M. Pozar, Microwave engineering, Wiley, 1998.5. M.J. Webster, B. Easter, and J.S. Hornsby, Accurate determination of

frequency dependent three element equivalent circuit for symmetric stepmicrostrip discontinuity, IEE Proc Microwaves, Antennas Propagat 137H (1990).

6. A.F. Thomson and A. Gopinath, Calculation of microstrip discontinuityinductances, IEEE Trans Microwave Theory Tech 23 (1975), 648–655.

© 2005 Wiley Periodicals, Inc.

A COMPREHENSIVE INVESTIGATIONOF THERMAL TREATMENT EFFECTSON RESONANCE CHARACTERISTICS INFBAR DEVICES

Linh Mai, Hae-Il Song, Le Minh Tuan, Pham Van Su, andGiwan YoonSchool of EngineeringInformation and Communications University (ICU)119 Munjiro, Yusong-guDaejon 305-732, Korea

Received 27 May 2005

ABSTRACT: This paper presents some methods to improve the res-onance characteristics of film bulk acoustic-wave resonator (FBAR)devices. The FBAR devices were fabricated on Bragg reflectors.Thermal treatments were done using sintering and/or annealing pro-cesses. The measurement shows a considerable improvement of re-turn loss (S11) and quality factor (Qs/p). These thermal treatmentsseem very promising for enhancing the FBAR’s resonance perfor-mance. © 2005 Wiley Periodicals, Inc. Microwave Opt Technol Lett47: 459 – 462, 2005; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.21199

Figure 5 Measured insertion loss (top) and relative phase shift (bottom)

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 47, No. 5, December 5 2005 459

Key words: FBAR device; Bragg reflector; thermal annealing; Q-fac-tor; return loss; interfabrication annealing; post-annealing

1. INTRODUCTION

With the rapid growth of wireless communications in the range from0.5 to 6 GHz, there has been a strong demand for the integration ofhigh-frequency devices on a silicon wafer. Thin-film bulk acousticwave resonator (FBAR) devices seem very suitable for microwavemonolithic integrated circuits (MMICs) since they can be realized onSi or GaAs substrates [1]. The basic FBAR structure consists of apiezoelectric thin-film sandwiched between top and bottom elec-trodes, where a resonance occurs when an electric field is applied ontoelectrodes [2]. Therefore, the piezoelectric thin-film plays a criticalrole in determining the resonance characteristics of the FBAR de-vices. The solidly mounted resonator (SMR) [3] has a Bragg reflectoras a mirror, usually fabricated by alternately depositing both high- andlow-impedance materials. In spite of some efforts to improve theFBAR device characteristics [4], no comprehensive investigation ofvarious thermal treatments on the FBAR devices has been reported.

In this paper, a comprehensive study on the thermal treatmenteffects on the resonance characteristics of the FBAR devices ispresented for the first time.

2. EXPERIMENT

Figure 1 shows the schematic structure of the FBAR device, wherea Bragg reflector (BR) of seven-layered SiO2 and W (SiO2/W)films was alternately deposited on a silicon wafer by using an RFmagnetron sputtering technique. The 0.6-�m-thick W films weredeposited under Ar gas pressure of 15 mTorr with DC power of150 W, and the 0.6-�m-thick SiO2 films were deposited under Argas pressure of 4 mTorr with RF power of 300 W. Then, the siliconwafer with the BR was divided into five samples (N1–N5). Thesamples were used to fabricate the FBAR devices. In order toinvestigate the thermal treatment effects, four samples (N2–N5)were thermally treated under different annealing conditions,whereas one sample N1 was not thermally treated in order to use

it as a reference sample. The first thermal annealing was carriedout as follows: only two samples (N3 and N4) were sintered for 30min at 400°C in ambient air [4] while the sample (N5) was sinteredfor 30 minutes at 400°C, but in Ar gas that was ambient in anelectric dehydrate furnace (EDF) equipment. Then, 0.2-�m-thickcobalt (Co) bottom electrodes (used for floating grounds) weresimultaneously deposited on all samples under 20-mTorr Ar gaspressure and with 150-W DC power. On the bottom electrode,1.2-�m-thick ZnO film was deposited under 10 mTorr of Ar/O2,and with RF power of 300 W. Immediately after the ZnO filmdeposition, the second thermal annealing (called interfabricationannealing) was performed for the three samples (N2, N4, and N5)all for 60 min at 200°C in the ambient Ar. The deposition andpatterning of the top Co electrodes (0.2 �m) on top of the ZnO film

TABLE 1 Thermal Processes

Thermal Steps

Samples

R1 R2 R3 R4 R5

First BR annealing at 400°C/30min — — Air Air Ar

Second interfabrication annealing at200°C/60 min — Ar — Ar Ar

Third post-annealing at 200°C/120min — Ar Ar Ar Ar

Figure 2 Return-loss characteristics vs. frequency for various thermalprocesses: (a) pattern 1; (b) pattern 2; (c) pattern 3

Figure 1 FBAR device structures: (a) cross-section view of the fabri-cated FBAR device; (b) 3D schematic structure of the FBAR device

460 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 47, No. 5, December 5 2005

completed the FBAR device fabrication. As a result, the fiveFBAR device samples (R1, R2, R3, R4, and R5), each correspond-ing to the BR samples (N1, N2, N3, N4, and N5), were obtained.

All FBAR devices were measured in order to extract the returnloss (S11) using a probe station and a Hewlet Packard/HP 8722Dnetwork analyzer. Then, the four samples (R2, R3, R4, and R5)were additionally post-annealed in EDF equipment at 200°C for120 min and characterized. The thermal treatment conditions forthe five samples are summarized in Table 1.

3. RESULTS AND DISCUSSION

Figure 2 shows the three resonator patterns and their return-loss(S11) characteristics versus frequency for various annealing con-ditions. Figures 2(a)–2(c) compare the return loss characteristics ofthe five FBAR devices (R1–R5) with the same resonator pattern,fabricated each on the Bragg reflectors (N1–N5), respectively. TheS11 values comparison clearly shows the advantage of the BRannealing effect, thereby confirming the result reported in [4]. Thereturn-loss values for the three resonator patterns have the sameincreasing trend from the resonators R1 and R2 (nonannealed BR),the resonators R3 and R4 (annealed BR in air), and the resonatorR5 (annealed BR in Ar). Undoubtedly, thermal annealing can beone of important factors to enhance the return-loss characteristics.From Figure 2(c), the resonators R1 and R2 have smallest return-loss values (S11 � �16.55 dB and S11 � �17.20 dB, respec-tively). Meanwhile, the return-loss values of the samples R3, R4,and R5 are S11 � �20.34, �24.7, �31.9 dB, respectively. Thereason why the return loss values of the R1 and R2 are smaller thanR3, R4, and R5 can be explained as in [4]. The resonator R3(fabricated on the BR N3) has about 3-dB smaller return loss thanthe resonator R4 (fabricated on the BR N4 and added interfabri-cation annealing process). The return-loss values of R3 and R4 aresmaller than that of R5 by about 10 and 7 dB, respectively. Table2 shows the extracted values S11 at a frequency of 1.833 GHz.According to [4], the quality of BRs influences the FBAR char-acteristics. Inside the original SiO2/W multiplayer BR, there mayexist some physical imperfections in the film microstructuresand/or some imperfect adhesions at interfaces between the physi-cally deposited films, thus degrading the device performances.

These physical imperfections and imperfect adhesions also exist inthe physical structure of the resonator. In order to resolve theseissues, the first step of the BR-annealing process and the secondstep of the interfabrication annealing process were applied. As aresult, the resonators showed better resonance characteristics. Fi-nally, there is one more thermal-process step (called the post-annealing process) for further improving the FBAR device char-acteristics in this experiment. To investigate any possible influenceof the post-annealing on the resonator properties, the four resona-tor samples with the same layout pattern 1 (R2 to R4) werepost-annealed in EDF equipment in Ar gas that was ambient for120 min. The return losses of the samples were extracted, as shownin Table 3. The return loss of sample R1 is shown for reference.

The post-annealing process seems to significantly improve theresonance characteristics of the four samples (R2–R5). The in-creased value (��S11� � S11�after � S11�before) of the return lossfrom samples R2 to R4 are 4.85, 7.15, 7.31, and 8.35 dB, respec-tively. Clearly, the post-annealing process seems to affect thesandwiched structure of resonator. It is speculated that the sand-wiched structure of the as-deposited multilayers (Co/ZnO/Co) mayhave some physical imperfections, such as incomplete adhesionsand micro defects in the film itself. The additional use of thepost-annealing process is believed to be able to eliminate or reducethe physical imperfections occurring during the thin-film deposi-tion. This eventually leads to the improvement of the FBAR deviceperformances.

The performance of the FBAR devices can be determined bythe figure of merit (FOM) [5] in terms of Q-factor. Based on anempirical definition [6], the series/parallel resonance Q-factors(Qs/p) were calculated, as shown in Table 4. The resonatorspost-annealed show larger Q-factor, as compared to those nonpost-annealed.

4. CONCLUSION

In this paper, the resonance characteristics of ZnO-based FBARdevices were comprehensively studied for various thermal treat-ments (the Bragg-reflector annealing process, interfabrication an-nealing process, and post-annealing process) and their combina-tions. The optimal combination of the various thermal treatmentsappears to significantly improve the return loss and Q-factors ofthe FBAR devices.

REFERENCES

1. J.H. Collins, A short history of microwave acoustics, IEEE TransMicrowave Theory Tech MTT-32 (1984), 1127–1140.

2. S.V. Krishnaswamy, J.F. Rosenbaum, S.S. Horwitz, and R.A. Moore,Film bulk acoustic wave resonator and filter technology, IEEE MTT-SDig, Albuqueque, NM (1992), 153–155.

3. K.M. Lakin, K.T. McCarron, and R.E. Rose, Solidly mounted resona-tors and filter, IEEE Proc Ultrasonics Symp (1995), 905–908.

4. M. Yim, D.H. Kim, D. Chai, and G. Yoon, Significant resonancecharacteristic improvements by combined used of thermal annealing

TABLE 2 Return Loss Values of Five Types of Resonatorwith Different Patterns

Patterns

Samples

Return Loss S11 [dB]

R1 R2 R3 R4 R5

Pattern 1 �18.40 �22.43 �24.60 �26.61 �33.02Pattern 2 �19.34 �21.83 �24.18 �27.10 �30.69Pattern 3 �16.55 �17.20 �21.34 �24.70 �31.90

TABLE 3 Return-Loss Values Comparison

Samples

Conditions

��S11�

Return Loss S11 [dB]

NonannealingBefore Post-Annealing

After Post-Annealing

R1 �17.40 — — —R2 — �22.43 �27.01 4.58R3 — �23.60 �30.75 7.15R4 — �25.61 �32.92 7.31R5 — �32.26 �40.61 8.35

TABLE 4 Effect of Thermal Annealing on Quality Factors

Samples

Conditions

Before Post-Annealing After Post-Annealing

Qs Qp Qs Qp

R2 4018 4453 4219 4649R3 4391 6919 4639 7053R4 4719 5482 5073 5984R5 5248 5693 6475 5956

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 47, No. 5, December 5 2005 461

and Co electrode in ZnO-based FBARs, IEE Electron Lett 39 (2003),1638–1640.

5. K.M. Lakin, G.R. Kline, and K.T. McCarron, High-Q microwaveacoustic resonators and filters, IEEE Trans Microwave Theory Tech 41(1993), 2139–2146.

6. S.H. Park, B.C. Seo, H.D. Park, and G. Yoon, Film bulk acousticresonator fabrication for radio frequency filter applications, JapaneseJ Appl Phys 39 (2000), 4115–4119.

© 2005 Wiley Periodicals, Inc.

A COMPACT PSEUDO-INTERDIGITALSIR BANDPASS FILTER WITHIMPROVED STOPBAND PERFORMANCE

Min-Hung Weng,1 Hung Wei Wu,2 and Cheng-Yuan Hung2

1 National Nano Device LaboratoriesTainan, Taiwan2 Department of Computer and CommunicationSHU TE UniversityKaohsiung, Taiwan

Received 19 May 2005

ABSTRACT: A pseudo-interdigital stepped impedance resonator (SIR)bandpass filter (BPF) using tapped I/O is designed and fabricated. Byusing a pseudo-interdigital structure and the SIR, the proposed BPFgains advantages of compactness and high selectivity. In this paper, thefirst and second spurious responses are shifted to higher than 3f0 (thedesigned center frequency) due to the SIR effect and the effect of usingthe tapped I/O, and fast passband-edge attenuation is obtained as a re-sult of the transmission zeros in the passband edge created by interstagecoupling. The experimental results of the fabricated filter show goodagreement with the simulation results. © 2005 Wiley Periodicals, Inc.Microwave Opt Technol Lett 47: 462–464, 2005; Published online inWiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.21200

Key words: transmission zero; tapped I/O; microstrip filter; steppedimpedance resonator; pseudo-interdigital

1. INTRODUCTION

Conventional microstrip interdigital bandpass filters (BPFs) arecompact, while they require short-circuit connections with viaholes, which is not quite compatible with planar fabrication tech-niques [1–4]. A new type of miniaturized microstrip BPF using apseudo-interdigital structure without via holes ground was re-ported in [5]. The pseudo-interdigital BPF has a more compactsize, and an attenuation pole due to the effect of interstage cou-pling. However, there is still a problem due to its requiring moreresonators in order to improve the skirt characteristics. In a generalfilter-design procedure, greater numbers of resonators are requiredto obtain a more rapid attenuation rate outside the passband.However, an increase in the resonator numbers increases not onlythe insertion loss in the passband, but also the filter size [1]. Inaddition, pseudo-interdigital BPFs have been prepared; however,the wideband characteristics have not been reported in detail [5]. Itis expected that the pseudo-interdigital BPFs using half-wave-length resonators will suffer from spurious response at 2f0 and 3f0

( f0 is the designed center frequency). It is therefore desirable toimprove the pseudo-interdigital BPFs which actually meet therequirements of both small size and wider stopband.

In this paper, we report a pseudo-interdigital BPF using tappedinput/output (I/O) and stepped impedance resonator (SIR) to im-prove the skirt characteristics and stopband rejections, as shown in

Figure 1. The BPF is able to place transmission zeros near thepassband edge so that higher selectivity with less resonators can beobtained. Using SIRs as the building blocks, the first and secondspurious responses can be shifted to a higher frequency [6]. Also,by using tapped-line I/O, space and cost are saved compared toconventional filter types, since the first and last end-sections of thefilter can be reduced [7]. This design filter is then implemented ondouble dielectric layers of standard FR4 in order to verify theprediction of our design.

2. DESIGN OF PSEUDO-INTERDIGITAL SIR FILTER

Figure 2 illustrates development of the pseudo-interdigital filterstructure. Each resonator element in a conventional interdigitalfilter is a quarter-wavelength long at the midband frequency and isshort-circuited at one end and open-circuited at the other end, asshown in Figure 2(a). In the past, a via-hole to the ground planewas normally used for the short-circuit connection on the micro-strip. Hong suggested that the short-circuit connection may bethusly connected without severe distortion of the bandpass fre-quency response, so as to achieve the modified interdigital filter, asshown in Figure 2(b). As the via-holes are removed, the modifiedinterdigital filter formed becomes a pseudo-interdigital filter, asshown in Figure 2(c). Theoretically, the important feature of thepseudo-interdigital BPF is that there is not much change in thevoltage and current distributions in the vicinity of the midbandfrequency. To further improve the stopband performance, theresonators in the pseudo-interdigital filter use the SIR; therefore,the proposed pseudo-interdigital filter can be seen as a combina-tion of the interdigital structure and the SIRs.

Figure 3 illustrates the typical structure of the SIR underconsideration. The fundamental resonance condition of SIRs iswell known [6]. The admittance of the resonator from the openend, Yi, is given by

Yi � jY2

�2�K tan �1 � tan �2� � �K � tan �1 � tan �2�

K�1 � tan2�1� � �1 � tan2�2� � 2�1 � K2� � tan �1 � tan �2, (1)

where the stepped impedance ratio K � Z2/Z1. The resonancecondition can be obtained as Yi � 0. The resonance condition andthe spurious-suppression ability are designed by setting a suitableelectrical length of the SIR and the K value. As K � 1, the SIRwill have a larger physical length and poor spurious-shifting abil-

Figure 1 Practical layout of the designed BPF

462 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 47, No. 5, December 5 2005