RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes

Transcription

1 RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes R. Kulke, W. Simon, C. Günner, G. Möllenbeck, D. Köther, M. Rittweger IMST GmbH Carl-Friedrich-Gauss-Str. 2 D Kamp-Lintfort, Germany kulke@imst.de, Abstract LTCC multilayer ceramic has been tested in several RF applications in the past. The benefits are obvious: high level of integration, buried components, low losses, robustness and others. LTCC has been established in mobile communication techniques up to a few GHz like GSM power amplifiers and frontend modules or Bluetooth transceiver circuits. RF design engineers are discovering LTCC more and more for higher frequencies. Some examples are WLAN at 5 GHz, RADAR sensors for automotive at 24GHz or even at 77GHz and several frequency bands for digital radio networks operating from about 20GHz to 60GHz. However, the RF-behaviour of LTCC waveguides in the upper GHz-range has barely been tested. The industry is missing a comparative test between various LTCC substrates. Therefore the consortium of the German R&D project EASTON, funded by the German Ministry of Education and Research (BMBF) and coordinated by the German Space Agency (DLR), decided to launch such a comparative evaluation of LTCC substrate systems up to 40GHz. The consortium has got great support from material suppliers, foundry services and research institutions, which are acknowledged at the end. Key words: LTCC, Multilayer Ceramic, RF and Microwave Tests, Material Characterization Design and Manufacturing The design of test patterns has been developed by the authors. The goal was to achieve the same conductor line width for all different LTCC substrate systems in case of microstrip (MS) and stripline (SL) waveguides to ensure a reliable comparison of the materials. The test layout covers 3 MS and 3 SL ring-resonators, a number of calibration standards, meander lines for DC characterization and further test patterns. The 50 MS lines have a line width of 390 m, while the buried 50 SLs (shielded with via fences) have got a width of 145 m. Figure 1 illustrates the test layout that has been delivered for manufacturing. Three substrate layers are required, one for MS and two for SL waveguides. Each LTCC system has its specific dielectric constant as well as several standard thickness. The authors performed a number of numerical simulations to determine the required tape thickness for each LTCC system. The example of DuPont s tape 951 is illustrated in figure 2. With a permittivity of 7.8 and the standard tapes A2 and AX a substrate thickness of h MS = 330 m for MS lines and h SL = 660 m for SLs has been calculated. This results in a characteristic line impedance for MS: Z L = 50.1 and for SL: Z L = All other LTCC systems, which have been investigated are listed in table 1 and 2 (calculated with Agilent s LineCalc). Again, only standard tapes, which are available on the market have been taken into account. h / m k = r eff Z L / DuPont DuPont Heraeus CT Ferro A6M,S Samsung G NIKKO Ag Table 1. Summary of MS line properties: width = 390 m, f = 20GHz h / m k = r = eff Z L / DuPont DuPont Heraeus CT Ferro A6M,S Samsung G NIKKO Ag Table 2. Summary of SL line properties: width = 145 m, f = 20 GHz R. Kulke et al.: RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes 1

2 Microstrip Calibration Lines Micostrip Ring- Resonators w = 390 m Coupled Striplines X = 66 mm Y = 66 mm Stripline Ring- Resonators w = 145 m Microstrip to Stripline Transitions Conductivity Measurement X = 0 Y = 0 Stripline Calibration Kit Resolution Test Lines Figure 1: Layout of microstrip and buried stripline test patterns (left), photos DuPont 943 (right) h = 130 m h = 200 m h = 200 m h = 130 m h = 130 m h = 200 m microstrip line ground stripline ground DuPont 951: r = 7.8 Conductors Vias Cavities Figure 2. Layer set-up for MS and SL waveguides for DP951-A2 and AX tape LTCC Tape Conductor Manufacturer Samsung G6 silver Samsung: Korea Nikko Ag3 gold Nikko: Japan Ferro A6-S silver Kyocera: USA Ferro A6-M gold Kyocera: USA DuPont 943 silver VTT: Finland DuPont 943 gold VTT: Finland DuPont 951 silver TU-Ilmenau, IMST: Germany DuPont 951 gold TU-Ilmenau, IMST: Germany Table 3. Delivered LTCC systems for characterization and evaluation A requirement for this investigation was to obtain interest and support of the LTCC tape suppliers, because there was no budget in the project to order the fabrication of the test patterns. Great response came from the LTCC manufacturer who organized the fabrication in their own labs or with partners. Finally 8 different LTCC systems have been delivered for characterization and evaluation. Table 3 lists the available parts sorted by delivery time. CT2000 is awaited. Characterization and Evaluation All parts have been delivered to IMST for characterization and evaluation. The investigation starts with the measurement of the geometry parameters: substrate height, MS line width and thickness, SL width and thickness, via diameters and positions, conductor diffusion into the substrate s surface and circuit geometry like ring, gap or line length dimensions. The surface roughness will be measured by partner TU- Ilmenau. The electrical tests cover the DCconductivity and the S-parameters up to 40GHz (on-wafer). The method for the evaluation of ring-resonators is established and already published in details [1 5]. Four main results can be obtained from this investigation: 1. Quality factor: Q 2. Effective permittivity of MS line: eff 3. Relative permittivity of SL: r 4. Total line losses for MS (w = 390 m) and SL (w = 145 m): MS and SL Figure 6 and 7 show the results of MS and SL total line losses. This includes different kinds of losses resulting from conductivity, dielectric losses, surface roughness and radiation. A theoretical partitioning of the loss contributions is published in [1]. This first evaluation here R. Kulke et al.: RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes 2

3 concentrates on the low loss LTCC systems: Ferro A6, DuPont 943 and Samsung G6. It is well known, that the losses depend on the line width of the conductors. With an increasing width the losses decrease. The goal was to achieve a width of 390 m for the MS waveguide. The tolerances of the delivered parts can be read from figure 6, too. A range from 385 m to 445 m is covered (average values). In the case of the SL losses the evaluation of the measurements is limited by the influence of the MS-to-SL transition. The on-wafer probe is contacted on a microstrip line in a cavity of the ceramic tapes. Figure 7 illustrates this configuration in a cross-view. This transition has been optimised with the 3D full-wave analysis tool EMPIRE from IMST. The transition works fine from DC to 30GHz. Beyond this limit the return losses (S 11 ) increase drastically. The buried stripline itself shows excellent properties even at higher frequencies. However it is difficult to find one common transition, which works for all LTCC tapes with different dielectric constant and different tape thickness. This problem limits the upper frequency of evaluation for a comparative test of various LTCC stripline waveguides. Nevertheless, the results are very smooth and promising. In addition the authors could demonstrate the benefits of buried and shielded stripline components in the frequency band up to 30GHz: low losses, no radiation, negligible cross-talk and high degree of integration. Figure 9 and 10 illustrate the determined permittivities of the MS and SL ring resonators. In case of the MS configuration the electromagnetic field is guided in air and in the dielectric substrate. Thus, the dielectric constant of a MS line is always greater than the effective permittivity ( r > r,eff ). r,eff (f) also depends of the frequency. An increasing frequency drags the wave into the substrate, which is called dispersion. A SL is a dispersion free waveguide, because the wave is guided in the homogenous substrate only. That s why the measured effective permittivity is equal to the dielectric constant, which has been demonstrated with the data in figure 10. Meander lines have been designed with a width of 150 m and 600 m to measure the DC conductivity of the screen printed conductors. Each meander has an average length of 15cm. The real geometry dimensions (l = length, w = width) and the line resistances R have been determined. As a result the square (surface) resistivity R = (Rw)/l is listed in table 4 split into silver and gold conductors. The line thickness has not been measured and therefore not taken into account. The surface roughness, the metal s granularity and the conductor edge smoothness may influence the DC measurement. See the microscope photos in figure 3 and 4 to get an impression of screen printed lines. LTCC Tape Cond. R / m R / m w=150 m w=600 m Samsung G6 silver Ferro A6-S silver DuPont 943 silver DuPont 951 silver Nikko Ag3 gold Ferro A6-M gold DuPont 943 gold DuPont 951 gold Table 4. Measured surface resistivity of Au and Ag screen printed conductors Conclusion and Outlook A test procedure has been launched to compare the microwave behaviour of various LTCC material systems. This includes standard ceramic tapes from different suppliers as well as the matched conductor pastes with gold and silver. A common test pattern design has been developed and delivered to manufacturing labs. This enables the comparison of the LTCC systems under nearly the same conditions. A number of measurements has been performed to determine the geometry parameters, the DC conductivity and the S-parameters up to 40GHz. The results of low loss LTCC tapes have been evaluated to achieve the quality factor, the effective permittivity and the total line losses, which are in a range of 0.1 to 0.6dB/cm from 5 to 40GHz. This investigation makes obvious, that LTCC has a strong potential in RF and microwave applications. It also shows, that various substrate materials are available on the market, which fulfil the demands. Worldwide foundry services and even smaller companies and institutions offer an accurate and reliable manufacturing of modules from sample to volume quantities. 40GHz is not the limit for LTCC as a substrate system for microwave modules. This upper frequency was dictated by the selected line resolution of screen printed conductors. It has been decided to utilize microwave on-wafer probes with a pitch distance of 450 m, which are specified up to 40GHz. This RF-benchmark initiative should be a first step of an ongoing investigation, where increased frequencies are only one goal to characterise and improve R. Kulke et al.: RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes 3

4 LTCC. New technologies for fine line conductors should be tested among other things: There are enhanced screen printers to print conductors with 30 m line width and spacing. There are photoimageable pastes available for processing inner and outer photodeveloped conductors. And there is the possibility for etched and thin-film outer conductors on LTCC. These promising developments should be evaluated with scientific care to open the field of applications up to 80 GHz or even 110 GHz?! An essential requirement for this exploration is the cooperation of the industry with material suppliers, foundry services and research institutions, which benefits all kind of applications. References [1] R. Kulke et al.: Investigation of Ring- Resonators on Multilayer LTCC, workshop contribution International Microwave Symposium IMS 2001, Phoenix, May 2001 [2] E. Kemppinen: Determination of the Permittivity of some Dielectrics in the Microwave and Millimeterwave Region, PhD. Thesis, Ouolu 1999 [3] R. K. Hoffmann, Intergrierte Mikrowellenschaltungen, Springer Verlag 1983 [4] P. Barnwell et al.: An Investigation of the Properties of LTCC Materials and Compatible Conductors for their use in Wireless Applications, IMAPS 2000, Boston [5] S. Vasudevan et Al.: The Effect of Processing Parameters and Surface Roughness of Conductor Traces on Microwave Properties of LTCC Structure, ISHM Proceedings 1996 Acknowledgement This investigation is supported by the German Ministry of Education and Research (BMBF) and the German Space Agency (DLR) in the research project EASTON (FKZ 50 YB 0007): Divider-Network Development with Integrated Antenna on Multilayer LTCC. Siegfried Voigt: DLR, Germany Heiko Thust, Karl-Heinz Drüe and Ralph Münnich: University of Ilmenau, Germany Matthias Spinnler: TESAT Spacecom, Germany A powerful support in manner of material delivery, test pattern manufacturing and helpful discussions was given by leading LTCC companies and namely the dedicated people: Vern Stygar and Liang Chai: Ferro, USA Eric Ness: Kyocera (Vispro Division), USA Erich Polzer and Gerard Vanrietvelde: DuPont, Germany and France Kari Kautio: VTT, Finland Yun-Hwi Park: Samsung, Korea Akira Jyogo: Nikko, Japan Peter Barnwell: Heraeus, England Figure 3. Corners of screen printed gold conductors with a 500-fold magnification Figure 4. Corners of screen printed silver conductors with a 500-fold magnification R. Kulke et al.: RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes 4

5 Figure 5. Microscope photo of coupling gap in MS gold ring resonator MS Line Losses [db/cm] Ferro A6M, Au w = 390 m Ferro A6S, Ag w = 445 m Samsung G6, Ag w = 370 m DuPont 943, Au w = 385 m DuPont 943, Ag w = 385 m MS Target Line Width = 390 m Figure 6. Measured line losses of microstrip lines SL Line Losses [db/cm] Ferro A6M, Au Ferro A6S, Ag Samsung G6, Ag DuPont 943, Au DuPont 943, Ag SL Target Line Width = 145 m 0.3 Cross Section Photo MS SL Figure 7. Measured line losses of buried striplines R. Kulke et al.: RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes 5

6 Figure 8. Microscope photo of coupling gap in MS silver ring resonator MS: Effective Permittivity eff = 5.35 w = 385 m 5 eff = 4.3 w = 445 m w = 390 m eff = 4.6 Ferro A6M, Au Ferro A6S, Ag Ferro A6S, Ag-II Samsung G6, Ag w = 370 m w = 440 m r = 5.9 r = 6.3 DuPont 943, Au MS Target Line Width = 390 m r = 7.5 DuPont 943, Ag Figure 9. Measured effective permittivity of microstrip lines 8 SL Dielectric Constant SL Target Line Width = 145 m Ferro A6M, Au Ferro A6S, Ag-II Samsung G6, Ag DuPont 943, Au DuPont 943, Ag r = 5.9 r = 6.3 r = Figure 10. Measured dielectric constant of buried striplines R. Kulke et al.: RF-Benchmark up to 40 GHz for various LTCC Low Loss Tapes 6