LTCC Technology for Planar Microwave Antenna Systems · LTCC is a well established ceramic multilayer packaging technology for high reliability applications. Advanced low-loss material

LTCC Technologyfor

Planar Microwave Antenna Systems

Von der Fakultät für Ingenieurwissenschaften,Abteilung Elektrotechnik und Informationstechnik

der Universität Duisburg-Essen

zur Erlangung des akademischen Grades

Doktor der Ingenieurwissenschaften

genehmigte Dissertation

von

Peter Uhligaus

Amtzell

Gutachter: Prof. Dr. Daniel ErniGutachter: Prof. Dr. Ingo Wol↵Gutachter: Prof. Dr. Jens Müller

Tag der mündlichen Prüfung: 14. November 2018

Contents

Abstract iii

Zusammenfassung 1

1 Introduction 31.1 Low Temperature Co-fired Ceramic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 Technical Ancestry of LTCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.3 State of the Art in LTCC Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.4 Objective of this Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101.5 Outline of this Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2 DBF Antenna for Ka-Band Satellite Communication 152.1 Antenna Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.2 Antenna Realization in LTCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3 Cavities and Micro-Channels 253.1 Requirements of Microwave Antennas Regarding Cavities and Micro-Channels . . . . . . 253.2 Manufacturing Cavities and Micro-Channels in LTCC . . . . . . . . . . . . . . . . . . . . 273.3 Liquid Cooling with Micro-Channels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.3.1 Switched Mode Amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.3.2 LTCC Implementation of the Switched Mode Amplifier . . . . . . . . . . . . . . 373.3.3 Thermal Test Die, Measurement Method . . . . . . . . . . . . . . . . . . . . . . 393.3.4 Measurement Results with Thermal Test Die . . . . . . . . . . . . . . . . . . . . 393.3.5 Infrared Thermography, Measurement Method and Results . . . . . . . . . . . . . 413.3.6 Liquid Cooling: Conclusion and Outlook . . . . . . . . . . . . . . . . . . . . . . 44

4 Stacking Accuracy 474.1 E↵ect of Layer-to-Layer Registration on the Electrical Performance of the Antenna . . . . 484.2 Process Improvements and Limitations Regarding Layer-to-Layer Registration . . . . . . 544.3 Design for Manufacturability Applied to RF-LTCC Modules . . . . . . . . . . . . . . . . 57

5 LTCC Process Adaptation for High Layer Count 61

6 Integrated Resistors 676.1 Microstructure of Thickfilm Resistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686.2 Conduction Mechanism in Thickfilm Resistors . . . . . . . . . . . . . . . . . . . . . . . 706.3 Electrical Properties of Thickfilm Resistors . . . . . . . . . . . . . . . . . . . . . . . . . 716.4 A Novel Concept for mm-Wave Matched Loads and Power Splitters in LTCC . . . . . . . 746.5 Qualification of Resistor Paste for LTCC . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

6.5.1 LTCC Resistor Paste on di↵erent Tapes and Top and Buried compared . . . . . . . 806.5.2 LTCC Resistors Comparative Study . . . . . . . . . . . . . . . . . . . . . . . . . 83

7 Conclusion and Outlook 897.1 Achievements and Contributions of this Work . . . . . . . . . . . . . . . . . . . . . . . . 897.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 907.3 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

i

ii Contents

Bibliography 93

A Appendix 111

Acronyms 111

Symbols 113

List of Figures 115

List of Tables 117

Glossary 119

AbstractLTCC (Low Temperature Cofired Ceramic) is a proven ceramic multilayer packaging technology forhigh reliability applications. This encompasses automotive, medical, security, aeronautics and aerospaceapplications. Advanced low-loss material systems and improved manufacturing technology extend theoperating range to microwave frequencies (up to 250 GHz). One of the virtues of LTCC is the option tointegrate passive components into the multilayer. Matching networks, couplers, and filters can be realized ininner layers, by this means the density of integration is further increased. In this way top and bottom surfacefeature more room for bare dies and SMT (Surface-Mounting Technology) components to accomplish activemicrowave systems.

LTCC multilayer substrates o↵er three-dimensional integration of components and routing of wave-guidesfor microwaves. This is facilitating new solutions which are impossible in planar technologies like thin-film hybrids or semiconductors. On the other hand, the accuracy requirements of the complex circuitryare a challenge. For the frequency range of the SANTANA antenna module for the satellite Ka-band(uplink 27.5 GHz to 31 GHz), manufacturing tolerances are critical since the tolerances for the dimensionsof transmission line components (especially for their length and width) have to be put in relation to thewavelength in the dielectric. In order to attain a design that is not only functional but can be produced withgood yield, DFM (design for manufacturability) is an important guideline throughout the entire developmentand design process.

Within the SANTANA projects [1], a succession of transmit/receive modules was developed with increas-ing density of integration. They utilize the above integration techniques and demonstrate the fitness of theLTCC technology for Ka-band multimedia terminals employing DBF (Digital Beam-Forming). The commu-nication link between a SANTANA system and mobile platforms with a medium size system demonstratedsuccessfully the system functionality. The SANTANA antenna module is a circular polarised, 8 8 elementarray that comprises four 4 4 element building blocks. This sub-array is arranged in a grid to form a digitalbeam forming terminal (transmit system) for the Ka-band uplink. It consists of 17 LTCC layers with 18metallisation layers. The front-end module integrates the antenna elements with branch-line couplers, thecalibration network, active RF circuits, LO power distribution networks, voltage supply and the pipework ofa liquid cooling system.

These complex and highly integrated LTCC antenna modules are far from standard and require severaldedicated solutions. This expository will present the specific structural and technological requirements andhow they have been addressed successfully in the course of the SANTANA projects.

iii

ZusammenfassungDie keramische Mehrlagentechnologie LTCC (Low Temperature Cofired Ceramic) hat sich überall dortbewährt, wo hohe Anforderungen an die Zuverlässigkeit, insbesondere auch an die mechanische undklimatische Belastbarkeit, gestellt werden. Das sind zum Beispiel Anwendungen im Automobil, in derMedizin, in der Sicherheitstechnik und in der Luft- und Raumfahrt. Die Verfügbarkeit von verlustarmen undhochfrequenzgeeigneten Materialsystemen (bestehend aus den LTCC Grünfolien und zugehörigen Pasten fürLeiterbahnen, Durchkontaktierungen, Widerstände und Kondensatoren) und eine verbesserte Fertigungstech-nologie erweitern den Anwendungsbereich für LTCC-Applikationen bis zu Millimeterwellenfrequenzen (bis250 GHz). Ein wesentlicher Vorteil der LTCC-Technologie ist dabei die Möglichkeit, passive Komponentenauch in Innenlagen zu integrieren ("vergrabene Bauteile"). So können Anpassnetzwerke, Koppelstrukturenund Filter im Innern des Mehrlagenaufbaus realisiert werden, was die Integrationsdichte weiter erhöht.Auf diese Weise wird in den Außenlagen mehr Platz für ungehäuste Bauteile und SMT-Komponenten(oberflächenmontierte Bauteile) gescha↵en.

Die dreidimensionale Aufbau- und Verbindungstechnik der LTCC-Substrate umfasst neben den obenerwähnten passiven Komponenten (Kondensatoren, Widerstände und Spulen) auch planare Wellenleiterwie Mikrostreifenleiter, Koplanarleitungen in der Außenlage und geschirmte Streifenleiter in den Innen-lagen. Dazu gehören auch impedanzkontrollierte und geschirmte Übergänge zwischen den Ebenen desMehrlagenaufbaus. Substratgefülle Hohlleiter (SIW) stellen, vor allem im Frequenzbereich zwischen 30 und80 GHz, eine interessante, weil verlustarme Alternative zu den Streifenleitern dar. Damit sind neue Lösun-gen möglich, die in herkömmlichen planaren Aufbau- und Verbindungstechniken wie Dickschichttechnik,Dünnfilmtechnik oder Halbleiterintegration (MMIC) nicht realisierbar sind. Diese komplexen Aufbautenmüssen vor allem bei Millimeterwellenanwendungen sehr genau und reproduzierbar gefertigt werden. Beider in dieser Arbeit vorgestellten Anwendung in der Satellitenkommunikation im Ka-band (Uplink 27.5 GHz- 31 GHz) gelten enge Fertigungstoleranzen für die LTCC-Module, da die Genauigkeitsanforderungen fürdie Dimensionen von Leitungsbauelementen (das gilt vor allem für deren Länge und Breite) auf die Wellen-länge im Dielektrikum bezogen werden. Das DFM (design for manufacturability) ist in allen Phasen derModulentwicklung eine wichtige Richtschnur, um ein reproduzierbar fertigbares Antennenmodul zu erzielen.

Im Verlauf der SANTANA-Projekte [1] wurde eine Reihe von Sende- und Empfangsmodulen mitzunehmender Integrationsdichte entwickelt. Dabei wurden die oben genannten Integrationstechnikeneingesetzt und eindrucksvoll die Eignung der LTCC-Technologie für hochkomplexe Satellitenterminals imKa-band demonstriert. Die Systemfunktionalität (d.h. die Breitbanddatenübertragung und elektronischeStrahlschwenkung) wurde durch eine erfolgreiche Kommunikationsverbindung zwischen einer Funkbakeund einem mobilen Terminal bewiesen. Das SANTANA Antennenmodul ist ein zirkular polarisiertes Anten-nenarray mit 8 8 Elementen, welches aus vier 4 4 Elementen zusammengesetzt ist. Dieses Sub-Array istder Grundbaustein für eine wesentlich größere Antennenapertur. Beispielsweise werden 88 solcher Module(4096 Strahler) in einem Raster angeordnet, um die Sendeantenne eines DBF-Terminals (Digital Beam-Forming) für den Ka-Band Uplink zu bilden. Das Antennenmodul selbst besteht aus 17 LTCC-Ebenen mit 18Metallisierungsschichten. Das Frontend-Modul beinhaltet die Antennenelemente mit Branchline-Kopplern,das Kalibrier-Netzwerk, die aktiven Mikrowellenschaltungen, die Leistungsteiler für das LO-Signal (localoscillator), die Spannungsversorgung und die Mikrokanäle für die Flüssigkeitskühlung.

Diese komplexen und hochintegrierten LTCC-Antennenmodule stellen die Substrattechnologie vor eineReihe neuer Herausforderungen, die maßgeschneiderte Lösungen erfordern. Die vorliegende Arbeit führtaus, worin die spezifischen strukturellen und technologischen Anforderungen dieser keramischen Mehrla-genschaltungen bestehen und wie sie im Rahmen der SANTANA-Projekte erfolgreich gelöst und umgesetztwurden.

1

1 Introduction1.1 Low Temperature Co-fired Ceramic

In the electronic packaging hierarchy Multi-Chip-Modules (MCMs) belong to first level integration. This isthe level of packaging that connects the semiconductor chip (IC, MMIC, ASIC, etc.) to the printed circuitboard (PCB). MCMs can provide an integrated solution where monolithic integration is not (yet) availabledue to lack of volume or not possible due to di↵erent semiconductor technologies. They generally fall intothree main categories [2]:

• MCM-L (Laminate),

• MCM-D (Deposited), or

• MCM-C (Ceramic).

PCB technology with copper conductors and thin organic laminate-based dielectrics is the basis of MCM-L. This technology is also referred to as "chip-on-board". Thin-film technology (MCM-D) [3] providesa high density of interconnects with a high pattern resolution typically in one conductor layer. Limitedmultilayer capability can be added when a photo-imageable polymer dielectric is used to fabricate conductorcrossovers and capacitors [4, 5]. High Temperature Co-fired Ceramic (HTCC) and Low Temperature CofiredCeramic (LTCC) belong to MCM-C. They have in common that sheets of unfired ceramic are screen printedwith thick-film conductor paste, laminated, and then co-fired. The dielectric layer of HTCC is typically 96 %alumina, and that of LTCC is glass-ceramic.

LTCC is a well established ceramic multilayer packaging technology for high reliability applications.Advanced low-loss material systems and improved manufacturing technology extend the application rangeto microwave and millimeter-wave applications. One of the virtues of LTCC is the option to integrate passivecomponents into the multilayer. Matching networks and filters can be realized in inner layers, by this meansthe density of integration is further increased. In this way top and bottom surface feature more room for baredies and SMT (Surface-Mounting Technology) components.

The dielectric of the LTCC multilayer stack is a glass-ceramic composite with a relative permittiv-ity "r ranging from 4 to 8. Special high-k materials range to a relative permittivity "r of 250 [6]. Thesemi-manufactured product for these dielectric layers is called LTCC tape or "green tape". It is the resultof a tape casting process [7, 8] and has the appearance of thick plastic foil. Fig. 1.1 shows LTCC tapes

Fig. 1.1: LTCC tapes. Fig. 1.2: LTCC pastes.

3

4 1 Introduction

from different manufacturers cut into coupons ready for further processing. The raw material for the tapecasting process is a slurry that contains glass and ceramic powder which will form the dielectric of theLTCC multilayer after sintering. The purity of these primary materials and their homogeneous distribution isvery important, but also the grain size distribution [9] of the glass and ceramic powders, since it influencesthe sintering behaviour and the final micro-structure of the dielectric [10]. For conductors and vias typicallygold, silver and alloys of these metals with palladium and/or platinum are used. Vias are filled by stencilprinting and conductors are structured by screen printing. Powders of these metals and alloys are blendedwith an organic vehicle and glass frit to form a paste (see Fig. 1.2) that is suitable for screen printing andcompatible with the tape.

In LTCC technology, layers are processed in parallel as illustrated in Fig. 1.3, which is an advantagecompared to other multilayer technologies where the layer stack is built up subsequently. Layers canbe inspected and eventually reworked before assembling them, which is increasing manufacturing yieldsignificantly [11]. This fact and screen-printed conductors make it an economic technology suitable forvolume production. Passives like resistors, capacitors and inductors can be integrated in three dimensions aswell as passive microwave components and antennas. The robust and hermetic substrate provides housingfunctions like feed-throughs, package walls and radomes for antennas [12]. Fig. 1.4 illustrates an examplewhere LTCC is used as a microwave substrate which is an integral part of the hermetic housing concept for a20 GHz synthesizer. High reliability applications in medical, industrial, automotive, aeronautics and spacetechnology [13] depend on LTCC as the robust substrate material with a wide operating temperature rangeand good thermal conductivity.

"Low Temperature" refers to a sintering temperature of 850 C to 900 C. This is in contrast to HTCC (HighTemperature Co-fired Ceramic) with a sintering temperature of 1400 C to 1600 C. The sintering temperatureof < 900°C facilitates the use of high conductivity metals like gold, silver or copper for conductors andvias whereas HTCC is using refractory metals in inner layers like tungsten, titanium, molybdenum or

Fig. 1.3: Schematic overview of the LTCC process.

1.1 Low Temperature Co-fired Ceramic 5

Fig. 1.4: 20 GHz fractional N Synthesizer in an LTCC module for space application. The Kovar frame and theLTCC substrate are part of the hermetic package for the module.

platinum with much higher melting point but also higher resistivity. Fig. 1.5 shows the relation betweenconductivity and melting point and the sintering temperatures of LTCC and HTCC. The sintering temperatureof the respective material system needs to be lower than the melting point of the metal utilized for vias andconductors.

"Co-fired" is indicating the fact that dielectric and metallization (vias and conductors) are fired (i.e.sintered) in one process step. This is in contrast to conventional thick-film hybrid circuits where multilayerdesigns require multiple firing steps. Therefore LTCC is saving process time and energy but also presents aninherent challenge as the requirements for physical and chemical material compatibility (tape and di↵erentpastes) are far more stringent.

6 1 Introduction

Fig. 1.5: Conductivity of metals vs. melting point. The yellow line at 875 °C indicates the sintering tempera-ture of LTCC and the red line at 1400 °C that of HTCC. Metals marked with an asterisk (*) require sintering ina non-oxidising atmosphere.

1.2 Technical Ancestry of LTCC

Multilayer ceramics have a lot in common with multilayer chip capacitors (MLCC). Both of them usetape casting to form the raw material for their dielectric. It was discovered very early in the history ofelectricity that glass and ceramic like porcelain are good insulators. When eventually the mechanism thatallowed to store electricity in a Leyden Jar [14] was understood, the target was to increase capacity pervolume with thin dielectric layers. Thanks to Michael Faraday, whose extensive experimental work andconclusions [15] led to a more systematic understanding of the physics behind it. The dielectric layersof early capacitors were wax impregnated paper [16] or mica [17]. Mica capacitors were e.g. used toreplace the bulky Leyden jars in spark-gap transmitters. The proliferation of radio communication andbroadcasting in the nineteen-twenties created a demand for alternative dielectrics since the availability of thenatural material mica was limited. The invention of the vacuum tube [18, 19] contributed to this progress.Ceramic capacitors made from titanium dioxide (rutile) became commercially available in 1926 [20]. Thesewired components were typically metallized ceramic discs or tubes. Erwin Schrödinger coined the term"Ferroelectricity" [21] already in 1912 but it was not before the early nineteen-forties that Barium Titanatebecame available as high-k dielectric [22] for capacitors. Capacitance can also be increased by stackingcapacitor discs in one package and connecting them in parallel. In 1960 Warren J. Gyurk filed a patent titled"methods for manufacturing multilayered ceramic bodies" [23]. He was not the first to propose alternatinglayers of ceramic and metal for capacitors. However, in this patent process steps are described that arealso characteristic for LTCC: tape casting with a doctor blade, screen printing of conductors and co-firingdielectric and conductor layers. When transistors became commercially available, telephone exchangesand computers were the applications that fuelled the progress of microelectronics packaging with theirdemand for high density of integration and high reliability. Standardized logic gates were realized e. g. inIBM’s Solid Logic Technology (SLT) as hybrid circuits on small ceramic substrates [24]. Eventually this

1.3 State of the Art in LTCC Technology 7

technology moved on from stacking such modules to ceramic multilayer circuits [25]. The peak sintering forthis High Temperature Co-fired Ceramic (HTCC) was 1550 °C [26], thus the conductor material had to berefractory metals like molybdenum, tungsten and molybdenum-manganese. Powerful mainframe computerswith increasing demands for packaging density, thermal management and computing speed remained thetechnology drivers in monolithic semiconductor integration, packaging and substrate technology for the nextdecades. With larger modules and higher density of integration conductors become longer and narrowerwhich means that their resistance becomes significant. In order to facilitate the use of conductors with higherconductivity like copper, silver or gold, it became necessary to reduce the peak sintering temperature belowthe melting point of these metals. Low Temperature Cofired Ceramic (LTCC) was developed in the 1980s byDuPont and IBM [27] based on thick-film conductors and dielectric paste that were already available andcould be sintered at 850 °C [28]. The Thermal Conduction Module (TCM) [29] that constituted the coreof the large IBM 308X computers, was certainly a milestone in this development worth mentioning. On asquare of 6” 6”, the TCM accommodated up to 133 ICs, each with 704 circuits. Spring loaded aluminumpistons on the backside of the flip-chip mounted circuits were employed to conduct the dissipated heat (up to300 W per module) to the water cooled housing. Traditionally LTCC is not on the short list of microwavesubstrates [30, 31, 32], but in the nineteen-nineties low-loss materials became available [33, 34, 35] andin conjunction with fine line conductors [36, 37] micro- and millimetre-wave applications increasinglybenefited from the potential of high density, three-dimensional integration.

1.3 State of the Art in LTCC Technology

The LTCC multilayer technology has several advantages:

• high reliability,

• extended operating temperature range,

• mechanical robustness,

• dimensional stability,

• density of integration,

• coecient of thermal expansion (CTE) matched to semiconductors,

• good thermal conductivity and power handling capabilities,

• embedded passive components,

• hermetic substrate material, and

• high circuit quality at moderate cost.

For these reasons it is well established in automotive, telecom, medical and security electronics.Photonic Integrated Circuits (PIC) in 100G and 200G systems require a packaging technology for mass

production of complex modules with a small form factor. LTCC provides the platform for these systemswith the capability to route microwave signals and to combine optoelectronic components in di↵erentsemiconductor technologies like indium-phosphide and silicon. Precise alignment of these componentsand their optical interfaces as well as dimensional stability under all operating conditions are particularlydemanding requirements [38].

In medical electronics applications of LTCC range from complex implantable devices like cardiac pacemakers [39] to very small injectable glucose sensors [40] which continuously monitor blood sugar levelswithin the body and transmit the measured data to an external wearable device.

8 1 Introduction

Engine control units, powertrain and brakes are typical automotive applications for LTCC [41, 42] whereextreme mechanical and climatic operating conditions are combined with a high demand for reliability insafety relevant functions. Advanced Driver-Assistance Systems (ADAS) use a multitude of di↵erent sensorsto capture the environment of the vehicle. LTCC is the basis of of various automotive radar sensors becauseit o↵ers good microwave performance, reliability and high density of integration for compact and robustmodules. Short range radar sensors [43] are used e.g. for lane change assistants and parking aids. Thefrequency band for this application is moving from 24 GHz towards 77 to 81 GHz because of regulatorybandwidth limitations at 24 GHz and because MMICs at 77 to 81 GHz are becoming more a↵ordable. Longrange radar sensors for Adaptive Cruise Control (ACC) operate at 76 to 77 GHz [44].

An interesting industrial radar sensor in LTCC was developed in the project Honsensor [45, 46]. Precisemeasurement of the phase-shift at 77 GHz facilitates a resolution in the µm-range at a distance of 5 to 60mm. It is intended for process control in machine tools where surface properties of rotating shafts or othermoving parts need to be measured contact-free.

The increasing demand for high volume digital communication is a motivator for telecommunicationto move towards the terahertz range (frequency 0.3 THz to 3 THz, wavelength 1 mm to 0.1 mm). Theundeniable technological challenges are rewarded with free frequency bands that o↵er high bandwidth forthe requirements of nowadays high data-rate communication [47]. Radar systems operating in this areabenefit from the huge bandwidth particularly in terms of precision [48]. LTCC is used successfully in thisfrequency range for both, the module and the antenna [49, 50]. Challenges known from millimeter-waveapplication like dielectric losses, printing resolution, and transitions from substrate (e.g. microstrip lines) towaveguides demand even more attention and creative solutions. The upper end of the frequency range forLTCC is still to be determined.

LTCC radar modules are also used in aerospace applications: The most recent radar of the EurofighterTyphoon is an AESA (active electronically steered array) system equipped with LTCC TX/RX modules[51, 52]. The principle of Digital Beam-Forming (DBF) requires the whole aperture of the antenna to becovered in a 15 mm grid with these X-band TX/RX modules. 384 very similar modules [53] are integratedinto the X-band SAR-antenna-subsystem of the two earth observation satellites TerraSAR-X (launch 2007)[54] and TanDEM-X (launch 2010) [55]. These modules feed 12 panels with 32 rows in vertical andhorizontal polarization. Both satellites fly in a tandem formation [56] and are delivering three-dimensionalradar images of the earth’s surface of unprecedented acuity [57]. Within the KERAMIS project the modulein Fig. 1.4 [58] together with a set of other LTCC modules underwent all qualification tests for the operationon a geostationary satellite [59]. After passing successfully, these modules were installed in the payloadof the Technology Experiment Carrier (TET-1) [60] and went into orbit in 2012. The LTCC modules wereoperated and monitored in orbit over one year as part of the OOV programme (On-Orbit Verification of newtechniques and technologies) of the German Aerospace Center (DLR) to gain valuable space heritage for theLTCC technology.

Embedded passives in LTCC modules like resistors, capacitors and inductors are increasing the densityof integration. Apart from that, their parasitic e↵ects like TCR (temperature coecient of resistance) orGF (gauge factor, i.e. sensitivity to mechanical stress) can be exploited to measure e.g. temperature orforce. Particularly in combination with micro-fluidic channels new solutions are available in LTCC wherethe substrate is hermetic, temperature stable and resistant against most media. Fluidic sensors [61] andflow sensors [62] are the obvious applications [63, 64, 65]. A capacitive pressure sensor in LTCC [66]uses the cavity and metallized membranes of LTCC layers to form the pressure sensing capacity in an LCresonator circuit. In a piezo-resistive pressure sensor [67, 68] this membrane carries a strain gauge to senseits deflection.

LTCC multilayer substrates o↵er three-dimensional integration of components and routing of wave-guidesfor microwaves. This is facilitating new solutions which are impossible in planar technologies like thin-filmhybrids or semiconductors. With these new possibilities a demand for new design and simulation techniques

1.3 State of the Art in LTCC Technology 9

CAD model for EM-simulation. Photograph.

Fig. 1.6: 800 MHz SMD low pass filter.

is generated. The LTCC low pass filter depicted in Fig. 1.6 [69] shows how passive components can befolded and stacked to form a very compact filter module (SMD1206 package: 3.2 × 1.6 × 1.2 mm3).

3D-EM-simulation with FDTD (finite element time domain) [70] is a powerful tool for this paradigm shiftfrom planar microwave circuits to three dimensional multilayer LTCC structures. The dual-band amplifier formobile phones in Fig. 1.7 is a good example for the successful optimization of a complex LTCC module. TheLTCC multilayer is simulated with a resolution of 25 µm. SMD components and bond-wires are included inthe EM-model. Active devices are embedded in the EM-model using 32 ports to connect them. Simulationtime is 4 min/port on an Intel Xeon 5350 dual-processor (3 Mcells, 150 MByte RAM, 530 Mcells/s). AnLTCC antenna module with integrated RF-frontend for high data rate at 60 GHz is a further example forthe efficient use of EM-simulation in the design and optimization of millimeter-wave LTCC modules. Itis a phased array antenna for WPAN (Wireless Personal Area Network) with a feed network in substrateintegrated waveguides (SIWs) [71]. In subsection 3.2 the special manufacturing requirements of this antennamodule are discussed.

CAD model for EM-simulation. Photograph.

Fig. 1.7: GSM power amplifier.

10 1 Introduction

1.4 Objective of this Thesis

LTCC multilayer modules are appreciated for their flexibility in realising a virtually arbitrary number oflayers with integrated circuit components like vias, cavities, thick film resistors, SMT components andchip devices. On the other hand, accuracy requirements of the complex circuitry are a challenge. For thefrequency range of the satellite Ka-band (uplink 27.5 GHz to 31 GHz, downlink 17.7 GHz to 21.2 GHz),manufacturing tolerances become even more critical since the tolerances for the dimensions of transmissionline components have to be put in relation to the wavelength in the dielectric. However, the demand for highintegration is especially increasing for this frequency range. The increasing demand for mobile access tofast data services is one of the drivers for future broadband satellite systems in Ka-band. Steerable antennasemploying Digital Beam-Forming (DBF) [72] provide fast and flexible reconfiguration capabilities withoutany moving parts and the wear and inertia associated with mechanical antenna positioners.

Within the SANTANA projects [1], highly integrated transmit/receive modules were developed to demon-strate the technology of Ka-band multimedia terminals employing DBF. The communication link between aSANTANA system and mobile platforms with a medium size system demonstrated successfully the systemfunctionality. The SANTANA antenna module is a circular polarised, 8 8 element array that comprisesfour 4 4 element building blocks. This sub-array is arranged in a grid to form a digital beam formingterminal (transmit system) for the Ka-band uplink. It consists of 17 LTCC layers with 18 metallisation layers.These system functionalities of the front-end are integrated into the LTCC module:

• Antenna elements with branch-line couplers,

• Calibration network,

• Active RF circuits,

• LO distribution networks,

• Voltage supply, and

• Liquid cooling system.

The antenna side of the module shown in Fig. 1.8 has 64 patches, and since it is a DBF antenna, eachof them requires a complete microwave front-end. The fully populated component side in Fig. 1.9 showstheir MMICs (Monolithic Microwave Integrated Circuits), filters and connectors. The complex and highlyintegrated LTCC antenna module is far from standard and required several dedicated solutions. Thisexpository will present the specific structural and technological requirements and how they have beenaddressed successfully in the course of the SANTANA project.

1.4 Objective of this Thesis 11

Fig. 1.8: Antenna side of the SANTANA module.

Fig. 1.9: Fully populated component side of the SANTANA module.

12 1 Introduction

1.5 Outline of this Thesis

This thesis was carried out at the laboratories of IMST GmbH, an associated research institute of theUniversity of Duisburg-Essen. In several research projects microwave antennas and modules for satellitecommunication were designed and developed. These projects, funded by the German Aerospace Center(DLR) on behalf of the German Federal Ministry of Economics and Technology (BMWi), address the spacesegment as well as the ground segment of the satellite link. In KERAMIS (Ceramic Microwave Circuitsfor Satellite Communications) [73], MultiFeed (Configurable Multi-Feeding System for Ka-band reflectorantennas in satellite communications) [74], and ActiveMultiFeed (Integration of active control circuitsinto the MultiFeed beam-forming network), LTCC modules for satellite payloads were developed whereasSANTANA (Smart Antenna Terminal) [1] originated hardware for the mobile ground segment. This stillongoing research work includes the construction of hardware demonstrators and the proof of concept in fieldtests. The present work is focussing on the technological solutions and advancements that are prerequisitefor the realization of LTCC modules which meet the requirements of microwave antennas and front-ends forspace application. The results of this long technological development process have been published in thecourse of these projects in conference papers and journal articles. To further readability and to harmonizethe appearance of the thesis, it was decided to rephrase text and edit drawings where required rather thanciting paragraphs and figures from the original publications. The thesis is organized as follows:

First, the concept of the DBF antenna for Ka-band satellite communication will be introduced andexplained in chapter 2. The antenna architecture with its key features and functions will be discussed insection 2.1. Subsequently, the multilayer realization in LTCC is described in section 2.2.

The multilayer construction includes cavities and micro-channels with di↵erent functions. Chapter 3begins with the requirements of the DBF antenna regarding cavities and micro-channels. Various techniquesto fabricate cavities and micro-channels will then be introduced and their suitability for di↵erent applicationsdiscussed.

Section 3.3 is dedicated to liquid cooling with micro-channels. The routing of cooling channels in acrowded multilayer in the space between microwave-, LO-, IF- and DC-supply-connections of 64 activefront-ends is presented and its e↵ectiveness tested. With a thermal test die as the standardized vehicle, liquidcooling with micro-channels in LTCC will be characterized and compared to another approach to cooling inLTCC-modules, die attach to a heat-spreader on a heat-sink.

For optimum scanning performance the patch antenna elements are arranged in a grid with a spacing ofhalf the wavelength in free space. This grid defines the lateral limitation for one element of the DBF antennaand its circuitry. Three-dimensional integration is the solution for this constraint, but vertical transitions formicrowave signals are adding a further requirement to the LTCC process: layer-to-layer registration or, inother words, stacking accuracy. In chapter 4 an 11-layer LTCC antenna module is analysed for the e↵ects ofmanufacturing tolerances using the 3-D full wave FDTD EM simulation software EMPIRE® [75]. Section4.1 will expand on the e↵ect of layer-to-layer registration on the electrical performance of the antenna. Twoextreme examples are selected from this extensive tolerance analysis, one because it has proven to be verysensitive to layer-to-layer misalignment, and one that was quite robust with regard to this manufacturingtolerance. The corresponding simulation results are presented and discussed. Microwave measurementresults of the first samples are added to show sample-to-sample variations. A millimeter-wave antennamodule of similar complexity was transferred from prototyping to manufacturing. The first batch showedchanges in the reflection coecient (S 11) at the antenna port as well as in the radiation pattern. Processimprovements and limitations regarding layer-to-layer registration are subject of section 4.2.

The LTCC process facilitates complex modules with a high number of layers. Fifty layers have beenrealized without reaching the limit of the technology [76], but the standard process parameters apply formodules of 0.4 mm to 2 mm fired thickness. An LTCC multilayer with 17 dielectric layers and a firedthickness of 3.4 mm requires careful adjustment of process parameters. However, most of the LTCC processshown in Fig. 1.3 is not a↵ected, since single layers are processed prior to stacking. Chapter 5 will dealin detail with the two process steps that need to be adapted for LTCC-modules with high layer count:

1.5 Outline of this Thesis 13

de-bindering and sintering. In this particular application de-bindering encompasses the sacrificial materialused for the formation of cooling channels in addition to the organic contents of tape and pastes. Analysistools will be introduced to examine the outcome and the adjustments of the process parameters derivedthereof.

In phased array antennas the level under the actual radiating element is the feed network or beam formingnetwork (BFN) which drives the antenna patches with the correct phase for the desired polarization ordistributes the signal to transmit antennas or combines the received signal of an array of antenna elements.These networks very often comprise resistive components e.g. in matched loads and power dividers.Chapter 6 will show how resistors for these microwave components are integrated in LTCC structures. Thecharacteristics of embedded resistors in LTCC are discussed. Since these thick-film resistors are co-fired(i.e. sintered together) with conductors and dielectric tape their mutual chemical and physical interactionis considerable. In that way LTCC resistors are di↵erent from other resistors in terms of process and itsinfluence on electrical properties. A comparison of resistors from the same paste in di↵erent situationsand on di↵erent tapes illustrates this (subsection 6.5.1). The properties of LTCC resistors depend ona plethora of process parameters. The question what variations are to be expected if a given design istransferred from one manufacturing site to another leads to a comparative study within the consortium of theiKERSATEC project. Subsection 6.5.2 describes how a resistor test coupon is manufactured with the samescreens and stencils in di↵erent facilities. All materials like LTCC tape, conductor and resistor paste arefrom the same manufacturing lot. Best care and attention is applied to have identical process parametersfor all participants. The evaluation of samples printed and fired at di↵erent manufacturing sites shows thevariation which is to be expected e.g. when a design is transferred from prototyping into volume production.

Since topics are discussed in this work from both an electrical and process engineering point of view, themixture of terminology from both areas may be confusing. In order to further clarity, a glossary of terms isprovided at the end of this work.

2 DBF Antenna for Ka-Band Satellite CommunicationThis DBF antenna was developed within the SANTANA project at IMST GmbH. The following chapterdescribes the microwave and antenna design background of this antenna module as it was developed by theproject team.

Phased array antennas allow controlling the direction and shape of the beam by distributing the signal toeach of the antenna elements (or a small subgroup), and then controlling phase and amplitude individually.The combination of the radiated fields of each of the array elements results in the desired radiation pattern.Christiaan Huygens’ theory on the nature of light with wavelets representing the elementary radiators and theresulting wave front, gives a very nice visualisation of the basic principle [77]. If all the elementary radiatorsare excited in phase as shown in Fig. 2.1, the resulting wave front is parallel to the plane in which theelements are arranged, and the wave is propagating perpendicular to it. A phase gradient across the antennaarray will tilt the plane of the wave front and thus change the direction of propagation. The application of anamplitude taper like the functions proposed by Taylor [78] on the antenna array allows to control the shapeof the beam and to reduce the level of the side lobes [79] .

Traditionally phased arrays are fed by power distribution networks followed by phase shifters and anamplifier or attenuator for every element to control phase and amplitude respectively. Digital beam forming(DBF) means that phase and amplitude shifts are applied by digital signal processing (DSP) at baseband level.In case of the transmitter the individual digital signal for each of the antenna elements (patches) undergoesa digital-to-analogue conversion to generate the intermediate frequency (IF). The IF is up-converted in amixer with the local oscillator (LO) and amplified to achieve the required power level of the transmit signalfor the antenna element. For the receiver the signal path is reversed. The incoming signal of each antennaelement is amplified by a low noise amplifier (LNA), down-converted in a mixer with the LO to the IF and

Fig. 2.1: Principle of phased array beam steering. Left image: All elements in phase and main lobe at 0°. Rightimage: Phase shift applied and main lobe tilted.

15

16 2 DBF Antenna for Ka-Band Satellite Communication

then analogue-to-digital converted to the digital base-band. The DSP is used to apply phase and amplitudeshifts to the digital signal. Therefore, each antenna element is equipped with a complete RF front-end, IFcircuits, DA- and AD-converters respectively, and dedicated digital logic. In conjunction with fast algorithmsfor beam forming and beam steering this defines a very flexible and versatile system for broadband mobilecommunication [72]. Linear polarised antennas require, in addition to the standard tracking capabilities,also polarisation tracking for maximum performance. Circular polarization eliminates the necessity ofpolarisation tracking in mobile platforms. For the SANTANA antenna module, circular polarization of theantenna elements is achieved by a branch-line coupler which produces two 90° phase shifted signals togenerate the circular polarization at the antenna [80, 81]. The antenna transmit and receive modules of thefirst generation consist of 16 (4 4) patches each, arranged in-plane according to the sequential rotationprinciple [82] [83]. The final development stage of the transmit antenna module features 64 (8 8) patcheswith 17 layers of LTCC tape. The antenna side of this module is shown in Fig. 1.8 and the component side inFig. 1.9. The antenna elements and the active RF front-ends of the LTCC module are visible on the outerlayers. Integrated in inner layers (see Fig. 2.10 and Fig. 2.11) there are the branch-line couplers, a calibrationnetwork, two LO distribution networks, DC power supply and a liquid cooling system.

The modular concept allows to integrate these sub-arrays to form larger antenna panels side by side,maintaining the optimum /2 distance between elements in each direction. The optimum here is referring toantenna performance, the radiation pattern for all envisaged scanning angles with a minimum of grating lobesand good eciency. With this modularity aperture size and performance of the antenna array can be scaledto fit the actual application. The envisaged number of elements required for a satellite link is 4096 (64 64).Modules of manageable complexity are an important prerequisite for the successful implementation of suchlarge systems and for the maintenance of the same.

2.1 Antenna Architecture

LTCC-multilayer technology provides the necessary degree of vertical integration for the high-densitymicrowave circuit in this application. The area defined by the cell size of one antenna element which is/2 (half a wavelength) in X and Y direction and accommodates the transmit branch-line coupler feed andthe complete RF-circuitry thereof. In further layers the complex calibration network is located. The arraybuilding block, consisting of 16 antenna elements, features a calibration network to enable an automaticarray calibration of each building block. The area of one calibration network is limited to the size ofthe 4 4 element array building block, i. e. a square of 2 edge length. Terraced cavities for the antennapatches with conductive walls (via fences) enable the patented calibration method (o↵-line as well ason-line) [84], and improve the radiation pattern by reducing unwanted mutual coupling between adjacentpatches. A "lossy line termination" for the branch-line coupler provides a good match even with a toleranceof ± 30 % which is to be expected from buried resistors. Furthermore, it is essential to have a low losssubstrate material with good microwave performance for this application. A low permittivity is certainly anadvantage for microwave antennas.

Fig. 2.2 shows the complete construction of such an antenna element. The structure consists of elevenlayers of FERRO A6 LTCC tape [85]. Four di↵erent functional blocks can be identified: The antennablock is depicted in blue, the green block marks the branch-line coupler, the violet parts form a part of thecalibration network and the RF-to-antenna interface is shown in grey. The transmit signal path is markedin red, while the yellow strip lines show the calibration signal path. In Fig. 2.3 top and bottom view ofthe 4 4 array, consisting of 16 of such antenna cells, are visualised.

2.2 Antenna Realization in LTCC 17

Fig. 2.2: Exploded view of one antenna element in eleven LTCC layers.

Top view. Bottom view.

Fig. 2.3: EM-model of the 4 × 4 array.

2.2 Antenna Realization in LTCC

Starting with the top layer, the details of the functional blocks are described in the following paragraphs [86].

Antenna Element

The antenna element is recessed in a cavity, as depicted in Fig. 2.2 and, in more detail, in Fig. 2.4. Thispatch antenna element with two excitation ports is operated in a frequency range from 29.5 GHz to 30 GHz.The phase shift between the excitation ports is 90° to ensure circular polarization. In close proximity to theantenna element there are two near field probes. These calibration probes are realized as vias in the dielectricof the antenna patch. They are receiving only a negligible part of the antenna signal for the calibration


network without disturbing the antenna performance significantly. The placement of the antenna in a cavitythat is surrounded by via fences allows for precise external and internal calibration [84, 87, 88].

Branch-Line Coupler

The branch-line coupler is used to excite the circular polarization of the patch field, while absorbing thecross-polar components of the antenna. This is an important feature, particularly for antenna arrays withhigh element count. The 3D EM simulation model of the branch-line coupler within the EMPIRE® fieldsolver [75] is shown in Fig. 2.5. Located beneath the antenna block, the branch-line coupler is realizedin stripline. A good matching is achieved in combination with an equal power division between the twoantenna feeding ports. Only a negligible amount of power has to be absorbed by port 4, which is matched bya buried thickfilm resistor.

Calibration Network

The complete calibration network of the 4 × 4 array is visualised in Fig. 2.6. For the calibration process, onesingle receiver is used for the 16 elements. Due to the symmetrical structure of the network, every calibrationprobe is connected with the same path length to the receiver. This ensures that the signal of every calibrationprobe has identical phase and attenuation.

Verification of 4 × 4 Antenna Array

The actual far field behaviour of the electronically steered 4 × 4 antenna array building block has beendetermined using an active measurement set-up as depicted in Fig. 2.7. For this purpose, the antenna isconnected to the dedicated Digital Beam-Forming transmit system, consisting of the RF circuitry, the IFcircuitry, the base band system and a computer with graphical user interface for calibration control andmeasurement monitoring. All elements of the building block are digitally controlled in amplitude and phaseto perform the Digital Beam-Forming (DBF). Thus, the 4 × 4 antenna array is set to different scanningdirections and the respective far field radiation patterns are recorded.

Fig. 2.8 and Fig. 2.9 show the measurement results for Right-Hand Circular Polarization (RHCP), that isthe co-polar component for this design, and the Left-Hand Circular Polarization (LHCP), the cross-polarcomponent of this antenna for different scanning angles in elevation θ. For the co-polar component shown inFig. 2.8, the scanning behaviour is good for the whole specified scanning range of −60° to 60°, maintaininga good side lobe level of about -9 dB without any amplitude tapering. Applying a bell shape amplitude

Fig. 2.4: Antenna element in cavity shielded with via fences.


Fig. 2.5: Branch-line coupler.

Fig. 2.6: Calibration network of the 4 × 4 antenna array.


Fig. 2.7: Measurement setup for far field measurements of the 4 4 antenna array in the anechoic chamber ofIMST.

distribution following the Taylor Technique [78] the side lobe level can be reduced further. Fig. 2.9 showsa very low level of the cross polar components independent from the dedicated scanning angle. Goodpolarization behaviour is accomplished by specific features of the design: First, circular polarization ofthe antenna element’s fields is attained using a branch-line coupler to feed the patch with appropriatephase shift. This results in a very good axial ratio of each single element. In addition, feeding with abranch-line coupler also attenuates the cross-polar component resulting from coupling e↵ects. Furthermore,the sequential rotation of the elements is even more improving the axial ratio of the overall array [80] [83].The measurement results for other planes (not depicted) are very similar. The results match very well withthe simulated patterns and confirm the validity of the FDTD simulation model (Fig. 2.3).

After verifying the design of the antenna patch, its feed network and the interaction of four building blocksof 16 patches each in a field test, the next development step is to integrate all this in an 8 8 array. In orderto reduce uncertainties in phase and amplitude of the 30 GHz transmit signal, the active RF-frontend is alsointegrated into the LTCC module. Each of the 64 antenna elements requires a power amplifier, a mixer andan IF filter. Neither the antenna side (Fig. 1.8) nor the component side (Fig. 1.9) of the complex LTCCmodule leaves enough space for cooling fins or similar elements to manage the envisaged power dissipationof 64 W from the active components (mainly the amplifiers). Liquid cooling is chosen to take care of the


Fig. 2.8: Right-Hand Circular Polarization (RHCP) measurement results of the 4 × 4 antenna array buildingblock scanned in elevation θ for azimuth φ = 90°.

Fig. 2.9: Left-Hand Circular Polarization (LHCP) measurement results of the 4 × 4 antenna array building blockscanned in elevation θ for azimuth φ = 90°.


heat dissipation of the module. The cutaway drawing in Fig. 2.10 shows the antenna side of the module tovisualize the internal construction. The LO feed network consists of two 1:32 power dividers supplying 64mixers with their LO signal of equal phase and amplitude.

The liquid cooling pipework is fanned out into eight parallel channels running directly under the poweramplifiers. The component side of the same partial CAD model in Fig. 2.11 depicts the position of theamplifiers along the cooling channels. Fig. 3.2 shows the cross section of the LTCC implementation. Therouting of cooling channels, DC supply lines, IF connections and LO distribution to each of the active frontends requires additional layers in the module and increases the layer count to seventeen. Water has a highheat capacity and is also a hazard-free heat transfer medium. This is one of the rare occasions when a safeand economic solution is also physically optimal.

The SANTANA antenna module is the basic building block for the integration of a planar DBF antennafor high data rate satellite communication in the Ka-band. The functionality of the technology demonstratorswas confirmed in land-mobile and airborne field tests. In one of these tests a ground station equipped withan 8 8 SANTANA antenna connected to a beacon on a research plane that flew di↵erent manoeuvres atan altitude of 2000 m to 6000 m to test the quality of the data link and the tracking capability of the wholesystem. The bi-directional transmission of a video-conference over this channel demonstrated the success ofthis flight experiment [89].

Mobile satellite communication is the key application for the SANTANA antenna. Particularly in scenarioswhere mechanical positioning of antennas is not desirable. An antenna that can be integrated into the fuselagerather than sitting on top of it, like the solutions currently in the market, has an obvious advantage forpassenger aircrafts where fuel consumption is an important figure.

The SANTANA project provides the cornerstone for an active antenna terminal for satellite communicationin the Ka-band. With an increasing number of Ka-band communication satellites like KA-sat and iridiumin orbit, the chicken-and-egg question of ground segment and space segment for a new frequency band isresolved. The modular concept of the SANTANA module with 64 elements and their front-ends enablesflexible combination of larger antenna arrays. This modularity is possible through vertical integration ofmicrowave components and active cooling in an LTCC module with unprecedented complexity and density.


Fig. 2.10: EM model of the SANTANA 8 × 8 antenna array, antenna side partly deconstructed.

Fig. 2.11: EM model of the SANTANA 8 × 8 antenna array, component side partly deconstructed.

3 Cavities and Micro-ChannelsMerriam-Webster defines a cavity as "an unfilled space within a mass; especially: a hollowed-out space" [90].In an LTCC multilayer construction a cavity designates a hollow inside the module without dielectric. Thiscan be e.g. a recessed area for active components as depicted in Fig. 1.7 or a volume completely enclosed bythe multilayer like the micro-channels discussed in this chapter (Fig. 3.2). The potential of LTCC for 3Dpackaging is even enhanced when cavities are implemented. In RF- and microwave-modules such cavitiesprovide direct access to RF-lines in inner layers, reduced thermal resistance under semiconductors andshorter wire-bonds to MMICs [91, 92]. Cavities add a further degree of freedom to the design of planarantennas. E.g. it is possible to recess the antenna patch in a cavity surrounded by via fences as shown inFig. 3.1 to de-couple the transmit signal of adjacent patches in an array. The SANTANA DBF-antenna withthis solution is explained in detail in section 2.1. The radiating elements of the SIW (substrate integratedwaveguide) array antenna presented in subsection 3.2 use narrow cavities of a very distinct shape (see Fig.3.9) to match the impedance of the rectangular waveguide end to free space.

3.1 Requirements of Microwave Antennas Regarding Cavities andMicro-Channels

The SANTANA 4 4 module needs cavities on both sides: each antenna patch is recessed in a separatecavity (Fig. 3.1) to de-couple adjacent patches during operation and to facilitate and improve the calibrationof the antenna array. On the bottom side, cavities are placed in order to make room for the active componentsof the front-end. Since LTCC multi-layers are laminated on a flat plate, sacrificial material is used to protectthe cavities facing to the lamination plate from sagging during lamination and sintering. Cavities may alsobe used as fluidic channels inside the LTCC module, e.g. for liquid cooling in the SANTANA 8 8 moduleintegration (Fig. 3.2, Fig. 2.10 and Fig. 2.11) [93], for active cooling of high power devices [94] or even asheat pipes [95, 96].

25

26 3 Cavities and Micro-Channels

Fig. 3.1: Cross section of antenna cavity in the SANTANA 8 × 8 module.

Fig. 3.2: Cross section of cooling channel in the SANTANA 8 × 8 module.

3.2 Manufacturing Cavities and Micro-Channels in LTCC 27

3.2 Manufacturing Cavities and Micro-Channels in LTCC

Cavities and micro-channels have not been very common in LTCC circuitry, therefore the technology toproduce them with the required quality and repeatability had to be developed from scratch within theSANTANA project.

Typically the formation of cavities and channels starts with cutting out the appropriate area of the relevantlayers. Due to the organic binder in the tape it is machinable in green state by drilling, milling, embossing[97, 98, 99], laser cutting and punching [100]. The latter two are the most common techniques employedfor via and cavity formation [101]. Cut-outs for cavities and windows can be punched along with via holesand registration holes to combine process steps and to improve registration accuracy. However, machiningcavities in a separate step after screen printing means that the tapes are still intact and present a plane surfaceduring via filling and screen-printing of conductors. This is a major advantage with regard to the precisionof via filling and conductor printing. Improved mechanical stability of the tape against elongation is afurther benefit for the alignment accuracy of these process-steps, particularly when having extensive cut-outs.During lamination cavities need to be protected against deformation. The cross-sections in Fig. 3.3 illustratethe need for process adaptation when channels and cavities are involved. The common denominator ofthe methods presented here is to provide counter pressure against the tendency towards excessive materialflow when temperature and high pressure are applied during lamination. One way to avoid deformation isto laminate the LTCC stack first and then punch or machine the laminate before sintering. The alignmentof milling traces with respect to printed features causes additional tolerances. A possible limitation ofthis method is the restriction to structures that can be milled from the outside of the laminate. Also, it isvery dicult to have metallization on the bottom of milled cavities and even more challenging to get thatconductor layer, which is printed after lamination, connected to the internal circuitry of the multilayer stack.

Three approaches to the realization of cavities in LTCC shall be discussed below:

• Sacrificial material,

• soft inserts, and

• hard inserts.

The motivation for hard inserts and their implementation shall be explained using a 60 GHz substrateintegrated waveguide (SIW) antenna as an example.

Sacrifical Material

The tape area removed for the cavity can be replaced by sacrificial material like carbon tape [103], which willdecompose and be removed along with the organics of tape and paste in the de-bindering phase of the ovenschedule. Fig. 3.4 shows the SANTANA 44 antenna with carbon inserts in the antenna cavities. Machiningthe inserts and assembling them in the stack with the tapes is quite laborious. Since the inserts are made of

Fig. 3.3: Cavity deformation and delamination, left: no insert, right: insucient insert [102].


sacrificial material and remain in the stack after lamination, this method facilitates internal cavities as showne.g. for the cooling channels depicted in Fig. 3.2. Carbon [104] is also available as screen-printable pastewhich is an advantage for complex, small patterns and automated production. The paste will lose volumewhen dried, this needs to be taken into account for the set-up of the process and the design of tools. Carbon oralternative sacrificial materials like wax [95] di↵er in their TGA (thermogravimetric analysis) from the innateorganics of the LTCC tapes and pastes. The debindering phase (up to 600°C for carbon based materials) isbordering on or even overlapping with the onset of sintering in the LTCC composite. The temperature profileand gas flow in the oven has to be to be adjusted carefully for this process variant. This applies especially forthe case where the cavity is completely enclosed and the only exit route for gaseous decomposition productsis the porosity of the LTCC stack before sintering. Pores in the laminate [105] develop during de-binderingwhere organic binders have been in the compound. Likewise, porosity in the compound is essential for thecompletion of de-bindering inside thick multi-layer stacks (see chapter 5). The useful process window forde-bindering of sacrificial material starts with the development of sucient porosity and ends with the onsetof sintering. Temperature gradient and composition of the sacrificial material are parameters that cruciallyinfluence the timing. Since the temperature gradient in this part of the sintering curve is also determined by acareful balance of the shrinkage of metal pastes and dielectric tape, there is only a very limited degree offreedom. Burn-out or pyrolysis of sacrificial material will consume oxygen of the oven atmosphere. Thiscan be in competition with sintering and bonding reactions of paste and tape and thus lead to unwanted sidee↵ects on properties like e.g. adhesion. Some of the chemical reactions involved are exothermic [106] andmake it dicult to control the temperature distribution where larger volumes of sacrificial material are used.

Soft Inserts

Soft inserts are very useful to produce cavities in LTCC modules [107]. Silicon rubber is cast in a mold toform a negative of the module surface. This mat or individual plugs are inserted into the cavities to distributethe lamination pressure. This is a very simple and straight forward method to get well defined cavity shapes.The soft inserts need to be removed before de-bindering and sintering as shown in Fig. 3.5 (Fig. 3.6 showsthe same tile after sintering). Thus it is limited to cavities which are accessible from outside and withoutundercuts. Although the inserts are flexible, they need to match exactly the dimensions of the respectivecavity to ensure homogeneous distribution of the lamination pressure. Since the inserts are removed afterlamination they have no influence on the sintering process whatsoever.

Hard Inserts

The extreme aspect ratio and the small dimensions of cavities in the LTCC substrate integrated waveguide(SIW) array antenna in Fig. 3.7 [108] for Wireless Personal Area Network (WPAN) make it an interesting

Fig. 3.4: LTCC module with carbon inserts.


Fig. 3.5: Silicon rubber insert is removed after lamina-tion.

Fig. 3.6: Cavities after sintering.

case study to demonstrate a further method to produce precisely controlled small cavities: hard inserts. Ashort introduction to the antenna concept shall explain the module architecture that leads to the structuralrequirements and the realization in an LTCC module.

Substrate Integrated Waveguide Antenna Concept

The electronically steerable antenna in Fig. 3.7 was designed to achieve the bandwidth of 10% in the60 GHz band, a maximum scan range of ±30°, and a total antenna gain of 18-20 dBi. Bandwidth and gainrespective eciency requirements determine the choice of the single element for this 4 6 array antenna.An open-ended rectangular waveguide radiator was selected for its large bandwidth, low cross polarization,small front to back radiation ratio and high eciency. However, the impedance of the rectangular waveguide

Fig. 3.7: Electronically steerable antenna for 60 GHz WPAN with lead frame.


has to be matched to the impedance of free space at the radiating end and a one-in-four power distributionnetwork with equal phase and amplitude has to be realized connecting the component side to the antennaside of the multilayer module. The realization of this feeding network with rectangular waveguides haslower losses than the corresponding stripline network (a binary tree of Wilkinson power dividers [109] or aH-tree) and thus has an important advantage for high-efficiency antennas. Microstrip and stripline technologyeasily connect to antenna-elements, feeding network and active circuits, but the surface resistance related tothese transmission lines increases with the square root of frequency [110], which becomes an issue at highmicrowave frequencies. This is one more reason to use a waveguide radiator and a feeding network builtwith substrate integrated waveguide (SIW). Via fences form the side walls of the rectangular waveguidesin LTCC and conductor metallization forms their upper and lower boundaries (Fig. 3.10). In each layer,conductor metallization connects the vias to provide a continuous current path along the side walls of thewaveguide. LTCC technology provides both, a good microwave dielectric and the capability to handle a(nearly) arbitrary number of layers (including vias and conductor metallization). Moreover, material andprocessing costs are competitive to other substrate systems like HTCC and printed circuit boards at the farend of the microwave frequency range [111].

Fig. 3.8 shows the EM-model of the proposed waveguide radiator and Fig. 3.9 the realization in LTCC.The topology of the feed network in Fig. 3.12 requires 90°-bends and T-junctions perpendicular to thesubstrate layers. Thus part of the unconventional SIW is perpendicular to the substrate with via fences on allfour sides as conductive wall. The radiating element is designed to have the same radiation properties as anopen-ended waveguide with the dominant propagating H10-mode. The electrical field component (parallel tothe E-plane) for the H10-mode of a rectangular waveguide (see Fig. 3.11) is

E¯ y = U

¯(z)

!2ab

sin"π

xa

#. (3.1)

This is also valid at the open end, but the solution for an SIW radiator with a cavity in the dielectric is alot more complex. Therefore a 3D full wave analysis with the EMPIRE® FDTD solver [75] was performedto simulate and optimize these radiators and their feed network.

Dielectric filled waveguide radiators in general are mismatched to free space. However, the transitionfrom a waveguide mode to a free space mode can be improved by reducing the effective dielectric constantin the antenna aperture by partially removing the dielectric. Fig. 3.8 and 3.9 depict the bone-shaped cavitieswhich are used to minimize reflections at the antenna’s interface to air. Another challenge of SIW in generalis leakage through the gaps between the vias; ideally the vias should represent conducting waveguide walls,which means total reflection or, in other words, no loss of radiated energy through the vias. To achieve thiswith a via fence, the via pitch is usually chosen as small as design rules permit. With enough available space,double (sometimes even triple) rows of vias are used for maximum shielding. On the other hand, it has beenshown [112] that the shielding efficiency of a via fence can be improved by forming a band stop structurewith the via pattern in the desired frequency band. The design goal is to minimize leakage by shifting the

Periodic metallic posts (via fences)

Cavity forelement matching to free space

E- planeMetallization between the layers

Fig. 3.8: EM model of the SIW antenna element. Fig. 3.9: Antenna side of the LTCCmodule.


Fig. 3.10: Substrate integrated wave-guide (SIW) in LTCC, cross-section.

Fig. 3.11: Electrical field in a waveguide, fundamen-tal H10-mode.

stop band to the desired frequency. The characteristics of the LTCC manufacturing process and the thicknessof available tapes further limit the degree of freedom for substrate height, via spacing and via diameter.Considering all this, the single element is designed, simulated and optimized. The beam forming network forthe array consists of four 1:4 power dividers. Fig. 3.12 shows one of these columns. The four elements in inthe centre of one column share one antenna feed, one element at each end of the column is terminated with amatched load to reduce the side lobe level. The waveguides of the feed network are also realized as SIW,using via fences as waveguide walls. Symmetry and equal length of the branches will ensure that all theantenna elements of one column radiate in phase. Beam steering in the plane perpendicular to the columns(H-plane) is accomplished by controlling the phase of the four antenna feeds (one per column).

Fig. 3.12 illustrates in detail one column of the array consisting of four active elements with their one-in-four feeding network. The scan specifications require that this antenna is steerable in the H-plane, thus theelement distance can be increased in the E-plane (along the column). This helps to reduce the complexityof integration, mutual coupling and improves array performance. At both ends end of each column thereare absorbing elements that help to improve the side lobe level and reduce the back radiation. They areterminated with thickfilm resistors as matched loads, thus the reflection from these elements is minimized. Amatched interface to the standard rectangular waveguide WR-15 is also implemented for the test antenna.

Fig. 3.12: One column of the array consisting of four active (radiating) elements and two passive (terminated)elements with the corresponding SIW feed network.


Fig. 3.13: The 3D-FDTD simulation of the electrical field distribution in the waveguide feeding network andthe active antenna elements. The plane of the cross-section is along the centre of one column (E-plane).

Fig. 3.13 shows the 3D-FDTD simulation of the electrical field distribution in the feeding network of theactive antenna elements for one column. The dynamic range in this diagram is 30 dB, thus the energy whichcouples to the passive elements is not visible here. The waveguide feed of this test antenna provides thetest port for the network analyser used to perform the far field measurements at 60 GHz in the anechoicchamber. In the final product, each antenna will be connected to an MMIC with a coplanar output via amatched transition from coplanar to waveguide. The total number of LTCC layers in this design is 14. Usinga waveguide (SIW) network in horizontal and vertical configuration as depicted in Fig. 3.12 for the beamforming network (BFN) increases the layer count compared to a planar BFN and thus the total thickness butit allows placing adjacent columns in the appropriate spacing. This results in a better scan characteristic andelimination of grating lobes from the scan range for wide scan angles in H-plane. The final column designis used to integrate the four-column array. In order to optimize the design, the array is simulated with allfour columns excited. For the verification of the design by measurement test antennas were built with awaveguide interface. The flange of the rectangular waveguide used here (WR-15) is too large to connect allfour columns simultaneously. The active impedance method is suggested by Pozar [113] [83] to characterizelarge antenna arrays by measuring the radiation pattern of one active element in the array while the otherelements are terminated with matched loads. The resulting active element patterns for all four columns inthe antenna module can be superimposed to obtain the pattern of the fully excited (scanned) phased array.This method includes mutual coupling e↵ects and thus the scanning performance of the whole phased arraycan be predicted including variation of array input impedance with the scan angle and scan blindness. Theantenna array is fully characterized by this procedure without the need for complex beam steering electronicslike phase shifters. The array input impedance has been optimized to achieve the optimum bandwidth forall specified scanning angles (here up to 30°). In order to minimize the coupling e↵ect, the columns wereconstructed with separated shielding (i.e. the waveguide side walls), to avoid ohmic contact between adjacentcolumns. This will restrict the coupling e↵ects to aperture coupling only, which is tolerable considering thespacing and scanning requirements. The decoupling of columns with grooves can be seen in Fig. 3.9.

LTCC Implementation of the SIW antenna module

The antenna elements and their feeding networks in SIW require 14 layers of dielectric. The millimeter-wavefront-end for RX and TX respectively requires five more layers for RF and DC routing. This results in astack of 19 layers and 3.8 mm fired thickness. Governed by antenna design one side of the LTCC modulehas a three dimensional topography as shown Fig. 3.9. The purpose of the grooves is to de-couple adjacentantenna elements by introducing an abrupt change in the permittivity. The I-shaped cavities in the antennaelements help to improve the VS WR at the open end of the SIW. The size of these cavities (250 µm wide and800 µm deep) and their tolerance requirements led to a new approach towards cavity formation. Soft insertswith the required dimension are dicult to handle in the process whereas hard inserts can be machined veryprecisely, even in small dimensions. However, during lamination, LTCC tapes tend to adhere to hard insertsand make the removal of the inserts risky for the product. On the other hand the virtues of this method

3.3 Liquid Cooling with Micro-Channels 33

were enough motivation to find a way to overcome the latter disadvantage. A hard insert with a stripperplate to separate insert and LTCC laminate without damage looked like a promising idea. The cavities arepunched into individual layers with a special punch and die. A dedicated stacking table was designed to holdalignment pins and hard inserts (Fig. 3.14). A stencil with cut-outs for alignment holes and cavities (i.e. the"stripper plate") goes on top of that to facilitate separation of laminate and stacking table after lamination.The resulting cavities inside the antenna elements are well defined and reproducible within narrow tolerances(Fig. 3.9).

Fig. 3.14: Stacking plate with hard inserts.

Grooves in the dielectric (see Fig. 3.9) to reduce coupling between the antenna elements cannot be punchedin individual layers but have to be cut into the surface of the sintered module with a dicing saw. It is importantto have precise control over the cut depth relative to the metallization in the inner layers. The laminationpressure is adjusted to compensate lot-to-lot variations of the shrinkage and to keep sintered dimensionswithin narrow limits to ensure good alignment of flip-chip mounted semiconductors, the lead-frame and thefired LTCC module.

3.3 Liquid Cooling with Micro-Channels

When liquid cooling is considered for the thermal management of the SANTANA antenna front-end, a proofof concept is required to decide whether to pursue this solution further. In order to verify the effectiveness ofthese cooling channels, the active components were replaced by resistors that simulate the thermal load inoperation. A power dissipation of up to 1 W per patch is envisaged, so the thermal mock-up is tested with atotal power dissipation of 64 W. The cutaway drawing in Fig. 2.10 shows the antenna side of the module tovisualize the internal cooling channels. The component side of the same partial CAD model in Fig. 2.11illustrates the position of the amplifiers along the cooling channels. They are the main contributors of thethermal load. The result of this test is plotted in Fig. 3.15, the temperature gradient from inlet to outletis palpable. However, the cooling performance exceeds the requirements even at 64 W of total dissipatedpower.

With the integration of micro-channels into the LTCC-multilayer-stack the microwave-substrate becomespart of the piping for liquid cooling and brings the coolant in close contact to the heat source. As anexample for this approach a switched mode amplifier shall be presented that uses liquid cooling for the powertransistor [114]. Two different designs for the routing of the cooling channels inside the LTCC are comparedto the reference design, where the power transistor is mounted on a metal block that is kept at constant


Fig. 3.15: Temperature distribution in the SANTANA 8 × 8 module.

temperature. For the exact determination of all relevant thermal parameters, such as the dissipated power andthe temperature on the die (junction temperature), a thermal test die is used instead of the transistor for thetest vehicles. Firstly the amplifier concept is introduced. Then the LTCC implementation of the same moduleis followed by the description of two measurement methods for the thermal characterization of the module.The results of the different approaches to handle heat dissipation are compared and evaluated according totheir cooling effectiveness.

3.3.1 Switched Mode Amplifier

Conventional solid state power amplifiers (class A, AB, B, C) as depicted schematically in Fig. 3.16 use thetransistor as a controllable impedance to amplify the input signal. Suitably designed, this amplifier concept(A, AB) has good linearity with limited efficiency (max. 50 %), where the latter can only be increased at theexpense of output power (class B, C).

As the name suggests, the switched mode amplifier (class D, E) [115] uses the transistor as a switch.Ideally, power is dissipated only during transition, namely within the interval when the product of currentand voltage is maximum. The simplified diagram of this type of amplifier is shown in Fig. 3.17. However,with increasing frequency and finite switching speed, the transition interval is requiring a larger share of thecycle leading to an increase in the overall power dissipation. An output resonator is required to restore thefundamental wave form of the input signal at the output. With class E amplifiers efficiencies of 70 % to 80 %can be achieved.

The simulation results of the class E amplifier show drain voltage and current of the transistor. The blueboxes in the time domain plot (Fig. 3.18) indicate the transition intervals where relevant dissipation occurs.It is apparent that the signals are non-sinusoidal. The graph of the same signals in the frequency domain(Fig. 3.19) shows the high content of harmonics (marked blue) and emphasizes the need for an outputresonator to restore the output signal.


Fig. 3.16: Conventional power-amplifier. Fig. 3.17: Switched-mode amplifier.

Fig. 3.18: Class E amplifier, simulation of drain voltage and current in time domain. The blue boxes indicatethe transition intervals where relevant dissipation occurs.

For this application a GaN (Gallium Nitride) HEMT (High Electron Mobility Transistor) (Cree CGH60015D[116]) was selected because of its advantages over Si (Silicon) and GaAs (Gallium Arsenide) for this appli-cation, which are high values of:

• breakdown voltage,

• saturated electron drift velocity,

• thermal conductivity,

• power density, and

• bandwidth.

Notwithstanding the operation in switched mode, the transistor has a power dissipation of 4.2 W on a chiparea of 0.9 mm2. For this reason thermal management demands particular attention.


Fig. 3.19: Class E amplifier, simulation of drain voltage and current in frequency domain. The high content ofharmonics, marked blue in the graph, emphasizes the need for an output resonator to restore the output signal.

Fig. 3.20: GaN transistor Cree CGH60015D (size: 1 x 0.9 mm2), the red rectangle indicates the active area.


3.3.2 LTCC Implementation of the Switched Mode Amplifier

The amplifier module was built in three versions with identical electrical design. The variation is in thedesign of the cooling channels.

• V1: The reference design of the amplifier module has no cooling channels but the transistor is mountedon a metal plate that is kept at constant temperature for the thermal measurements.

• V2a: The transistor of this amplifier module is mounted on top of an LTCC multilayer with internalcooling channels. The channel width is limited to 700 µm. The flow of the coolant is directedperpendicular against the footprint of the transistor (impingement cooling).

• V2b: The cooling channels of this amplifier module are wider (1000 µm) and follow a simpler route,the flow of the coolant is parallel to the mounting base of the active device.

The LTCC amplifier modules were mounted on an aluminum base plate together with two PCBs for theDC voltage supply (see Fig. 3.21). In version V1 the base plate was water cooled to serve as a temperaturecontrolled heat sink.

Amplifier module V1 has a stepped cavity for the transistor which is mounted on a pedestal of CE6F, aSiAl alloy with a CTE of 7 ppm/K and a thermal conductivity of 110 W/mK. The layer stack shown in Fig.3.22 encompasses of 6 layers of 254 µm thick green tape. The transistor is brazed with gold-tin solder to thepedestal which is brazed together with the LTCC substrate to a base plate of the same SiAl alloy. This is awell established construction for high-power modules in LTCC, thin-, and thickfilm technology and willserve as bench mark for the thermal performance.

Fig. 3.23 shows the micro-channel design for amplifier module V2a inside the LTCC multilayer in anisometric drawing together with the thermal relevant metallization under the transistor chip. Thermal viasconnect the transistor die thermally to the cooling channel. The cylindrical cavities at each end of thecooling channel define the inlet (left) and the outlet (right) of the cooling channels. They are in predefinedplaces to facilitate the use of a standardized fluidic connector on the thermal test bench [117]. One targetof the channel design in module V2a is to avoid long cuts in the tape layers for improved handling of thesingle layers through the process. Another goal is to limit the cross section of the channels to minimizeaggregations of sacrificial material. Larger volumes of the sacrificial material have led to uncontrolledtemperature development during de-bindering in previous applications. Parallel channels are provided to

Fig. 3.21: Isometric view of the amplifier module V1 breadboard model.


Fig. 3.22: LTCC layer stack (production drawing) of module V1 on the left and V2b on the right (Courtesy ofMSE).

Fig. 3.23: Micro-channel design of module V2a with transistor chip.

reduce the flow resistance. Finally the flow of the coolant should be directed perpendicular against thefootprint of the transistor. The right side of Fig. 3.23 shows the detail of the transistor proximity with arrowsindicating the flow of the coolant. The cooling channels of 215 µm height and 700 µm width are routedthrough the RF-design and its ground planes and thus demonstrate the potential to integrate the presentedmicro-channel technology in high density electrical circuitry. The resulting layer stack consists of 8 layers of254 µm thick green tape.

Whereas the electrical design and layout of amplifier V2b is exactly the same as in V2a, the design ofthe cooling channel in Fig. 3.24 is straightened for the following reasons: Manufacturing is simplified by astraight-lined channel, this applies for the cut outs in the tape layers but even more for the carbon inlays,i.e. the sacrificial material that is used to support the channels during lamination in order to maintain theirshape. In combination with a wider cross section (215 µm height and 1000 µm width) the flow resistance isexpected to be reduced. The LTCC layer stack in Fig. 3.22 shows that 6 layers of 254 µm thick green tapeare used compared to 8 layers in V2a. The right side of Fig. 3.24 shows the vicinity of the transistor whichis again mounted on a pad with thermal vias to connect it to the cooling channel.


Fig. 3.24: Micro-channel design of module V2b with transistor chip.

3.3.3 Thermal Test Die, Measurement Method

A thermal test die is an approach to analyse thermal packaging and material problems by replacing the heatreleasing device with a component that allows for a controlled power dissipation and precise measurementof the temperature at the chip surface [118]. Here the transistor of the power amplifier is the componentunder examination. The following specifications of the thermal test die should be close to those of the chipbeing emulated by the thermal test die:

• size (including thickness),

• thermal conductivity (ideally made of the same semiconductor bulk material),

• thermal capacity,

• heating area (active area),

• temperature profile across the heating area, and

• backside metallization (suitable for the required technique of die-attach).

For the thermal metrology one temperature sensor in the centre of the chip’s active area is mandatory,whereas multiple sensors are an advantage for larger chips in order to get a temperature profile acrossthe chip surface. Four-wire connections (Kelvin connections) to the temperature sensors will improve themeasurement accuracy. The thermal test die G 423B from Infineon [119] is available in a size close tothe transistor die under examination and was chosen because a custom tailored thermal test die was notavailable. In contrast to the GaN transistor, the thermal test die consists of a silicon die with a thicknessof 643 µm compared to 100 µm in case of the transistor. However, it is still an excellent vehicle for acomparative study of the di↵erent cooling concepts presented here. Fig. 3.25 shows the layout of the thermaltest die G 423B. The die contains a temperature sensing diode, a planar resistor for heating and a daisy chain.In addition to the switch mode amplifiers with the GaN transistor all three versions of the amplifier module,V1, V2a and V2b were assembled with the thermal test die in lieu of the transistor.

3.3.4 Measurement Results with Thermal Test Die

With the thermal test die, liquid cooling using micro-channels in LTCC is characterized for modules V2aand V2b. Module V1, with the thermal test die mounted on a metal plate at constant temperature, serves as areference for the comparison of the thermal measurements. The thermal characterization is performed on a


Fig. 3.25: Thermal test die (1 x 1 mm2).

thermal test bench where volumetric flow and temperature of the coolant can be controlled [117]. Thermalresistance Rθ is determined by the following equation:

Rθ =∆ϑ

Q(3.2)

where ∆ϑ is the temperature difference between transistor junction and cooling water at the inlet. The heatflow Q corresponds to the power dissipation in the active device.

Fig. 3.26: Thermal resistance of LTCC module with cooling channels vs. volumetric flow.


The results of the measurements with thermal test dies are summarized in Fig. 3.26. The coolant at theinlet is at room temperature and the chip temperature is kept at ≈ 100 °C. The chain line marks the thermalresistance of the reference design V1 with the thermal test die mounted on a pedestal in a window in thesubstrate. The influence of the coolant flow through the base plate is negligible in case of the referencedesign since the massive aluminium base plate is also at room temperature. The graph shows the significantdifference in thermal resistance between the two different channel designs in modules V2a and V2b. Thepositive effect of the coolant flow upward against the base of the transistor in V2a becomes apparent whencomparing the performance of the same sample with reversed flow. The thermal resistance is 4 K/W to5 K/W lower if the coolant flows perpendicular against the footprint of the active device. The comparison ofthe pressure drop between module V2a and V2b shows a bigger difference than the variation in the channeldesign would suggest. The pressure drop in a smaller channel is more prone to deviations in channel crosssection due to process tolerances, such as variations in width and height or layer-to-layer misalignment.

3.3.5 Infrared Thermography, Measurement Method and Results

Thermal imaging cameras exploit the effect that infrared radiation is emitted by all objects with a temperatureabove absolute zero according to Planck’s law [120]. They are used to detect infrared radiation (λ = 9 µm to14 µm) and produce images of that radiation, called thermograms. The thermal images are two-dimensionalrepresentations of the amount of infrared energy emitted, transmitted, and reflected by an object. Assumingthat the emission of infrared energy from the device under test is predominant and the contribution bytransmission and reflection can be neglected, these colour maps can be interpreted as temperature distributionon the surface of the device. In this way thermal imaging cameras are useful tools for the non-contactmeasurement of temperature in electronics, machinery, buildings etc.. Here the camera is used to monitorthe temperature of the transistor in amplifier modules V1, V2a and V2b. The optical measurement is notinterfering with the operation of the microwave circuit. A macro lens is attached to get an effective imageresolution of 18 µm per pixel.

The thermogram in Fig. 3.27 shows the temperature distribution in the vicinity of the transistor in amplifiermodule V2a. It is operated at 3.5 W and cooled with 50 ml/min. In order to convert the received infraredradiation to temperature, a factor ε (emissivity) is required [121], that expresses the material’s ability to emit

Fig. 3.27: Thermogram of the GaN transistor in amplifier module V2a operated at 3.5 W. The dotted red line inthe photograph on the left indicates the area shown in the thermogram. The active area of the transistor, wherethe temperature readings are taken, is marked with the small black square in the thermogram. The colour baron the right is the temperature scale of the thermogram.


thermal radiation and is an optical property of the surface, in this case the active area of the GaN transistoras depicted in Fig. 3.20.

To calibrate the temperature scale, the GaN transistor is heated externally to discrete temperature stepsand thermograms of the chip surface are recorded along with temperature readings of a PT1000 sensor. Theemissivity is then calculated with the received infrared radiation of the active area of the GaN transistor(the small black rectangle in Fig. 3.27) and the temperature of the surface given by the PT1000 sensor.Fig. 3.28 shows the variation of the emissivity ε over temperature. Since the camera software can handleonly a constant value of ε, the temperature at the transistor junction was kept at ≈ 100 °C for this set ofmeasurements in order to reduce measurement uncertainty. The value for ε was determined by logarithmicextrapolation to be 0.232 at 100 °C.

Thermograms were taken at different flow rates. The according thermal resistance is calculated withequation 3.2. Here the junction temperature is determined by the camera software for the active area definedin the thermogram, which is given by the small black rectangle in the centre of the transistor in Fig. 3.27.The resulting thermal resistance is plotted against the flow rate in Fig. 3.29 together with the pressure dropin the cooling channel. It can be observed that an increase of volumetric flow slightly reduces the thermalresistance between 10 and 100 ml/min. It is worth mentioning that even a low volumetric flow of the coolantwill effect good (i.e. low) thermal resistance.

The LTCC modules V2a with GaN transistor in Fig. 3.29 and V2a with thermal test die in Fig. 3.26 areassembled on identical LTCC substrates from the same batch. Within the same channel design (see Fig.3.23) the same pressure drop could be expected. The direct comparison of these samples in Fig. 3.30 (thedark blue curve and the light blue one) points out that there is a significant difference in the pressure drop.The Hagen-Poiseuille equation [122, 123]

V =dVdt=π · r4

8 · η∆pl= −π · r

4

8 · η∂p∂z

(3.3)

˚

Fig. 3.28: Infrared emissivity of GaN vs. temperature.


Fig. 3.29: Thermal resistance of LTCC module with cooling channels vs. volumetric flow.

Fig. 3.30: Direct comparison of cooling channels: pressure drop vs. volumetric flow.


for volumetric flow rate V in round tubes shows that the pressure drop p is inversely proportional to thefourth power of the radius r of the tube. l is the length of the tube, z the direction of flow is the dynamicviscosity of the liquid in the tube. Even with the simplification of laminar flow of an incompressible fluid ina straight tube the equation shows the leverage of the dimensions defining the cross-section of the coolingchannel. The cooling channel is defined by the fired tape thickness and the cut-out punched in the tapelayer, but there are also the manufacturing tolerances of the carbon inserts which are used to stabilize thecavity during lamination, and the shrinkage tolerance that are influencing the final channel dimensions.Layer-to-layer misalignment (see Chapter 4) has an impact on pressure drop in a cooling channel that mustnot be neglected. A vertical connection from one tape layer to the next will act as a choke when misaligned.All these e↵ects are amplified when the channel dimensions are reduced.

3.3.6 Liquid Cooling: Conclusion and Outlook

Two di↵erent measurement methods were used to examine substrate integrated liquid cooling in a complexLTCC module. The approach with a thermal test die allows for an easy control of the dissipated powerand the proper measurement of temperature. However, this requires a dedicated assembly designed andbuilt exclusively for the thermal characterization and a thermal test die having the same properties as theactive component in the circuit. The thermogram taken with the infrared camera is not interfering withthe electrical circuitry since it is an optical method. It requires that the active component(s) are accessiblefor infrared photography, this may be a limitation where a shielding or package is necessary to include itselectrical and/or thermal e↵ect. Further, the emissivity of the surfaces under examination needs to be knownor determined for the calibration of the thermogram. It is a positive side e↵ect that the whole field of view ismonitored for possible thermal overload. The variance in thermal resistance R for the two measurementmethods is explained by the di↵erent thermal conductivity of silicon and GaN and the considerable deviationin thickness of the thermal test die from the transistor chip.

The antenna module and the power amplifier presented here demonstrate that liquid cooling can beintegrated into a LTCC substrate, combined with complex microwave and DC circuitry. This approachprovides a solution for thermal management where passive cooling is too bulky or provides insucientthermal flow. An LTCC module with liquid cooling has not only a low thermal resistance but the junctiontemperature of the active device can be lowered further by lowering the temperature of the coolant. Liquidcooling allows to transport heat out of constricted module configurations, but there is an extra e↵ort neededto provide the module with the required coolant flow at the desired temperature. Furthermore, the coolingchannel needs to be shielded from the microwave circuitry or routed at a safe distance from RF-lines becausethe relative permittivity of water is 80 (at room temperature and 3 GHz). This would certainly influenceRF-circuits, if the channel is close enough.

Directing the flow of the coolant perpendicular against the footprint of the transistor (jet impingementheat transfer [124]) in amplifier module V2a (see Fig. 3.23) is reducing the thermal resistance significantlyas shown in Fig. 3.26. The dimensions of the fluidic channels in modules V2a and V2b were guidedby envisaged feasibility of the LTCC process and by experience with previous projects utilizing micro-channels. The demonstrators presented here are the basis for future work which will analyse further thefluid mechanics of the cooling channel and optimize the precision of simulation results with the thermalperformance measured in this elaboration. In addition to the simulation of the electrical behaviour of thecircuit at microwave frequencies [75], the heat transfer in solids [125] and the flow of the coolant have to besimulated to predict the thermal characteristics of the module [126, 127, 128]. Scaling down the dimensionsof fluidic channels increases the surface-to-volume ratio and the e↵ect of viscous forces becomes moresignificant [129]. The substrate thickness between channel and transistor was designed according to standarddesign rules ( 400 µm). Future experiments should answer questions like how thin this can be made withoutsacrificing hermeticity and flatness of the chip pad. Clearly this is a design parameter with a significantinfluence on the thermal resistance. The carbon tape that is used as sacrificial material allows to form vias init with the same method as in LTCC tape. After de-bindering the vias are free standing metal columns in the


channel. They extend the thermal vias into the cooling channel and provide an excellent heat exchange withthe coolant [117, 130]. However, the influence of the vias on substrate flatness and hermeticity should beinvestigated.

Future design tools might include CFD (Computational Fluid Dynamics) in the same way as FDTDEM-simulation and thermal simulation are already implemented. Thus the design flow of the microfluidicsystem and that of the electrical design in the module are merged and the optimization of the design ofcooling channels and the analysis of manufacturing tolerances are accelerated. In analogy to electricalcontinuity tests in-line with flying probe test systems, the substrate needs to be tested for the functionality ofthe cooling channels before assembly. In volume production in-line tests with short cycle times have to beinstalled that will ensure that the circuits meet the quality standards (e.g. hermeticity and hydraulic flowresistance) for their micro-channels in the same way as electrical and dimensional requirements.

4 Stacking AccuracyLTCC dielectrics have a higher permittivity than most of the common organic circuit boards. Together withgood RF performance and nearly arbitrary layer count these are excellent prerequisites for high densityof integration in hybrid RF circuits. In this context, good RF performance means primarily low lossesat microwave frequencies and constant permittivity vs. frequency. The well-established screen-printingprocess used in LTCC is a further benefit in the volume production of the SANTANA antenna module.However, microwave circuits like filters, matching networks, and antennas will change their properties whenvias in the signal path or ground vias are shifted from their nominal position. The associated geometricalchanges will also a↵ect the radiation pattern of the SANTANA antenna module. The high permittivity of thedielectric helps to integrate all required components into the available area. But it leads also to challengingdemands in terms of manufacturing tolerances. Apart from tolerances in material properties (permittivity,shrinkage and layer thickness) deviations in the geometry influence the overall system performance. In orderto relate expected manufacturing tolerances to their influence on antenna performance, the eleven-layerLTCC module that has been introduced in chapter 2 is investigated thoroughly using IMST’s 3-D full waveFDTD EM simulation software EMPIRE® [75]. This chapter will concentrate on the e↵ect of layer-to-layer misalignment which is particularly interesting where microwave transitions perpendicular to substratelayers are necessary for vertical integration [131]. The antenna is operated at 27.5 GHz to 31 GHz wheremanufacturing tolerances become more critical due to the small wavelength (i.e. 10 mm in air and 4 mmin the dielectric), an e↵ect that is increased by the high permittivity of the LTCC dielectric. Two extremeexamples of possible layer-to-layer misalignment are selected from the extensive tolerance analysis anddiscussed in detail. The complete architecture of one antenna element has already been depicted in Fig. 2.2.The di↵erent functional blocks of the antenna element and their specific design features have been describedin detail in section 2.1.

Layer-to-layer misalignment in a multilayer stack is the result of finite accuracy in several process steps:

• dimensional stability of the tape through all process steps,

• accuracy of the position of via holes (precision of punching machine),

• alignment of conductor print to via pattern,

• alignment and tacking of tape layers during stacking.

Catch-pads (also referred to as annular rings) are one way to maintain connectivity when layers are slightlymisaligned. This is of course at the expense of integration density. It also adds parasitic capacitance to signalvias and thus has to be taken into account for accurate EM-modelling. The dimensional stability of the tapethrough all process steps can be ensured by a process flow where the backing tape (Mylar) remains with thetape layers and is removed only immediately before stacking or even during the stacking operation. This iseven more recommended when coupon sizes like 8” 8” or larger are processed. Tool wear in the punchingmachine will not only influence the quality and diameter of the punched hole but also the accuracy of thepunch position. It is thus important to monitor tool wear also to maintain precise positioning of vias [100].

47

48 4 Stacking Accuracy

4.1 Effect of Layer-to-Layer Registration on the Electrical Performance of theAntenna

The functionality and performance of passive microwave circuits depends amongst other things on theimpedance and the electrical length of the waveguides. This applies also for patch antennas and theirfeed networks, of course. In planar microwave circuitry typically microstrip lines, striplines and coplanarwaveguides are used as transmission lines. The impedance of such a planar waveguide is determined byits cross section (i.e. line width, substrate thickness), the substrate permittivity, and the frequency. Theelectrical length is a function of mechanical length and e↵ective permittivity. Within the permittivity rangeof commercially available LTCC material systems Ferro A6S is on the low end with a relative permittivity of5.9. It still helps to miniaturize the RF circuitry but also still increases the influence of variations in geometrysuch as line width, and gap width, and line length. For the planar waveguides (i.e. signal lines in one layer)the accuracy of printed patterns is most important followed by the substrate thickness and the toleranceof the permittivity. In a 3-D integrated microwave circuit like the SANTANA antenna module, transitionsbetween layers or even vertical waveguides are essential. In addition to the above mentioned tolerances,relative position of vias to lines, cavities to structures and alignment of conductor patterns from layer tolayer become increasingly importance. Coupling gaps in filters and distances of functional elements toneighbouring shielding walls are further dimensions which strongly influence the performance of microwavecircuits and antennas.

Layer-to-Layer Alignment Accuracy

The multilayer structure illustrated in Fig. 2.2 consists of eleven layers of LTCC dielectric and twelveconductor layers which can be laterally displaced against each other. This investigation is focussed onlayer-to-layer alignment accuracy and thus, mutual displacement of multiple prints in the same layer (whichare required for functional pastes such as thickfilm resistors, solderable conductors, or fine-line patterns)are not taken into account. Further, the e↵ect of distortion of individual layers and shrinkage tolerancesdue to di↵erent coverage of tape layers with conductor paste are considered to be included in an increasedrange for linear displacement in X- and Y-direction. For the simulation, the nominal position plus fourvariations in X- and Y-direction are assumed. If all possible displacements of eleven tape layers with twelvemetallization layers were investigated, it would result in 922 9.8 · 1020 possibilities. This would exceedthe performance of even very powerful computers and the perseverance of any design engineer. In order tokeep the simulation manageable, the tolerance analysis is concentrating on layers which are considered to besensitive by the design team. The results for the most sensitive layer will be presented and discussed, alongwith one layer that was also analysed and turned out to be quite insensitive to misalignment.

Tolerance Analysis and Simulation Results

LTCC stackers with automated optical alignment are typically specified to be accurate within ± 5 µm [132].This is the mechanical performance of the apparatus and does not account for inaccuracies in shape andposition of the fiducials on the tape or distortion of the green tape etc.. Stacking with mechanical registrationcan be accurate within ± 30 µm. In order to include the e↵ect of outliers that were observed in the practicalrealization of the initial test structures, the antenna is simulated with a lateral layer-to-layer misalignment of± 100 µm in steps of 25 µm. The antenna performance of the complete stack of eleven layers, includinghybrid ring couplers, RF interfaces and calibration network is simulated by 3-D EM modelling of the wholestructure using the FDTD field solver EMPIRE® [75]. For clarity of the diagrams for reflection coecientand radiation pattern, only the results for ± 50 µm displacement are plotted. The metallization layer C3,that carries the antenna patch (third from top in Fig. 2.2) is considered sensitive because displacement ofthis patch relative to its feeding points strongly influences the impedance and cross polar suppression of the

4.1 Effect of Layer-to-Layer Registration on the Electrical Performance of the Antenna 49

antenna element. This kind of misalignment disturbs also the symmetry of the antenna patch with respect tothe surrounding via fences and thus is prone to degrade the radiation pattern.

Fig. 4.1: Reflection coecient S 11 vs. frequency for several lateral displacements of layer C3. 3-D EM simula-tion with EMPIRE®.

The simulation results in Fig. 4.1 show the e↵ect of the lateral displacement of the metallization layer C3on the reflection coecient S 11 of the antenna. The black line in the diagram shows S 11 for the nominalposition of C3 aligned according to the design. The displacement of C3 in either X- or Y-direction has anoticeable influence on S 11. However, the resulting reflection coecient of the antenna module with theassumed stacking tolerance is still acceptable for the envisaged application.

Fig. 4.2 shows the simulated radiation pattern with its variations due to displacement of layer C3 inX-direction. Fig. 4.3 is the corresponding plot for displacement of C3 in Y-direction. The upper set ofcurves, marked RHCP (Right-Hand Circular Polarization) shows the amplitude for the co-polar component.The lower set of curves, marked LHCP (Left-Hand Circular Polarization) is the cross-polar componentin this antenna design. Both diagrams show that the e↵ect on the RHCP is negligible for the amount ofdisplacement considered here. However, the LHCP is strongly a↵ected yielding distinct variations for thefour directions (+X, -X, +Y, -Y) that are presented in the two plots. Cross-polar suppression is the di↵erencebetween RHCP and LHCP. This important antenna specification is clearly influenced by misalignment ofthe metallization layer C3, but it is considered still sucient for the expected tolerance in the alignment oflayer C3.

The metallization layer C2 (second from top in Fig. 2.2) contains conductor frames that connect thevia fence shielding the antenna cavity. In the tolerance analysis by EM-simulation this layer turned outto be robust against misalignment in the envisaged range. When plotted in the same scale as the previoussimulations, the variations in the reflection coecient of Fig. 4.4 and the radiation pattern of Fig. 4.5 are sominute that the di↵erence is hardly visible.


Fig. 4.2: Radiation pattern for several lateral displacements of layer C3 in X-direction. 3-D EM simulationwith EMPIRE®.

Fig. 4.3: Radiation pattern for several lateral displacements of layer C3 in Y-direction. 3-D EM simulationwith EMPIRE®.


Fig. 4.4: Reflection coecient S 11 vs. frequency for several lateral displacements of layer C2. 3-D EM simula-tion with EMPIRE®.

Fig. 4.5: Radiation pattern for several lateral displacements of layer C2 in X-direction. 3-D EM simulationwith EMPIRE®.


RF Reflection Measurement of the 4 4 Array Building Block

The 4 4 antenna array building blocks are manufactured using the LTCC prototype line at IMST. Prior tocombining the array building blocks with the RF circuitry, each element of the building block is characterizedusing a probe station measurement setup, as shown in Fig. 4.6. Measurements of reflection coecients ofthe individual antenna elements in one 4 4 building block are given in Fig. 4.7. Four of these were used toassemble the 8 8 antenna array. The comparison of the reflection coecient vs. frequency in Fig. 4.7 forseveral samples shows a very good reproducibility of the antenna elements within one building block. Theremaining di↵erences between the elements have two main reasons: Firstly there is a systematic di↵erencein the symmetry of the inner and outer antenna elements of the 4 4 antenna array building block in thetop right corner of Fig. 4.6, centre elements have immediate neighbours in eight directions, edge elementshave immediate neighbours in five directions, and corner elements have immediate neighbours only in threedirections. Secondly there are stochastic manufacturing tolerances within one building block (e.g. slightvariations in the quality of via connections or printing accuracy within one substrate layer). The measurementresults of antennas from di↵erent manufacturing batches exhibit also good lot-to-lot reproducibility.

The previous section 4.1 is describing how manufacturing tolerances are taken into account from the firstconcept of this highly integrated 30 GHz DBF antenna module through all development steps. The antennamodule originated by this design flow is robust against the inevitable tolerances of a manufacturing process.Furthermore all process steps are carefully investigated to identify potential issues and to improve precision.With 3D FDTD EM simulation the influence of single process deviations and manufacturing tolerances canbe simulated and studied faster and more economic, isolated from other e↵ects. When test-structures aremeasured and analysed, the e↵ect of interest is very often superimposed and obscured by other manufacturingtolerances or variations in the properties of the material. The tolerance analysis is an important part of thedesign process required to limit accuracy requirements to a feasible and economic degree.


Fig. 4.6: Probe station measurement of the 4 × 4 antenna array building block. The top left image shows theprobe station with the VNA (vector network analyser) in the background. The photograph bottom right is thedetail of the two G-S-G microwave on-wafer probes that were used to contact the RF-ports of the antennamodule.

Fig. 4.7: Measurement results for the reflection coefficient of the single elements of a 4 × 4 antenna arraybuilding block. Several LTCC-runs have been fabricated and compared. "R4 K3 4A" designates the run, thepanel, and the position of the antenna module in the panel. "11" to "44" refers to the position of the antennaelement within the module.


4.2 Process Improvements and Limitations Regarding Layer-to-LayerRegistration

Before stacking and collating, each layer of the LTCC stack can be inspected and reworked or discarded ifnot within specification. The final multilayer after lamination and sintering is usually not easily accessiblefor inspection of the layer-to-layer registration. E.g. the Ka-band power divider module of the "MultiFeed"project [74] in Fig. 4.8 does not reveal much about possible issues with alignment in inner layers. Layer-to-layer registration can be monitored by placing test structures (i.e. alignment monitors) like identicalconductor patterns in each layer in unused areas of the LTCC panel ("tile") and examining them in a polishedcut image. However, this is too cumbersome and time consuming for an in-line process control. Moreover,the result is only available after cutting of the tile and polishing of the cut image.

Industrial radiography as shown in Fig. 4.9 is a non-destructive inspection that allows insight into themodule because of the good X-Ray contrast between the conductor metal and the dielectric. In this imagingtechnique dense areas above or below the area of interest will obscure details. An X-Ray CT-scan results ina three dimensional voxel representation of the sample where every point in the 3-D grid is attributed witha density value. Fig. 4.10 depicts a thin horizontal CT slice that shows precisely the conductor pattern ofthe Wilkinson power dividers [109] inside the centre of the LTCC module in Fig. 4.8. The dark spots arethe vias, they are darker because of their higher X-Ray density. This image is generated with an imagingsoftware that allows to compute arbitrary cross-sections or isometric views [133] from the above mentionedthree dimensional voxel representation of the sample.

The CT-scan in Fig. 4.11 allows a close inspection of the vias and their quality. It is apparent that the viasare functional but small voids are visible in their structure.

Pre-production samples of the V-band antenna module in Fig. 4.12 exhibited massive deviations in theelectrical performance (i.e. reflection coecient at the antenna port and radiation pattern) compared to proto-types and the EM-simulation. They were analysed with a CT-scan to find the reasons for their malfunction.The CT-scan identified the issue of misaligned layers but also inhomogeneities in the microstructure of thevias. The CT allows to identify the optimum position for the polished cut image. The micro-photograph ofthe polished cut image at this position confirms the findings in the CT. The larger magnification of Fig. 4.13illustrates the alignment issue in detail and shows also horizontal cracks in the vias and detachment of viaand conductor metallization.

Top. Bottom.

Fig. 4.8: MultiFeed power divider.

4.2 Process Improvements and Limitations Regarding Layer-to-Layer Registration 55

Fig. 4.9: Radiograph overview of the MultiFeedpower divider.

Fig. 4.10: A thin horizontal CT slicealong one layer (voxel size 10 µm) of theMultiFeed power divider.

Fig. 4.11: CT-scan detail of MultiFeed power divider: isometric view of vias.


Fig. 4.12: CT-scan compared to micro-photograph of polished cut image.

When the volume manufacturing process is scrutinized to identify possible deviations from the prototypingprocess, the stacking method is found to be different and not accurate enough. An interview of the operatorsexhibits the use of incompatible via fill paste. Consequently the manufacturing process, particularly thestacking and collating method is reviewed and it is ensured that the stacking tool is used with registrationpins and matching registration holes in the tapes, and compatible metallization pastes are used as specified.

Fig. 4.13: Detailed micro-photograph of polished cut image. The red arrows mark vias with inner cracks anddelaminations from the conductor metal.

4.3 Design for Manufacturability Applied to RF-LTCC Modules 57

4.3 Design for Manufacturability Applied to RF-LTCC Modules

Dovetailing of design and manufacturing capabilities (i.e. DFM, design for manufacturability) opens newoptimization potentials for microwave circuits. E.g. coupling structures separated by one tape layer, whichdemand high accuracy in their relative position, can be printed on top and bottom side of the same tape ratherthan printing them on the top side of two subsequent tapes. There is still the tolerance due to the registrationof the screen print on the tape, but the stacking tolerance is eliminated and no longer relevant for the 1.8 GHzto 18 GHz 10 dB directional coupler [134] built with this method.

Microwave bandpass filters in LTCC multilayer technology are compact, robust, and reliable componentsthat can be integrated in complex systems without further tuning. However, they are very susceptibleto manufacturing tolerances [135]. In principle, a bandpass filter is a chain of coupled resonators. Thenumber of resonators, their resonant frequency, their quality factor, and their coupling determine the filter’scharacteristics. At microwave frequencies lumped element filters (i.e. composed of discrete capacitors,inductors and resistors) are dicult to realize because the size of the components is no longer negligiblecompared to the wavelength. A distributed element approach will represent capacitances and inductances ofthe filter design with transmission lines of di↵erent length and impedance [136]. Whereas this concept isused with various types of waveguides, the filter can be built very compact with striplines in a multilayermodule. The relative permittivity of LTCC (typically in the range of 6 to 8) is further reducing the size ofthe filter [137]. The inherent shielding of the stripline filter is a positive side e↵ect of this construction. Asignificant di↵erence between a distributed element filter and its lumped-element counterpart is that theformer will have multiple passbands, because transmission line transfer characteristics repeat at harmonicintervals [138]. These spurious passbands have to be taken into account for the overall system design.

The conventional design of an interdigital bandpass filter utilizes /4 (quarter wavelength) resonators.These resonators are grounded on one end with vias as shown in Fig. 4.14 for bandpass filter 3 (BPF3). Theposition of these vias with respect to the conductor pattern influences their resonant frequency because itdetermines the electrical length of the resonator. Manufacturing tolerances regarding registration of viasand conductors will thus e↵ect the filter performance. The measurement results of twelve samples of BPF3in Fig. 4.15 show the susceptibility of the filter concept to manufacturing tolerances. Some of the samplesshow the quality fall-o↵ in the centre of the pass band, i.e. too high insertion loss S12 and an insucientreflection coecient S11. This is an e↵ect that is also observed in the production of similar LTCC filters inlarger quantities.

Bandpass filter 2 (BPF2) is optimized for manufacturability and yield in LTCC multilayer technology.The layout of the signal layer in Fig. 4.16 with /2 resonators shows that there is no ground connectionrequired for the resonators because they are open ended on both sides. Here the resonance is determinedby the accuracy of the conductor metallization only. The /2 resonators are approximately twice as longas the /4 resonators for the same centre frequency which increases the width of the filter module. Forthe /4 design the first spurious passband appears at 3 fc whereas the /2 design has its first spuriouspassband at 2 fc. The twelve samples of the /2 design (BPF2) were manufactured in the same batch,i.e. with the same manufacturing tolerances as the samples of the /4 design (BPF3) in figure 4.15. Themeasurements of BPF2 in figure 4.17 are within the specifications and prove the superior manufacturabilityof this filter concept.

The two di↵erent filter designs for 20 GHz with the same target specifications have been built and evaluated.The conventional interdigital filter with /4 resonators proved to be dependent on the exact registration ofvias and conductors. The alternative design with /2 resonators is able to deliver higher yield and to complywith more stringent requirements regarding passband insertion loss and return loss when they are producedwith the same material and under the same conditions [139].


Fig. 4.14: Bandpass filter 3 with λ/4 resonators, layout of the signal layer.

Fig. 4.15: Bandpass filter 3 with λ/4 resonators, measurement results

4.3 Design for Manufacturability Applied to RF-LTCC Modules 59

Fig. 4.16: Bandpass filter 2 with λ/2 resonators, layout of the signal layer.

Fig. 4.17: Bandpass filter 2 with λ/2 resonators, measurement results

5 LTCC Process Adaptation for High Layer CountThe LTCC process facilitates complex modules with a high number of layers. Fifty layers have beenrealized without reaching the limit of the technology [76]. The SANTANA 8 × 8 antenna module isan LTCC multilayer with 17 dielectric layers and a fired thickness of 3.4 mm and thus requires carefuladjustment of process parameters, because the standard parameters apply for modules of 0.4 mm to 2 mmfired thickness. However, most of the LTCC process shown in Fig. 1.3 is not affected, since single layers areprocessed prior to stacking.

Photograph. Cross-sectional view.

Fig. 5.1: Dual band GNSS (Global Navigation Satellite System) antenna. The photograph shows the tuningstubs of the L1 patch antenna on the top surface and the cavities to access the tuning stubs of the E5a patch an-tenna in an inner layer. The cross-sectional view shows the position of the two antenna patches in the top layersand the two corresponding branch line couplers sandwiched between ground planes in the lower part. Branchline couplers are required to feed the antenna patches with 90° phase shift for optimum circular polarization.

This chapter will focus on the two process steps that need to be adapted for LTCC-modules with highlayer count: de-bindering and sintering. For the SANTANA antenna module de-bindering encompasses thesacrificial material used for the formation of cooling channels (see section 3.2 for details) in addition tothe organic contents of tape and pastes. Two analysis tools, TGA (thermogravimetric analysis) and opticaldilatometry [140, 141], are introduced to examine the critical process steps. Adjustments to the sinteringcurve (oven schedule) are derived thereof and implemented.

The dual band patch antenna for the Galileo GNSS (Global Navigation Satellite System) depicted in Fig.5.1 [76] with forty layers of LTCC shall serve as an example to demonstrate the adaptation of the LTCCprocess for an unusual high number of layers. Lamination, de-bindering and co-firing of the LTCC stackare the process steps requiring particular attention. Changes in lamination and sintering parameters willpotentially influence electrical properties, mainly the permittivity. The antenna design needs to be adjustedaccording to the measured results of the precursor test coupons with resonators or antenna patches in theintended frequency range. Usually the number of layers is determined by the complexity of the circuit (i.e.the number of signals, voltage supplies etc. connected between the ICs integrated in the module). In thisparticular application the substrate height is required to achieve the required bandwidth for each of the twoantenna patches. The dual band antenna for two of the Galileo bands [142], L1 at 1.58 GHz in the upper

61

62 5 LTCC Process Adaptation for High Layer Count

L-band, and E5a at 1.18 GHz in the lower L-band is realized as a combination of two coupled patches.Extending the bandwidth of a single antenna to cover both frequency bands would compromise performance,mainly antenna gain and eciency for both operating frequencies. A stacked patch antenna with individuallyfed elements is the solution of choice to combine both frequency bands in one antenna module with optimumperformance for each of them. Circular polarization is attained by separately quadrature feeding each patchthrough a branch-line coupler. These features add further layers to an already considerable substrate height.

From the antenna specifications and a draft design thereof the following requirements for the multilayerare derived:

• Relative permittivity in the range of 7 to 10,

• Low RF loss,

• Thickness of dielectric: 10 mm,

• Multilayer with four dielectric layers.

Adaptation of the LTCC Process

The LTCC process in Fig. 1.3 is parallel for the tape layers until the stack is assembled ("stacking"). Upto this step individual tape layers are processed, i.e. the parameters for these steps remain unchanged withrespect to the standard process. After stacking, the whole stack has to be processed. Taking into account thatthe maximum fired thickness for a standard tape layer is 230 µm, 40 to 50 layers need to be stacked to get afired thickness of 10 mm. Stacking of a thick LTCC substrate is primarily a matter of overcoming mechanicalrestrictions in the stacker, like the length of registration pins. In the next step the stack is laminated underhigh pressure (21 MPa) and temperature (70 C) into a solid block, the laminate. Pressure is distributedevenly in the isostatic press but care must be taken to ensure that the whole mass has reached the requiredtemperature during dwell time before applying lamination pressure. The stack needs to be vacuum baggedbefore inserting it into the water bath of the isostatic laminator. The shape of the stack together with thestacking tray requires optimization to reduce the stress on the lamination bag and avoid leaks. Even thecomposite foil of the bag itself has to be reinforced to withstand the increased tensile stress around the edgesof the stack and the stacking plate.

The composition of LTCC tapes and pastes varies significantly from system to system [143]. Also thesintering mechanism di↵ers considerably [144, 145]. The blue curve in Fig. 5.2 shows the standard sinteringprofile for Heraeus CT700 suggested by the material manufacturer [146] in comparison to other standardsintering profiles. The standard sintering schedule is developed and recommended typically for moduleswith 4 to 8 layers. It starts with de-bindering up to around 500 C, followed by a well defined ramp up tosintering temperature. The gradient of this ramp is adjusted for optimal co-firing of LTCC tape, via fill paste,conductor paste, and possibly further functional pastes. After sucient dwell time at sintering temperature,the oven is cooled with the correspondingly defined rate. With a stack of approximately ten times thestandard thickness a much longer de-bindering time is to be expected. A thermogravimetric analysis (TGA)is one method to look closer into the de-bindering phase by measuring the weight loss over temperature.Concerning the LTCC process, it is an accurate measurement of solvents and organics removed of the greenlaminate during de-bindering [147].

The result is displayed in Fig. 5.3, it shows the change of mass over time for Heraeus CT 700. The greencurve is the output of a recording micro-balance which measures the weight of the LTCC specimen in thefurnace. A horizontal portion, or plateau indicates constant sample weight. In a curved portion, the steepnessof the curve indicates the rate of mass loss. The red curve, the DTG, is the first derivative thereof whichshows more clearly discerned the features in the TGA curve. Its relative minima at 247.9 C, 296.9 C and403 C indicate the temperatures where changes in mass (i.e. loss) are pronounced. These are exactly thetemperature steps where extra time is required to de-binder an LTCC module of unusual height. The peaks

63

Fig. 5.2: Sintering profiles for different LTCC material systems.

TG, W

eigh

t Ret

aine

d (%

)

DTG

, Der

ivat

ive

of W

eigh

t Ret

aine

d (%

/min

)

Change of mass: -0.49 %



Temperature (°C)

Remaining mass: 91.46 % (990.2 °C)

Fig. 5.3: Thermogravimetric Analysis of Heraeus CT700. The green curve is the TGA (left X-axis) whichrecords the change in mass and the red chain line is the first derivative thereof, the DTG (differential thermo-gravimetric analysis, right X-axis).


will move slightly to lower temperatures if the heating rate is lower and to higher temperatures for thickerspecimen.

Fig. 5.4: Shrinkage of DP 951. Shrinkage is measured with an optical dilatometer so that the measurement isobscured and interrupted by fumes from 30 min to 80 min during the de-bindering phase.

Whereas the TGA looks at the relative change of mass in the specimen, an optical dilatometer measuresdimensional change over temperature [141]. The optical dilatometer is a furnace with windows to illuminate(backlit) and observe the sample with a camera to measure dimensional change over time during de-binderingand sintering. Fig. 5.4 shows the results for DuPont 951. Fumes generated during de-bindering obscure theoptical measurement, but the data before and after this phase show only the difference that is to be expectedfrom thermal expansion. The actual lateral and thickness shrinkage occurs in the beginning of the sinteringphase from just below 800 C to 875 C. Fig. 5.3 and Fig. 5.4 in combination show that the laminate loosesmass mainly in several steps during de-bindering (200 C to 500 C) when organic contents decompose.Dimensions do not change much in this period, the LTCC becomes porous [105]. With the onset of sinteringthe substrate starts to densify and shrinkage occurs in all three dimensions for unconstrained sintering.

To further facilitate de-bindering and gas flow, an extremely porous setter is used. By implementing theresults of the above analyses, the total oven schedule is increased from 8 hours to 60 hours (see Fig. 5.5).The better part of this increase in duration is due to a prolonged de-bindering phase. After this optimization itis possible to sinter a stack of 41 tape layers (10 mm height) without cracks or delamination. An examinationby cross sectioning shows that the sample is also densely sintered. Vias through the whole stack were slightlyrecessed in the top layer which is perfectly acceptable for this application.

Lamination and sintering conditions influence shrinkage and density of the LTCC module. With theprocess changes necessary for the high layer count it is to be expected that nominal values for shrinkagewill not apply. A test tile with a simple patch antenna shows that the shrinkage in thickness (Z) is onepercentage point less than nominal whereas the lateral shrinkage (X, Y) is one percentage point higher thannominal. Since the sintered density of the tile is affected, permittivity will also be influenced by the alteredoven schedule. The geometry of the same test tile (dielectric thickness and patch size) is measured and

65

Fig. 5.5: Sintering profile for Heraeus CT700 with 41 tape layers (black) compared to standard sintering curve(red). The main difference is the prolongation of the de-bindering phase from 200 C to 500 C for the highnumber of layers. Both curves show the programmed settings of the furnace.

entered into IMST’s FDTD software EMPIRE™ [75] to simulate the reflection coefficient of the antenna. Thepermittivity in the simulation is then adjusted to bring simulation results into agreement with the measuredreflection coefficient. The resulting relative permittivity of 6.6 serves as a basis for the final antenna design.Measurement results for the radiation patterns in the E5a band (Fig. 5.6) and the L1 (Fig. 5.7) band confirmsimulation results and field tests with the antenna in a GNSS instrument prove a positioning accuracycomparable to much larger antennas.


Fig. 5.6: Measurement results for the radiation pattern of dual band GNSS antenna in the E5a band.

Fig. 5.7: Measurement results for the radiation pattern of dual band GNSS antenna in the L1 band.

6 Integrated ResistorsCompared to surface mounted components, embedded resistors in LTCC modules require less space in theouter layers and reduce the parasitic impedance associated with an SMD package and the line length requiredto route the component. A planar resistor used as an embedded passive gives the designer full control overthe geometry of the component and it provides the solution for high integration density and good microwaveperformance. Integrated resistors are also improving the MTBF (mean time between failures) of a moduleby reducing the number of interconnects [148, 149, 150]. Typical applications for resistors in microwavecircuits are terminating resistors, attenuators and power splitters. A fully resistive power splitter is compactand has a very high bandwidth. The input port and all output ports are connected via a resistor to a centralnode. If all ports have the impedance ZL and the number of outputs is n then the resistors Ra are equal andcalculate to:

Ra = ZLn 1n + 1

. (6.1)

However, there is a considerable attenuation associated with this construction:

a = 20 log10

1n

!(6.2)

expressed in dB. Wilkinson suggests a matched and symmetric 1:N power splitter, that is lossless (assumingideal conductors) if all output ports are at equal phase and amplitude [109]. Fig. 6.1 shows a micro-striplinerealization of the 1:2 version. The black area in the photograph in Fig. 6.1 is the screen printed resistor of thedepicted Wilkinson power divider. The /4 lines transform the impedance of the parallel outputs ≠ and Æ tomatch the impedance at port ¨. This impedance transformer can be replaced by multiple /4 line sectionsto increase the bandwidth of the component [151]. If operated as power divider with equal impedance atthe output ports ≠ and Æ, no power is dissipated in the resistor R. If there is an unsymmetric signal, e.g. areflection from a defective amplifier or antenna input, this is absorbed in the resistor. By design the outputports are isolated. Both properties provide operating reliability for the overall system. Particularly the 1:2and 1:3 dividers are suitable for planar circuits. Furthermore these components can be cascaded to formlarge beam forming networks (BFN) like the 1:6 power divider in Fig. 4.8 [74] or the 1:32 microstrip powerdivider which is used in the TerraSAR-X and TanDEM-X earth observation satellites [152]. The SANTANAantenna module uses integrated resistors for the lossy line termination [81] of the branch line coupler and asprinted absorbers in inner layers to suppress parallel plate modes in the dielectric.

Along with the conductor pastes LTCC technology has adapted screen printed resistors from thickfilmtechnology. Like conductor pastes, resistor pastes have three principal constituents:

• The functional phase is a powder of conductive oxides like ruthenates or silver palladium and eventuallya small proportion of metal.

• The adhesion element (also called binder): In the paste the adhesion element is glass powder likelead borosilicate glass, a crystalline oxide or a combination of the two. The adhesion element willembed the functional phase and provide adhesion to the LTCC substrate after sintering. The binderalso influences the mechanical properties of the final film.

• The paste vehicle consists of an organic binder, and a solvent or thinner. The combination of theorganic binder and thinner is often referred to as the vehicle, since it acts as the transport mechanismof the active and adhesion elements to the substrate. The paste vehicle provides for the thixotropy theresistor paste needs to be screen printed to the LTCC tape.

67

68 6 Integrated Resistors

Photograph Schematic

Fig. 6.1: Wilkinson power divider.

6.1 Microstructure of Thickfilm Resistors

The electrical performance of thickfilm resistors strongly depends on the microstructure of the fired resistor[153, 154]. The peculiar thermal behaviour and the current noise thereof are only comprehensible witha conception of the physical structure [155, 156]. Furthermore the influence of process parameters onelectrical properties will be illustrated by means of micro-structural development during the process. Forthickfilm resistors with a sheet resistance of 10 Ω/! to 100 kΩ/!, Ruthenate based pastes are most common.Ruthenium oxide RuO2, lead ruthenate Pb2Ru2O7 or bismuth ruthenate Bi2Ru2O7 are used as the functionalphase in volume fractions of 5 % to 60 % [157] to formulate resistor pastes with a wide range of sheetresistance. The particle size of the functional phase in resistor pastes (e.g. ruthenium oxide) ranges from10 nm to 200 nm and that of the adhesion element (glass) from 1 µm to 3 µm. Figure 6.2 visualizes the sizerelation of ruthenium oxide and glass particles in typical resistor pastes.

In a well matched LTCC material system [158] differential shrinkage of resistor paste and tape is minimizedby selecting the appropriate glass in the resistor paste. The residual tendency of the multilayer substrateto warp during sintering is mitigated by careful adjustment of the temperature profile in this process step.Infiltration of the tape glass into the resistor and vice versa is a further challenge of co-firing. Devitrifiableglasses in tape and resistor paste are an approach to mitigate infiltration and excessive chemical reactionsbetween resistor, tape and termination metal [159]. At temperatures above 820 °C the highly viscous glassrapidly devitrifies from glass to ceramics. The glasses of post-fired resistor paste usually remain in glassystate after melting.

Like conductors, resistors are usually screen printed. Typical screen specifications are 280 to 325mesh/in and 10 to 15 µm emulsion build up. The target for the dried thickness of the printed resistor is20 to 30 µm depending on the specifications of the paste. This is to ensure optimum repeatability andperformance of the sintered resistor. Microwave circuits frequently demand for resistors of very small size(L ≪ 500 µm or W ≪ 500 µm) where the printing resolution of resistor pastes becomes a limiting factor fordimensional accuracy. Changing the traditional printing sequence by printing resistors first and terminationsecond helps to mitigate the inaccuracies in the print geometry of the resistor. Further improvement of

6.1 Microstructure of Thickfilm Resistors 69

Fig. 6.2: Particle size comparison in resistor pastes.After Imanaka [27].

Fig. 6.3: Microstructural changes in thickfilm resistors associated with Ostwald ripening:a) initial sintering stage (t < 60 min) b) optimum microstructure (t = 60 min)c) extended ripening (t > 60 min). After Yang [155].

definition for miniaturized resistors can be achieved by utilizing the fine line techniques that were developedfor LTCC conductors [36]. Whereas direct write methods like dispensing with a micropen [160], inkjet [161]and aerosol printing [162] are available for co-fired and post-fired resistors, thin film resistors [163, 164, 165]can only be applied post fired and are thus limited to outer layers.Before lamination, the better part of the solvents is removed in a drying step. De-bindering at 300 C to600 C will remove organic components completely. At this stage only the functional phase and the adhesionelement remain. The following sintering process changes the glass in the resistor tremendously. The glasspowder melts and wets the RuO2 particles. Large glass particles devitrify. While RuO2 is chemically stableduring sintering, its distribution in the compound changes considerably. X-ray data show that smaller particlesdissolve and redeposit on the surface of larger parts [155] over prolonged time at sintering temperature. Thiseffect is known as Ostwald ripening [166]. Figure 6.3 depicts the crystallite size change in a schematic way.Fine oxide particles which are present in the initial resistor paste are represented by small grey dots, mediumsized particles are shown as open circles and large particles as solid circles. At the beginning of the ripeningprocess (a) part of the fine particles are dissolved and additional medium size particles appear (b). At thisstage medium size particles dominate and the resistance is lowest. Prolonged sintering (c) leads to furthergrain growth and large particles dominate the structure. As the total amount of RuO2 is constant, the distancebetween conducting particles in the resistor increases with grain coarsening. The resistivity is consequently


increased. The fired resistor can be envisaged as three-dimensional network of percolation paths [167]formed from conducting particles with thin layers of glass in between [157]. These thin glass layers are themain cause for the piezoresistive behaviour of thickfilm resistors, for their noise characteristic and for theirpeculiar temperature dependence of resistance. The sensitivity of thickfilm resistors to firing conditionsin many respects (R, TCR, noise, etc.) suggests to meticulously keep all process parameters within thelimits defined during process qualification. This is particularly true for the sintering profile, temperaturegradient and atmosphere in the furnace. Reversely a reference substrate of known behaviour can be used tocharacterize the uniformity of the sintering furnace in terms of temperature distribution and gas flow. Screenprinting parameters and sintering profile can be used to fine tune the performance of a resistor. However,for co-fired resistors in LTCC the sintering profile is of course identical to that of the whole LTCC stack.Therefore the degree of freedom for optimizing the sintering profile with regard to resistor properties islimited.

6.2 Conduction Mechanism in Thickfilm Resistors

During sintering RuO2 particles become coated with glass which isolates them from adjacent particles.Thus electrical conduction is more complex than in a film of the same metal oxide. The ’glaze resistor’was already in industrial use for several decades [168] before a satisfactory explanation of the nature ofits conductivity was provided in the late seventies of the past century. Pike and Seager [169] proposea tunnelling-barrier model to describe the electrical conduction mechanisms in thickfilm resistors. Theresistance of the conductive particles and the resistance of the separating glass barrier contribute to theexpression which characterises the temperature dependence of resistance R(T ) of a thickfilm resistor under alow electrical field:

R(T ) =12

Rbo

sin(aT )

aT

! 1 + e( E

kT ) + Rmo(1 + bT ) (6.3)

where T is the absolute temperature, a accounts for the tunnelling probability, E is the electrostatic chargingenergy, k is Boltzman’s constant, Rbo is a factor determined by two di↵erent contributions to the overalltunnelling mechanism and Rmo is the metallic-phase resistance if the temperature was extrapolated to absolutezero. The term sin(aT )

aT describes the weak temperature dependence of tunnelling through insulating films.a is the first coecient of a Taylor expansion of the natural logarithm of the tunnelling probability andcan be estimated by the insulator barrier height [170]. Typically a increases as the tunnelling probabilitydecreases. Based on the Fermi-Dirac distribution, e( E

kT ) accounts for the equilibrium number of chargecarriers with a charge energy of E, which is the minimum energy required to make the tunnelling transition.The temperature dependence of the resistance of metallic oxides is expressed by Rmo(1 + bT ). At lowtemperatures, according to this model, the limited availability of charge carriers with enough energy totunnel through the glass barrier dominates the resistance. As temperature rises there are more charge carriersavailable and resistance drops. When temperature increases further, resistance due to thermal scattering ofcharge carriers within the conductive oxide particles (the rightmost part in equation (6.3)) becomes dominantand the overall resistance increases again. These e↵ects lead to the near parabolic shape of the typical R(T )curve of thick-film resistors (see Fig. 6.4). Whereas this model explains the peculiar behaviour of thick-filmresistors over temperature, there are two more factors that will influence the temperature characteristic. Oneis the change in height of the tunnelling barrier due to thermal expansion of glass, the second factor isthe piezoresistivity of RuO2-based resistors [171]. The latter makes the resistor susceptible to mechanicaltensions which arise from thermal mismatch within the compound. This is not only an unwanted e↵ect forfixed resistors but it facilitates also sensor applications like strain gauges and pressure sensors [172].

6.3 Electrical Properties of Thickfilm Resistors 71

6.3 Electrical Properties of Thickfilm Resistors

Thick- and thin-film resistors are characterized by their sheet resistance R!. This is a design-orientedsimplification derived from the general formula for resistance

R =ρ

tLW, (6.4)

assuming homogeneous thickness t in the resistive film with length L and width W. ρ is the resistivity(specific electrical resistance), an intrinsic property of the material. By introducing sheet resistance

R! =ρ

t, (6.5)

the general formula is reduced to a two-dimensional representation

R = R!LW. (6.6)

The sheet resistance R! is the resistance of a square resistor and thus is in Ω. For enhanced clarity it iscommon to use Ω/! (’Ohms per square’). Further explanations will show that the accuracy of this model islimited due to variations in thickness and the influence of termination metallization. However, it is a veryuseful tool for the layout of LTCC resistors. In a simple and straight forward method the resistor is composedof squares where the the square in the corner of a meander is counted half a square. The resistance resultsfrom the product of the number of squares and the sheet resistance R!.

The change of resistance over temperature is expressed as temperature coefficient of resistance (TCR)

TCR =R(ϑ2) − R(ϑ1)R(ϑ1)(ϑ2 − ϑ1)

106 (ppm). (6.7)

‰

˚

Fig. 6.4: Change of resistance versus temperature for a thickfilm resistor. The parabolic shape of the curve istypical for thickfilm resistors and has its reason in their conduction mechanism.


Fig. 6.4 shows the change of resistance vs. temperature for a typical thickfilm resistor. To account for thevarying gradient two numerical values are used to characterize the thermal coecient of resistance:

• HTCR (Hot Temperature Coecient of Resistance), the reference temperatures for this chord slopeare chosen depending on application and operating temperature range.E.g. in Fig. 6.4: #1 = 25 °C and #2 = 75 °C, HTCR = 25 ppm/K

• CTCR (Cold Temperature Coecient of Resistance), accordingly this is describing the characteristicsbelow ambient temperature.E.g. in Fig. 6.4: #1 = -25 °C and #2 = 25 °C, CTCR = -100 ppm/K

Measurement methods for TCR and TCR-tracking are defined in the following IEC and MIL standards:IEC 60115-1 4.8.4.2 [173], IEC 60195 [174], MIL standard 202 Method 304 [175] and 308 [176].

Thermal noise in passive conductors is the measurable evidence of random movement of charge carriersinside the same conductor. The e↵ect was first described by Johnson [177] and Nyquist [178] and is thereforealso referred to by their name. The spectral distribution of thermal noise is flat which is why it is also labeledas "white" noise in analogy to optical spectra. The power of thermal noise

Pn = 4kT B (6.8)

is independent of externally applied voltage and material of the conductor. It is the product of absolutetemperature T , bandwidth B and Boltzmann’s constant k = 1.3807 10=23 J K=1.

The power spectral density S T is constant for thermal noise:

S T = 4kTR with the unit [S T ] = V2/Hz. (6.9)

The voltage of thermal noise is

vrms =p

4kT BR (6.10)

with R the resistance of the component.Conduction over tunnelling barriers and along percolation paths [179] causes current fluctuations when

current is passed through a thickfilm resistor. Noise in a resistor is divided into thermal noise and excessnoise attributed to this bunching and releasing of electrons due to fluctuating conductivity, also called currentnoise. Current noise is directly proportional to the current flowing through the resistor. The power spectraldensity of current noise

S E =bU2

fwith the unit [S E] = V2/Hz. (6.11)

was found to be inversely proportional to frequency [180, 181]. b is a constant that depends on resistor pasteand process parameters. 1/ f noise is also labeled "pink" noise according to the above mentioned analogyto optics. Resistors used in LTCC systems show the same characteristic noise behaviour as their thickfilmancestors [182].

Low frequency noise is also a good instrument for reliability assessment in complete modules [183].Noise from di↵erent sources is considered uncorrelated and thus added geometrically. Figure 6.5 showsthe resulting noise spectral density (in log. scale) for thickfilm resistors. The frequency where currentnoise and thermal noise are of the same level depends on material, geometry and thermal situation of theresistor. RF and microwave circuits are operated at frequencies where 1/ f noise is not expected at firstglance. However, 1/ f noise is upconverted in mixers and synthesizers and appears as phase noise at highfrequencies [184, 185].

6.3 Electrical Properties of Thickfilm Resistors 73

10 100 1 k 10 k 100 k 1 M 10 M

nois

e sp

ectra

l den

sity

frequency (Hz)

current noise predominates

thermal noise predominates

Fig. 6.5: Spectral density of total noise in thick-film resistors.

Current noise in resistors is specified as noise index

NI = 20 log10

vrms

VDC

!. (6.12)

The noise index NI is expressed as dB in a decade with vrms the root mean square noise voltagein µV and VDC the voltage drop across the resistor in V [186]. Figure 6.6 compares the noise indexof thick film (LTCC) resistors to other resistor technologies [187].

Resistors change their resistance when they are exposed to tensile or compressive stress. As explainedearlier, the conduction mechanism of thick film resistors makes them particularly useful for strain gauges

Wire WoundMetal Foil

Deposited CarbonCarbon Composition

Thin FilmLTCC and Thick Film

-50 -40 -30 -20 -10 0 10Noise Index (dB)

Integrated Resistors

Discrete Resistors

Fig. 6.6: Noise index of di↵erent resistor types after [187]


where this e↵ect is exploited to measure this mechanical quantity. Sensitivity to stress is expressed as GaugeFactor GF. In equation (6.13) "1 and "2 denominate the the stress of two di↵erent load situations (" = l/l).

GF =R("2) R("1)R("1)("2 "1)

. (6.13)

The maximum operation voltage of a resistor is specified as RCWV (Rated Continuous Working Voltage)and is determined by

RCWV =p

PmaxR. (6.14)

This value needs to be reduced if the dielectric strength of the resistor is defining a lower limit for theoperating voltage [173]. S TOL (Short Time Overload) is specifying the permanent change in the resistanceafter applying 2.5 times RCWV for 5 seconds. The relative resistance change is then measured after 30minutes (specification e.g. |R| 5 (0.02 R + 0.1 ) [188]).

6.4 A Novel Concept for mm-Wave Matched Loads and Power Splitters inLTCC

The matched load at port 4 of the branch-line coupler in Fig. 2.5 is using a lossy transmission line followedby a termination with a 50 thickfilm resistor. The lossy transmission line is realized by printing thestripline with resistor paste instead of metal paste. The EM-wave is already attenuated when it arrives at theresistor termination and reflections caused by mismatch are further attenuated on their way back. With thismethod a buried termination is created with good microwave performance (i.e. low reflection) even when theresistor cannot be laser trimmed. Furthermore this concept can be utilized to spread the power dissipationover a larger area and allow for better cooling.

The Wilkinson power divider in Fig. 6.1 [109] is a very common component in planar BFNs (beamforming networks) [54]. It can be integrated in complex multilayer circuits [74] and is suitable for symmetricand unsymmetric power dividers (respectively combiners). Because of their good isolation, a BFN withWilkinson power dividers is robust against defects in individual branches. The resistor in the centre of thepower divider is a lumped element that should be very small compared to the wavelength . To be consideredas a discrete component, the size of the resistor should not exceed /10. This is demanding higher printingresolution and accuracy with higher frequencies, eventually with the consequence of unacceptably lowmanufacturing yield. Above all, the miniaturisation of the resistor is in conflict with the power rating thatmay be a requirement in some applications like BFNs in the transmit path of a system.

The schematic diagram of the two-way Gysel power splitter [189] in Fig. 6.7 shows that the resistorsrequired in this concept are two external 50 terminations (RL) connected to ground. Choosing 50 ter-minations has the advantage that they can be connected to the power splitter with transmission lines(Z = Z0 = 50 ) of arbitrary length. This allows the use of physically larger terminations which can handlehigher power and can be adapted to ease of manufacturing. The signal input at port 1 is split into two outputswith equal phase and amplitude at ports 2 and 3.

Within the iKersatec project a new approach to microwave terminations was developed and realized[190, 191]. The topology of the Gysel power splitter allows to utilize this concept in multilayer BFNs [192].The underlying idea of this matched load is to move the resistor which is absorbing the incident wave fromthe centre conductor of the stripline to the ground plane. In this way the resistor becomes larger and lesssusceptible to manufacturing tolerances. Fig. 6.8 is an isometric view of the multilayer construction. Thestripline in the centre layer ends in a stub that is exciting a cavity resonator which is confined laterallywith rows of vias. Top and bottom metallization are replaced by a thickfilm resistor layer to create a lossy

6.4 A Novel Concept for mm-Wave Matched Loads and Power Splitters in LTCC 75

Fig. 6.7: Schematic of a two-way Gysel power splitter. The signal input at port 1 is split into two outputs withequal phase and amplitude at ports 2 and 3. The transmission lines with impedance Z1...4 have an electricallength of /4.

resonator. The FDTD EM-simulation result for S 11 of the lossy resonator 1 in Fig. 6.9 shows very goodperformance around 38 GHz even with a realistic tolerance of the sheet resistance. The three curves indicateresults for the nominal sheet resistance R of 120 and the tolerance analysis for a deviation of ± 20%. Thelength of the stub is the parameter to tune the operating frequency range. However, there is no DC returnpath. On one hand this can be advantageous in some applications, on the other hand the ground connectionmay be required in active circuits or to avoid ESD (electrostatic discharge) issues.

Lossy resonator 2 in Fig. 6.10 features a via at the end of the resonator stub which provides the groundconnection and extends the useful frequency range at the lower end as demonstrated in the EM-simulationresult for S 11 of the lossy resonator 2 in Fig. 6.11.

The third approach of the lossy resonator includes a u-shaped aperture in the thickfilm resistor layer in thevicinity of the via connection at the end of the resonator stub as depicted in Fig. 6.12. This adds a furtherdegree of freedom in the design which has been used successfully to increase the useful frequency rangefrom DC to 50 GHz (see Fig. 6.13). Because the resistor is in the top layer, this additional feature in thelayout can be laser trimmed to tune the circuit in operation.

Fig. 6.15 shows the layout of a symmetrical two-way Gysel power splitter that comprises two lossyresonators (third approach) as matched loads (in the upper half of the image). As a comparison the layout ofthe Wilkinson power divider in the MultiFeed BFN is shown in Fig. 6.14 [74]. In this case it was not feasibleto keep the size of the resistor (i.e. the pale rectangle in the layout) within the dimension of /10. Thereforethe overall electrical length of resistor plus connecting lines was determined to be /2 in order to have theequivalent phase relationship (of a lumped element) at the respective nodes.

The comparison of the tolerance analysis (FDTD EM-simulations) of the isolation between port 2 and3 for the Wilkinson and the Gysel divider in Fig. 6.16 shows that there is a significant influence of thesheet resistance on the performance of both types of dividers. However, the isolation of the Gysel divideris sucient for the considered tolerances over the whole operating frequency range of the developed BFN(27.5 GHz to 31 GHz).


Fig. 6.8: Isometric view of lossy resonator 1.

Fig. 6.9: Reflection coecient S 11 vs. frequency of the lossy resonator 1. The three curves show simulationresults for the nominal sheet resistance R of 120 and the tolerance analysis for a deviation of ± 20%. Thered line marks the specified limit for S 11 of -10 dB.


Fig. 6.10: Isometric view of lossy resonator 2.



Fig. 6.12: Isometric view of lossy resonator 3. The structure of the inner layers under the resistor area is identi-cal to Fig. 6.10.



Fig. 6.14: Layout of the Wilkinson powerdivider in the MultiFeed BFN. The pale rectan-gle is the thickfilm resistor.

Fig. 6.15: Layout of the Gysel powerdivider with lossy resonator 3 as matchedloads. The turquoise rectangle is thethickfilm resistor of RL in Fig. 6.7.

Fig. 6.16: Comparison of the isolation S 23 vs. frequency of a Gysel and a Wilkinson power splitter with vari-ation of sheet resistance as parameter in the FDTD simulation with EMPIRE®. The three curves per dividershow results for the nominal sheet resistance R of 120 and the tolerance analysis for a deviation of ± 20%.


The concepts for millimeter-wave matched loads and power splitters in LTCC presented in the previousparagraphs demonstrate the flexibility of the ceramic multilayer technology to cater for the di↵erent require-ments of microwave circuits in terms of power handling capability and high yield combined with ambitiousRF specifications and the progress which has been achieved in the development projects of the electronicallysteerable antennas.

6.5 Qualification of Resistor Paste for LTCC

The Ishikawa diagram in Fig. 6.17 illustrates the interdependencies of material, process and electricalproperties of LTCC resistors. Ideally all material (tape and paste) and process (print and fire) parameters arein a window that will allow small variations without any e↵ect on the final result. The resistor properties arethen only influenced by the design (length L and width W). In reality the process window is not as wide asthe engineer would like it to be and a qualification procedure is needed to establish the basis for the finaldesign and to verify the whole production process regularly. Test samples with resistors in realistic (i.e. closeto the final design in length and width) dimensions are fabricated with production parameters to get verifieddata for sheet resistance and all other relevant electrical specifications including the influence of terminationmetal. Use of overglaze and the position in the stack [193] (i.e. top layer or buried and how many layers oftape are on top of the buried resistor) have a significant e↵ect on resistance as well as on TCR. Refiring maybe needed for post-fired resistors and metallizations. This will a↵ect resistance and TCR also and thereforedetermination of refiring drift is part of the qualification procedure. If a new batch of paste or tape is used inproduction it is good engineering practice to do a "send ahead". In other words a small quantity is producedand carefully checked against all relevant specifications in order to adjust parameters to compensate fordeviations before volume production continues with the new material batch.

6.5.1 LTCC Resistor Paste on different Tapes and Top and Buried compared

Since these thickfilm resistors are co-fired (i.e. sintered together) with conductors and dielectric tapetheir mutual chemical and physical interaction is considerable during this process step. In this respectLTCC resistors are di↵erent from other film resistors in terms of process and its influence on electricalproperties. The following comparison of resistors from the same paste in di↵erent situations and on di↵erenttapes demonstrates the strong interdependency for the parameter sheet resistance [194]. In the course of the

Resistor Properties:ResistanceToleranceTCRCurrent NoiseStabilityPower RatingFrequency ResponseVoltage Coefficient

Design

Print

Paste

Fire Trim

LengthWidth

Furnace Profile

Furnace Atmosphere

No. Of Re-Firings

Peak TemperatureDwell Time

Furnace Load

Temperature Distribulion

Method (e.g. Laser)Trim Cut

Overglaze

Termination MetallurgyOrientationrelative to printing direction

Particle Size

Paste ReologyScreen Mesh

Emulsion Thickness

Print Parameters Speed, Force, Snap Off etc.

Glass ContentPaste Composition

Min. Remaining Width

Print Thickness

Outer Layer or.Buried Pot life

Homogenisation

Fig. 6.17: Process and material dependencies of resistor properties shown in an Ishikawa diagram.

6.5 Qualification of Resistor Paste for LTCC 81

MultiFeed project, the low loss LTCC material DuPont 943 [195] was discontinued and replaced by DuPont9k7 [196]. For designs in the new material combination a re-qualification of the LTCC material systemincluding resistors was required. Wilkinson power splitters [109] play an important role in the beam formingnetwork of the MultiFeed module [74]. Fig. 6.1 shows that the thickfilm resistor is not accessible for DCmeasurement of resistance. This is a further reason to have a resistor with the same size and orientation (i.e.angle to the printing direction) on the tile (panel) to measure the actual resistance and to monitor the process.The resistor pastes of the DuPont 943 system [197], HFB12 with a nominal sheet resistance R of 25 andHFB22 with a nominal sheet resistance R of 200 are also recommended by the manufacturer for theDuPont 9k7 system. The investigation is extended to include further resistor dimensions, they are listed intable 6.1.

The supplier is performing material characterization of resistor pastes with a square L = W = 1 mm[197, 198] or rectangular resistor L = 2 mm, W = 1 mm [199]. In microwave and millimeter-wave circuitsresistors are typically much smaller. For optimum RF performance the dimensions should be very smallcompared to the wavelength . Practical limitations for the miniaturization of resistors are the maximumpower density and the minimal size that can be screen printed with good reproducibility. The printingresolution of thickfilm resistor paste is much lower than that of conductor paste. The printing accuracy of theresistor is also hampered by the fact that it has to be printed over the edge of the termination to provide thenecessary overlap with the conductor metal. Furthermore the resistance at the interface between terminationmetal (i.e. the conductor paste connecting to the resistor) and the resistor is part of the electrical resistanceof the component. This interface is also influenced by chemical reactions during sintering [200] like silvermigration into the resistor material [201]. In the qualification procedure, these e↵ects are summarized in thelength e↵ect of sheet resistance and TCR [198].

Fig. 6.18 and 6.19 show the measurement results for resistors manufactured in di↵erent batches withthe same screens and with paste and tape from the same manufacturing lot. Fig. 6.18 compares the resultsfor sheet resistance R of the DuPont HFB12 paste as buried resistor in DuPont 943 and 9k7. The finalmodule requires solderable conductors on both sides. The AuPtPd (gold, platinum, palladium) paste for thatpurpose is post-fired, thus the module will be exposed to two refiring cycles. The nominal sheet resistanceR of 25 is adjusted accordingly by the refiring drift specified in the data sheet (R after 2 refires: 38 %to 48 % [197]) and indicated as horizontal chain line in the diagram. The length of the error bar showsthe variation of the samples and the centre mark the mean value. Even though the variation in size isvery moderate (factor 2), there is a clear influence of the resistor size on sheet resistance. Resistor R3 is150 µm wide (this dimension should be easily printable) and shows considerably longer error bars than thewider samples. The comparison of the resistors in the di↵erent LTCC materials shows a significant di↵erencein sheet resistance and confirms the interaction of tape material and resistor paste.

Fig. 6.19 compares the results for sheet resistance R of the DuPont HFB12 paste as buried resistor inDuPont 9k7 and as resistor on the top layer of the same LTCC substrate. They are both co-fired. Analogueto the previous diagram there is a considerable e↵ect of the resistor dimensions on the sheet resistance. For aburied resistor, the LTCC tape is on both sides and the surface for chemical reaction between both materialsis roughly double. On the other hand, a top resistor is more exposed to the oven atmosphere which will alsoreact with the paste during sintering. The measurements confirm that the position of the resistor in the stack(top or buried) has a strong influence on its sheet resistance.

Table 6.1: Resistor dimensions.

Resistor R1 R2 R3 R4

length ( µm) 1300 1000 700 1235

width ( µm) 300 400 150 330

aspect ratio L/W 4.33 2.5 4.66 3.74


Ω

Fig. 6.18: Sheet resistance R! of DuPont HFB12 paste buried in two different LTCC substrates, DuPont 943and 9k7.

Ω

Fig. 6.19: Sheet resistance R! of DuPont HFB12 paste buried and on top of DuPont 9k7.


6.5.2 LTCC Resistors Comparative Study

The properties of LTCC resistors depend on a plethora of process parameters (see Fig. 6.17 for a short list).The question what variations are to be expected if a given design is transferred from one manufacturingsite to another leads to a comparative study within the consortium of the iKERSATEC project [192]. Thefollowing project partners with LTCC manufacturing capabilities participated in the comparative study:Technical University of Ilmenau (TUI), Micro Systems Engineering GmbH in Berg (MSE) and IMST GmbHin Kamp-Lintfort. This section describes how a resistor test coupon is manufactured with the same screensand stencils in these di↵erent facilities. All materials like LTCC tape, conductor and resistor paste are fromthe same manufacturing batch. Best care and attention is applied to have identical process parameters for allparticipants. The evaluation of samples printed and fired at di↵erent manufacturing sites shows the variationwhich is to be expected e.g. when a design is transferred from prototyping into volume production.

The resistors in the test layout are designed to include typical dimensions and aspect ratios (length/width ofresistor area) plus the size that is used by material suppliers to characterize electrical properties of thickfilmresistor paste. These dimensions are summarized in table 6.2 and designed into the layout shown in Fig. 6.20.All resistor terminals are accessible for a four-wire measurement of the resistance. 16 cells of this test layoutare arranged on the test panel as depicted in Fig. 6.21. The screens and stencils for the manufacturing of thetest panels are designed to be suitable for top layer and buried resistors.

The design of experiment in table 6.3 and Fig. 6.21 should include the following variations:

1. Position of the resistor in the stack, RTL: top layer or RIL: inner layer.

2. Resistor paste: DuPont HFB12 with a nominal sheet resistance R of 25 / or DuPont HFB22 witha nominal sheet resistance R of 200 / [197].

3. LTCC facility that is printing, stacking and laminating the LTCC stack: Technical University ofIlmenau (TUI), Micro Systems Engineering GmbH in Berg (MSE) or IMST GmbH in Kamp-Lintfort.

Table 6.2: Resistor dimensions for the iKERSATEC resistor X-check.

Resistor R1 R2 R3 R4 R5 R6

length ( µm) 500 250 6000 500 1000 1000

width ( µm) 250 250 300 500 500 1000

aspect ratio L/W 2 1 20 1 2 1

Fig. 6.20: Resistor test layout (top layer). The resistor terminals are designed for four-wire measurement.Dimensions are according to table 6.2.


Fig. 6.21: Layout of the resistor test panel (top layer), sintered dimensions, dimensions in parenthesis arescaled ("un-shrunk") for screens and stencils.

4. LTCC facility that is sintering the LTCC laminate: Technical University of Ilmenau (TUI), MicroSystems Engineering GmbH in Berg (MSE) or IMST GmbH in Kamp-Lintfort.

The LTCC tape for this study is DuPont 9k7 [196], all tapes are from the same manufacturing lot. Table6.3 lists the 36 combinations that are fabricated, measured and evaluated. There are two samples of eachcombination in the experiment.

The evaluation of the resistor samples after sintering encompasses measurement of the thickness of the toplayer resistors with a laser profilometer, shrinkage measurement, measurement of the actual length and width,and the four-wire measurement of the electrical resistance. For the resistor dimension and the followingdetermination of the sheet resistance R the resistor is approximated to be rectangular. A semi-automatedmeasurement set-up with a four-wire Ohmmeter and a switch matrix connected to a prober with a dedicatedprobe card as depicted in Fig. 6.22 is used to perform the multitude of measurements eciently. A selectionof the measurement results is presented and discussed. In Fig. 6.23 to 6.25 the height of the column marks


Table 6.3: Design of experiment for the iKERSATEC resistor X-check.

sampledesignation

stackposition

resistorpaste

printed &laminated @

fired@

RIL: inner layerRTL: top layer

HFB 12HFB 22

IMST, TUI, MSE A: IMSTB: TUIC: MSE

A_IMST_RIL_12 inner layer HFB 12 IMST IMSTB_IMST_RIL_12 inner layer HFB 12 IMST TUIC_IMST_RIL_12 inner layer HFB 12 IMST MSEA_TUI_RIL_12 inner layer HFB 12 TUI IMSTB_TUI_RIL_12 inner layer HFB 12 TUI TUIC_TUI_RIL_12 inner layer HFB 12 TUI MSEA_MSE_RIL_12 inner layer HFB 12 MSE IMSTB_MSE_RIL_12 inner layer HFB 12 MSE TUIC_MSE_RIL_12 inner layer HFB 12 MSE MSEA_IMST_RIL_22 inner layer HFB 22 IMST IMSTB_IMST_RIL_22 inner layer HFB 22 IMST TUIC_IMST_RIL_22 inner layer HFB 22 IMST MSEA_TUI_RIL_22 inner layer HFB 22 TUI IMSTB_TUI_RIL_22 inner layer HFB 22 TUI TUIC_TUI_RIL_22 inner layer HFB 22 TUI MSEA_MSE_RIL_22 inner layer HFB 22 MSE IMSTB_MSE_RIL_22 inner layer HFB 22 MSE TUIC_MSE_RIL_22 inner layer HFB 22 MSE MSEA_IMST_RTL_12 top layer HFB 12 IMST IMSTB_IMST_RTL_12 top layer HFB 12 IMST TUIC_IMST_RTL_12 top layer HFB 12 IMST MSEA_TUI_RTL_12 top layer HFB 12 TUI IMSTB_TUI_RTL_12 top layer HFB 12 TUI TUIC_TUI_RTL_12 top layer HFB 12 TUI MSEA_MSE_RTL_12 top layer HFB 12 MSE IMSTB_MSE_RTL_12 top layer HFB 12 MSE TUIC_MSE_RTL_12 top layer HFB 12 MSE MSEA_IMST_RTL_22 top layer HFB 22 IMST IMSTB_IMST_RTL_22 top layer HFB 22 IMST TUIC_IMST_RTL_22 top layer HFB 22 IMST MSEA_TUI_RTL_22 top layer HFB 22 TUI IMSTB_TUI_RTL_22 top layer HFB 22 TUI TUIC_TUI_RTL_22 top layer HFB 22 TUI MSEA_MSE_RTL_22 top layer HFB 22 MSE IMSTB_MSE_RTL_22 top layer HFB 22 MSE TUIC_MSE_RTL_22 top layer HFB 22 MSE MSE


the mean value of the resistance, the box shows the standard deviation and the vertical line connects theminimum and the maximum value of the respective measurement.

The bar graph in Fig. 6.23 compares the sheet resistance of all resistors of the test layout in two tiles.Both were printed, laminated and sintered at IMST. Since R2 is the resistor with the smallest dimensions,it was expected that printing tolerances would lead to the largest standard deviation for R2 in this com-parison. However, there is no significant result pointing at screen printing issues with one of the resistorshapes. Whereas the concept of a sheet resistance is indispensable in the design of film resistors, it has itsshortcomings when applied to thickfilm resistors as this comparison of components with di↵erent shapeand otherwise identical parameters demonstrates. Fine tuning of the layout is required in the design flowafter initial tests with the intended geometry. The following graphs compare results for resistor R2 with thesmallest dimensions (L = W = 250 µm) which is most likely to appear in a practical microwave design.

The trusting cooperation within the consortium of the iKERSATEC project enabled the exchange of verydetailed process descriptions. But even with identical screens, similar printing equipment, and parametersthat were adjusted to conform as good as possible, there are big variations in the resistance. The resultsfor the other resistors and material combinations not shown here exhibit similar patterns. For sintering theresearch laboratories of Technical University of Ilmenau (TUI) and IMST use the same model of a box-typefurnace with a chamber that is small compared to the bottom loader furnace used for production at MicroSystems Engineering (MSE). Both diagrams, Fig. 6.24 and Fig. 6.25 show a systematic di↵erence betweenthe samples sintered in the smaller furnaces of the research labs (TUI and IMST) and those sintered in theproduction furnace. The di↵erence in resistance is most noticeable, but more importantly, there is also amuch smaller standard deviation for the samples sintered at MSE. There are two conclusions from thiscomparison: firstly, sintering resistors is the acid test for sintering furnaces and secondly, a bigger furnacehas a larger "golden zone", i.e. the volume inside the furnace chamber with very uniform temperature andgas distribution. A closer look at the standard deviation of the resistance values in Fig. 6.24 reveals anothersystematic correlation between the sample values: Samples printed at MSE have a much lower standarddeviation than those of the research labs at TUI and IMST. This can also be observed in Fig. 6.25 but thedi↵erence is not as pronounced.

Fig. 6.22: Probe card and needles with the resistor test substrate.


Fig. 6.23: Resistor Test Comparison of sheet resistance for R1 - R6, top layer, resistor paste: DuPont HFB12with a nominal sheet resistance R! = 25 Ω/!.

Fig. 6.24: Resistor Test Comparison of R2, outer layer, resistor paste: DuPont HFB12 with a nominal sheetresistance R! = 25 Ω/!, the left designation of the column is the facility to sinter the tile and the right designa-tion is where it was printed and laminated.


Fig. 6.25: Resistor Test Comparison of R2, outer layer, resistor paste: DuPont HFB22 with a nominal sheetresistance R! = 200 Ω/!, the left designation of the column is the facility to sinter the tile and the right desig-nation is where it was printed and laminated.

The following control elements are available for adjusting the resistance value to compensate for thevariations which are to be expected when a design is transferred:

• Screen printing parameters: printing speed, squeegee type and force, to control the print thickness ofthe resistor.

• Screen parameters: emulsion thickness, and theoretical ink volume of the mesh. These parametersalso control the print thickness of the resistor.

• Blending of the resistor paste to achieve the desired sheet resistance.

• Adjustment of the resistor geometry.

The transfer process must take into account the time and effort required to make adjustments, and in thelatter case also the flexibility in the layout to change dimensions.

This extensive study clearly highlights the adjustments necessary for the transition of an LTCC moduledesign from prototype to production. Within the scope of this study, the process parameters were transferredin the best possible way. However, with a regular transition, slightly larger deviations are to be expecteddue to different material lots of pastes and tapes. In spite of the observed variations between differentmanufacturing sites, the reproducibility within resistors from the same facility is very good.

7 Conclusion and OutlookThe aim of this work was to provide the ceramic multilayer technology that facilitates the construction of avery complex DBF antenna for the uplink of the satellite Ka-band. Digital Beam-Forming implies a full RFfront-end for each of the 64 antenna elements. These antenna elements are arranged in a 8 8 grid of /2(i.e. 5 mm @ 30 GHz) distance. The RF front-end of the TX antenna encompasses a branch-line coupler,calibration probes, a power amplifier, an image rejection filter and a mixer. All these components fit in thesquare of 5 5 mm2 that is defined by the /2 distance of the antenna elements. For the 8 8 sub-arrayin one antenna module the calibration network, the LO network, liquid cooling of the power amplifiers,the distribution of the DC supply, and the IF connection are also integrated into the multilayer substrate.LTCC multilayer technology was applied successfully to accomplish the implementation of the requiredfunctionality in vertical integration. To fit into the underlying modular concept, the 8 8 sub-array has aform factor that allows to integrate it side by side into a much larger array, maintaining the grid of /2 acrossthe whole antenna aperture.

7.1 Achievements and Contributions of this Work

Within the scope of the SANTANA project, the work of this thesis provides solutions for the electrical andstructural requirements of the Ka-band SatCom antenna module. These concepts in have been implementedin the functional LTCC antenna module as part of a SatCom terminal demonstrator [1].

The manufacturing technology for the cavities of the antenna patches has been developed and processparameters have been adjusted to fabricate these features with the required precision. The cavities withconductive walls reduce coupling between adjacent elements and improves thus the scanning performanceof the complete antenna array. Moreover the same cavities facilitate a novel and patented array calibrationtechnique [84].

Since both sides of the antenna module are densely populated, on one side with antenna elements (seeFig. 1.8) and on the other side with the active components of the front-end (see Fig. 1.9), liquid coolingwas chosen to dissipate the heat from the power amplifiers. For this purpose, micro-channels have beenintegrated into the inner layers of the multilayer by means of sacrificial material. This technique maintainsthe shape of the micro-channels during lamination and sintering of the LTCC substrate. Among the possiblesolutions for channels in the LTCC multilayer, this approach was chosen because of its special suitability fordimension and channel routing in the module. The manufacturing process was developed and the optimumprocess parameters, in particular the sintering conditions, were determined. The routing of cooling channelsin a crowded multilayer in the space between microwave-, LO-, IF- and DC-supply-connections of 64 activefront-ends has been optimized for manufacturability and e↵ectiveness. In the concluding thermal tests theperformance of this cooling concept proved to exceed the system requirements by a decent margin (i.e. morethat a factor of two in the dissipated power).

The SANTANA antenna module consists of 17 layers, that have to be aligned and stacked accurately ontop of each other. In course of this project, the suitable process has been installed and refined to ensureoptimum precision. The e↵ect of layer-to-layer registration and the inherent tolerances on the reflectioncoecient of the antenna and its radiation pattern has been analysed by FDTD EM-simulation and verifiedby microwave measurements. Both, simulations and measurements show that the manufacturing processmeets the requirements of the DBF antenna module.

One of the main challenges in the production of LTCC modules with a very high number of layers isde-bindering and sintering of the green body. Exact analysis of the process (e.g. with TGA and opticaldilatometry) as well as repeated test cycles were applied to define the process window with optimum

89

90 7 Conclusion and Outlook

parameters (e.g. oven schedule and gas flow) to get reproducibly good results. The method presented in thiswork is also suitable as a general approach to optimize de-bindering and sintering in the LTCC process.

Embedded passive components like resistors are an important part of beam forming networks. Thereproducibility of embedded resistors has been investigated thoroughly with focus on resistor pastes anddimensions that are suitable for millimeter-wave LTCC modules. Concepts for power dividers and matchedloads have been realized and characterized that achieve both, good microwave performance and good yieldwith the standard production technique (i.e. screen printing).

7.2 Conclusions

The SANTANA antenna module is the basic building block for the integration of a planar DBF antennafor high data rate satellite communication in the Ka-band. This planar AESA (active electronically steeredarray) antenna is easier to integrate into the body of a vehicle or into the fuselage of an aircraft than anMSA (mechanically steered array) or a reflector antenna [202]. Thanks to its electronic steerability, it is notexposed to the wear and tear of its mechanically positioned counterparts. The functionality of the technologydemonstrators was confirmed in land-mobile and airborne field tests [89].

The SANTANA project provides the cornerstone for an active antenna terminal for satellite communicationin the Ka-band. With an increasing number of Ka-band communication satellites like KA-sat [203, 204]in orbit, the chicken-and-egg question of ground segment and space segment for a new frequency band isresolved. The modular concept of the SANTANA module with 64 elements and their front-ends enablesflexible combination of larger antenna arrays. This modularity is possible through vertical integrationof microwave components and active cooling in an LTCC module with unprecedented complexity anddensity. Design for manufacturability (DFM) was the guideline through the development of components andprocesses. The result is a highly integrated complex antenna module in LTCC multilayer technology that canbe manufactured reproducibly.

7.3 Outlook

Apart from airborne and maritime applications where the high data rate is attractive, compact Ka-bandsatellite ground terminals for land-mobile applications provide a reliable solution for search and rescue, anddisaster management particularly when the telecommunication infrastructure is disrupted [205, 206, 207].DBF for millimeter-waves also plays an important role in the development of the emerging next generationof mobile communication 5G [208, 209].

Even with today’s powerful computers and distributed computing in clusters of powerful processors,the tolerance analysis of multilayer circuits is still a time consuming task. It would be desirable to have asoftware environment that facilitates the input of manufacturing tolerances and helps to centre the design inthe tolerance range. Experienced engineers achieve good results following their "gut feeling" through thedesign process [210]. Space mapping [211] is a technique which tries to emulate this intuitive process bylinking a coarse model (e.g. a circuit simulation in "SPICE" or "ADS") to a fine model (e.g. a 3D full waveanalysis with the EMPIRE® FDTD solver [75]). This bidirectional link allows to refine the coarse modelwith the results from the fine model and to optimize the design with less cycles of the fine model. Toleranceanalysis, yield prediction, and design centering are tasks which could be accelerated by this approach in asimilar matter as filter design and optimization [212].

LTCC has the potential to produce three dimensional circuits [213, 108]. Whereas traditional concepts arebuilding the module in layers, the combination with injection moulding [214, 215, 216] opens up completelynew degrees of freedom for the construction of LTCC modules. Package walls or dielectric lenses canbecome integral parts of the ceramic substrate.

3D printing or additive manufacturing (AM) is an umbrella term for manufacturing techniques thatcreate three-dimensional parts by depositing structured layers of material with various methods. Additive

7.3 Outlook 91

manufacturing means that material is added to a part rather than removed from from a stock of raw material.This is particularly advantageous when precious metals or other expensive primary materials are involved.Almost arbitrary shapes or geometries can be generated following a three-dimensional CAD model. 3Dprinting of ceramic material, particularly LTCC [217, 218, 219] is a technique that will further enhance thedesign options and enable rapid prototyping.

In a wider sense, direct writing technologies for conductor and resistor inks also belong to additivemanufacturing. Aerosol and ink jet printing [220, 36, 221] o↵er additional flexibility for design changes (i.e.no screens or stencils required), finer lines, and printing on tilted or curved surfaces.

LTCC multilayer technology provides a range of unique solutions in many fields where high density ofintegration and robustness against environmental and mechanical stress is required. In combination withgood microwave and millimeter-wave properties there is a potential for module and antenna integrationthat goes even beyond the examples presented in this work. There are already quite a few universitiesthat teach this interesting technique in theory and practise, but the majority of the graduates of electricalengineering does not know the enormous possibilities and sticks to the more common PCB technology. Thisis also influencing strategic decisions when the same people get into management positions. Education,dissemination of this technology and further development of software tools are the fields to be worked on.

Bibliography[1] S. Holzwarth, A. F. Jacob, A. Dreher, C. Hunscher, H. Fischer, A. Stark, B. Rohrdantz, A. Geise,

K. Kuhlmann, R. Gieron, O. Litschke, D. Lohmann, W. Simon, P. Buchner, M. V. T. Heckler, andL. A. Greda, “Active antenna arrays at Ka-band: Status and outlook of the SANTANA project,” inProc. Eur. Conf. Antennas Propag. 4th, April 2010.

Cited on pages iii, 1, 10, 12, 89, and 124.

[2] S. J. Lehtonen, C. R. Moore, A. S. Francomacaro, R. L. Edwards, and G. V. Clatterbaugh, “Microwavemultichip modules,” Johns Hopkins APL Tech. Dig., vol. 15, no. 1, pp. 48 – 56, 1994.

Cited on page 3.

[3] H. Delfs, Hybridschaltungen. Kohlhammer Stuttgart, 1973.Cited on pages 3 and 126.

[4] T. Lane, F. Belcourt, and R. Jensen, “Electrical characteristics of copper/polyimide thin-film multilayerinterconnects,” IEEE Trans. Compon., Hybrids, Manuf. Technol., vol. 10, no. 4, pp. 577–585, Dec1987.

Cited on page 3.

[5] G. Carchon, K. Vaesen, S. Brebels, W. D. Raedt, E. Beyne, and B. Nauwelaers, “Multilayer thin-film MCM-D for the integration of high-performance RF and microwave circuits,” IEEE Trans.Compon. Packag. Technol., vol. 24, no. 3, pp. 510–519, Sep 2001.

Cited on page 3.

[6] A. H. Feingold, M. Heinz, and R. L.Wahlers, “Compliant dielectric and magnetic materials for buriedcomponents,” in Proc. Intl. Symp. Microel., vol. 35. IMAPS, September 4-6 2002.

Cited on page 3.

[7] W. Rath and K. Schricker, “Verfahren zum Herstellen dünner Plättchen aus keramischer Masse,”German Patent 7201, October, 1942.

Cited on page 3.

[8] R. E. Mistler, “Tape casting: Past, present, potential,” Amer. Ceram. Soc. Bull., vol. 77, no. 10, pp.82–86, 1998.

Cited on page 3.

[9] T. Allen, Particle Size Measurement. Chapman and Hall/CRC, 1996.Cited on page 4.

[10] M. Eberstein, R. Müller, S. Reinsch, T. Rabe, W. A. Schiller, A. Thiel, and J. Deubener, “Kineticmodeling of LTCC shrinkage: E↵ect of alumina content,” J. Microelectron. Electron. Packag.,vol. 4, no. 4, pp. 173–180, 2007.

Cited on page 4.

[11] J. Müller, J. Klein, and J. Rayho, “In-process verification of MLC substrates,” Proc. Eur. Microel.Packag. Symp., 16-18 June 2002.

Cited on page 4.

[12] P. Uhlig, S. Holzwarth, and M. Geissler, “Antenna concepts for ceramic multilayer modules,” in Proc.Int. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol. 5th, 4 2009.

Cited on page 4.

[13] I. Wol↵, P. Uhlig, R. Kulke, C. Günner, and J. Kassner, “LTCC modules for future communicationsatellites in Ka-band,” IEEE MTT-S Int. Microw. Symp. Dig., 5 2016.

Cited on page 4.

93

94 Bibliography

[14] J. Priestley, The history and present state of electricity: with original experiments. London: Printedfor J. Dodsley, J. Johnson and T. Cadell, 1767.

Cited on page 6.

[15] M. Faraday, “Experimental researches in electricity,” Philos. Trans. Roy. Soc. London, vol. 128, no. 1,pp. 1–40, jan 1838.

Cited on page 6.

[16] D. G. Fitzgerald, “Improvements in electrical condensers or accumulators,” UK Patent 3466/1876,September 2, 1876.

Cited on page 6.

[17] W. Dubilier, “Development, design and construction of electrical condensers,” J. Franklin Inst., vol.248, no. 3, September 1949.

Cited on page 6.

[18] R. von Lieben, “Kathodenstrahlenrelais,” German Patent DRP179 807, 19. Nov., 1906.Cited on page 6.

[19] L. D. Forest, “The audion; a new receiver for wireless telegraphy,” Trans. Am. Inst. Electr. Eng., vol.XXV, pp. 735 – 763, 1906.

Cited on page 6.

[20] J. Ho, T. R. Jow, and S. Boggs, “Historical introduction to capacitor technology,” IEEE Elect. Insul.Mag., vol. 26, no. 1, pp. 20–25, January 2010.

Cited on page 6.

[21] E. Schrödinger, “Studien über Kinetik der Dielektrika, den Schmelzpunkt, Pyro- und Piezoelektrizität,”Sitzungsber. der Kaiserl. Akad. der Wissensch. in Wien. Mathem.-Naturw. Klasse, vol. CXXI, Abt.IIa, pp. 1937 – 1972, 1912.

Cited on page 6.

[22] A. R. von Hippel, “Ferroelectricity, domain structure, and phase transitions of barium titanate,” Rev.Mod. Phys., vol. 22, pp. 221 – 237, Jul 1950.

Cited on page 6.

[23] W. J. Gyurk, “Methods for manufacturing multilayered monolithic ceramic bodies,” US PatentUS3 192 086 A, 1965.

Cited on page 6.

[24] E. M. Davis, W. E. Harding, R. S. Schwartz, and J. J. Corning, “Solid logic technology: Versatile,high-performance microelectronics,” IBM J. Res. Dev., vol. 8, no. 2, pp. 102 – 114, April 1964.

Cited on page 6.

[25] H. W. Stetson, “Method of making multilayer circuits,” US Patent US3 189 978 (A), 1965-06-22,1962.

Cited on page 7.

[26] A. J. Blodgett, “A multilayer ceramic multi-chip module,” IEEE Trans. Compon., Hybrids, Manuf.Technol., vol. 3, no. 4, pp. 634 – 637, 1980.

Cited on page 7.

[27] Y. Imanaka, Multilayered Low Temperature Cofired Ceramics (LTCC) Technology. Springer, 102010.

Cited on pages 7 and 69.

[28] P. J. Holmes and R. G. Loasby, Handbook of Thick Film Technology. Ayr: ElectrochemicalPublications, 1976.

Cited on page 7.

Bibliography 95

[29] R. R. Tummala, E. J. Rymaszewski, and A. G. Klopfenstein, Eds., Microelectronic PackagingHandbook. Van Nostrand Reinhold Co., 1989.

Cited on page 7.

[30] T. C. Edwards, Foundations for Microstrip Circuit Design, 1st ed. John Wiley & Sons, 10 1981.Cited on page 7.

[31] R. K. Ho↵mann, Handbook of Microwave Integrated Circuits, ser. Artech House Microwave Library.Artech House, 9 1987.

Cited on page 7.

[32] T. Laverghetta, Microwave Materials and Fabrication Techniques, 3rd ed. Artech House, 9 2000.Cited on page 7.

[33] R. L. Brown, P. W. Polinski, and A. S. Shaikh, “Manufacturing of microwave modules using low-temperature cofired ceramics,” in IEEE MTT-S Int. Microw. Symp. Dig., 1994, pp. 1727 – 1730.

Cited on page 7.

[34] J. Müller, “Entwurf, Herstellung und HF-Charakterisierung gedruckter passiver Bauelemente undStrukturen in LTCC,” Ph.D. dissertation, Technische Universität Ilmenau, 1996.

Cited on page 7.

[35] D. I. Amey, M. Y. Keating, M. A. Smith, S. J. Horowitz, P. C. Donahue, and C. R. Needes, “Low losstape materials system for 10 to 40 GHz application,” in Proc. Intl. Symp. Microel., 2000, pp. 654 –658.

Cited on page 7.

[36] P. Uhlig, D. Stöpel, S. Mosch, and J. Müller, “Fine line conductors in LTCC,” in Proc. IMAPS NordicAnnu. Conf., 2012, pp. 124 –132.

Cited on pages 7, 69, and 91.

[37] E. K. Polzer, “Feinststrukturierung von Dickschichtleitern mittels Ätztechnik,” Fortschrittsber. derDKG, vol. 81, pp. 119 – 123, 2004.

Cited on page 7.

[38] K. Wörho↵, A. Prak, F. Postma, A. Leinse, K. Wu, T. J. Peters, M. Tichem, B. Amaning-Appiah,V. Renukappa, G. Vollrath, J. Balcells-Ventura, P. Uhlig, M. Seyfried, D. Rose, R. Santos, X. J. M.Leijtens, B. Flintham, M. Wale, and D. Robbins, “Photonic hybrid assembly through flexiblewaveguides,” in Proc. SPIE, Silicon Photonics and Photonic Integrated Circuits V, L. Vivien,L. Pavesi, and S. Pelli, Eds., vol. 9891. SPIE, May 13 2016.

Cited on page 7.

[39] J. Burdon and J. Yamamoto, “Miniaturized co-fired electrical interconnects for implantable medicaldevices,” US Patent US20 070 060 970 A1, March, 2007.

Cited on page 7.

[40] E. Johannessen, O. Krushinitskaya, A. Sokolov, and B. Akselsen, “Toward an injectable continuousosmotic glucose sensor,” J. Diabetes Sci. Technol., vol. 4, no. 4, pp. 882 – 892, 2010.

Cited on page 7.

[41] M. Eberstein, “Gläser als Schlüsselkomponenten für hochperformante Steuerelektronik im Automobil,”presented at 6. CTI Konferenz Glas im Automobil,, 23. / 24. April 2012.

Cited on page 8.

[42] W. Röthlingshöfer, S. Albers, and J. Schaich, “Low Temperature Cofire Ceramic (LTCC), eine Tech-nologie für elektronische Steuergeräte unter extremen Belastungen,” Symp. ’Verfahrenstechnik’der DKG, April 2004.

Cited on page 8.

96 Bibliography

[43] P. Uhlig, C. Günner, S. Holzwarth, J. Kassner, R. Kulke, A. Lauer, and M. Rittweger, “LTCC shortrange radar sensor for automotive applications at 24 GHz,” Advancing Microelectronics, vol. 32,no. 2, pp. 10 – 12, 2005.

Cited on page 8.

[44] D. Schwanke, T. Haas, M. Harnack, A. Bittner, U. Schmidt, E. Feurer, S. Hoyler, and B. Moll, “LTCCDünnfilm-Polymer Kombination für 77-81 GHz Radarsysteme,” VDE-Kongress Aachen, 2006.

Cited on page 8.

[45] Y. C. Venot, “Entwicklung und Integration eines Nahbereichsradarsensorsystems bei 76,5 GHz,” Ph.D.dissertation, Universität Karlsruhe, 2003.

Cited on page 8.

[46] T. Klein, M. Faassen, R. Kulke, and C. Rusch, “A 77 GHz radar frontend in LTCC for small range,high precision industrial applications,” in Proc. Eur. Microw. Integr. Circ. Conf. 7th, Oct 2012, pp.905–908.

Cited on page 8.

[47] H.-J. Song and T. Nagatsuma, “Present and future of terahertz communications,” IEEE Trans. THzSci. Technol., vol. 1, no. 1, pp. 256–263, Sept 2011.

Cited on page 8.

[48] M. Pauli, B. Goettel, S. Scherr, A. Bhutani, S. Ayhan, W. Winkler, and T. Zwick, “Miniaturizedmillimeter-wave radar sensor for high-accuracy applications,” IEEE Trans. Microw. Theory Techn.,vol. 65, no. 5, pp. 1707–1715, May 2017.

Cited on page 8.

[49] T. Tajima, T. Kosugi, H.-J. Song, H. Hamada, A. E. Moutaouakil, H. Sugiyama, H. Matsuzaki,M. Yaita, and O. Kagami, “Terahertz mmics and antenna-in-package technology at 300 ghz forkiosk download system,” J. Infrared, Millimeter, Terahertz Waves, vol. 37, pp. 1213 – 1224, 2016.

Cited on page 8.

[50] M. Ihle, S. Ziesche, C. Zech, and B. Baumann, “Compact ltcc packaging and printing technologiesfor sub-thz modules,” Proceedings of the Electronics System-Integration Technology ConferenceESTC, 2018.

Cited on page 8.

[51] P. Schuh, R. Leberer, H. Sledzik, M. Oppermann, B. Adelseck, H. Brugger, R. Behtash, H. Leier,R. Quay, and R. Kiefer, “20W GaN HPAs for next generation X-band T/R-modules,” in IEEEMTT-S Int. Microw. Symp. Dig., June 2006, pp. 726–729.

Cited on page 8.

[52] R. Rieger, A. Klaassen, P. Schuh, and M. Oppermann, “GaN based wideband T/R module formulti-function applications,” in Proc. Eur. Microw. Conf., Sept 2015, pp. 514–517.

Cited on page 8.

[53] M. Oppermann, J. Schroth, and F. Thurow, “Transmit/receive (T/R) modules - key elements for phasedarray antennas,” in Proc. Intl. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol. 9th, 42013, p. TP24.

Cited on page 8.

[54] P. Uhlig, R. Kulke, D. Köther, J. Kassner, A. Lauer, and M. Rittweger, “Power distribution networksin space applications,” in Proc. Intl. Hi-Tech 3D Packag. Workshop, 6 2006.


[55] W. Pitz and D. Miller, “The TerraSAR-X satellite,” IEEE Trans. Geosci. Remote Sens., vol. 48, no. 2,pp. 615–622, Feb 2010.

Cited on page 8.

Bibliography 97

[56] S. Buckreuss and M. Zink, “The missions TerraSAR-X and TanDEM-X: Status, challenges, futureperspectives,” in Proc. URSI Gen. Assy. Sci. Symp. 30th, Aug 2011, p. 1.

Cited on page 8.

[57] S. Suchandt and H. Runge, “High-resolution surface current mapping using TanDEM-X ATI,” in Proc.Eur. Conf. SAR 10th, June 2014, pp. 1–4.

Cited on page 8.

[58] P. Uhlig, C. Günner, J. Kassner, and R. Kulke, “Keramische Mehrlagenschaltungen für die neueGeneration von Kommunikationssatelliten,” in Proc. Symp. Keram. Mehrlagentech. der DKG,December 2016.

Cited on page 8.

[59] R. Kulke, G. Möllenbeck, C. Günner, P. Uhlig, K. H. Drüe, S. Humbla, J. Müller, R. Stephan,D. Stöpel, J. F. Trabert, G. Vogt, M. A. Hein, A. Molke, T. Baras, A. F. Jacob, D. Schwanke,J. Pohlner, A. Schwarz, and G. Reppe, “Ceramic microwave circuits for satellite communication,”J. Microelectron. Electron. Packag., vol. 6, pp. 27 – 31, 2009.

Cited on page 8.

[60] A. Nicolai, S. Roemer, and S. Eckert, “The TET satellite bus - future mission capabilities,” in Proc.IEEE Aerosp. Conf., March 2014, pp. 1–6.

Cited on page 8.

[61] N. Ibanez-Garcia, C. S. Martinez-Cisneros, F. Valdes, and J. Alonso, “Green-tape ceramics. newtechnological approach for integrating electronics and fluidics in microsystems,” Trends Anal.Chem., vol. 27, no. 1, pp. 24 – 33, 2008.

Cited on page 8.

[62] D. Jurkow, Technology and properties of integrated LTCC sensors. PIAP, 2013, pp. 9 – 18.Cited on page 8.

[63] E. S. O. Fakunle and I. Fritsch, “Microchannel immunoassays with electrochemical detection indevices constructed from low temperature co-fired ceramic and screen printed gold,” NSTI-Nanotech, vol. 1, pp. 466 – 469, 2005.

Cited on page 8.

[64] L. J. Golonka, “Technology and applications of low temperature cofired ceramic (LTCC) basedsensors and microsystems,” Bull. Polish Acad. Sci., vol. 54, no. 2, pp. 221 – 231, 2006.

Cited on page 8.

[65] H. Bartsch, C. Rensch, T. Thelemann, J. Müller, and M. Ho↵mann, “Fully integrated bridge-typeanemometer in LTCC-based microfluidic systems,” Adv. Sci. Technol., vol. 54, pp. 401–404, 92008.

Cited on page 8.

[66] M. A. Fonseca, J. M. English, M. von Arx, and M. G. Allen, “High temperature characterization ofceramic pressue sensors,” Georgia Institute of Technology, School of Electrical and ComputerEngineering, Atlanta, GA, 30332, Tech. Rep., 2001.

Cited on page 8.

[67] T. Maeder, V. Fahrny, S. Stauss, G. Corradini, and P. Ryser, “Design and characterisation of low-costthick-film piezoresistive force sensors for the 100 mN to 100 N range,” in Proc. IMAPS PolandAnnu. Conf. 29th, 9 2005.

Cited on page 8.

[68] S. Ziesche, L. Rebenklau, and U. Partsch, “Sensoren in keramischer Mehrlagentechnik - aktuelleEntwicklungen am Fraunhofer IKTS,” PLUS, vol. 13, no. 5, pp. 1167–1171, 2011.

Cited on page 8.

98 Bibliography

[69] R. Kulke, G. Möllenbeck, P. Uhlig, K. Maulwurf, and M. Rittweger, “Designing SMD lowpass filtersin multilayer LTCC technology,” in Proc. Eur. Microel. Packag. Conf., June 2007, pp. 40–44.

Cited on page 9.

[70] A. Wien, A. Lauer, and I. Wol↵, “Multi-PC FDTD: Solving large scale EM problems,” in IEEE MTT-SInt. Microw. Symp. Dig., May 2010, pp. 972–975.

Cited on page 9.

[71] B. Sanadgol, S. Holzwarth, P. Uhlig, A. Milano, and R. Popovich, “60 GHz SIW steerable antennaarray in LTCC,” ZTE Commun., vol. 10, no. 4, pp. 29 – 60, 2012.

Cited on page 9.

[72] A. Dreher, D. Heberling, S. Holzwarth, C. Hunscher, A. F. Jacob, N. Niklasch, H. Pawlak, L. Richard,A. Schroth, L. C. Stange, and M. Thiel, “Ka-band DBF terminal concepts for broadband satellitecommunications,” in Proc. ESA Workshop Satell. Antenna Technol. 25th, September 2002, pp. 95– 102.


[73] R. Kulke, C. Günner, G. Möllenbeck, P. Uhlig, and M. Rittweger, “Protoflight model developmentfor spaceborne LTCC RF-modules,” in Proc. Intl. Symp. Microel. 41st, November 2008, pp.1001–1006.


[74] T. Klein, C. Günner, J. Kassner, R. Kulke, and H. Wolf, “Multiple feed per beam networks for Ka-bandsatellite communication systems in LTCC technology,” in IEEE MTT-S Int. Microw. Symp. Dig., 62011, pp. 1 –4.

Cited on pages 12, 54, 67, 74, 75, 81, 123, and 128.

[75] User and Reference Manual for the 3D EM Time Domain Simulator EMPIRE XPU, Version 7.51 ed.,IMST GmbH, Kamp-Lintfort, Germany, October 2016. [Online]. Available: http://www.empire.de/media/pdf/Manual/EMPIRE-Manual-750.pdf

Cited on pages 12, 18, 30, 44, 47, 48, 65, and 90.

[76] P. Uhlig, D. Manteu↵el, and S. Malkmus, “High layer count in LTCC dual band antenna for GalileoGNSS,” J. Microelectron. Electron. Packag., vol. 5, pp. 161 – 168, April 2008.


[77] C. Huygens, Traité de la lumière - où sont expliquées les causes de ce qui luy arrive dans la réflexion.Leiden: Pieter van der Aa, 1690.

Cited on page 15.

[78] T. T. Taylor, “Design of line-source antennas for narrow beamwidth and low side lobes,” IRE Trans.Antennas Propag., vol. 3, no. 1, pp. 16–28, Jan 1955.


[79] H. Schippers, J. Verpoorte, P. Jorna, A. Hulzinga, A. Meijerink, C. Roelo↵zen, R. G. Heideman,A. Leinse, and M. Wintels, “Conformal phased array with beam forming for airborne satellitecommunication,” in Proc. Intl. ITG Workshop Smart Antennas, 2 2008, pp. 343–350.

Cited on page 15.

[80] T. Teshirogi, M. Tanaka, and W. Chujo, “Wideband circularly polarised array antenna with sequentialrotations and phase shift of elements,” in Proc. Intl. Symp. Antennas and Propag. ISAP, vol. 4, 81985, pp. 117–120.


[81] O. Litschke, W. Simon, and S. Holzwarth, “A 30 GHz highly integrated LTCC antenna element fordigital beam forming arrays,” in IEEE Intl. Symp. Antennas and Propag. Dig., vol. 3B, 7 2005, pp.297–300.

http://www.empire.de/media/pdf/Manual/EMPIRE-Manual-750.pdf

http://www.empire.de/media/pdf/Manual/EMPIRE-Manual-750.pdf

Bibliography 99


[82] L. Baggen, S. Holzwarth, W. Simon, and O. Litschke, “Phased array using the sequential rotationprinciple: analysis of coupling e↵ects,” in Proc. IEEE Intl. Symp. Phased Array Syst. Technol., 102003, pp. 571–576.

Cited on page 16.

[83] W. Wiesbeck, “Antennen und Antennensysteme,” Skriptum zur Vorlesung, Universität Karlsruhe(TH), Institut für Höchstfrequenztechnik und Elektronik, 2007.


[84] S. Holzwarth and O. Litschke, “Kalibrierung einer elektronischen steuerbaren Planarantenneund elektronisch steuerbare Antenne mit einer Messsonde im reaktiven Nahfeld,” PatentDE102 005 011 128A1, 9, 2006.


[85] LTCC Design Guide Data Sheet, Ferro Electronic Materials, 2002.Cited on page 16.

[86] P. Uhlig, S. Holzwarth, O. Litschke, W. Simon, and R. Baggen, “A digital beam-forming antennamodule for a mobile multimedia terminal in LTCC-multilayer technique,” in Proc. Eur. Microel.Packag. Conf. 15th, 6 2005.

Cited on page 17.

[87] H. Pawlak, D. Notel, and A. F. Jacob, “Compact calibration scheme for planar DBF transmit antennaarrays using circular polarization at Ka-band,” in IEEE Intl. Symp. Antennas and Propag. Dig.,vol. 2B, 7 2005, pp. 752–755.

Cited on page 18.

[88] H. Pawlak and A. F. Jacob, “An external calibration scheme for DBF antenna arrays,” IEEE Trans.Antennas Propag., vol. 58, no. 1, pp. 59–67, 1 2010.

Cited on page 18.

[89] A. Geise, A. F. Jacob, K. Kuhlmann, H. Pawlak, R. Gieron, P. Siatchoua, D. Lohmann, S. Holzwarth,O. Litschke, M. V. T. Heckler, L. Greda, A. Dreher, and C. Hunscher, “The SANTANA project,”in Proc. Eur. Air Space Conf. 1st, 9 2007, pp. 481 – 487.


[90] (2017) cavity. [Online]. Available: https://www.merriam-webster.com/dictionary/cavityCited on page 25.

[91] R. Kulke, G. Möllenbeck, W. Simon, A. Lauer, and M. Rittweger, “Point-to-multipoint transceiver inLTCC for 26 GHz,” in Proc. IMAPS Nordic Annu. Conf., 10 2002.

Cited on page 25.

[92] A. Schulz, T. Welker, N. Gutzeit, D. Stöpel, F. Wollenschläger, and J. Müller, “Optimized cavities formicrowave applications using the new low loss LTCC material Du Pont 9k7,” in Proc. Intl. Conf.Exhib. Ceram. Interconnect Ceram. Microsyst. Technol. 8th, 4 2012.

Cited on page 25.

[93] S. Holzwarth, O. Litschke, W. Simon, H. Fischer, J. Kassner, and A. Serwa, “Highly integrated 8 x 8antenna array demonstrator on LTCC with integrated RF circuitry and liquid cooling,” in Proc.Eur. Conf. Antennas Propag. 4th, April 2010.

Cited on page 25.

[94] F. Barlow, J. Wood, A. Elshabini, E. Stephens, R. Feeler, G. Kemner, and J. Junghans, “Fabricationof precise fluidic structures in LTCC,” in Proc. Intl. Conf. Exhib. Ceram. Interconnect Ceram.Microsyst. Technol., 4 2008, pp. 183 – 188.

Cited on page 25.

https://www.merriam-webster.com/dictionary/cavity

100 Bibliography

[95] W. K. Jones, Y. Liu, and M. Gao, “Micro heat pipes in low temperature cofire ceramic (LTCC)substrates,” IEEE Trans. Compon. Packag. Technol., vol. 26, no. 1, pp. 110 – 115, march 2003.


[96] M. Mach and J. Müller, “3d-fluidic cooling structures in LTCC,” in Proc. Eur. Microel. Packag. Conf.,6 2007.

Cited on page 25.

[97] A. Albrecht, J. Botiov, M. Fischer, K.-H. Drüe, M. Hintz, and H. Wurmus, “Alternative way to highcurrent structures in LTCC,” in Proc. Eur. Microel. Packag. Conf., June 2003.


[98] T. Rabe, P. Kuchenbecker, and B. Schulz, “Hot embossing: An alternative method to produce cavitiesin ceramic multilayer,” Int. J. Appl. Ceram. Technol., vol. 4, no. 1, pp. 38–46, 2007.


[99] H. Bartsch, “Mikrostrukturierung von LTCC-Grünfolien durch Prägen,” Ph.D. dissertation, Fakultätfür Maschinenbau der Technischen Universität Ilmenau, 2009.


[100] G. Hagen and C. Kohlberger, “Quality aspects in ceramic green tape micro structuring,” in Proc. Intl.Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol. 11th, 4 2015.


[101] K.-J. Wolter, L. Rebenklau, and G. Hagen, “Via formation in LTCC tape: A comparison of technolo-gies,” in Proc. Intl. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol., 4 2005, pp.369–376.


[102] R. Follmann, R. Kulke, S. Holzwarth, P. Uhlig, and M. Rittweger, “LTCC packaging and integrationtechnologies,” in Proc. Eur. Microw. Conf., 10 2009.

Cited on page 27.

[103] High Purity Carbon Tape Data Sheet TCS-CARB-1, Harmonics, Inc., 9524 Roosevelt Way NE, SeattleWA 98115, 2000.

Cited on page 27.

[104] T. Maeder, Y. Fournier, S. Wiedmer, H. Birol, C. Jacq, and P. Ryser, “3D structuration of LTCC /thick-film sensors and fluidic devices,” in Proc. Intl. Symp. Microel. 40th, 1 2007.

Cited on page 28.

[105] H. Birol, T. Maeder, and P. Ryser, “Processing of graphite-based sacrificial layer for microfabricationof low temperature co-fired ceramics (LTCC),” Sens. Actuators, A, vol. 130-131, pp. 560–567,2006.


[106] K. A. Peterson, K. D. Patel, C. K. Ho, S. B. Rohde, C. D. Nordquist, C. A. Walker, B. D. Wroblewski,and M. Okandan, “Novel microsystem applications with new techniques in low-temperatureco-fired ceramics,” Int. J. Appl. Ceram. Technol., vol. 2, no. 5, pp. 345 – 363, 2005.

Cited on page 28.

[107] N. E. McNeal and R. A. Nagy, “Thermoplastic plug method of fabricating an integrated circuitpackage having bonding pads in a stepped cavity,” U.S. Patent 4,680,075, July 14, 1987, 1987.

Cited on page 28.

[108] P. Uhlig, S. Holzwarth, B. Sanadgol, and A. Serwa, “Three-dimensional surface in LTCC for aMM-wave antenna,” in Proc. Intl. Symp. Microel. 44th, 10 2011.


Bibliography 101

[109] E. J. Wilkinson, “An n-way hybrid power divider,” IRE Trans. Microwave Theory Tech., vol. 8, no. 1,pp. 116–118, 1 1960.

Cited on pages 30, 54, 67, 74, 81, and 128.

[110] G. Ponchak, N. Dib, and L. Katehi, “Design and analysis of transitions from rectangular waveguide tolayered ridge dielectric waveguide,” IEEE Trans. Microwave Theory Tech., 7 1996.

Cited on page 30.

[111] R. Kulke, M. Rittweger, P. Uhlig, and C. Günner, “LTCC-Mehrlagenkeramik für Funk- und Sensor-Anwendungen,” PLUS, pp. 2131–2136, 12 2001.

Cited on page 30.

[112] F. Xu and K. Wu, “Guided-wave and leakage characteristics of substrate integrated waveguide,” IEEETrans. Microwave Theory Tech., vol. 53, no. 1, pp. 66 – 73, 1 2005.

Cited on page 30.

[113] D. M. Pozar, “A relation between the active input impedance and the active element pattern of aphased array,” IEEE Trans. Antennas Propag., vol. 51, no. 9, pp. 2486 – 2489, 9 2003.

Cited on page 32.

[114] P. Uhlig, A. Serwa, U. Altmann, T. Welker, J. Müller, D. Schwanke, J. Pohlner, and T. Rittweg,“Liquid cooling in an LTCC-module for a switched mode amplifier,” in Proc. IMAPS Nordic Annu.Conf., jun 2017, pp. 90–96.

Cited on page 33.

[115] N. O. Sokal and A. D. Sokal, “Class E – a new class of high-eciency tuned single-ended switchingpower amplifiers,” IEEE J. Solid-State Circuits, vol. 10, no. 3, pp. 168–176, 6 1975.

Cited on page 34.

[116] CGH60015D data sheet, Rev 3.1 ed., Cree, Inc., 4600 Silicon Drive Durham, North Carolina, USA27703, 2012.

Cited on page 35.

[117] T. Welker and J. Müller, “Design and fabrication of integrated fluidic channels for liquid cooling of aLTCC device,” in Proc. Intl. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol. 10th, 42014, pp. 1–5.


[118] Thermal Test Chip Guideline (wire bond type chip), JEDEC Std. JESD51-4, 2 1997.Cited on page 39.

[119] Infineon G 423B Thermal Test Chip, Version 2 ed., Infineon, March 2003.Cited on page 39.

[120] M. Planck, “Zur Theorie des Gesetzes der Energieverteilung im Normalspektrum,” Verh. Dtsch. Phys.Ges., vol. 2, no. 17, pp. 237 – 245, 1900.

Cited on page 41.

[121] Manual: ThermoVision A40M, a72 ed., FLIR Systems, Inc., North Billerica, MA. 01862, USA, 102004.

Cited on page 41.

[122] G. H. L. Hagen, “Über die Bewegung des Wassers in engen cylindrischen Röhren,” Ann. Phys. Chem.,vol. 122, no. 3, pp. 423 – 442, 1839, iSSN 0003-3804.

Cited on page 42.

[123] J.-L. M. Poiseuille, “Recherches expérimentales sur le mouvement des liquides dans les tubes de trèspetits diamètres,” Comptes Rendus Acad Sci, vol. 11, pp. 1041–1048, 1840.

Cited on page 42.

102 Bibliography

[124] N. Zuckerman and N. Lior, “Jet impingement heat transfer: Physics, correlations, and numericalmodeling,” Adv. Heat Transfer, vol. 39, pp. 565–631, 2006.


[125] J.-P. Sommer, B. Michel, U. Goebel, and J. Jelonnek, “Thermally optimised millimeter wave packageon an LTCC ceramic board,” in Proc. Intersoc. Conf. Therm. Thermomech. Phenom. Electron.Syst. 9th, 6 2004.

Cited on page 44.

[126] T. Welker and J. Müller, “Design, simulation and fabrication of liquid cooled LTCC devices utilizingintegrated channels,” in Proc. Intersoc. Conf. Therm. Thermomech. Phenom. Electron. Syst. 15th,6 2016, pp. 830 – 835.

Cited on page 44.

[127] A. Pietrikova, T. Girasek, P. Lukacs, T. Welker, and J. Müller, “Simulation of cooling eciency viaminiaturised channels in multilayer LTCC for power electronics,” J. Electr. Eng., vol. 68, no. 2, 12017.

Cited on page 44.

[128] J. Zhang, Y. F. Zhang, M. Miao, Y.-F. Jin, S.-L. Bai, and J. Q. Chen, “Simulation of fluid flow andheat transfer in microchannel cooling for LTCC electronic packages,” in Proc. Intl. Conf. Electron.Packag. Technol. High Dens. Packag., 9 2009, pp. 327 – 330.

Cited on page 44.

[129] N.-T. Nguyen, Mikrofluidik: Entwurf, Herstellung und Charakterisierung, 1st ed. Vieweg+TeubnerVerlag, 2004.

Cited on page 44.

[130] H. Adluru, M. A. Zampino, Y. Liu, and K. W. Jones, “Embedded heat exchanger in LTCC substrate,”in Proc. Intl. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol., 4 2003.

Cited on page 45.

[131] P. Uhlig, S. Holzwarth, O. Litschke, A. Serwa, and D. T. Tran, “On the influence of layer-to-layermisalignment on the microwave performance of LTCC antenna modules,” in Proc. Intl. Symp.Microel. 39th, 10 2006.

Cited on page 47.

[132] J. E. Ellis, “Manual vs. automatic stacking of LTCC layers.” Credence Engineering, Inc., Tempe,Ariz., Tech. Rep., July 2010. [Online]. Available: http://www.credenceeinc.com/files/Download/ManualvsAutomaticStacking7-14-2010.pdf

Cited on page 48.

[133] R. Dudek, E. Noack, R. Döring, and M. Hildebrandt, “Zuverlässigkeitsaspekte hinsichtlich des Ein-satzes keramischer Bauteile in elektronischen Baugruppen,” in Proc. Symp. Keram. Mehrlagentech.der DKG, 12 2016.

Cited on page 54.

[134] B. Lopez-Berrocal, J. de Oliva-Rubio, E. Marquez-Segura, A. Moscoso-Martir, I. Molina-Fernandez,and P. Uhlig, “High performance 1.8-18 Ghz 10 dB low temperature co-fired ceramic directionalcoupler,” Progr. Electromagn. Res., vol. 104, pp. 99 – 112, 2010.

Cited on page 57.

[135] P. Uhlig, J. Kassner, C. Günner, A. Serwa, and M. Tinnefeld, “Ausbeute bei kritischen Mikrowellen-modulen in LTCC und der Einfluss des Siebzustandes,” KOENEN Technologietag, Ottobrunn,2017.

Cited on page 57.

Bibliography 103

[136] W. P. Mason and R. A. Sykes, “The use of coaxial and balanced transmission lines in filters and wideband transformers for high radio frequencies,” Bell Syst. Tech. J., vol. 16, no. 3, pp. 275–302,1937.

Cited on page 57.

[137] J. Kassner and W. Menzel, “A drop-on band-pass filter for millimeter-wave multichip modules onLTCC,” IEEE Microw. Guided Wave Lett., vol. 9, no. 11, pp. 456–457, 11 1999.

Cited on page 57.

[138] G. L. Matthaei, L. Young, and E. M. T. Jones, Microwave Filters, Impedance-Matching Networks,and Coupling Structures, ser. Artech House Microwave Library, 1964, vol. 1.

Cited on page 57.

[139] P. Uhlig, J. Kassner, C. Günner, and E. Noack, “Microwave filter design optimized for ceramicmultilayer technique,” in Proc. Int. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol.12th, 2016.

Cited on page 57.

[140] U. Deisinger, A. Stiegelschmidt, A. Roosen, D. Schwanke, F. Bechtold, and C. Schmaus, “Charakter-isierung und Schwindungsverhalten von LTCC - Grünfolien,” PLUS, vol. 3, no. 3, 2001.

Cited on page 61.

[141] M. Wagner, A. Stiegelschmitt, A. Roosen, F. Bechtold, and D. Schwanke, “Optical dilatometer forinsitu measurements of warpage e↵ects during firing of LTCC multilayer structure,” in Proc. Intl.Symp. Microel. 35th, 9 2002.


[142] J. A. Avila-Rodriguez, “On generalized signal waveforms for satellite navigation,” Ph.D. dissertation,University FAF Munich, Faculty of Aerospace Engineering, Munich, Germany, 2008.

Cited on page 61.

[143] W. K. Jones, Y. Liu, B. Larsen, P. Wang, and M. Zampino, “Chemical, structural, and mechanicalproperties of the LTCC tapes,” J. Microelectron. Electron. Packag., vol. 23, no. 4, pp. 469–473,2000.

Cited on page 62.

[144] W. A. Schiller, “Zur Funktion der Gläser in der LTCC-Technologie,” DGG- Glasforum 2005 bei derSchott AG, 2005.

Cited on page 62.

[145] R. Müller, M. Eberstein, S. Reinsch, J. Deubener, A. Thiel, and W. A. Schiller, “E↵ect of rigidinclusions on sintering of LTCC,” Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B, vol. 48,no. 4, pp. 259–266, August 2007.

Cited on page 62.

[146] HERATAPE CT700 Materials System, data sheet, Heraeus Circuit Materials Division, 10 2003.Cited on page 62.

[147] A. W. Coats and J. P. Redfern, “Thermogravimetric analysis. a review,” The Analyst, vol. 88, pp.906–924, 1963.


[148] Reliability Prediction of Electronic Equipment, USA, DoD Std. MIL-HDBK-217, 1995.Cited on page 67.

[149] Electronic Components Reliability - Reference-Condition for Failure Rates and Stress: ModelsConversions, IEC Std. IEC 61 709, 1997.

Cited on page 67.

104 Bibliography

[150] A. Meyna and B. Pauli, Taschenbuch der Zuverlässigkeitstechnik. Hanser Fachbuchverlag, 2010.Cited on page 67.

[151] H. H. Meinke and F.-W. Gundlach, Taschenbuch der Hochfrequenztechnik, 4th ed. Springer, 1986.Cited on pages 67 and 128.

[152] D. Köther, A. Lauer, A. Wien, P. Uhlig, G. Pautz, G. Möllenbeck, and J. Berben, “X-band 1:32 dividernetwork for space application with excellent amplitude and phase balance,” in Proc. Eur. Microw.Conf. 33rd, 10 2003, pp. 899–902.


[153] D. J. Pedder, “Techniques for the examination of thick film resistor microstructures by T.E.M.”Electrocomponent Sci. Technol., vol. 4, no. 2, pp. 85–88, 1 1977.

Cited on page 68.

[154] S. M. Chitale and R. W. Vest, “Critical relationships between particle size, composition, and mi-crostructure in thick-film resistors,” IEEE Trans. Compon. Hybrids Manuf. Technol., vol. 11, no. 4,pp. 604–610, 12 1988.

Cited on page 68.

[155] P. Yang, M. A. Rodriguez, P. Kotula, B. K. Miera, and D. Dimos, “Processing, microstructure, andelectric properties of buried resistors in low-temperature co-fired ceramics,” J. Appl. Phy., vol. 89,no. 7, pp. 4175–4182, 4 2001.


[156] A. Kolek, A. W. Stadler, Z. Zawislak, K. Mleczko, and A. Dziedzic, “Noise and switching phenomenain thick-film resistors,” J. Phys. D: Appl. Phys., vol. 41, 2008.

Cited on page 68.

[157] Y.-M. Chiang, L. A. Silverman, R. H. French, and R. M. Cannon, “Thin glass film between ultrafineconductor particles in thick-film resistors,” J. Am. Ceram. Soc., vol. 77, no. 5, pp. 1143–1152,1994.


[158] A. Roosen, D. Pohle, and M. Wagner, “Characterization of the shrinkage behavior of pure, screenprinted thick film pastes and LTCC green tapes,” in Proc. Intl. Conf. Exhib. Ceram. InterconnectCeram. Microsyst. Technol., 4 2004.

Cited on page 68.

[159] J. R. Rellick and A. P. Ritter, “Non-trimmed buried resistors in green tape circuits,” in Proc. Intl. Conf.High Dens. Packag. MCMs, 1999.

Cited on page 68.

[160] P. Yang, D. Dimos, M. A. Rodriguez, R. F. Huang, S. Dai, and D. Wilcox, “Direct-write precisionresistors for ceramic packages,” in Proc. Mat. Res. Soc. Symp., vol. 542, 12 1998, pp. 159–164.

Cited on page 69.

[161] M. Wassmer, W. Diel, and K. Krüger, “Inkjet printing as technology for in-situ blending of thick-filmresistor inks,” in Proc. Intl. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol. 8th, 42012, pp. 142–151.

Cited on page 69.

[162] U. Partsch, S. Mosch, and M. Ihle, “Aerosol printed co- and post-fired functional films for LTCCmultilayer components,” in Proc. IMAPS Poland Annu. Conf., 9 2011.

Cited on page 69.

[163] G. Reppe, A. Rebs, and A. Schwarz, “Thick- and thin-film technologies on fired LTCC,” in Proc.MacroNano-Colloq. Ceram. Microsyst. 2nd, 2009.

Cited on page 69.

Bibliography 105

[164] J. Müller, D. Schwanke, T. Haas, E. Feuer, B. Schweizer, and A. Klaassen, “Design capabilities andmicrowave performance of thin film on LTCC,” in Proc. IMAPS Nordic Annu. Conf., 9 2003.

Cited on page 69.

[165] E. Feurer, “Thin film in combination with other substrate technologies,” in Proc. Eur. Microel. Packag.Conf., 6 2003.


[166] W. Ostwald, “Über die vermeintliche Isomerie des roten und gelben Quecksilberoxyds und dieOberflächenspannung fester Körper,” Z. Phys. Chem., vol. 34, pp. 495–503, 1900.

Cited on page 69.

[167] F. Forlani and M. Prudenziati, “Electrical conduction by percolation in thick film resistors,” Electro-component Sci. Technol., vol. 3, pp. 77–83, 1976.

Cited on page 70.

[168] R. D. Smith, “Metallizing composition for glass,” US Patent 2 103 598, 12, 1937.Cited on page 70.

[169] G. E. Pike and C. H. Seager, “Electrical properties and conduction mechanisms of Ru-based thick-filmcermet resistors,” J. Appl. Phy., vol. 48, pp. 5152–5169, dec 1977.

Cited on page 70.

[170] R. Stratton, “Volt-current characteristics for tunneling through insulating films,” J. Phys. Chem. Solids,vol. 23, no. 9, pp. 1177–1190, September 1962.

Cited on page 70.

[171] C. Grimaldi, P. Ryser, and S. Straessler, “Piezoresistive anisotropy of thick-film resistors,” J. Eur.Ceram. Soc., vol. 24, pp. 1893–1896, 2004.

Cited on page 70.

[172] M. Hrovat, A. Bencan, D. Belavic, J. Holc, and G. Dražic, “The influence of firing temperature on theelectrical and microstructural characteristics of thick-film resistors for strain gauge applications,”Sens. Actuators, A, vol. 103, no. 3, pp. 341–352, 2 2003.

Cited on page 70.

[173] Fixed resistors for use in electronic equipment - Part 1: Generic specification, IEC Std. IEC 60 115-8,Rev. Edition 2.0, 01 2009.


[174] Method of measurement of current noise generated in fixed resistors, IEC Std. IEC 60 195 TC/SC 40,Rev. 1.0, 1 1965.

Cited on page 72.

[175] Electronic and Electrical Component Parts; Method 304: Resistance temperature characteristic,USA, DoD Test Method Standard MIL-STD-202, Rev. G, February 2002.

Cited on page 72.

[176] Electronic and Electrical Component Parts; Method 308: Current-Noise Test for Fixed Resistors,USA, DoD Test Method Standard MIL-STD-202, Rev. G, February 2002.

Cited on page 72.

[177] J. B. Johnson, “Thermal agitation of electricity in conductors,” Phys. Rev., vol. 32, no. 1, pp. 97–109,7 1928.

Cited on page 72.

[178] H. Nyquist, “Thermal agitation of electric charge in conductors,” Phys. Rev., vol. 32, no. 1, pp.110–113, 7 1928.

Cited on page 72.

106 Bibliography

[179] R. W. Vest, “Conduction mechanisms in thick film microcircuits,” Purdue Research FoundationLafayette, Final technical report ADA024825, 12 1975.

Cited on page 72.

[180] J. Ringo, E. Stevens, and D. Gilbert, “On the interpretation of noise in thick-film resistors,” IEEETrans. Parts Hybrids Packag., vol. 12, no. 4, pp. 378–380, 12 1976.

Cited on page 72.

[181] L. K. J. Vandamme, “Criteria of low-noise thick-film resistors,” Electrocomponent Sci. Technol.,vol. 4, pp. 171–177, 1 1977.

Cited on page 72.

[182] A. Kolek, P. Ptak, and A. Dziedzic, “Noise characteristics of resistors buried in low-temperatureco-fired ceramics,” J. Phys. D: Appl. Phys., vol. 36, pp. 1009–1017, 2003.

Cited on page 72.

[183] L. K. J. Vandamme, “Noise in materials: Low frequency noise as a diagnostic tool for reliability andquality assessment of electronic devices,” in Fluctuations et bruit: seminaire, M. A. Py, M.-O.Hongler, and J.-P. Laedermann., Eds., 2000, pp. 67–96.

Cited on page 72.

[184] R. Best, Theorie und Anwendung des Phase-locked Loops. AT-Verlag, Aarau, 1982.Cited on page 72.

[185] U. L. Rohde, Microwave and Wireless Synthesizers. John Wiley and Sons Inc., 1997.Cited on page 72.

[186] J. Conrad, G., N. Newman, and A. Stansbury, “A recommended standard resistor-noise test system,”IRE Trans. Compon. Parts, vol. 7, no. 3, pp. 71–88, 1960.

Cited on page 73.

[187] G. Vasilescu, Electronic Noise and Interfering Signals: Principles and Applications. Springer, 12005.

Cited on page 73.

[188] Fixed resistors for use in electronic equipment - Part 8: Sectional specification - Fixed surface mountresistors, IEC Std. IEC 60 115-8, Rev. Edition 2.0, 01 2009.

Cited on page 74.

[189] U. H. Gysel, “A new n-way power divider/combiner suitable for high-power applications,” in IEEEMTT-S Int. Microw. Symp. Dig., May 1975, pp. 116–118.

Cited on page 74.

[190] T. Klein, P. Uhlig, C. Günner, and R. Kulke, “Substrate-integrated divider networks in LTCC withoptimized tolerance / isolation properties for Ka-band satellite systems,” in Proc. Intl. Symp.Microel. 47th, 2014.

Cited on page 74.

[191] J. Balcells-Ventura, T. Klein, P. Uhlig, C. Günner, and R. Kulke, “Tolerance-optimized RF structuresin LTCC for mm-wave frequencies applications,” J. Ceram. Sci. and Technol., vol. 6, no. 4, 2015.

Cited on page 74.

[192] P. Uhlig, “Innovative keramische Schaltungsplattformen für künftige Satellitentechnologien - iKER-SATEC,” Schlussbericht des DLR/BMBF-Verbundprojektes "iKERSATEC"; FörderkennzeichenBMBF 50 YB 1307. - Verbund-Nummer 01138125, 2016.


[193] H. Thust, K.-H. Drüe, T. Kirchner, T. Thelemann, and E. Polzer, “Behavior and performance of buriedresistors in green tape,” Appl. Microw. Wireless, pp. 30–36, 10 1999.

Cited on page 80.

Bibliography 107

[194] P. Uhlig, C. Günner, T. Klein, and R. Kulke, “LTCC Materialsysteme fuer Mikrowellenanwendungen:Umstellung von DuPont 943 auf 9k7 bei einem Ka-Band Beam-Forming-Network,” in Proc.IMAPS Germany Annu. Conf., 10 2011.

Cited on page 80.

[195] DuPont 943 Green Tape Data Sheet, data sheet, DuPont Microcircuit Materials, 2001.Cited on page 81.

[196] DuPont 9k7 Green Tape Data Sheet, data sheet, DuPont Microcircuit Materials, 2009.Cited on pages 81 and 84.

[197] HFB Series Buried Resistors Data Sheet, data sheet, DuPont Microcircuit Materials, November 2006.Cited on pages 81 and 83.

[198] R 8900 T Series Air-Fired Resistor System for HERATAPE CT700 Data Sheet, 120802nd ed., datasheet, Heraeus Circuit Materials Division, 10 2002.

Cited on page 81.

[199] FX 87 Series LTCC Resistors Data Sheet, Rev 09/07 ed., data sheet, Ferro Electronic Materials, 2007.Cited on page 81.

[200] K. Mleczko, Z. Zawislak, A. W. Stadler, A. Kolek, A. Dziedzic, and J. Cichosz, “Evaluation ofconductive-to-resistive layers interaction in thick-film resistors,” Microelectron. Reliab., vol. 48,no. 6, pp. 881–885, 6 2008.

Cited on page 81.

[201] A. Cattaneo, M. Cocito, F. Forlani, and M. Prudenziati, “Influence of the metal migration from screen-and-fired terminations on the electrical characteristics of thick-film resistors,” ElectrocomponentSci. Technol., vol. 4, no. 3-4, pp. 205–211, 1 1977.

Cited on page 81.

[202] M. Lisi, P. Angeletti, C. Catallo, A. Florio, G. Lucchi, and A. Cucci, “Airborne antennas for broadbandsatellite communications,” in Proc. ESA Workshop Satell. Antenna Technol. 25th, 9 2002.

Cited on page 90.

[203] J. Letschnik, H. Enke, F. Schurig, L. Schelde, and U. Walter, “Measurement of a weak satellite beaconlevel at 27.5 GHz with a dual circular antenna array [measurements corner],” IEEE AntennasPropag. Mag., vol. 56, no. 5, pp. 174–183, 10 2014.

Cited on page 90.

[204] H. Fenech, Amos, Tomatis, and Soumpholphkakdy, “Ka-SAT and future HTS systems,” in Proc. IEEEIntl. Vac. Electron. Conf. 14th, May 2013, pp. 1–2.

Cited on page 90.

[205] C. A. Grazia, M. Klapez, N. Patriciello, M. Casoni, A. Amditis, E. Sdongos, H. Gierszal, D. Kanakidis,C. Katsigiannis, K. Romanowski, P. Simplicio, A. Oliveira, S. Sonander, and J. Jackson, “Integra-tion between terrestrial and satellite networks: the PPDR-TC vision,” in Proc. IEEE Intl. Conf.Wireless Mobile Comput. Netw. Commun. 10th, 10 2014.

Cited on page 90.

[206] G. Baraglia, “Ground segment of the future: An innovative SatCom toolbox,” in IET Sem. Mil. Satell.Commun., 6 2015, pp. 1–25.

Cited on page 90.

[207] H. Wakana, H. B. Li, A. Miura, Y. Nirei, and M. Arakida, “Airborne imaging and Ka-band satellitecommunications in an experimental disaster management system,” Space Commun., vol. 18, no.3,4, pp. 157–166, 12 2002.

Cited on page 90.

108 Bibliography

[208] P. Kela, M. Costa, J. Turkka, M. Koivisto, J. Werner, A. Hakkarainen, M. Valkama, R. Jantti, andK. Leppanen, “Location based beamforming in 5G ultra-dense networks,” in Proc. IEEE Veh.Technol. Conf. 84th, 9 2016, pp. 1–7.

Cited on page 90.

[209] S. Han, C. I, Z. Xu, and C. Rowell, “Large-scale antenna systems with hybrid analog and digitalbeamforming for millimeter wave 5G,” IEEE Commun. Mag., vol. 53, no. 1, pp. 186–194, 1 2015.

Cited on page 90.

[210] J. W. Bandler, “Space mapping – have you ever wondered about the engineer’s mysterious "feel" for aproblem? [speaker’s corner],” IEEE Microw. Mag., vol. 19, no. 2, pp. 112–122, 3 2018.

Cited on page 90.

[211] ——, “The journey to automated design optimization and a vision for the future,” in IEEE MTT-S Int.Microw. Symp. Dig., 6 2017, pp. 1517–1519.

Cited on page 90.

[212] J. E. Rayas-Sanchez and V. Gutierrez-Ayala, “EM-based monte carlo analysis and yield prediction ofmicrowave circuits using linear-input neural-output space mapping,” IEEE Trans. Microw. TheoryTech., vol. 54, no. 12, pp. 4528–4537, 12 2006.

Cited on page 90.

[213] R. W. Johnson, M. Strickland, and D. Gerke, “3-D packaging: A technology review,” NASA ElectronicParts and Packaging Program, Report 071805, 6 2005, retrieved June 15th, 2017. [Online].Available: http://nepp.nasa.gov/docuploads/EA7E7EA1-BD30-4DA4-BD615FEA1A7F5AE9/3DPackagingReport071805.pdf

Cited on page 90.

[214] A. Baumann, M. Brieseck, S. Hoehn, T. Moritz, and R. Lenk, “Development in multi-componentpowder injection moulding of steel-ceramic compounds using green tapes for inmould labelprocess,” Powder Inject. Mould. Intl., vol. 2, no. 1, pp. 55–58, 2008.

Cited on page 90.

[215] V. Piotter, E. Honza, A. Klein, T. Müller, and K. Plewa, “Powder injection moulding of multi-materialdevices,” Powder Metallurgy, vol. 58, no. 5, pp. 344–348, oct 2015.

Cited on page 90.

[216] S. Ziesche, C. Lenz, and A. Müller-Köhn, “Mehrlagenkeramik und Keramikspritzguss - eine tech-nologische Kombination zur Herstellung dreidimensionaler funktioneller Keramikkomponenten,”Keram. Z., vol. 69, no. 1, pp. 29–33, 2017.

Cited on page 90.

[217] A. Petajajarvi, “3D ceramics by laser drawing,” IMAPS Nordic: Microw. Mater. Appl., 2006.Cited on page 91.

[218] W. Kollenberg, Additive Fertigung von Keramiken. HvB Verlag, 2016, ch. 3.4.2.9, pp. 1 – 26.Cited on page 91.

[219] D. Polsakiewicz and W. Kollenberg, “LTCC-based packaging by 3D-multi-material-printing (3Dmm),”in Proc. Fraunhofer Direct Digit. Manuf. Conf., 3 2014.

Cited on page 91.

[220] S. Mosch, “Finelinestrukturierung von Dickschichtpasten,” 54. Tre↵en des Sächsischen ArbeitskreisesElektronik-Technologie, 2010.

Cited on page 91.

[221] J. Balcells-Ventura, J. Leiss, P. Uhlig, M. Ihle, S. Ziesche, A. Bisognin, and C. Luxey, “Substrateintegrated MM-wave antenna utilizing metalized cavity walls,” in Proc. Intl. Conf. Exhib. Ceram.Interconnect Ceram. Microsyst. Technol., 2015, pp. 221 – 228.

Bibliography 109

Cited on page 91.

[222] R. D. Stewart and A. L. Stewart, Proposal Preparation. Wiley-Interscience, 1992.Cited on page 119.

[223] C. A. Balanis, Antenna Theory: Analysis and Design, 3rd Edition, 3rd ed. Wiley-Interscience, 42005.

Cited on page 119.

[224] O. C. Woodard, “High density interconnect verification of unpopulated multichip modules,” in Proc.IEEE/CPMT Int. Electron. Manuf. Technol. Symp., 11th, 9 1991, pp. 434–439.

Cited on page 121.

[225] Letter Designations for Radar-Frequency Bands, IEEE Std. 521-2002, 2003.Cited on pages 122 and 127.

[226] R. E. Trease and R. L. Dietz, “Rheology of pastes in thick-film printing,” Solid State Technol., vol. 15(I), pp. 39–43, 1972.

Cited on page 126.

[227] G. Wang, E. C. Folk, F. Barlow, and A. Elshabini, “Fabrication of microvias for multilayer LTCCsubstrates,” IEEE Trans. Electron. Packag. Manuf., vol. 29, no. 1, pp. 32 – 41, 1 2006.

Cited on page 127.

[228] B. Dziurdzia, “Photoimageable thick-films in microwaves,” Advancing Microelectronics, vol. 29,no. 2, 2002.

Cited on page 128.

[229] N. Nagai, E. Maekawa, and K. Ono, “New n-way hybrid power dividers,” IEEE Trans. Microw. TheoryTech., vol. 25, no. 12, pp. 1008–1012, 12 1977.

Cited on page 128.

A Appendix

AcronymsACC Adaptive Cruise Control 8

ActiveMultiFeed Integration of active control circuits into the MultiFeed beam-forming network 12

AD analogue-to-digital converter 16

ADAS Advanced Driver-Assistance System 8

ADS Advanced Design System 90

AESA active electronically steered array 8, 90

AM additive manufacturing 90, 91

ASIC Application Specific Integrated Circuit 3

BFN beam forming network 13, 32, 67, 74, 75, 79, 81, 90, 128

Bi2Ru2O7 Bismuth ruthenate 68

BMWi German Federal Ministry of Economics and Technology 12

CAD Computer-Aided Design 22, 91

CFD Computational Fluid Dynamics 45

CT Computed Tomography 54

CTE Coecient of Thermal Expansion 37

CVD chemical vapour deposition 126

DA digital-to-analogue converter 16

DBF Digital Beam-Forming i, iii, 8, 10, 12, 15, 18, 22, 25, 52, 89, 90

DC direct current 16, 22, 32, 37, 44, 75, 81, 89

DFM design for manufacturability i, iii, 57, 90

DLR German Aerospace Center 8, 12

DSP digital signal processing 15, 16

DTG di↵erential thermogravimetric analysis 62, 63

EM electro-magnetic 12, 47–51

ESD electrostatic discharge 75

111

112 Acronyms

FDTD Finite-di↵erence time-domain 12, 20, 30, 32, 45, 47, 48, 52, 65, 75, 79, 89, 90

GaAs Gallium Arsenide 35

GaN Gallium Nitride 35, 39, 42, 44

GNSS Global Navigation Satellite System 61, 65

HEMT High Electron Mobility Transistor 35

HTCC High Temperature Co-fired Ceramic 3–7, 30

IC Integrated Circuit 3, 61

IF intermediate frequency 15, 16, 20, 89

iKERSATEC Innovative ceramic circuits for future satellite technologies 13, 83, 86

KERAMIS Ceramic Microwave Circuits for Satellite Communications 12

LHCP Left-Hand Circular Polarization 18, 21, 49

LNA low noise amplifier 16

LO local oscillator i, iii, 10, 15, 16, 22, 89

LTCC Low Temperature Cofired Ceramic i, iii, 3–10, 12, 13, 16, 20, 22, 25, 27–34, 37, 38, 42, 44, 47, 48,52, 54, 57, 61, 62, 64, 67–70, 72, 80, 81, 83, 84, 88–91, 119

MCM Multi-Chip Module 3

MCM-C Multi Chip Module Ceramic Substrate 3

MCM-D Multi Chip Module Deposited (metal and dielectric) 3

MCM-L Multi Chip Module Laminate Substrate 3

MMIC Monolithic Microwave Integrated Circuit i, 3, 8, 10, 25, 32

MSA mechanically steered array 90

MTBF mean time between failures 67

MultiFeed Configurable Multi-Feeding System for Ka-band reflector antennas in satellite communications12

OOV On-Orbit Verification of new techniques and technologies 8

Pb2Ru2O7 Lead ruthenate 68

PCB printed circuit board 3, 37, 91

PIC Photonic Integrated Circuit 7

PVD physical vapour deposition 126

radar RAdio Detection And Ranging 8

RF radio frequency iii, 10, 16, 20, 25, 32, 44, 47, 48, 52, 62, 80, 81, 89

RHCP Right-Hand Circular Polarization 18, 21, 49

RuO2 Ruthenium oxide 68–70

RX Receive 8, 32

SANTANA Smart Antenna Terminal 10, 12, 15, 16, 25, 47, 48, 67, 89, 90

SAR synthetic aperture radar 8

SatCom satellite communication 12, 22, 89, 90

Si Silicon 35

SIW substrate integrated waveguide i, 9, 25, 27, 28, 30–32

SMD Surface-Mount Device 9, 67

SMT Surface-Mounting Technology iii, 3

SPICE Simulation Program with Integrated Circuit Emphasis 90

TEM-mode Transverse Electromagnetic Mode 120, 123, 126

TET Technology Experiment Carrier 8

TGA thermogravimetric analysis 28, 61, 62, 64, 89

TX Transmit 8, 32, 89

VNA vector network analyser 53

WPAN Wireless Personal Area Network 9, 28, 29

SymbolsB Bandwidth

CTCR Cold Temperature Coecient of Resistance

E Electrostatic charging energy

GF Gauge Factor

HTCR Hot Temperature Coecient of Resistance

L Length

NI Noise index

Pn Power of thermal noise

113

R Resistance

RCWV Rated Continuous Working Voltage

R Sheet resistance

R Thermal resistance

S TOL Short Time Overload

S E Power spectral density of current noise

S T Power spectral density of thermal noise

T Absolute temperature

TCR Temperature Coecient of Resistance

V Voltage

VS WR Voltage Standing Wave Ratio

W Width

ZL Load Impedance

p Pressure drop

V Volumetric flow rate

k Boltzman’s constant, k = 1.3807 10=23 J K=1

t Thickness

" In the context of infrared thermography: Emissivity

"0 Vacuum permittivity, "0 8.854 187 817 6 10=12 F m=1.

"r Relative permittivity

" Permittivity, " = "r"0

Dynamic viscosity

Wavelength.

Azimuth angle in the radiation pattern of an antenna.

resistivity (specific electrical resistance)

Standard deviation

Elevation angle in the radiation pattern of an antenna.

# Temperature

List of Figures

1.1 LTCC tapes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 LTCC pastes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Schematic overview of the LTCC process. . . . . . . . . . . . . . . . . . . . . . . . . . . 41.4 LTCC hermetic package for a 20 GHz Synthesizer . . . . . . . . . . . . . . . . . . . . . . 51.5 Conductivity of metals vs. melting point . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.6 SMD low pass filter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.7 GSM power amplifier. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.8 Antenna side of the SANTANA module . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.9 Component side of the SANTANA module . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.1 Principle of phased array beam steering . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.2 Antenna element, exploded view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.3 EM-model of the 4 4 array. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.4 Antenna element in cavity shielded with via fences. . . . . . . . . . . . . . . . . . . . . . 182.5 Branch-line coupler. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192.6 Calibration network of the 4 4 antenna array. . . . . . . . . . . . . . . . . . . . . . . . 192.7 Measurement setup for far field measurements in the anechoic chamber . . . . . . . . . . 202.8 Radiation pattern of the 4 4 antenna array RHCP . . . . . . . . . . . . . . . . . . . . . 212.9 Radiation pattern of the 4 4 antenna array LHCP . . . . . . . . . . . . . . . . . . . . . 212.10 SANTANA 8 8 antenna array EM model, antenna side, partly deconstructed . . . . . . . 232.11 SANTANA 8 8 antenna array EM model, component side, partly deconstructed . . . . . 23

3.1 Cross section of antenna cavity in the SANTANA 8 8 module . . . . . . . . . . . . . . 263.2 Cross section of cooling channel in the SANTANA 8 8 module . . . . . . . . . . . . . . 263.3 Cavity defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.4 LTCC module with carbon inserts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283.5 Silicon rubber insert . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.6 Cavities after sintering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.7 60 GHz LTCC WPAN antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.8 EM model of the SIW antenna element . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.9 Antenna side of the LTCC module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303.10 Substrate integrated waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.11 Electrical field in a waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.12 SIW array column with the corresponding feed network . . . . . . . . . . . . . . . . . . . 313.13 Electrical field distribution of SIW feeding network and antenna elements. . . . . . . . . . 323.14 Stacking plate with hard inserts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.15 Temperature distribution in the SANTANA 8 8 module. . . . . . . . . . . . . . . . . . . 343.16 Conventional power-amplifier. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.17 Switched-mode amplifier. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.18 Class E amplifier, simulation of drain voltage and current in time domain. . . . . . . . . . 353.19 Class E amplifier, simulation of drain voltage and current in frequency domain. . . . . . . 363.20 GaN transistor Cree CGH60015D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.21 Isometric view of the amplifier module V1 breadboard model. . . . . . . . . . . . . . . . 373.22 LTCC layer stack (production drawing) of the amplifier module . . . . . . . . . . . . . . . 383.23 Micro-channel design of module V2a . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

115

116 List of Figures

3.24 Micro-channel design of module V2b . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.25 Thermal test die, chip photograph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.26 Thermal resistance of LTCC module with cooling channels vs. volumetric flow. . . . . . . 403.27 Thermogram of the GaN transistor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.28 Infrared emissivity of GaN vs. temperature. . . . . . . . . . . . . . . . . . . . . . . . . . 423.29 Thermal resistance of LTCC module with cooling channels vs. volumetric flow. . . . . . . 433.30 Direct comparison of cooling channels: pressure drop vs. volumetric flow. . . . . . . . . . 43

4.1 Reflection coecient vs. frequency for displacement of layer C3 . . . . . . . . . . . . . . 494.2 Radiation pattern for displacement of layer C3 in X-direction . . . . . . . . . . . . . . . . 504.3 Radiation pattern for displacement of layer C3 in Y-direction . . . . . . . . . . . . . . . . 504.4 Reflection coecient vs. frequency for displacement of layer C2 . . . . . . . . . . . . . . 514.5 Radiation pattern for displacement of layer C2 in X-direction . . . . . . . . . . . . . . . . 514.6 Probe station measurement of the 4 4 antenna array building block . . . . . . . . . . . . 534.7 Reflection coecient of the single elements of a 4 4 antenna array building block . . . . 534.8 MultiFeed power divider. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544.9 Radiograph of the MultiFeed power divider . . . . . . . . . . . . . . . . . . . . . . . . . 554.10 CT slice of the MultiFeed power divider . . . . . . . . . . . . . . . . . . . . . . . . . . . 554.11 CT-scan detail of MultiFeed power divider: vias . . . . . . . . . . . . . . . . . . . . . . . 554.12 CT-scan compared to polished cut image . . . . . . . . . . . . . . . . . . . . . . . . . . . 564.13 Microphotograph of defective vias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564.14 Bandpass filter 3 layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584.15 Bandpass filter 3 S-parameters measurement . . . . . . . . . . . . . . . . . . . . . . . . . 584.16 Bandpass filter 2 layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594.17 Bandpass filter 2 S-parameters measurement . . . . . . . . . . . . . . . . . . . . . . . . . 59

5.1 Dual band GNSS antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615.2 Sintering profiles for di↵erent LTCC material systems . . . . . . . . . . . . . . . . . . . . 635.3 Thermogravimetric Analysis of Heraeus CT700 . . . . . . . . . . . . . . . . . . . . . . . 635.4 Shrinkage of DP 951 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645.5 Sintering profiles for Heraeus CT700 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655.6 Radiation pattern of dual band GNSS antenna in the E5a band . . . . . . . . . . . . . . . 665.7 Radiation pattern of dual band GNSS antenna in the L1 band . . . . . . . . . . . . . . . . 66

6.1 Wilkinson Power Divider in LTCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 686.2 Particle size comparison in resistor pastes . . . . . . . . . . . . . . . . . . . . . . . . . . 696.3 Microstructural changes in thickfilm resistors associated with Ostwald ripening . . . . . . 696.4 Change of resistance vs. temperature for a thick-film resistor . . . . . . . . . . . . . . . . 716.5 Spectral Density of Total Noise in Thick-Film Resistors . . . . . . . . . . . . . . . . . . . 736.6 Noise index of di↵erent resistor types . . . . . . . . . . . . . . . . . . . . . . . . . . . . 736.7 Schematic of a two-way Gysel power splitter . . . . . . . . . . . . . . . . . . . . . . . . 756.8 Isometric view of lossy resonator 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 766.9 Reflection coecient of lossy resonator 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 766.10 Isometric view of lossy resonator 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776.11 Reflection coecient of lossy resonator 2 . . . . . . . . . . . . . . . . . . . . . . . . . . 776.12 Isometric view of lossy resonator 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 786.13 Reflection coecient of lossy resonator 3 . . . . . . . . . . . . . . . . . . . . . . . . . . 786.14 Layout of the Wilkinson power divider . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796.15 Layout of the Gysel power divider . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796.16 Comparison of Gysel and Wilkinson power splitter . . . . . . . . . . . . . . . . . . . . . 79

6.17 Process and material dependencies of resistor properties . . . . . . . . . . . . . . . . . . . 806.18 Sheet resistance of DuPont HFB12 paste in two di↵erent LTCC substrates . . . . . . . . . 826.19 Sheet resistance of DuPont HFB12 paste buried and on top of DuPont 9k7 . . . . . . . . . 826.20 Resistor test layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 836.21 Resistor test panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 846.22 Probe card and needles with the resistor test substrate. . . . . . . . . . . . . . . . . . . . . 866.23 Resistor Test Comparison R1 - R6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 876.24 Resistor Test Comparison R2 outer layer I . . . . . . . . . . . . . . . . . . . . . . . . . . 876.25 Resistor Test Comparison R2 outer layer II . . . . . . . . . . . . . . . . . . . . . . . . . 88

List of Tables

6.1 Resistor dimensions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 816.2 Resistor dimensions for the iKERSATEC resistor X-check. . . . . . . . . . . . . . . . . . 836.3 Design of experiment for the iKERSATEC resistor X-check. . . . . . . . . . . . . . . . . 85

117

GlossaryActiveMultiFeed

The goal of the ActiveMultiFeed (AMF) project is to develop a reconfigurable power divider networkfor a Multiple-Feed per Beam (MFB) antenna feed network for Ka-band satellite communicationapplications. In contrast to conventional network the AMF will be smaller and weight reduced. Thehigh complexity requires a combination of di↵erent circuit technologies. The core-cell will be againa honeycomb structure made of LTCC (see MultiFeed). In di↵erent branches of the network, activecomponents will be integrated for phase and amplitude setting. 12

batch

A number of items or a quantity of material (like paste or tape) manufactured in the same time frame,with the same equipment, the same recipe, the same parameters, and the same raw materials. Thebatch (or lot) number is used to identify it in the Certificate of Conformity and other documents forprocess control. 80, 83

breadboard

Stewart [222] defines a breadboard or a brassboard as "a laboratory or shop working model that mayor may not look like the final product or system, but that will operate in the same way as the finalsystem". 37, 115

circular polarization

Balanis [223]:"A time-harmonic wave is circularly polarized at a given point in space if the electric (ormagnetic) field vector at that point traces a circle as a function of time. The necessary and sucientconditions to accomplish this are if the field vector (electric or magnetic) possesses all of the following:a. The field must have two orthogonal linear components, and b. The two components must have thesame magnitude, and c. The two components must have a time-phase di↵erence of odd multiples of90°. The sense of rotation is always determined by rotating the phase-leading component toward thephase-lagging component and observing the field rotation as the wave is viewed as it travels away fromthe observer. If the rotation is clockwise, the wave is right-hand (or clockwise) circularly polarized; ifthe rotation is counter-clockwise, the wave is left-hand (or counter-clockwise) circularly polarized.The rotation of the phase-leading component toward the phase-lagging component should be donealong the angular separation between the two components that is less than 180°. Phases equal to orgreater than 0°and less than 180°should be considered leading whereas those equal to or greater than180°and less than 360°should be considered lagging." (see polarization) iii, 10, 16–18, 20, 62

co-fired

In LTCC (Low Temperature Co-fired Ceramic) technology, "Co-fired" is indicating the fact thatdielectric and metallization (vias and conductors) are fired in one process step. This is in contrastto conventional thick-film hybrid circuits where multilayer designs require multiple firing steps.Therefore LTCC is saving process time and energy but also presents an inherent challenge as therequirements for physical and chemical material compatibility (tape and di↵erent pastes) are far morestringent. 5, 68, 70, 80, 81

co-polarization

Balanis [223]:"Co-polarization represents the polarization the antenna is intended to radiate (receive)while cross-polarization represents the polarization orthogonal to a specified polarization, which isusually the co-polarization." (see polarization) 18, 49

119

120 Glossary

CPW

The coplanar waveguide (CPW) consists of a conductive strip on a dielectric substrate which isseparated from the surrounding ground-plane in the same level by two gaps. Since the electrical fieldof the CPW is partly in air and partly in the dielectric substrate, it is a quasi TEM-mode transmissionline. The CPW is particularly helpful where vias are not available. Having ground-plane and conductoron the same side of the substrate can be useful for the integration of active and passive componentsthat need to be connected to the transmission line and ground. 32, 48

cross-polarization

(see co-polarization, polarization) 18, 20, 49

CT-scan

Computed Tomography (CT) can produce three-dimensional representations of an object by processinga series of (usually X-Ray) images taken at di↵erent angles and positions. The output is a voxelbased three-dimensional image of the exterior and interior of the inspected sample. In non-destructivematerial testing X-Ray sources with higher acceleration voltage than in medical applications can beused with the e↵ect of very high spatial resolution. 54

de-bindering

The laminated stack still contains most of the organics that were useful in the process so far to makethe tape flexible and workable and helped to get the viscosity of the pastes right for the screen-printingprocess. The better part of solvents has been evaporated during several drying steps, while a smallpercentage is still found in tapes and pastes. All organics have to be removed out of the laminate priorto sintering. Carbon will act as a sintering inhibitor if it is not fully removed out of the stack beforesintering. Carbon residues of the organic compounds will also deteriorate RF-performance. The ovenprofile for LTCC de-bindering is chosen to provide enough time at discrete temperatures for the binderburnout. Naming this phase binder burnout is common but inaccurate since only part of the organics isactually "burnt" out. Organic components are also cracked and evaporated. Therefore the less commonexpression "de-bindering" is describing the process better. For the material in the inner layers of thestack the porosity of the stack is the only exit path. This porosity develops when the LTCC stack isde-bindered slowly form the surface towards the inner layers. The glasses with low melting point startto melt at approximately 600°C. This will seal the porosity required for the outgassing. De-binderingshould be accomplished at this point. Sacrificial materials as shown in Fig. 3.4 that decomposedi↵erently from the tape organics are an added challenge and have the potential to destroy the stackwith their evaporation pressure when burnt out too fast. A simple way to monitor the de-binderingprocess online is to measure the oxygen content of the exhaust gas. The de-bindering process isconsuming oxygen. Given a constant air flow in the furnace the completion of debindering can thus bededucted from a recovery of the oxygen content. A Thermo-Gravimetric-Analysis (TGA) is a morethorough insight in the development of debindering over temperature and time. 61

devitrification

The reaction when glass changes from glassy to ceramic, crystallization in a formerly crystal-free(amorphous) glass. In glass processing techniques devitrification is usually an undesired reactionbecause it changes optical and mechanical properties, e.g. the glass becomes opaque and cracksdevelop. However in LTCC materials devitrification increases re-firing stability and helps to minimizeglass infiltration in adjacent areas. 68

downlink

In the context of satellite communications, the downlink (DL) designates the link from a satellite to aground station. 10

Glossary 121

E-plane

In the radiation pattern of a linear polarized antenna this is the plane parallel to the E-field (electricalfield) of the antenna element. 30–32

emulsion build up

Emulsion build up specifies the thickness of the emulsion on a finished printing screen. It is definedas the height of emulsion above the screen (substrate side). It is not to be confused with the initialthickness of the dry emulsion laminated to the screen, since some of it will sink into the screen fabricwhen laminated to it. 68

flying probe test system

Flying probe test systems are utilized to test printed circuit boards and multi chip modules electrically.These testers are highly specialized industrial robots that can be programmed to position two ormore probes in predefined places and perform in-circuit tests in assembled units or continuity testsas manufacturing defect analysis on unpopulated boards. Flying probe testers are more flexible thanbed-of-nail testers and avoid the expense of dedicated equipment for each module. However, there isa trade-o↵ in throughput because measurements are made serially. For this reason flying probes areprimarily used for low to medium quantities [224]. 45

Galileo

Galileo is the name of the global navigation satellite system (GNSS) (global navigation satellitesystem) of the EU (European Union) that is currently being put into operation through the EuropeanSpace Agency (ESA) and the European GNSS Agency (GSA). 61, 62

GNSS

A global navigation satellite system (GNSS) is a satellite navigation system with global coverage.Examples are the United States NAVSTAR Global Positioning System (GPS), the Russian GLONASSand the European Union’s Galileo. 61

green

A green body or greenware is a ceramic workpiece before it is sintered. In LTCC technology this isthe laminated stack of tape layers. 27

H-plane

In the radiation pattern of a linear polarized antenna this is the plane parallel to the H-field (magneticfield) of the antenna element, perpendicular to the E-plane. 31, 32

iKersatec

The goal of this German BMBF/DLR project (2013 - 2016) is the development of new technologies forfuture communication satellites. A higher eciency (financial and electrical) as well as an improve-ment of the electric properties (signal-to-noise ratio, bandwidth) are the long-term objectives. Thecrucial task of the project is the development and research of new technologies based on KERAMIS®-LTCC. New circuit technologies and microwave modules are developed which satisfy the demandsof future satellites and partially include semiconductor elements. The focus of iKersatec lays on thehybrid integration of new functional elements. Project 50YB1307, Consortium 01138125. 13, 74

122 Glossary

isostatic lamination

Prior to isostatic lamination the LTCC stack is placed on top of a solid plate to maintain flatness duringlamination. A sheet of plastic foil (e.g. PET or PTFE) is stacked between plate and LTCC stack toavoid contamination of the LTCC with metal and to facilitate release of the stack after lamination.The whole arrangement is wrapped in rubber foil to improve pressure distribution and then put in avacuum bag. Now the LTCC stack is ready to go into the pressure chamber of the isostatic laminator.The chamber is flooded with hot water and the stack is left there long enough (dwell time typ. 10 to30 minutes @ 70°C) to heat completely through before pressure (typ. 21 MPa) is applied for 5 to 10minutes to laminate it. The physical principle of this method ensures excellent distribution of pressureand temperature in the stack during lamination. The extra e↵ort with vacuum bagging and unwrappingis rewarded with well controlled shrinkage and ease of cavity formation. 62

jet impingement heat transfer

Zuckerman and Lior: "Compared to other heat or mass transfer arrangements that do not employphase change, the jet impingement device o↵ers ecient use of the fluid, and high transfer rates. Forexample, compared with conventional convection cooling by confined flow parallel to (under) thecooled surface, jet impingement produces heat transfer coecients that are up to three times higher ata given maximum flow speed, because the impingement boundary layers are much thinner, and oftenthe spent flow after the impingement serves to turbulate the surrounding fluid. Given a required heattransfer coecient, the flow required from an impinging jet device may be two orders of magnitudesmaller than that required for a cooling approach using a free wall-parallel flow. For more uniformcoverage over larger surfaces multiple jets may be used. The impingement cooling approach alsoo↵ers a compact hardware arrangement." [124] 44

Ka-band

In satellite communication the letter Ka designates an up-link frequency range of 27.5 to 31 GHz anda down-link frequency range of 17.7 to 21.2 GHz.[225] i, iii, 10, 12, 22, 54, 89, 90

KERAMIS

Ceramic Microwave Circuits for Satellite Communications [73], a German BMBF/DLR researchproject (Oct. 03 - Sep. 06): KERAMIS® aims at the development of innovative and inexpensivecomponents for future applications in multimedia satellite communications. The principle of operationand potential benefit for satellite-based systems will be demonstrated. The rationale of the project is toexploit the possibility of integrating passive and active components in LTCC multilayer structures,and by minimising the complexity of the semiconducting components. LTCC technology o↵ers anumber of benefits such as low-cost production and assembly, compact size and hermetic sealing,which are most important for satellite-based applications. Three technological milestones are underdevelopment for the Ka-band satellite applications used for the down-link : 1. Solid state poweramplifier 2. Frequency synthesizer (VCO) 3. Reconfigurable 4x4 switch matrix 12

L-band

In satellite communication the letter L designates a down-link frequency range of 1.53 to 2.7 GHz[225]. The name is borrowed from the L-band in RADAR (1 to 2.6 GHz). 62

lamination

After stacking the LTCC layers in correct order and accurate position they need to be assembledpermanently, they need to be laminated. Pressure and temperature is applied to accomplish this.The organic part of the LTCC tape provides adhesion in the green (unfired) state. This is the reason

Glossary 123

why large metallized areas in inner layers should be avoided. Typical lamination parameters forcommercial LTCC material systems are around 70°C and 21 MPa for five to ten minutes. Whilesoftening of the tape is required to achieve reliable adhesion it may also cause undesired deformationin conjunction with the applied pressure. The influence of lamination pressure on shrinkage duringsintering emphasizes the importance of good control over lamination pressure and its uniformity. Thisrelationship is also used to compensate lot-to-lot variations of shrinkage and thus improve overallaccuracy of module dimensions. Lamination is mainly done in an isostatic press (autoclave) or auniaxial press. Green Laminate Density (GLD) is the specific weight of the laminated stack. It is aparameter that is suitable to monitor the result of the lamination process. 62

lith-film

Photographic film with very high contrast used in photo-lithography. This orthochromatic materialgives high density blacks and no grays. 124

microstrip

This planar waveguide consists of a conductive strip on a dielectric substrate with a conductiveground-plane. Since the electrical field of the microstrip line is partly in air and partly in the dielectricsubstrate, it is a quasi TEM-mode transmission line. 8, 30, 48, 67

microwave

Microwave designates the wavelength range from 300 mm to 1 mm corresponding to the frequencyrange of 1 GHz to 300 GHz. iii, 3, 4, 10, 16, 25, 30, 33, 44, 47, 48, 67, 68, 74, 80, 81, 86, 89–91

millimeter-wave

Millimeter-wave designates the wavelength range from 10 mm to 1 mm corresponding to the frequencyrange of 30 GHz to 300 GHz. The International Telecommunications Union (ITU) labels this frequencyrange extremely high frequency (EHF). The frequency bands available in this wavelength range o↵erlarger bandwidth e.g. for high data rate communication, and high resolution radar. 3, 8, 9, 12, 32, 80,81, 90, 91

MultiFeed

Configurable Multi-Feeding System for Ka-Band Reflector Antennas in Satellite Communications [74],A German BMWI/DLR project (Apr. 09 - Mar. 11): MultiFeed stands for the design, manufacturingand characterization of a multilayer LTCC beam-forming network to excite the feed array of a highgain multi spot beam antenna for Ka-band multimedia services. Multi spot beam scenarios, coveringthe respective service areas by a so called four colour beam topology seem to be the most suitabletechnology for this kind of satellite communication services [?]. 12, 75, 79, 81

Mylar

Mylar is one of the trade names for BoPET. BoPET (Biaxially-oriented polyethylene terephthalate) is apolyester film made from stretched polyethylene terephthalate (PET) and is used in the LTCC processas a backing film during tape casting for its high tensile strength, and its chemical and dimensionalstability. LTCC tape is delivered in rolls or sheets with Mylar backing. 47

optical dilatometry

An optical dilatometer measures dimensional changes over temperature [141]. The optical dilatometeris a furnace with windows to illuminate (backlit) and observe the sample with a camera in orderto measure dimensional changes over time. In the LTCC process, dimensional changes during de-bindering and sintering are particularly interesting to analyse material behaviour and helping tounderstand the influence of temperature profile and atmosphere in the sintering furnace. 61, 89

124 Glossary

piezoresistive

The piezoresistive e↵ect describes the change in electrical resistance of a component when mechanicalstress is exerted. In contrast to the piezoelectric e↵ect, there is no change in electrical potential. 70

post-fired

It is one of the key features of the LTCC process that all materials (tape, conductor paste etc.) aresintered (i.e. co-fired) in one process step. However, some resistor pastes, special conductors oroverglaze are printed on the fired LTCC substrate. These pastes are called post-fired to distinct themfrom the co-fired materials, and they are closer to conventional thick-film pastes as they are printed onglass-ceramic that will not shrink during sintering (re-firing). 68, 81

PT1000

The PT1000 temperature sensor is a resistance temperature detector (RTD) with a resistance of 1000Wat 0°C. The resistor material is platinum, which has a high accuracy and repeatability as precisiontemperature sensor. 42

refiring

It is one of the key features of the LTCC process that all materials (tape, conductor paste etc.) aresintered (i.e. co-fired) in one process step. However, some resistor pastes, special conductors oroverglaze are printed on the fired LTCC substrate. To sinter these materials repeated sintering cyclesare required. It is important that the LTCC substrate is mechanically stable when subjected to repeatedsintering ("refiring"). Resistors will change their value when refired, this is called "refiring drift" andit depends on the number of sintering cycles and the peak temperature thereof. 80, 81

SANTANA

A German BMWI/DLR project (phase 1 (2001 - 2003), phase 2 (2003 - 2007)): In the SANTANAproject, IMST developed transmit (Tx) and receive (Rx) modules for a mobile Ka-band terminal forhigh data rate communication channels for aircraft or ships. This terminal is equipped with electroni-cally steerable antennas based on digital beam-forming. These antennas o↵er the greatest flexibilitywithout the disadvantages inherent in mechanical systems [1]. http://www.smart-antennas.de/. iii, 10,12, 22, 33, 61

screen printing

For via filling and conductor printing, similar printing techniques are utilized: stencil printing andscreen printing. From the various techniques used in the graphic arts industry screen printing isselected for microelectronic circuits because it can handle the inks required for conductors with goodconductivity (i.e. sucient thickness of print) and printing resolution. For the screen printing processthe screen is the "printing plate". To fabricate this screen a tensioned open weave mesh is mounted ona rigid frame and coated with emulsion. The artwork on a lith-film is then copied with UV-light tothe emulsion. The developed emulsion provides openings where conductor patterns are to be printed.For the the precision and repeatability required in microelectronics o↵ contact screen printing hasproven most suitable. The characteristic of this method is the o↵-contact distance (also referred to as"snap o↵") at which the screen is positioned above the tape to be printed. At the start of the printingcycle paste is applied in a streak in front of the squeegee. The squeegee pushes the screen down ontothe printing surface and moves across the screen. During that movement the paste rolls in front ofthe squeegee and is pushed into the screen’s open mesh. The screen is charged with paste. Screentension and o↵ contact distance provide the force to separate the screen from the tape right behind thesqueegee. The screen is now discharged and the paste is in a well defined pattern on the tape. 4, 6, 28,67, 70, 81, 86, 88, 90

Glossary 125

sheet resistance

Thick and thin film resistors are characterized by their sheet resistance. This is assuming homogeneousthickness in the resistive film. The general formula for resistance (A.1) is thus reduced to a two-dimensional representation (A.3).

R =

t· L

W(A.1)

R =

t(A.2)

R = R ·LW

(A.3)

The sheet resistance R is the resistance of a square resistor (L =W) and thus is in . For enhancedclarity it is common to use / (’Ohms per square’). 68, 71, 75–84, 86–88

sintering

During sintering the constituents of the LTCC multilayer get their final functionality and properties.Conductor paste turns into a conductor, glass and ceramic powders become dielectrics and resistor pastegets its resistance. Sintering is a decisive process step for the electrical and mechanical properties ofthe final multilayer substrate. In technical terminology sintering is the solidification and densificationof a green body (or powder) into a compact matter through temperature treatment. It is characteristicfor this process that not all ingredients melt and thus the outer shape is preserved. The densificationshould merely lead to a uniform and reproducible shrinkage. The goal is a dense, homogeneous, crack-and pore-free compound with a fine crystalline structure. The driving force of sintering is the tendencyof the system to reduce surface and interface energy by grain growth and pore shrinkage. Very oftenchemical reaction potential and capillary forces play also a role. The sintering process strives towardsa minimum of free enthalpy. It can be promoted by applying external pressure as well as vacuum.Favourable sintering conditions help to get close to the theoretical density of the compound. Themost significant advantage of sintering is the ability to combine materials which cannot be combinedotherwise into compounds with new electrical and mechanical properties. 4, 5, 61

stacking

In order to assemble the multilayer, the LTCC tapes with vias and conductors are stacked in correctorder and in the precise position. A stacking plate with registration pins is a simple means to do thismanually. Mechanical registration requires a certain robustness of the tape and manual skills of theoperator. For thin or soft tape, robotic stacking with optical registration is preferred. The tape is placedon an alignment table and fiducials for layer-to-layer registration are captured by cameras. The tapeis then aligned in X-, Y- and -orientation. A vacuum pick-up transfers the tape to the stack on thestacking plate. In order to hold the layers in place until the stack is laminated, the last layer is eitherpre-laminated with moderate heat and pressure on the entire area or tacked with heated pins in theperiphery of the stack. It is also possible to tack tapes with solvent or glue. Tacking will be doneoutside the actual module area since it will deform the tape where heat or solvent is applied. In someLTCC-lines the tape remains on the carrier foil (BoPET) until it is stacked and pre-laminated. Thecarrier foil is removed from the top of the stack and the next layer can be stacked. 47, 62

126 Glossary

stripline

The conductive strip of this planar waveguide is sandwiched between two layers of dielectric whichare shielded by a two conductive ground-planes on top resp. bottom. The stripline is a true TEM-modetransmission line. 30, 48

TerraSAR-X

IMST developed and manufactured high-precision 1:32 RF-distribution networks for the earth obser-vation satellite TerraSAR-X. These devices provide the highly stable distribution of high-frequencysignals at 9.65 GHz. In this application, the stability of amplitudes and phases directly a↵ects theprecision of the satellite’s radar images. 8

thermo-gravimetric analysis

The thermogravimetric analysis (TGA) utilizes an instrument referred to as a thermogravimetricanalyser. This instrument consists of an inert and temperature-stable sample holder coupled to a micro-balance in a furnace with controlled temperature and atmosphere. This micro-balance continuouslyrecords changes of mass in the sample under investigation over time as temperature in the furnace ischanging. The TGA curve is the plot of percentage of initial mass vs. time (or temperature). The firstderivative of the TGA curve (the DTG curve) is used to determine inflection points that are used fordetailed interpretations as well as di↵erential thermal analysis [147]. In the field of LTCC technology,TGA is particularly useful to optimize process parameters (mainly the temperature profile) for thede-bindering of the laminate (i.e. the multilayer stack of LTCC layers), where the organic componentsof tapes and pastes are decomposed and removed. 61

thickfilm

Thickfilm technology is an additive process to manufacture microelectronic circuit boards. The mostcommon substrate is alumina (Al2O3). Conductors are fabricated by screen-printing pastes based ongold, silver or copper. Pastes for passive components have a functional phase according to the desiredperformance (e.g. metals or metal-oxides for resistors or ceramic and glass for dielectrics). Dryingand sintering of the printed structure are the next process steps to finalize the circuit. 18, 31, 37, 48,67, 68, 70–72, 74, 75, 80, 81, 83, 86

thinfilm

Thinfilm technology designates vacuum deposition of thin layers of di↵erent materials. Either CVD(chemical vapour deposition) or PVD (physical vapour deposition) is used to deposit layers with athickness ranging from a few nanometres to micrometres. It is used e.g. for optical coatings (such asantireflective coatings), corrosion protection and hard coatings on cutting tools. In microelectronics,PVD by sputtering is the most common vacuum deposition technique to produce conductors, adhesionlayers, di↵usion barriers and resistors. Thinfilm technology is well established for microwave circuitson alumina, sapphire, fused silica etc. [3]. It is also the standard metallization for semiconductors. Ona perfectly flat and smooth surface resolution depends on the wavelength of the light used for exposureand of course on the performance of masks, photo-resist and exposure unit. It is also suitable forLTCC metallizations [165]. Since it is a vacuum deposition technique applied on sintered substrates,it is post-fired and limited to outer layers. 37

thixotropy

Thixotropic liquids are non-newtonian i.e. the relationship of shear stress and shear rate is nonlinearand time variant. The longer the shear stress is exerted, the lower the viscosity gets. Typical pastes forscreen printing are thixotropic, they are shear thinning and their viscosity is time dependent. Treaseand Dietz [226] have shown how the viscosity of a thick-film paste develops during the printing

Glossary 127

process. When the paste is spatulated in the jar and applied on the screen, viscosity is already reduced.The paste is rolling in front of the squeegee as it is pushed across the screen with the e↵ect thatviscosity is further reduced. When the paste is forced through the screen, the shear stress is maximumand viscosity reaches its minimum. When the screen separates from the substrate, no further shearstress is exerted, the paste is levelling and viscosity is raising. Due to the thixotropy of the paste,viscosity measurement will depend on the level of shear stress and the time it is applied. Exactcompliance with test conditions is a prerequisite for meaningful measurements. For thixotropic pastesthere is no easy way to convert results from one test method to the other. Viscosity of thick film pastesis temperature dependent, so this is a further parameter to keep constant for measurements and theprinting process. Pastes for LTCC are typically expensive due to their high content of noble metals.Therefore a measurement method is to be preferred which can handle very small samples and allowsthe complete reuse of the examined specimen. 67

tile

"Tile" is a colloquial expression for the panel in the production of ceramic substrates. 54, 81, 86

uniaxial lamination

For uniaxial lamination the LTCC stack is put inside a lamination tool. This tool provides also pinsfor mechanical registration. The whole assembly is placed in the press and brought in contact withthe press plates to heat it through. When the whole stack is at lamination temperature pressure isapplied. It is good practice to rotate the tool by 180° in the press after half the time to compensatefor a possible wedge error in the set-up. Nevertheless it is essential to have press plates and toolingexactly parallel. The LTCC stack itself can contribute to inhomogeneous distribution of laminationpressure when screen printed metallization, capacitors or resistors add up to height di↵erences. 62

uplink

In the context of satellite communications, the uplink (UL) designates the link from a ground stationto a satellite. i, iii, 10

V-band

V-Band is the frequency range from 40 to 75 GHz [225]. It is used for crosslink communicationbetween satellites and high data rate Wi-Fi applications. 54

via filling

After via formation the tapes for each layer have via holes which need to be filled with via fill pasteto establish electrical connection between adjacent layers. Usually a screen-printer is used to fill thevias. While conductors are typically printed with a screen, vias are filled with a stencil. This stencil ismade from thin (30 to 50 µm) metal foil. Stainless steel and nickel are the most common materialsfor stencils in the LTCC process. The stencil is mounted in a cast aluminum frame within a frameof polyester fabric to protect the thin foil against excessive strain. Stencil printing is performed incontact, i.e. snap-o↵ distance is set to zero. The LTCC tape is placed with a paper liner on top of aporous work holder. Vacuum is applied to keep the tape flat and in place. Vacuum helps also to fill thevias completely by evacuating air from the via holes during filling. Via holes are filled through theapertures of the stencil with via fill paste. Drying of the via fill paste in a convection oven or a conveyorbelt dryer will complete this process step. Increasing density of integration is not only demandingsmaller line-width but also smaller via diameters and catch pads. Micro vias (50 to 100 µm) are notonly a challenge for the via-formation process but also for via filling [227] and stacking. 4

128 Glossary

via formation

Vertical and electrically conductive connections between adjacent levels of the ceramic multilayer arecalled vias. They are realized by filling the via hole with appropriate conductor paste. A ComputerNumerically Controlled (CNC) Punch is the typical equipment to form vias. However, due to theorganic binder in the tape it is machinable in green state by drilling, milling, embossing [97] [98] [99],laser cutting and punching. Dedicated tapes with photo-sensitive organics allow vias to be definedin a photo-lithographic process [228]. Punching and laser cutting are the most common techniquesemployed for via formation [100] [101]. 47

voxel

The three-dimensional analogy of a pixel. In a 3-D grid each position is given a set of parameters(density, colour etc.), these elements are called voxels (volume elements). 54

Wilkinson power divider

Wilkinson suggests a matched and symmetric 1:N power splitter, that is lossless (assuming idealconductors) if all output ports are at equal phase and amplitude [109]. Fig. 6.1 shows a micro-striplinerealization of the 1:2 version. The black area in the photograph in Fig. 6.1 is the screen printed resistorof the depicted Wilkinson power splitter. The /4 lines transform the impedance of the parallel outputs≠ and Æ to match the impedance at port ¨. This impedance transformer can be replaced by multiple/4 line sections to increase the bandwidth of the component [151]. If operated as power divider withequal impedance at the output ports ≠ and Æ, no power is dissipated in the resistor R. If there is anunsymmetric signal, e.g. a reflection from a defective amplifier or antenna input, this is absorbed inthe resistor. By design the output ports are isolated. Both properties provide operating reliability forthe overall system. Particularly the 1:2 and 1:3 dividers are suitable for planar circuits. Furthermorethese components can be cascaded to form large beam forming networks (BFN) like the 1:6 powerdivider in Fig. 4.8 [74] or the 1:32 microstrip power divider which is used in the TerraSAR-X andTanDEM-X earth observation satellites [152]. Nagai proposes a similar 1:N divider [229] that can berealized as a planar microwave circuit without cross-overs. 30, 54, 67, 74, 75, 81

Own Publications

Related Publications[1] D. Pohle, C. Schulz, M. Oberberg, P. Uhlig, A. Serwa, P. Awakowicz, and I. Rolfes, “Progression of

the multipole resonance probe - advanced plasma sensors based on LTCC-technology,” in Proc.Eur. Microw. Conf., 2018.

[2] A. Wien, P. Uhlig, J. Kassner, C. Günner, and E. Noak, “Compact 30 GHz LTCC band pass filterdesign,” in ESA Intl. Workshop Microw. Filters 7th, 2018.

[3] D. Pohle, C. Schulz, M. Oberberg, P. Uhlig, A. Serwa, P. Awakowicz, and I. Rolfes, “An advancedhigh-temperature stable multipole resonance probe for industry compatible plasma diagnostics,” inProc. German Microw. Conf. 11th. mar 2018.

[4] I. Wol↵, C. Günner, J. Kassner, R. Kulke, and P. Uhlig, “New heights for satellites: LTCC multilayertechnology for future satellites,” IEEE Microw. Mag., vol. 19, no. 1, pp. 36–47, Jan 2018.

[5] D. Pohle, C. Schulz, M. Oberberg, P. Uhlig, A. Serwa, P. Awakowicz, and I. Rolfes, “Realisierung einertemperaturstabilen Multipol-Resonanz-Sonde in LTCC-Technologie fuer die Plasmadiagnostik,”Kleinheubacher Tagung (U.R.S.I. Landesausschuss in der Bundesrepublik Deutschland e.V.), 2017.

[6] P. Uhlig, J. Kassner, C. Günner, A. Serwa, and M. Tinnefeld, “Ausbeute bei kritischen Mikrowellen-modulen in LTCC und der Einfluss des Siebzustandes,” KOENEN Technologietag, Ottobrunn,2017.

Cited on page 57.

[7] P. Uhlig, A. Serwa, U. Altmann, T. Welker, J. Mueller, D. Schwanke, J. Pohlner, and T. Rittweg,“Liquid cooling in an LTCC-module for a switched mode amplifier,” in Proc. IMAPS Nordic Annu.Conf., jun 2017, pp. 90–96.

Cited on page 33.

[8] P. Uhlig, C. Günner, J. Kassner, and T. Klein, “Konzepte für Verteilnetzwerke in LTCC-Mehrlagentechnik,” DLR Workshop: Keramische Schaltungsträger - eine innovative Technologienicht nur für die Satellitenkommunikation?, 2016.

[9] P. Uhlig, C. Günner, J. Kassner, and R. Kulke, “Keramische Mehrlagenschaltungen für die neueGeneration von Kommunikationssatelliten,” in Proc. Symp. Keram. Mehrlagentech. der DKG,December 2016.

Cited on page 8.

[10] P. Uhlig, “Innovative keramische Schaltungsplattformen für künftige Satellitentechnologien - iKER-SATEC,” Schlussbericht des DLR/BMBF-Verbundprojektes "iKERSATEC"; FörderkennzeichenBMBF 50 YB 1307. - Verbund-Nummer 01138125, 2016.


[11] I. Wol↵, P. Uhlig, R. Kulke, C. Günner, and J. Kassner, “LTCC modules for future communicationsatellites in Ka-band,” IEEE MTT-S Int. Microw. Symp. Dig., 5 2016.

Cited on page 4.

[12] K. Wörho↵, A. Prak, F. Postma, A. Leinse, K. Wu, T. J. Peters, M. Tichem, B. Amaning-Appiah,V. Renukappa, G. Vollrath, J. Balcells-Ventura, P. Uhlig, M. Seyfried, D. Rose, R. Santos, X. J. M.Leijtens, B. Flintham, M. Wale, and D. Robbins, “Photonic hybrid assembly through flexiblewaveguides,” in Proc. SPIE, Silicon Photonics and Photonic Integrated Circuits V, L. Vivien,L. Pavesi, and S. Pelli, Eds., vol. 9891. SPIE, May 13 2016.

Cited on page 7.

129

130 Related Publications

[13] P. Uhlig, J. Kassner, C. Günner, and E. Noack, “Microwave filter design optimized for ceramicmultilayer technique,” in Proc. Int. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol.12th, 2016.

Cited on page 57.

[14] J. Balcells-Ventura, T. Klein, P. Uhlig, C. Günner, and R. Kulke, “Tolerance-optimized rf structuresin LTCC for mm-wave frequencies applications,” in Proc. Intl. Conf. Exhib. Ceram. InterconnectCeram. Microsyst. Technol., 4 2015.

[15] J. Balcells-Ventura, J. Leiss, P. Uhlig, M. Ihle, S. Ziesche, A. Bisognin, and C. Luxey, “Substrateintegrated MM-wave antenna utilizing metalized cavity walls,” in Proc. Intl. Conf. Exhib. Ceram.Interconnect Ceram. Microsyst. Technol., 2015, pp. 221 – 228.

Cited on page 91.

[16] J. Balcells-Ventura, T. Klein, P. Uhlig, C. Günner, and R. Kulke, “Tolerance-optimized RF structures inLTCC for mm-wave frequencies applications,” J. Ceram. Sci. and Technol., vol. 6, no. 4, 2015.

Cited on page 74.

[17] M. Geissler and P. Uhlig, “Technologie-Konzepte für planare Satellitenkommunikation-Antennen,”SMT-ASIC-Packaging Nürnberg, 6. Mai 2015 Kongress Halbtag -Hochfrequenzbaugruppen-, 2015.

[18] P. Uhlig, D. Stöpel, J. Müller, and S. Mosch, “Feinleiter-Druck in keramischen Mehrlagenschaltungen(LTCC),” Hochschultag der Firma KOENEN, 2014.

[19] T. Klein, P. Uhlig, C. Günner, and R. Kulke, “Substrate-integrated divider networks in LTCC withoptimized tolerance / isolation properties for Ka-band satellite systems,” in Proc. Intl. Symp. Microel.47th, 2014.

Cited on page 74.

[20] J. Balcells-Ventura, J. Leiss, M. Ihle, S. Ziesche, and P. Uhlig, “Aerosol-printed horn-shaped antennaon LTCC,” in Proc. Eur. Conf. Antennas Propag. 7th, April 2013, pp. 2461–2464.

[21] J. Leiss, J. Balcells, P. Uhlig, and M. Sohling, “Umsetzung eines Mikrowellen- Sensormoduls inkonventioneller Siebdrucktechnik und Aerosoldruck,” in Proc. IMAPS Germany Annu. Conf.,October 2012.

[22] P. Uhlig, D. Stöpel, S. Mosch, and J. Müller, “Fine line conductors in LTCC,” in Proc. IMAPS NordicAnnu. Conf., 2012, pp. 124 –132.


[23] B. Sanadgol, S. Holzwarth, P. Uhlig, A. Milano, and R. Popovich, “60 GHz SIW steerable antennaarray in LTCC,” ZTE Commun., vol. 10, no. 4, pp. 29 – 60, 2012.

Cited on page 9.

[24] P. Uhlig, S. Holzwarth, B. Sanadgol, and A. Serwa, “Three-dimensional surface in LTCC for aMM-wave antenna,” in Proc. Intl. Symp. Microel. 44th, 10 2011.


[25] P. Uhlig, J. Leiß, and D. Köther, “Syntactic foam - a new dielectric for beam forming networks in spaceapplications,” VDE / ITG Diskussionssitzung des FA 9.1 "Messverfahren der Informationstechnik"zum Thema Materialcharakterisierung, Ruhr-Universität Bochum, March 2011.

[26] P. Uhlig, C. Günner, T. Klein, and R. Kulke, “LTCC Materialsysteme fuer Mikrowellenanwendungen:Umstellung von DuPont 943 auf 9k7 bei einem Ka-Band Beam-Forming-Network,” in Proc. IMAPSGermany Annu. Conf., 10 2011.

Cited on page 80.

[27] J.-P. Sommer, C. Günner, P. Uhlig, R. Kulke, and B. Michel, “Design of low loss beam formingnetworks - supported by numerical simulations and material characterisation,” in Proc. Intl. Conf.Electron. Packag. Technol. High Dens. Packag. 11th, 2010.

Related Publications 131

[28] R. Kulke and P. Uhlig, “LTCC technology for RF applications,” in Proc. Elect. and Electron. Eng. forCommun., 2010.

[29] W. Simon, J. Kassner, O. Litschke, H. Fischer, S. Holzwarth, and P. Uhlig, “Highly integrated Kaband antenna array,” IEEE MTT Soc. of the Long Island Sec., Lecture, June 2nd 2010. [Online].Available: http://www.ieee.li/mtt/index.htm

[30] M. Martinez-Vazquez, A. Bettray, R. Baggen, S. Holzwarth, O. Litschke, C. Oikonomopoulos-Zachos,B. Sanadgol, W. Simon, and P. Uhlig, “Challenges in practical design of planar arrays,” in Proc.Eur. Conf. Antennas Propag. 4th, April 2010.

[31] P. Uhlig, M. Geissler, S. Holzwarth, J. Leiss, D. Manteu↵el, and M. Martinez-Vazquez, “Anten-nenintegration in keramischen mehrlagenschaltungen,” PLUS, vol. Jan., pp. 166 – 175, Januar2010.

[32] B. Lopez-Berrocal, J. de Oliva-Rubio, E. Marquez-Segura, A. Moscoso-Martir, I. Molina-Fernandez,and P. Uhlig, “High performance 1.8-18 Ghz 10 dB low temperature co-fired ceramic directionalcoupler,” Progr. Electromagn. Res., vol. 104, pp. 99 – 112, 2010.

Cited on page 57.

[33] D. Manteu↵el, M. Arnold, and P. Uhlig, “Considerations on configurable multi-standard antennas formobile terminals realized in LTCC technology,” Radioengineering, vol. 18, no. 4, pp. 394 –, 122009.

[34] R. Follmann, R. Kulke, S. Holzwarth, P. Uhlig, and M. Rittweger, “LTCC packaging and integrationtechnologies,” in Proc. Eur. Microw. Conf., 10 2009.

Cited on page 27.

[35] P. Uhlig, S. Holzwarth, and M. Geissler, “Antenna concepts for ceramic multilayer modules,” in Proc.Int. Conf. Exhib. Ceram. Interconnect Ceram. Microsyst. Technol. 5th. 4 2009.

Cited on page 4.

[36] J.-P. Sommer, B. Michel, E. Noack, B. Seiler, and P. Uhlig, “Finite element simulation and microdeformation measurements - contributions to the development of advanced packages with hiddendies,” Smart Syst. Integr. 3rd, March 2009.

[37] R. Kulke, G. Möllenbeck, C. Günner, P. Uhlig, K. H. Drüe, S. Humbla, J. Müller, R. Stephan,D. Stöpel, J. F. Trabert, G. Vogt, M. A. Hein, A. Molke, T. Baras, A. F. Jacob, D. Schwanke,J. Pohlner, A. Schwarz, and G. Reppe, “Ceramic microwave circuits for satellite communication,”J. Microelectron. Electron. Packag., vol. 6, pp. 27 – 31, 2009.

Cited on page 8.

[38] R. Kulke, C. Günner, G. Möllenbeck, P. Uhlig, and M. Rittweger, “Protoflight model development forspaceborne LTCC RF-modules,” in Proc. Intl. Symp. Microel. 41st, Nov 2008, pp. 1001–1006.


[39] P. Uhlig, E. K. Polzer, E. Mclean, and G. Vanrietvelde, “LTCC packages for microwave applications,”IWPC Workshop - Next Generation Automotive Radar, Oct 2008.

[40] P. Uhlig, R. Follmann, O. Kersten, T. Kohl, R. Kulke, and G. Moellenbeck, “A 20 Ghz fractional-nsynthesizer module for satellite operation,” in Proc. IMAPS Nordic Annu. Conf., Sept 2008, pp.124–128.

[41] R. Kulke, G. Möllenbeck, P. Uhlig, and K. Maulwurf, “Entwurf und Optimierung von Tiefpass Filternals SMD-Bauteil in LTCC,” PLUS, vol. 10, Juni 2008, pp. 1244 – 1251.

[42] P. Uhlig, D. Manteu↵el, and S. Malkmus, “High layer count in LTCC dual band antenna for GalileoGNSS,” J. Microelectron. Electron. Packag., vol. 5, pp. 161 – 168, April 2008.


http://www.ieee.li/mtt/index.htm

132 Related Publications

[43] R. Kulke, G. Möllenbeck, C. Günner, P. Uhlig, and M. Rittweger, “LTCC multi-chip modules forKa-band multimedia satellite technology,” in Proc. German Microw. Conf., March 2008, pp. 1 –4.

[44] R. Kulke, G. Möllenbeck, P. Uhlig, K. Maulwurf, and M. Rittweger, “Designing SMD lowpass filtersin multilayer LTCC technology,” in Proc. Eur. Microel. Packag. Conf., June 2007, pp. 40–44.

Cited on page 9.

[45] P. Uhlig, R. Kulke, G. Möllenbeck, and K. Maulwurf, “Entwurf und Optimierung von Tiefpass Filternals SMD-Bauteil in LTCC,” in Proc. IMAPS Germany Annu. Conf., IMAPS Germany. IMAPSGermany, 2007.

[46] I. Wol↵, R. Kulke, P. Uhlig, and T. Mobley, “LTCC for micro- and millimeter-wave applications,” ShortCourse for IEEE MTT-S Int. Microw. Symp., 2007.

[47] J. Müller, R. Perrone, K.-H. Drüe, R. Stephan, J. Trabert, M. Hein, D. Schwanke, J. Pohlner, G. Reppe,R. Kulke, P. Uhlig, A. F. Jacob, T. Baras, and A. Molke, “Comparison of high-resolution patterningtechnologies for LTCC microwave circuits,” J. Microelectron. Electron. Packag., vol. 4, no. 3, 2007.

[48] R. Kulke, O. Kersten, J. Winkler, C. Günner, G. Möllenbeck, P. Uhlig, and M. Rittweger, “Packagedmicrowave components for Ka-band multimedia satellite communication: Amplifier, oscillator andswitch modules,” in Proc. MacroNano-Colloq. LTCC RF Microsyst. Interconn. 1st, 2006.

[49] R. Kulke and P. Uhlig, “Shielded waveguides for microwave applications in LTCC,” in Proc.MacroNano-Colloq. LTCC RF Microsyst. Interconn. 1st, November 2006.

[50] P. Uhlig, S. Holzwarth, O. Litschke, A. Serwa, and D. T. Tran, “On the influence of layer-to-layermisalignment on the microwave performance of LTCC antenna modules,” in Proc. Intl. Symp.Microel. 39th, 10 2006.

Cited on page 47.

[51] ——, “The influence of layer-to-layer misalignment on the microwave performance of LTCC antennamodules,” in Proc. Internat. Wiss. Kolloq. Ilmenau 51st, 2006.

[52] H. Thust, R. Perrone, J. Mueller, M. Hein, J. Trabert, C. Kutscher, R. Stephan, D. Schwanke, J. Pohlner,G. Reppe, R. Kulke, P. Uhlig, A. Jacob, and T. Barras, “Technology benchmarking of high resolutionstructures on LTCC for microwave circuits,” in Proc. Elect. Syst. Integr. Technol. Conf. 1st, vol. 1,September 2006, pp. 111 – 117.

[53] J. Müller, R. Perrone, H. Thust, K.-H. Drüe, C. Kutscher, R. Stephan, J. Trabert, M. Hein, D. Schwanke,J. Pohlner, G. Reppe, R. Kulke, P. Uhlig, A. F. Jacob, T. Baras, and A. Molke, “Technologybenchmarking of high resolution structures on LTCC for microwave circuits,” in Proc. Elect. Syst.Integr. Technol. Conf. 1st, September 2006, pp. 111 –117.

[54] P. Uhlig, R. Kulke, D. Köther, J. Kassner, A. Lauer, and M. Rittweger, “Power distribution networks inspace applications,” in Proc. Intl. Hi-Tech 3D Packag. Workshop, 6 2006.


[55] R. Kulke, C. Günner, S. Holzwarth, J. Kassner, A. Lauer, M. Rittweger, P. Uhlig, and P. Weigand,“24 GHz radar sensor integrates patch antenna and frontend module in single multilayer LTCCsubstrate,” in Proc. Eur. Microel. Packag. Conf. 15th, June 2005, pp. 239–242.

[56] R. Kulke, D. Koether, P. Uhlig, J. Kassner, A. Lauer, and M. Rittweger, “Technologies for reliablepower distribution networks in SAR earth observation and multimedia communication satellites,” inProc. ESA Antenna Workshop on Space Antenna Syst. and Technol. 28th, June 2005, pp. 198–204.

[57] P. Uhlig, S. Holzwarth, O. Litschke, W. Simon, and R. Baggen, “A digital beam-forming antennamodule for a mobile multimedia terminal in LTCC-multilayer technique,” in Proc. Eur. Microel.Packag. Conf. 15th, 6 2005.

Cited on page 17.

[58] P. Uhlig, C. Günner, S. Holzwarth, J. Kassner, R. Kulke, A. Lauer, and M. Rittweger, “LTCC shortrange radar sensor for automotive applications at 24 GHz,” Advancing Microelectronics, vol. 32,no. 2, pp. 10 – 12, 2005.

Cited on page 8.

[59] P. Uhlig and R. Kulke, “Flexibles Prototyping von LTCC-Schaltungen für Mikrowellenanwendungendurch die Laserbearbeitung von Grünfolien,” Fortschrittsber. der DKG, vol. 81, pp. 75 – 82,Dezember 2004, und "Technische Keramische Werksto↵e", 90. Ergaenzungslieferung, Januar 2006.

[60] S. Holzwarth, R. Kulke, J. Kassner, and P. Uhlig, “Antenna integration on LTCC radar module forautomotive applications at 24 ghz,” in Proc. IMAPS Nordic Annu. Conf., September 2004.

[61] M. Rittweger, R. Kulke, and P. Uhlig, “Multichip module design and fabrication in LTCCmultilayer technology,” March 2004, online, accessed 2-June-2012. [Online]. Available:http://www.ltcc.de/downloads/rd/pub/ltcc_at_imst.pdf

[62] P. Uhlig, C. Günner, S. Holzwarth, J. Kassner, R. Kulke, A. Lauer, and M. Rittweger, “LTCC shortrange radar sensor for automotive applications at 24 Ghz,” Proc. Intl. Symp. Microel. 37th, 2004.

[63] D. Köther, A. Lauer, A. Wien, P. Uhlig, G. Pautz, G. Möllenbeck, and J. Berben, “X-band 1:32 dividernetwork for space application with excellent amplitude and phase balance,” in Proc. Eur. Microw.Conf. 33rd, 10 2003, pp. 899–902.


[64] J. Kassner, R. Kulke, P. Uhlig, M. Rittweger, P. Waldow, R. Münnich, and H. Thust, “Highly integratedpower distribution networks on multilayer LTCC for Ka-band multiple-beam phased array antennas,”in Proc. IMAPS Nordic Annu. Conf., September 2003, pp. 51 – 54.

[65] R. Kulke, V. Wahle, D. Sollbach, P. Uhlig, M. Rittweger, S.-P. Schmitz, and P. Waldow, “High level ofintegration for bluetooth modules on LTCC,” in Proc. Eur. Microel. Packag. Conf., 2003, p. 16.

[66] R. Kulke, M. Rittweger, P. Uhlig, and C. Günner, “LTCC-Mehrlagenkeramik für Funk- und Sensor-Anwendungen,” PLUS, pp. 2131–2136, 12 2001.

Cited on page 30.

[67] R. Kulke, W. Simon, G. Möllenbeck, J. Kassner, P. Uhlig, S. Holzwarth, and P. Waldow, “Powerdistribution networks in multilayer LTCC for microwave applications,” in Proc. Intl. Symp. Microel.,October 2001, pp. 321–324.

Patent[1] C. Günner, T. Klein, and P. Uhlig, “Hochfrequenzmodule und Hochfrequenzsysteme,” German Patent

DE102 013 112 647A1, 2015.

Other Publications[1] P. Uhlig, J. Leiss, R. Marek, J.-P. Sommer, and H. Wolf, “Syntactic foam - a new approach to beam

forming networks,” in Proc. IMAPS Nordic Annu. Conf., September 2009, pp. 107 – 111.

[2] P. Uhlig, H. Wolf, R. Marek, and J.-P. Sommer, “VERSA: Leichte und verlustarme Verteilnetzwerkefuer Satellitenanwendungen,” Heinrich Hertz Mission, Workshop zur Technologieverifikation 8./9.Juni 2009, TESAT Backnang, 2009.

133

http://www.ltcc.de/downloads/rd/pub/ltcc_at_imst.pdf

134 Other Publications

[3] P. Uhlig, J. Leiss, R. Marek, J.-P. Sommer, and H. Wolf, “Light weight - low loss beam forming networksfor space applications,” in Proc. Eur. Conf. Antennas Propag. 3rd, march 2009, pp. 740 – 744.

[4] P. Uhlig and M. Geissler, “VERSA - Verteilnetzwerke fuer Satellitenanwendungen,” DLR- Workshop"Aktive Antennen" am 22. und 23. November 2005 in Oberkassel, 11 2005.

[5] D. Köther, A. Lauer, A. Wien, P. Uhlig, G. Pautz, G. Möllenbeck, and J. Berben, “Hochpräzise X-Band-Verteilnetzwerke für Erderkundungssatelliten,” in Proc. IMAPS Germany Annu. Conf., 2004.

[6] ——, “Advanced techniques for the production of high precision X-band power distribution / combina-tion networks,” in Proc. Eur. Microel. Packag. Symp. 3rd, June 2004.

[7] ——, “Reliable 1:32 power divider for space application,” in Proc. ESA Microw. Technol. and Techn.Workshop. Microwave and Millimetre-Wave Section (TOS-ETM) of the European Space Agency(ESA), May 2004.

Documents

LTCC Technology for Planar Microwave Antenna Systems · LTCC is a well established ceramic multilayer packaging technology for high reliability applications. Advanced low-loss material