J. Sens. Sens. Syst.

, 5, 73–83, 2016
www.j-sens-sens-syst.net/5/73/2016/
doi:10.5194/jsss-5-73-2016
© Author(s) 2016. CC Attribution 3.0 License.

Joining technologies for a temperature-stable integration

of a LTCC-based pressure sensor
J. Schilm1 , A. Goldberg1 , U. Partsch1 , W. Dürfeld2 , D. Arndt2 , A. Pönicke3 , and A. Michaelis1
1 Fraunhofer IKTS, Fraunhofer Institute for Ceramic Technologies and Systems, Winterbergstraße 28,
01277 Dresden, Germany
2 ALL IMPEX GmbH, Bergener Ring 43, 01458 Ottendorf-Okrilla, Germany
3 Modine Europe GmbH, Arthur-B.-Modine-Straße 1, 70794 Filderstadt, Germany

Correspondence to: J. Schilm ([email protected])

Received: 26 October 2015 – Revised: 13 January 2016 – Accepted: 26 January 2016 – Published: 9 March 2016

Abstract. Besides the well-known application as circuit boards and housings, multilayer low-temperature co-
fired ceramics (LTCC) offer a flexible and temperature-stable platform for the development of complex sensor
elements. Commercial LTCC qualities are usually available with a matching set of metallization pastes which
allow the integration of various electrical functions. However, for the integration of ceramic sensor elements
based on LTCC into standardized steel housings it is necessary to compensate the mismatching thermal expan-
sion behaviour. Therefore balancing elements made of Kovar® (Fe–29 wt% Ni–17 wt% Co) and alumina ceramic
(Al2 O3 ) can be used. These components have to be joined hermetically to each other and to the LTCC sensors.
In this study, brazing experiments were performed for combinations of Kovar–Al2 O3 and Kovar–LTCC with
Ag–Cu–Ti- and Ag–Cu–In–Ti-based commercial braze filler metals, Cusil-ABA® and Incusil® -ABA, respec-
tively. For both active braze filler metals, optimized processing parameters were investigated to realize hermetic
Kovar–Al2 O3 and Kovar–LTCC joints.

1 Introduction printed on the surface of the sensor cell, are connected to

a Wheatstone bridge and measure the strain caused by the
Low temperature co-fired ceramics (LTCC) is a well- deflection of the LTCC diaphragm due to pressure differ-
known technology for highly integrated, reliable, and high- ences in- and outside of the cell. Especially this sensor setup
temperature-stable microelectronic packages in mobile com- has been used for this study on a temperature-stable inte-
munication or for automotive, space, or medical applications gration attempt. The LTCC technology easily enables vari-
(Peterson et al., 2008). Due to its linear stress/strain be- able sensor geometries, e.g. different diaphragm thicknesses
haviour and its ability for integration of three-dimensional for different pressure ranges by using different tape thick-
shapes like diaphragms, channels, and cavities, according to nesses and thus all types of pressure sensors (relative, abso-
various authors, LTCC complies with all requirements for the lute, differential) can be constructed. According to Fournier
integration of mechanical structures, e.g. for pressure sen- et al. (2010) in comparison with classic steel- or ceramic-
sors (Zarnik et al., 2010; Partsch et al., 2012; Fournier et al., based pressure sensors, such LTCC-based pressure sensors
2010). A new piezo-resistive pressure sensor concept was de- have many advantages like a very linear pressure-dependent
veloped by Partsch et al. (2007). Figure 1 shows a overview signal behaviour together with an excellent high-temperature
of a completely assembled LTCC-based pressure sensor with and long-term stability. Furthermore, LTCC is a high-volume
housing, pressure port and wiring. This new design princi- technology which helps to produce sensor elements in a cost-
ple allows the sensor cell to be fully mechanically decoupled effective manner as all components of the sensor system
and stress-free inside the sensor frame. Only thin LTCC can- (sensor frame, electronic components) can be integrated in
tilevers containing microchannels are used for the pressure one LTCC-based multilayer substrate. However, for testing,
connection of the sensor cell. Thick film resistors, screen-

Published by Copernicus Publications on behalf of the AMA Association for Sensor Technology.
74 J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC)

Figure 2. Single LTCC pressure and steel connect (a) and scheme
for the stepwise integration of a LTCC-based sensor element into
the steel connect (b).

electron beam welding to other devices or which may be a

more complex device itself. The following step B is required
in order to create a joinable surface for step C and will be
described later in detail. In step B the bonding of a ceramic
layer made from alumina (Al2 O3 ) or LTCC offers the pos-
Figure 1. LTCC pressure transmitter and corresponding drawing of sibility to integrate the LTCC-based sensor elements in step
inner setup. C by sealing with a glass-based solder or other techniques.
A convenient way for the implementation of step B is the
so-called active metal brazing process.

calibration, and subsequent application, the LTCC sensor ele- 1.1 Joining of ceramics to metals by brazing (step B)
ments have to be connected gas-tight to the measuring equip-
ment. In most cases standardized steel connectors are used to Fernie et al. (2009) describes several direct bonding tech-
ensure a gas-tight connection of the LTCC sensors to the sys- niques to join ceramics and metals together for hermetic
tem. But, the integration of LTCC-based sensor elements for joints. Besides techniques without any liquid phase like dif-
high-temperature applications requires suitable interconnect fusion bonding or friction welding, direct bonding methods
technologies. For example, if the LTCC sensors and the steel utilizing a liquid phases based on adhesives, braze filler met-
connectors are glued directly together, the maximum opera- als, or glass solders can be used. For the desired application,
tion temperature of the sensor is limited to the glass transi- where in joining step B (Fig. 2) a planar joint between Kovar
tion temperature of the epoxy resin, which is in most cases and Al2 O3 or LTCC has to be realized which is suitable for
below 190 ◦ C. Also soldering offers no reliable option as the operation temperatures up to 300 ◦ C, brazing is the appropri-
joint strength decreases rapidly at higher temperatures due ate method. According to Nascimento et al. (2003) a com-
to interdiffusion processes, which in turn results in a loss of mon way to braze ceramics and metals is the metallization of
gas-tightness. For an increased thermal stability other join- the ceramic prior to the brazing process since then the met-
ing technologies like glass sealing or brazing can be used. allized ceramics can be brazed to metals without any active
However, in this case the mismatching coefficients of ther- braze filler metal. However, the metallization processes im-
mal expansion of LTCC and steel will limit the lifetime of ply several individual process steps, which makes them com-
the integrated sensor elements as thermal cycles will initi- plicated and expensive. In contrast to this quite old technol-
ate cracks along the sealing or inside the sensor. To over- ogy Walker and Hodges (2008) describe active metal braz-
come these problems, a stepwise integration of the LTCC- ing as a technique which allows brazing ceramics directly
based sensor elements into steel connectors was developed. to metals or themselves without any additional metallization
This approach offers the opportunity to outbalance the dif- steps. Active brazing alloys are based on filler metals like
ferent thermal expansion coefficients of LTCC and steel, and Ag, Ag–Cu, or Au and contain low fractions of so-called ac-
to increase the operation range of the sensor assembly to tive species (i.e. Ti, Zr, Hf). These elements enhance wet-
high temperatures up to 300 ◦ C. The integration concept is ting of the ceramic surface during brazing in an oxygen-
schematically shown in Fig. 2. The integration approach of free environment using protective atmospheres or vacuum
LTCC-based sensors can be divided into three steps. In step (p<10−4 mbar).
A a balancing element made of a nickel–cobalt ferrous alloy, In the literature active metal brazing of Kovar to Al2 O3
i.e. Kovar® (Fe–29 wt% Ni–17 wt% Co) with a coefficient of is much more extensively investigated than the brazing of
thermal expansion that closely matches that of the ceramic Kovar to LTCC. A good overview of the problem can be
materials at low temperatures, is brazed to the steel hous- found in Walker and Hodges (2013). When Al2 O3 ceram-
ing. Such a steel connection produced from cost-efficient and ics are brazed to themselves with Ag–Cu–Ti active braze
construction steel types are necessary for having a standard filler metals, the joint microstructure shows a bi-layered re-
interconnection interface which can be processed easily by action zone at the interface between the Al2 O3 and the braz-

J. Sens. Sens. Syst., 5, 73–83, 2016 www.j-sens-sens-syst.net/5/73/2016/

J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC) 75

ing alloy. This reaction layer consists of a Ti-rich oxide layer mina, less information on brazing of LTCC is available. One
with a thickness below 1 µm completely covering the alu- approach to braze LTCC is to use a metallized LTCC in com-
mina interface and a second, a few-microns-thick mixed ox- bination with a non-active braze filler metal. For this purpose
ide layer containing Ti, Cu, and Al (Stephens et al., 2003; Keusseyan and Dilday (1993) investigated the brazeability of
Lin et al., 2014). Al2 O3 –Al2 O3 joints prepared in this man- Cu-, Ag-, and Au-based thick film metallization layers and
ner are hermetic and reach joint strengths of > 95 % of the concluded that for LTCC–metal joints the brazing temper-
strength of the base material. If now one of the Al2 O3 pieces ature should be limited to 500 ◦ C in order to minimize the
is substituted by Kovar, the results change. Hahn et al. (1998) mechanical stresses caused by the mismatching thermal ex-
demonstrated in their work that Kovar–Al2 O3 joints achieved pansion coefficients. Another approach, followed by Walker
no high strength values and reached only 40 % of the ini- et al. (2006), was the investigation of PVD thin-film coat-
tial Al2 O3 –Al2 O3 joint strength. Additionally, Kovar–Al2 O3 ings like Ti–Au, Ti–Pt, and others. While brazing with a Ag–
joints made by active metal brazing showed poor hermetic- Cu–In braze filler metal, the Ti–Pt thin films yielded the best
ity. Stephens et al. (2000) revealed by microstructural analy- hermeticity performance and highest strength. However, on
sis that in non-hermetic joints no continuous Tix Oy reaction active metal brazing of LTCC without further metallization
layers were formed at the interface between Al2 O3 and the layers only one study was found in the literature. Further-
brazing alloy. According to Vianco et al. (2003a) this phe- more Walker et al. (2006) stated that hermetic joints of LTCC
nomenon of titanium scavenging can be explained by the dis- and Kovar were only be possible if the LTCC was ground
solution of nickel and iron from Kovar in the molten braze and re-fired prior to the brazing process. As shown, brazing
and the strong affinity of nickel to titanium. Arróyave and of LTCC only is less investigated, and no systematic results
Eagar (2003) describe that during the brazing process nickel are given in the literature which let one judge about the suit-
and titanium react to form intermetallic compounds while the ability of the active metal brazing technique for LTCC. In
activity of titanium in the melt is decreased and the formation the case of Al2 O3 , besides Ag–Cu–Ti active braze filler met-
of the necessary reaction zone at the Al2 O3 interface is sup- als, no brazing alloys with lower brazing temperatures were
pressed. To improve the joint strength of Kovar–Al2 O3 and to tested. Thus in the present study the commercially available
prevent the formation of intermetallic compounds, different active braze filler metal Incusil® -ABA is investigated to pro-
barrier layer concepts were developed. Mo, Ni, and Mo–Ni vide brazing parameters as a means to obtain hermetically
coatings on Kovar were tested by Hahn et al. (1998). They re- brazed Kovar–Al2 O3 and Kovar–LTCC joints. For compari-
port an increased bending strength of the joints by more than son with the literature, joining of Al2 O3 and LTCC to Kovar
80 % of the uncoated base material. Vianco et al. (2003b) in- with Cusil-ABA® was investigated as well.
vestigated the influence of Mo thickness and found that braz-
ing of Kovar–Al2 O3 with a 500 µm thick Mo barrier layer
2 Methods and materials
yielded the best hermeticity performance and strength. Mag-
netron sputtering of titanium layers on Al2 O3 was introduced 2.1 Ceramic materials
by Zhu et al. (2014) as an alternative method to improve
the joint strength and gas-tightness of Kovar–Al2 O3 joints. The LTCC sensors are based on DuPont’s GreenTape 951
The mechanical metallization of alumina surfaces was ap- system, as this material system is a fair compromise in
plied by Nascimento et al. (2007) to achieve a proper pre- comparison to other ceramic co-firing materials regarding
metallization of the ceramic component. Besides introducing Young’s modulus and fracture strength. While the exact com-
barrier layers, Wielage et al. (2012) showed that an improve- position of this LTCC quality is not published by DuPont,
ment of the joints is possible if induction brazing with much within this work it is important to know that the main crys-
shorter brazing times compared to conventional furnace braz- talline phase consists of Al2 O3 grains which are bonded by
ing is applied. Microstructural analysis showed a remarkable a PbO-based glass frit. For joining experiments LTCC sam-
reduction of intermetallic compounds and a decrease of the ples were made by laminating three tape layers followed by
reaction layer thickness between Kovar and the brazing alloy. a firing process with a peak temperature of 850 ◦ C similar
For the desired application and workflow in step B (Fig. 2) to the one described by Fournier et al. (2010). After firing
it seems to be interesting to use LTCC as a joining partner the sintered LTCC had a thickness of 630 µm and was cut
for Kovar instead of Al2 O3 . The main reason for the use into single samples (7 × 7 mm2 ) with a dicing saw. The sur-
of LTCC is given by minimized thermomechanical stresses face roughness of the as-fired LTCC was Ra < 0.36 µm. The
in the case of using a LTCC balancing element because it alumina ceramic was obtained in thick film standard qual-
has the same coefficient of thermal expansion (CTE) value ity with an Al2 O3 content of 96 % (Rubalit® 708 S, Cer-
as the sensor element itself. Another reason comes from the amTec, Marktredwitz, Germany), a thickness of 250 µm, and
idea that it could be possible to braze the LTCC sensor di- a surface roughness of Ra < 0.36 µm. Samples were prepared
rectly on the Kovar element if the required brazing tempera- by laser scribing and subsequent breaking along the scribed
ture is low enough to avoid changes in the microstructure of lines. As metallic joining partners, balancing elements made
the complex LTCC sensor. However, in comparison to alu- of Kovar were prepared from a massive rod according to the

www.j-sens-sens-syst.net/5/73/2016/ J. Sens. Sens. Syst., 5, 73–83, 2016

76 J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC)

Table 1. Active metal brazing filler metals with their compositions and brazing temperatures.

Brazing filler metal Ag (wt%) Cu (wt%) In (wt%) Ti (wt%) Brazing temperature

Incusil-25-ABA 43.6 29.1 24.3 3.0 650 ◦ C
Incusil-ABA 59.0 27.25 12.5 1.25 755 ◦ C
Cusil-ABA 63.0 35.25 1.75 810–850 ◦ C

sensor setup indicated in Fig. 1. The bottom side of the ele-

ment will be brazed to the steel interconnect and the cavity in
the upper side a ceramic balancing substrate made of Al2 O3
or LTCC will be brazed by an active metal brazing process.
The Kovar components additionally contain a channel struc-
ture that allows access of the pressurized gas to the sensor
membrane. The surfaces of all samples were degreased prior
to assembly and brazing processes.

2.2 Brazing and glass sealing

In accordance with the previously described integration strat-

egy, two brazing processes – one for joining the steel con-
nect to Kovar and one for joining Kovar to Al2 O3 – have
been selected. A one-step brazing process appears to be more
favourable but offers fewer opportunities for a process con- Figure 3. High temperature pressure measurement station for char-
trol regarding the hermeticity of the different interfaces. For acterizing sensors up to 200 bar and 650 ◦ C.
this reason the brazing processes were separated. For the first
brazing joint between steel and Kovar, a nickel-based braz-
ing foil (MBF-20 from Metglas Inc., Conway, SC, USA) is temperature at a rate of 5 K min−1 to 400 ◦ C with subsequent
applied which has a liquidus temperature well above the ac- furnace cooling. The brazing temperatures were varied be-
tive filler metal brazes used for the second brazing process. tween 810 and 850 ◦ C for Cusil-ABA in order to investigate
For this purpose the MBF-20 brazing foil was cut by a laser the influence of brazing temperature on gas-tightness and mi-
process into shapes matching strictly to the joined side of the crostructure of the joined assemblies. For the final integra-
components. The foils were placed in between both compo- tion of the LTTC sensor into this prepared steel connector, a
nents adjusted with an additional load on top of the arrange- commercially available, lead-free sealing glass from ASAHI
ment and brazed in vacuum (<1 × 10−5 mbar) with the fol- (4115DS-NY01) supplied as a ready-to-use paste with an ap-
lowing brazing cycle: from room temperature at 5 K min−1 to propriate firing profile having a peak temperature of 500 ◦ C
940 ◦ C with a hold for 15 min in order to achieve a homoge- for use in muffle furnace was screen-printed on the back side
nous furnace temperature and than again with 5 K min−1 up of the LTCC sensor element. For the joining process the sen-
to the brazing peak temperature of 1055 ◦ C with an addi- sor was placed on the ceramic balancing element together
tional hold time of 15 min. Cooling down to room temper- with a mechanical load of 10 g.
ature was conducted at 3 K min−1 .
For joining of the ceramics (Al2 O3 and LTCC) to Kovar, 2.3 Characterization
three types of braze filler metals provided by Wesgo Met-
als (Hayward, CA, USA) were used, which are listed in Ta- Upon brazing, the gas-tightness of each brazed assem-
ble 1. These alloys were applied in the form of laser cut bly in each integration step was measured using a helium
foils with a thickness of 50 µm. Brazing was carried out in leak detector (Phoenix XL30, Oerlikon Leybold Vacuum,
a full-metal vacuum furnace with molybdenum heating el- Cologne, Germany). A joint with a helium leak rate of
ements at a pressure <1 × 10−5 mbar. The following pro- 1 × 10−8 mbar s−1 or better was considered as hermetic. Mi-
cess cycle was used for brazing: heating from room temper- crostructural analysis of the joints was performed on pol-
ature to 550 ◦ C (for Incusil-ABA) or 700 ◦ C (Cusil-ABA) at ished cross-sectioned samples using a scanning electron mi-
a rate of 10 K min−1 , holding for 20 min to obtain a temper- croscope (abbreviation: SEM; NVision 40, Carl Zeiss SMT,
ature homogenization inside the furnace, further heating up Oberkochen, Germany). The micrographs shown in this pa-
to the desired brazing temperature (Table 1), holding of braz- per were recorded in the element specific back-scattered elec-
ing temperature for 10 min, and then cooling down to room tron mode. Additionally, the scanning electron microscope

J. Sens. Sens. Syst., 5, 73–83, 2016 www.j-sens-sens-syst.net/5/73/2016/

J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC) 77

Figure 4. Scanning electron micrographs of steel (1.4542)–Kovar joints brazed with MBF-20 showing an overview in (a) and details of the
microstructure in (b).

is equipped with an energy dispersive X-ray analysis sys- 3.1 Brazing of Kovar and steel
tem (abbreviation: EDX; Inca x-sight, Oxford Instruments,
Brazing of the balancing Kovar element into the steel hous-
Abingdon, England), which allows for a quantitative detec-
ing is the first step of the integration procedure. In accor-
tion of elements.
dance with brazing temperatures which are required for ac-
The characteristic of the pressure sensor is performed by a
tive filler braze between 750 and 850 ◦ C, it is necessary to
newly developed pressure measurement system. The sensors
perform this brazing step at a higher temperature which lies
can be measured in a special chamber oven KU70/07-A of
well above the formerly mentioned one. One must take care
the company THERMCONCEPT using a high-temperature-
on stability of the Kovar alloy and potential reactions with the
capable pressure rail and a ceramic-insulated electrical
filler braze. Based on these boundary conditions MBF-20, an
wiring for a temperature range of 25–650 ◦ C. At the same
amorphous nickel braze filler metal was chosen. The brazing
time, the sensors were applied by means of pressure con-
process, was performed at temperatures between 1040 and
troller PACE 5000 of the company GE Measurement & Con-
1060 ◦ C. The micrograph in Fig. 4a shows an overview of
trol in the range of 0–200 bar with and the characteristic
the brazing zone indicating a good and pore-free adhesion of
curve is measured with a computer-controlled system. Fig-
both components. A closer look at Fig. 4b reveals the forma-
ure 3 shows the inner setup of this newly developed measure-
tion of darker chromium borides which are brittle intermetal-
ment device which allows the simultaneous characterization
lic phases in the brazing alloy. This indicates that chromium
of 6 sensors at maximum in the range of 25–600 ◦ C.
from the steel slightly dissolves into the molten brazing al-
loy MBF-20, which also contains small amounts of boride
3 Results and discussion for reduced melting temperatures. Without going into much
into detail we can say that it was possible to achieve hermetic
As claimed in the Introduction and with respect to the chosen dense joints with this materials and the SEM investigations
integration strategy, especially the joining process of metals gave no hints for significant interfacial reactions. With these
to ceramics represent a challenge. Active metal brazing of results the CTE adjusted steel connector for the further inte-
Kovar and similar alloys to Al2 O3 by active metal brazes is gration of the ceramic components is available.
not generally a new topic, but in terms of reduced brazing
processing temperatures the use of indium-containing active
metal brazes appears attractive. In the case of the SiO2 - and 3.2 Brazing KOVAR to ceramic interlayer
PbO-containing LTCC material it can be supposed that the 3.2.1 Brazing with Incusil-ABA
active component titanium in the active metal brazing alloy
will undergo a redox reaction with these oxides. Possible re- Active metal brazing of LTCC and Al2 O3 to Kovar with
action products could be titanium silicides of titanium–lead Incusil-ABA at 755 ◦ C for 10 min yields in both cases to her-
intermetallic compounds. Especially the formation of sili- metically sealed assemblies. Surprisingly, we were able to
cides with different components of high-temperature-stable realize hermetic joints of as-fired LTCC and Kovar, which
brazing alloys is described by McDermid and Drew (1991) is in sharp contrast to the results of Walker et al. (2006).
or Liu et al. (2009). Such intermetallic phases have a brittle SEM images of the microstructures of Kovar–Al2 O3 and
character and can have disadvantageous effects on the adhe- Kovar–LTCC joints brazed with Incusil-ABA are shown in
sion of the braze at the ceramic surfaces. For these reasons Fig. 5a and b, respectively. These micrographs show the
special interest must be paid to the interfacial reactions be- typical structure of the Ag–Cu–In eutectic with a Ag-rich
tween the different brazing alloys and joint materials in order phase (white regions, with dissolved In and Cu) and a Cu-
to identify proper brazing alloys and brazing conditions. rich phase (grey regions) together with enclosed intermetal-
lic phases in the brazing alloy (dark grey regions) and reac-

www.j-sens-sens-syst.net/5/73/2016/ J. Sens. Sens. Syst., 5, 73–83, 2016

78 J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC)

Figure 5. SEM images of Kovar–Al2 O3 (a) and Kovar–LTCC (b) joints brazed with Incusil-ABA at 755 ◦ C for 10 min.

Figure 6. Enlarged SEM images of Fig. 5a showing the Al2 O3 /Incusil-ABA (a) and the Kovar/Incusil-12.5-ABA interfaces (b).

Figure 7. Enlarged SEM images of Fig. 5b showing the LTCC/Incusil-ABA (a) and the Kovar/Incusil-12.5-ABA interfaces (b).

tion layers on both interfaces. Figure 6a and b are the en- small band which meanders parallel to the Kovar surface.
larged images from Fig. 5a displaying the reaction layers at Further away from the interface in the brazing alloy Ni-Cu-
the interface between Al2 O3 and the brazing alloy, and be- Ti compounds are visible. In addition, down to a depth of
tween Kovar and the brazing alloy, respectively. At the inter- 25 µm, Ag, In, and Cu from the brazing alloy are found at
face between Al2 O3 and the brazing alloy a very thin reac- the grain boundaries of the Kovar and along Fe–Co grains
tion layer with submicron thickness was formed. This reac- which are depleted of Ni. In comparison with Kovar–Al2 O3
tion layer completely covers the alumina interface, yielding a joints, the microstructure of Kovar–LTCC joints with Incusil-
mean helium lake rate of 6 × 10−10 mbar s−1 . The main con- ABA looks similar. Figure 7a and b are the enlarged images
stituents of the reaction layer are Ti and O, but also elements from Fig. 5b showing the interfaces between LTCC and the
of the Kovar, i.e., Ni, Fe, and Co, are detected. This suggests brazing alloy, and between Kovar and the brazing alloy, re-
that the constituents of the Fe–Ni–Co alloy show a strong spectively. Again, through diffusion of Fe, Ni, and Co and
affinity to Ti even at lower temperatures than in the publi- their reaction with the active element Ti, intermetallic com-
cations of Stephens et al. (2000) and Vianco et al. (2003a). pounds were formed in the brazing alloy and along the inter-
The strong reactivity of Fe, Ni, and Co with Ti shaped the in- face to Kovar. However, in the micrographs two differences
terface between Kovar and the brazing alloy as several inter- in comparison with Kovar–Al2 O3 joints are found. Firstly,
metallic compounds like (Fe,Ni,Co)x Tiy with a high amount the very thin (Ti,Fe,Ni,Co)x Oy reaction layer at the inter-
of Ni (abbreviation: Ni–Co–Fe–Ti) or Fe (abbreviation: Fe– face LTCC/brazing alloy contains traces of Si and Pb, the
Ni–Co–Ti) are observed. These intermetallic phases form a main constituents of the glass phase of the LTCC. Secondly,

J. Sens. Sens. Syst., 5, 73–83, 2016 www.j-sens-sens-syst.net/5/73/2016/

J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC) 79

Figure 8. SEM images of Kovar–Al2 O3 joints brazed with Incusil-25-ABA at 650 ◦ C for 10 min.

Table 2. Hermeticity after brazing with Cusil-ABA as a function of

brazing parameters.

Joint setup Peak process Hermetic joints

temperature (fraction)
Al2 O3 –Kovar 810 ◦ C 2/6
Al2 O3 –Kovar 830 ◦ C 12/12
Al2 O3 –Kovar 850 ◦ C 6/6
LTCC–Kovar 810 ◦ C 2/6
LTCC–Kovar 830 ◦ C 6/6
LTCC–Kovar 850 ◦ C Not tested
Figure 9. SEM image of a Kovar–LTCC joint brazed with Cusil-
ABA at 810 ◦ C for 10 min.

the reaction layer is non-continuous with some pores where

the brazing alloy was not able to wet the LTCC completely. which may also be a reason for the stronger phase forma-
However, all brazed assemblies were hermetic with an aver- tion. This reaction captures at least a fraction of the active-
age helium lake rate of 4 × 10−9 mbar s−1 . phase titanium from the brazing alloy which is necessary to
Further experiments were conducted by using the Incusil- enable a wetting process on ceramic surfaces. Thus due to
25-ABA brazing paste. An apparent advantage of the brazing insufficient brazing results no further experiments were per-
alloy is the lower processing temperature between 640 and formed with this brazing alloy containing a high percentage
680 ◦ C. Within the scope of this study it was not possible of indium. The presented results showed that brazing of Ko-
to achieve hermetically brazed joints between Al2 O3 –Kovar var with indium-containing active braze filler metals leads to
and LTCC–Kovar while using the Incusil-25-ABA brazing considerable destruction of the Kovar microstructure. An op-
alloy. The wetting of the braze on the Kovar surface was ex- timization of the brazing cycle could help to minimize this
cellent, which in turn led to spreading of the melt out of the behaviour. However this was beyond the scope of the present
brazing gap all over the Kovar surface. As a consequence the study and will be addressed in the future. As explained in the
brazing joints contained numerous and quite large pores, and next section, the indium-free Cusil-ABA braze filler metal
a porous microstructure was formed as seen in Fig. 8a. Also leaves the microstructure of the Kovar nearly intact.
the reason for the excellent wetting can be taken form the
SEM images in Fig. 8. A strong interaction between the ac- 3.2.2 Brazing with Cusil-ABA
tive metal braze and the Kovar alloy leads to the destruction
of the microstructure beneath the Kovar surface and is quite Active metal brazing of Al2 O3 and LTCC to Kovar with
more pronounced than is the case for the Incusil-ABA braze. Cusil-ABA was performed at three different brazing temper-
This strong reactivity enables the wetting of the molten braz- atures for a minimum of six samples for each brazing con-
ing alloy on the Kovar. Also here the dissolution of the Ko- dition. Table 2 summarizes the obtained hermeticity data.
var into the brazing melt results in the formation of nickel– While brazing of Al2 O3 yielded to hermetically sealed as-
titanium-based phases in the brazing alloy, which can be rec- semblies in most cases, brazing of LTCC gave rather differ-
ognized as the darker disperse phase in the brazing zone. ent results. After brazing of LTCC–Kovar joints at a temper-
In accordance with the enhanced dissolution of the Kovar, ature of 810 ◦ C, only a few of the assemblies were hermetic
this phase formation seems to be more pronounced. However (Fig. 9). Microstructural investigation showed that the inter-
Incusil-25-ABA contains more titanium than Incusil-ABA, face between LTCC and the brazing alloy is weakly bonded

www.j-sens-sens-syst.net/5/73/2016/ J. Sens. Sens. Syst., 5, 73–83, 2016

80 J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC)

Figure 10. SEM images of Kovar–Al2 O3 (a) and Kovar–LTCC (b) joints brazed with Cusil-ABA at 830 ◦ C for 10 min.

Figure 11. Enlarged SEM images of Fig. 10a showing the Al2 O3 –Cusil-ABA (a) and the Kovar–Cusil-ABA interfaces (b).

because it seems that only a non-continuous and thin reaction The main phase comprises a Fe-rich intermetallic compound
layer was formed. Thus at this brazing temperature no reli- (Fe–Ni–Co–Ti) that covers the interface of the Kovar com-
able joining was possible. When the brazing temperature was pletely. Adjacent to the Fe–Ni–Co–Ti phase a second Ni-
increased to 830 ◦ C, all brazed LTCC–Kovar joints showed rich phase (Ni–Co–Fe–Ti) was found. A third intermetallic
gas-tightness due to the formation of a continuous reaction phase composed of Ni, Cu, and Ti is observed in the brazing
layer at the interface between LTCC and Kovar, which is alloy. The microstructural analysis of Kovar–Al2 O3 joints
shown later in detail. Due to the fact that this brazing tem- brazed with Cusil-ABA and Incusil-ABA showed that dur-
perature is close to the sintering temperature of the LTCC ing the active metal brazing processing similar phases were
tape, it was initially assumed that brazing is not possible be- formed in the brazing seam. However, in the case of Cusil-
cause of softening of the residual glassy phase in the LTCC. ABA the intermetallic compounds are located near the Kovar
This was not confirmed, and the results showed strong and surface (Fig. 11b), whereas in the case of Incusil-ABA a lace-
hermetic bonding. However, in contrast to Al2 O3 , brazing of work phase was formed (Fig. 6b). Additionally, while braz-
LTCC at 850 ◦ C was not tried as the LTCC is sintered at this ing with Cusil-ABA no penetration of the Kovar along the
temperature and the stability of the ceramic material is lim- grain boundaries by the brazing alloy was observed (Fig. 10a
ited. In fact a repeated heating of the DuPont 951 tape up to and b). This leads to the conclusion that the reactivity of Fe,
the processing temperature is possible without any degrada- Ni, and Co with the active element titanium in the braze filler
tion of the microstructure. Scanning electron micrographs of metals is enhanced because of the presence of indium or of
the microstructures of Kovar–Al2 O3 and Kovar–LTCC joints the lower melting temperature of the Ag–Cu–In eutectic. The
brazed with Cusil-ABA at 830 ◦ C are shown in Fig. 10a and observation that the Incusil-ABA braze filler metal with the
b, respectively. In these micrographs the brazing alloy dis- higher indium content leads to a stronger destruction of the
plays the typically structure of the Ag–Cu eutectic with a Kovar microstructure along the grain boundaries lets one as-
Ag-rich phase (white region) and a Cu-rich phase (grey re- sume that especially the grain boundary phases of the Kovar
gions). Furthermore, the formation of reaction layers on both consist of a alloy composition which forms low melting com-
interfaces is visible. positions with indium. A look at the binary phase diagrams
These reaction layers are shown in more detail in Fig. 11a iron–indium, cobalt–indium, and nickel–indium reveals that
and b. The active element titanium formed a continuous re- nickel and cobalt can form low melting phases under brazing
action layer with a thickness of 0.7–1 µm bordering the inter- conditions (Okamoto, 1997, 2003). In the case of iron this
face between Al2 O3 and the brazing alloy. The reaction layer behaviour is shifted to higher temperatures and should not
consists of titanium and oxygen with minor amounts of Ni, be pronounced below 800 ◦ C (Okamoto, 1990). On the other
Fe, and Co. The elemental composition is the same as ob- side according to Berry (1987) the grain boundary phases
served for Kovar–Al2 O3 joints brazed with Incusil-ABA. At of Kovar-based alloys tend to form oxide-rich phases, which
the interface between Kovar and the brazing alloy an up to makes them susceptible to corrosion processes. So we should
3 µm thick reaction layer with multiple phases was formed.

J. Sens. Sens. Syst., 5, 73–83, 2016 www.j-sens-sens-syst.net/5/73/2016/

J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC) 81

Figure 12. Enlarged SEM images of Fig. 10b showing the LTCC–Cusil-ABA (a) and the Kovar–Cusil-ABA interfaces (b).

Table 3. Compositions of interfacial reaction layers Cusil-ABA– Table 4. Compositions of interfacial reaction layers Cusil-ABA–
LTCC at 830 ◦ C for 10 min and Incusil-ABA–LTCC after brazing LTCC and Cusil-ABA–Al2 O3 after brazing at 850 ◦ C for 10 min.
at 755 ◦ C for 10 min.
Element Cusil-ABA– Cusil-ABA–
Element Cusil-ABA– Incusil-ABA– Ma. % LTCC Al2 O3
Ma. % LTCC LTCC
O 14.3 12.5
O 10.2 16.5 Al 1.2 4.14
Al 0.6 1.6 Si 4.7
Si 3.8 4.5 Ti 32.9 36.6
Ti 44.8 36.4 Fe 11.6 12.6
Fe 6.7 10.1 Co 5.6 4.7
Co – 3.9 Ni 24.0 25.0
Ni 8.2 10.1 Cu 4.5 4.3
Cu 25.8 13.0 Ag 1.0 1.3
In – –
Ag – –

brazed with Incusil-ABA. The comparison of the interface

between Kovar and the brazing alloy after brazing to Al2 O3
note that a more detailed investigation of the grain boundary (Fig. 11b) and LTCC (Fig. 12b) showed no difference in mi-
phase is necessary in order to clarify this behaviour. crostructural appearance like thickness, phases formed, or el-
The microstructure of Kovar–LTCC joints brazed with emental composition. Based on these results it was decided to
Cusil-ABA is similar to the microstructure of Kovar–Al2 O3 use the Cusil-ABA braze filler metal with a brazing tempera-
joints. Figure 12a and b are the enlarged images from ture of 830 ◦ C for the construction of the complete sensor as
Fig. 10b displaying the reaction layers at the interface be- shown in the next section.
tween LTCC and the brazing alloy, and between Kovar
and the brazing alloy, respectively. At the interface between
3.2.3 Joining of sensor and electrical connection
LTCC and the brazing alloy a nearly 1 µm thick reaction layer
was formed (Fig. 12a). Besides the main constituents of tita- In accordance with the integration procedure the last step in-
nium and oxygen, the EDX analysis revealed the presence of volves the soldering of the sensor element by a glass paste
Fe, Ni, and Co from Kovar and of minor traces of silicon and which was screen-printed on the back side on the sensor and
lead from the glass phase of the LTCC. It is noteworthy that fired at maximum temperature of 550 ◦ C. Joining and sealing
the compositions of the interfacial reaction layers bordering processes for packaging of ceramic-based sensor elements
the LTCC interface are quite similar to the one found after are established for quite a long time, and so numerous quali-
brazing of LTCC and Kovar with Incusil-ABA as seen by fied glass solders are available for this task. Figure 13a illus-
EDX data in Table 3, which compares the compositions of trates the final assembling steps of the sensor element into the
the interfacial layers brazed with Cusil-ABA at 830 ◦ C and steel connector with both brazed balancing elements made
with Incusil-ABA at 755 ◦ C. A large difference is only rec- of Kovar and alumina. In Fig. 13b a SEM image shows the
ognized for the copper content. However we should not for- joining zones Kovar–Al2 O3 and Al2 O3 –LTCC sensor of a
get the small thickness of the reaction layer in the case of completely assembled sensor. It can be seen that both zones
Incusil-ABA which adds an uncertainty to the spectral data. contain only few pores and are well attached to each other.
So additionally in Table 4 similar results are presented for After the soldering step of the sensor a final measurement of
compositions of two interfacial layers resulting from braz- the helium leakage rate was performed. Completely assem-
ing both LTCC and Al2 O3 with Cusil-ABA at 850 ◦ C for bled sensors were mounted and characterized in the afore-
10 min. However, the higher brazing temperature for Cusil- mentioned pressure rail at temperatures up to 300 ◦ C. As an
ABA yielded a much thicker reaction layer than for samples example, Fig. 14 displays a set of characteristic curves of

www.j-sens-sens-syst.net/5/73/2016/ J. Sens. Sens. Syst., 5, 73–83, 2016

82 J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC)

Figure 13. Joint components showing the stepwise integration procedure (a) and a SEM image of a cross section of a packaged LTCC
pressure sensor (b).

Figure 14. Temperature-dependent pressure-signal characteristics

of a completely packaged LTCC sensor. Figure 15. Completely assembled sensor with steel connect screw
and welded steel housing with wiring.

a completely assembled sensor based on an applied pres-

sure (bar) and the corresponding sensor signal (1mV / V) temperatures, but the strong interaction with the Kovar metal
between 25 and 300 ◦ C. The sensor signal shows a good lin- and the porous brazing seams result in unreliable joints.
earity in the investigated pressure and temperature range. The With Cusil-ABA, joints were hermetically sealed at brazing
sensitivity remains unaffected, and the particular curves are temperatures > 810 ◦ C. In all cases microstructural analysis
only shifted by a small offset, which can be compensated by revealed the development of intermetallic compounds that
an accompanied temperature measurement. might be brittle, but their influence on the joint strength is un-
clear and will be investigated in the future. The combination
of this metal-to-ceramic brazing step with additional joining
4 Conclusions processes allows the hermetic integration of a ceramic LTCC
pressure sensor into steel housing with an adapted standard-
The present work focused on the joining process of Ko- ized thread (Fig. 15).
var with alumina and LTCC as part of on approach to in-
tegrate LTCC-based sensors into steel connects. While us-
ing commercially available active braze filler metals (Cusil-
ABA, Incusil-ABA, Incusil-25-ABA) under certain condi-
tions, both ceramic types were hermetically sealed to Ko-
var. Hermetic joining of Al2 O3 to Kovar was possible with
Incusil-ABA and with Cusil-ABA for all investigated tem-
peratures. Additionally, brazing of LTCC to Kovar was possi-
ble and shown for the first time. At 755 ◦ C with Incusil-ABA
hermetic LTCC–Kovar joints were realized. The higher in-
dium content of Incusil-25-ABA would enable lower brazing

J. Sens. Sens. Syst., 5, 73–83, 2016 www.j-sens-sens-syst.net/5/73/2016/

J. Schilm et al.: Joining of Kovar to alumina and to low-temperature co-fired ceramics (LTCC) 83

Acknowledgements. The authors thank Felix Köhler, Birgit Okamoto, H.: In-Ni (Indium-Nickel), J. Phase Equilib., 24, p. 379,
Manhica, Maria Striegler, and Sabine Fischer for sample prepa- 2003.
ration, helium leak rate measurement, and scanning electron Partsch, U., Gebhardt, S., Arndt, D., Georgi, H., Neubert, H., Fleis-
microscopy. cher, D., and Gruchow, M.: LTCC based sensors for mechanical
quantities, Proceedings of the 16th European Microelectronics
Edited by: H. Fritze and Packaging Conference & Exhibition, Oulu, Finland, 17–20
Reviewed by: two anonymous referees June 2007, 381–388, 2007.
Partsch, U., Lenz, C., Ziesche, S., Lohrberg, C., Neubert, H., and
Maeder, T.: LTCC-based sensors for mechanical quantities, In-
formacije MIDEM, 42, 260–271, 2012.
References
Peterson, K. A., Knudson, R. T., Garcia, E. J., Patel, K. D., Okan-
Arróyave, R. and Eagar, T. W.: Metal substrate effects on the ther- dan, M., Ho, C. K., James, C. D., Rohde, S. B., Rohrer, B. R.,
mochemistry of active brazing interfaces, Acta Mater., 51, 4871– Smith, F., Zawicki, L. R., and Wroblewski, B. D.: LTCC in mi-
4880, 2003. croelectronics, microsystems, and sensors, Proceedings of the
Berry, K. A.: Corrosion Resistance of military microelectronics 15th International Conference on Mixed Design of Integrated
package at the lead-glass interface, in: Proceedings of the ASM’s Circuits and Systems, Poznan, Poland, 19–21 June 2008, 23–37,
3rd Conference on Electronic Packaging: Materials and Corro- 2008.
sion in Microelectronics, Minneapolis, MN, Materials Park, OH, Stephens, J. J., Vianco, P. T., Hlava, P. F., and Walker, C. A.:
ASM International, 28–30 April 1987, 55–61, 1987. Microstructure and performance of Kovar/alumina joints made
Fernie, J. A., Drew, R. A. L., and Knowles, K. M.: Joining of engi- with Silver-Copper base active metal braze alloys, in: Advanced
neering ceramics, Int. Mater. Rev., 54, 283–331, 2009. brazing and soldering technologies, edited by: Vianco, P. T. and
Fournier, Y., Maeder, T., Boutinard-Rouelle, G., Barras, A., Craque- Singh, M., ASM International (Materials Park), 240–251, 2000.
lin, N., and Ryse, P.: Integrated LTCC pressure flow temperature Stephens, J. J., Hosking, T. J., Headly, P. F., Hlava, P. F., and Yost,
multisensor for compressed air diagnostics, Sensors, 10, 11156– F. G.: Reaction layers and mechanisms for a Ti-activated braze
11173, 2010. on sapphire, Metall. Mater. Trans. A, 34, 2963–2972, 2003.
Hahn, S., Kim, M., and Kang, S.: A study of the reliability of brazed Vianco, P. T., Stephens, J. J., Hlava, P. F., and Walker C. A.: Tita-
Al2 O3 joint systems, IEEE Trans. Comp. Pack. Manuf. Tech. C., nium scavenging in Ag-Cu-Ti active braze joints, Weld. J., 82,
21, 211–216, 1998. 268S–277S, 2003a.
Keusseyan, R. L. and Dilday, J. L.: Development of brazing inter- Vianco, P. T., Stephens, J. J., Hlava, P. F., and Walker C. A.: A
connection to low thermal expansion glass-ceramics for high per- barrier layer approach to limit Ti scavenging in FeNiCo/Ag-Cu-
formance multichip packaging, Proc. of the 43rd Conference on Ti/Al2 O3 active braze joints, Weld. J., 82, 252s–262s, 2003b.
Electronic Components and Technology, Orlando, FL, USA, 1–4 Walker, C. A. and Hodges, V. C.: Comparing metal-ceramic brazing
June 1993, 896–903, 1993. methods, Weld. J., 87, 43–50, 2008.
Lin, K.-L., Singh, M., and Asthana, R.: Interfacial characteriza- Walker, C. A. and Hodges, V. C.: Metal-nonmetal brazing for
tion of alumina-to-alumina joints fabricated using Silver-Copper- electrical, packaging and structural applications, in: Advances
titanium interlayers, Mater. Char., 90, 40–51, 2014. in Brazing: Science, Technology and Applications, edited by:
Liu, Y., Huang, Z. R., and Liu, X. J.: Joining of sintered silicon Sekulić, D. P., Woodhead Publishing Oxford, 498–524, 2013.
carbide using ternary Ag-Cu-Ti active brazing alloy, Cer. Int., Walker, C. A., Uribe, F., Monroe, S. L., Stephens, J. J., Goeke, R. S.,
35, 3479–3484, 2009. and Hodges, V. C.: High-temperature joining of low-temperature
McDermid, J. R. and Drew, R. A. L.: Thermodynamic brazing alloy co-fired ceramics, Proceedings of the 3th International Brazing
design for joining silicon carbide, J. Am. Cer. Soc., 74, 1855– and Soldering Conference, San Antonio, TX, USA, 24–26 April
1860, 1991. 2006, 54–59, 2006.
Nascimento, R. M., Martinelli, A. E., and Buschinelli, A. J. A.: Wielage, B., Hoyer, I., and Hausner, S.: Induction brazing of alu-
Review Article: Recent advances in metal-ceramic brazing, mina and zirconia with various metals, Proc. 5th International
Cerâmica, 49, 178–198, 2003. Brazing and Soldering Conference, Las Vegas, NV, USA, 22–25
Nascimento, R. M., Martinelli, A. E., Buschinelli, A. J. A., and April 2012, 101–108, 2012.
Sigismund, E.: Interface microstructure of alumina mechanically Zarnik, M. S., Mozek, M., Macek, S., and Belavic, D.: An LTCC-
metallized with Ti brazed to Fe-Ni-Co using different fillers, based capacitive pressure sensor with a digital output, Informa-
Mater. Sci. Eng. A, 466, 195–200, 2007. cije MIDEM, 40, 74–81, 2010.
Okamoto, H.: Co-In (Cobalt-Indium), J. Phase Equilib., 18, p. 315, Zhu, W., Chen, J., Jiang, C., Hao, C., and Zhang, J.: Effects of
1997. Ti thickness on microstructure and mechanical properties of
Okamoto, H.: Fe-In (Iron-Indium), Binary Alloy Phase Diagrams, alumina-Kovar joints brazed with Ag-Pd/Ti filler, Ceram. Int.,
2nd Edn., edited by: Massalski, T. B., 2, 1712–1714, 1990. 40, 5699–5705, 2014.

www.j-sens-sens-syst.net/5/73/2016/ J. Sens. Sens. Syst., 5, 73–83, 2016