(LTCC) and Multilayer Capacitors

Proc. 43rd International Scientific Colloquium, Ilmenau (Germany), Sept.21-24 1998 Vol. 2, p.198-202

A. Dziedzic/ L.J. Golonka/ M. Henke

Temperature Properties of Low Temperature Cofiring Ceramics (LTCC) and Multilayer Capacitors 1. INTRODUCTION The use of passive components (resistors, capacitors, and inductors) is increased in modern microelectronics. For example, the PC microprocessor speed increased over the last 15 years from 4 to 400 MHz. During this time the microprocessor voltages have dropped by about 50%. Both these trends independently require much more passives [1]. Therefore there is a growing need for integration of passive components. The Low Temperature Cofiring Ceramics (LTCC) technology is one of the most suitable for fabrication of integrated passive components, especially with the high level of integration, i.e. combination of a wide variety of capacitors, resistors, and inductors into a single package [2]. Temperature very often determines the usability range of different materials and/or devices. In the case of ceramic materials this quantity affects the permittivity as well as dielectric losses. Moreover it is well known that these parameters are frequency and voltage dependent. Therefore this paper presents the temperature characteristics of multilayer capacitors fabricated in LTCC technology. The same dependencies measured for multilayer capacitors permit to compare LTCC with other ceramic materials applied in electronic circuits. 2. SAMPLE FABRICATION AND MEASUREMENTS DESCRIPTION The Ferrotape A6-M based on a crystallizing glass, the key component in the Ferro A6 LTCC material system, with thickness 125 m (white) and 200 m (dark brown) have been used to study the temperature properties. The manufacturer information about dielectric properties of this tape is rather very modest - the dielectric constant, K, at 10 MHz is 5.9 and dissipation factor (loss tangent, tan ) measured at the same frequency is less than 0.2% [3]. Therefore the monolithic structures of a multilayer capacitors have been prepared to characterize the temperature behaviour of this dielectrics. The 5-layer samples with 3 parallel-connected capacitors, each with area of 130 mm2, have been made in order to increase the measured capacitance. The Ag conductor paste FX-33-229 from Ferro all silver based system, compatible with the mentioned tape, served as internal electrodes. The test structure has been prepared as follow [4]: vias fabrication for parallel connection of capacitors, screen printing of capacitor electrodes and drying in a box oven at 70oC for 10 minutes, lamination in an hydrostatic press at 210 bar for 10 minutes held at 70 oC, firing the laminates in Nabertherm L3/S furnace with the following recipes I - v1 = 2oC/min, T1 = 450oC/120 min, v2 = 4oC/min, T2 = 850oC/10 min II - v1 = 2oC/min, T1 = 450oC/120 min, v2 = 4oC/min, T2 = 800oC/10 min

- v1 = 2oC/min, T1 = 450oC/120 min, v2 = 4oC/min, T2 = 900oC/10 min - v1 = 2oC/min, T1 = 450oC/120 min and then firing in BTU belt furnace with an overall 60 min cycle and 10 minutes soaking at 850oC peak temperature. The capacitance of the test samples has been equal about 300-400 pF for 125 m tape and 120-180 pF for 200 m one. The HP 4263A LCR Meter interfaced to an IBM PC for data acquisition and presentation has been used for measurements the dependence of capacitance and dissipation factor as a function of temperature in the range from -170oC to 130oC. In order to avoid influence of humidity on presented results the temperature properties have been measured starting from the room temperature too. Because of the initial capacitance of test samples the parallel equivalent circuit i.e. C p-tan mode has been used for measurements and further data analysis and presentation. In the experimental set-up both C = f(T) and tan = f(T) characteristics have been taken every 10 degrees at 1, 10 and 100 kHz frequency for signal with 1 V amplitude. Similar measurements and analysis have been made for100 pF, 1nF, 10 nF (all with 0805 size) and 100 nF (1206 size) multilayer chip capacitors as well. III IV

3. TEMPERATURE PROPERTIES OF MULTILAYER CHIP CAPACITORS The first multilayer ceramic capacitors have been produced in the early 1950s. It seems that there are the most wide used capacitors at present [5, 6]. Because of the necessary capacitance range the different dielectric materials, known under NPO, X7R, Z5U or Z5V categories, are used for their fabrication. In general their temperature behaviour is presented only in the range from -55 to 125oC. Base on results of our measurements (Figs. 1 and 2) we can say that 100 pF capacitor is made from NPO (low-K) dielectrics, 1 nF and 10 nF ones from X7R (medium-K) ceramics, whereas 100 nF one represents Z5U (high-K) ferroelectric ceramics. An interesting run of dissipation factor versus temperature is shown in Fig. 2. The dielectric losses of 100 pF capacitor are almost temperature independent but they are approximately one order larger than presented in literature for NPO dielectrics [6]. The dissipation factor for the other devices is relatively high at low temperatures but diminishes gradually in the useful temperature range.

0.10

DISSIPATION FACTOR

1.0

C/C25

0.8 0.6 0.4

cap 100 pF cap 1 nF cap 10 nF cap 100 nF

0.2 0.0 -200

-100

0

100 o

TEMPERATURE [ C]

Fig. 1. Comparison of relative capacitance changes versus temperature for multilayer chip capacitors.

0.08 cap. 100 pF cap. 1 nF cap. 10 nF cap. 100 nF

0.06 0.04 0.02 0.00 -200

-100

0

100 o

TEMPERATURE [ C]

Fig. 2. Variation of dissipation factor as a function of temperature for different chip capacitors.

It is interesting to note that below -100oC the real capacitance of 100 nF chip capacitor is less than for 10 nF one (Figs. 1 and 3). The dependencies of capacitance and loss tangent versus temperature for 100 nF device (Figs. 3 and 4) are similar to those reported for relaxor ceramics [7] where the position of dielectric constant peak and its magnitude are dependent on frequency.

140.0n

0.12

100.0n 80.0n

DISSIPATION FACTOR

CAPACITANCE [F]

120.0n f = 1 kHz f = 10 kHz f = 100 kHz

60.0n 40.0n 20.0n 0.0 -200

-100

0

100

0.09

0.06

0.03

0.00 -200

f = 1 kHz f = 10 kHz f = 100 kHz

-100

0

100 o

o

TEMPERATURE [ C]

TEMPERATURE [ C]

Fig. 3. Temperature dependence of capacitance for 100 nF capacitor at different frequencies.

Fig. 4. Temperature dependence of loss tangent for 100 nF capacitor at various frequencies.

4. TEMPERATURE BEHAVIOUR OF FERRO A6-M TAPE The monotonous, almost linear increase in capacitance as a function of temperature is visible in many cases (e.g. Figs. 5 and 7). This behaviour is characteristic for temperature stable, low-K inorganic dielectrics. The dielectric losses in these cases are rather insignificant (less than 1% in the whole temperature range) and temperature-independent - Figs. 6 and 8). The Temperature Coefficient of Capacitance (TCC), TCC = (C2 - C1)/C1(T2 - T1) where C2 - capacitance at T2 = 125oC (Hot TCC) or -55oC (Cold TCC) and C 1 - capacitance at T1 = 25oC can be calculated based on measured C(T) characteristics. For example, the values of HTCC are equal 120 ppm/K (f = 1 kHz), 170 ppm/K (f = 10 kHz) or 260 ppm/K (f = 100 kHz) for 200 m tape fired using II recipe in comparison with 310 ppm/K (f = 1 kHz), 370 ppm/K (f = 10 kHz) or 400 ppm/K (f = 100 kHz) obtained for 100 pF capacitor. However there are also situations where local peaks appear both on C(T) and tan (T) curves (Figs. 9-12). The dielectric losses, even for samples without mentioned peaks, are larger than manufacturer’s data. Probably our measurement conditions differ considerably from those used by Ferro (not only in frequency but we suppose also in test sample geometry). Therefore one should remember that thick-film capacitors with larger size (electrode area) are characterized by larger dissipation factor [8]. However this does not explain the peaks reported in Figs. 9-12. It seems that the test structure is not fully hermetic because of the mismatch of coefficient of thermal expansion (CTE) for tape and electrode material. The thermal stresses, especially introduced during cooling, can cause microcracks or even delamination. It is known that the CTE of A6 tape is 7 ppm/K whereas the CTE of silver is 19

ppm/K [4] and even the addition of low thermal expansion materials to the applied conductive composition can minimize their difference only in part.

0.03 f = 10 kHz f = 100 kHz f = 10 kHz f = 100 kHz

310.0p

DISSIPATION FACTOR

CAPACITANCE [F]

315.0p

305.0p

300.0p -200

-100

0

0.02

0.01

0.00 -200

100 o

TEMPERATURE [ C]

Fig. 6. Dielectric losses as a function of temperature for 125 m tape fired in I recipe.

DISSIPATION FACTOR

CAPACITANCE [F]

f = 10 kHz f = 100 kHz f = 10 kHz f = 100 kHz

-100

0

0.04 0.02

-100

0

100 o

Fig. 8. Dielectric losses as a function of temperature for test sample from 200 m tape fired in II recipe.

355.0p

160.0p f = 10 kHz f = 100 kHz f = 10 kHz f = 100 kHz

CAPACITANCE [F]

CAPACITANCE [F]

0.06


TEMPERATURE [ C]

Fig. 7. Capacitance versus temperature for test structure made from 200 m tape and fired in II recipe.

345.0p

340.0p

335.0p -200

0.08

0.00 -200

100 o

TEMPERATURE [ C]

350.0p

100

0.10

165.0p

160.0p -200

0 o

180.0p

170.0p

-100

TEMPERATURE [ C]

Fig. 5. Capacitance versus temperature for 125 m tape fired according to I recipe.

175.0p


-100

0

100 o

TEMPERATURE [ C]

Fig. 9. Plot of capacitance for 125 m tape fired in IV recipe.

150.0p


140.0p

130.0p

120.0p -200

-100

0

100 o

TEMPERATURE [ C]

Fig. 10. Changes in capacitance with temperature for 200 m tape fired according to I recipe.

0.02

0.10 f = 10 kHz f = 100 kHz f = 10 kHz f = 100 kHz

DISSIPATION FACTOR

DISSIPATION FACTOR

0.03

0.01

0.00 -200

-100

0

100 o

TEMPERATURE [ C]

Fig. 11. Temperature dependence of tan for 125 m tape fired in IV recipe showing the presence of loss peak.

0.08 0.06


0.04 0.02 0.00 -200

-100

0

100 o

TEMPERATURE [ C]

Fig. 12. Temperature dependence of dissipation factor for 200 m tape fired in I recipe.

Acknowledgements This work has been supported by Polish State Committee for Scientific Research, Grant no 8T11B02913 References [1] R. Lasky; Growth continues for passive components, Electronic Packaging and Production, March 1998, p.77 -78 [2] R.A. Ladew, S.Makl; Integrating passive components, Proc. 1995 Int. Microelectronics Symp. (ISHM -USA), p.59-65 [3] FERRO A6 LT CC Materials System Product Information, 1996 [4] T . Williams, A. Shaikh; Silver via metallization for Low T emperature Cofired Ceramic T ape, Proc. 1994 Int. Microelectron ics Symp. (ISHM-USA), p.324-329 [5] M. Kahn, D.P. Burks, I. Burn, W.A. Schulze; Ceramic capacitor technology, Chapt. 4 in Electronic Ceramics, Ed. by L.M. Levinson, Marcel Dekker, 1988, p.191-274 [6] G. Goodman, R.C. Buchanan, T .G. Reynolds; Ceramic capacitor materials, Chapt. 2 in Ceramic Materials for Electronics, E d. by R.C. Buchanan, Marcel Dekker, 1991, p.69-127 [7] G.A. Smolenskii, V.A. Isupov, A.I. Agranovskaya, S.N. Popov; Ferroelectrics with diffuse phase transitions, Soviet Physic s - Solid State, vol.2 (1961), p.2584 [8] J. Steinberg, B. Kistler, R. Cooper; Materials and applications for thick film RC networks, Proc. 1990 Int. Microelectronics Symp. (ISHM-USA), p.276-284

Authors: Dr. Andrzej Dziedzic Prof. Leszek J. Golonka Marcin Henke Institute of Microsystem Technology, Wrocław University of Technology Wybrzeże Wyspiańskiego 27 50-370 WROCŁAW, POLAND tel.: +48 - 71 - 55 48 22, fax: +48 - 71 - 55 48 22 or +48 - 71 - 328 35 04, e-mail: [email protected]