CHARACTERIZATION OF THICK FILM CAPICITORS

Teaching activities of the "Institut Universitaire de Technologic de Lille" (France) were previously reported. The present paper reports measurement results obtained by students during their last training period for about a hundred capacitors made with commercial dielectrics (EMCA CDP 400, EMCA CDP 2000, and ESL K 1000). We examine the following points: reproducibility of capacitors during fabrication, correlation between electrical parameters of components and firing conditions, measurements of capacitance and loss factor between 10 kHz and 20 MHz for temperatures ranging from 20C to 100 C.


FOREWORD
Teaching activities of the Institut Universitaire de Technologic de Lille (Dpartement "Gnie Elec- trique") in the thick film field have previously been reported.1-3The principle of operation and fabrica- tion of hybrid circuits is taught to the students as a part of their training in electronics.Moreover, every year several students spend two months in our thick film laboratory for their training period which take place at the end of the two year course.In this paper the results obtained in 1976 are reported.
We studied the electric characterization of thick film capacitors for the following reasons.Discussions with thick film specialists lead to the conclusion that study of reproducibility of capacitors during fabrica- tion and correlation between electrical parameters of components and firing conditions is required.Effects produced by firing conditions have to be known either to specify the temperature and time margin authorized for getting the same permittivity, or to' adjust this parameter to a desired value.Values of capacitances and of their loss factor versus frequency and temperature may also be very useful.Specialized papers provide a good deal of data concerning complex permittivity of glasses, titanates and other dielectrics 4 but unfortunately this information is useless for the designer who is concerned with pastes whose composition is generally unknown to him.On the other hand, the growing use of screened capaci- tors necessitates the knowledge of a greater number of parameters.However it is obvious that each ink manufacturer cannot supply all the desired parameters 125 for all the possible applications.Specific information can be investigated through particular studies for example the one presented here.

PRESENTATION OF STUDIED COMPONENTS
Results and conclusion reported in this paper come from an experimental study carried out on about a hundred capacitors made with commercial dielectric pastes screened with 200 mesh grids.These capacitors were fired in a four zone Piezo-ceram oven which allows systematic experiments versus time and firing temperature.Several screen printing types of opera- tion were considered.The materials used were EMCA CDP 400, EMCA CDP 2000, Electro Science K 1000 and Cermalloy 9100 which are mean and high per- mittivity materials and a cross over dielectric.The conductor paste used was EMCA 4249.
We successively examined the problem of repro- ducibility in capacitor fabrication, the results con- cerning capacitance and loss factor measurements versus manufacturing parameters, frequency and temperature.

REPRODUCTIVITY OF FABRICATION
We considered a screened, fiat, rectangular capacitor whose thickness, length and width are e, a and b, respectively.
The dispersion on capacitances Ac can be related to the variations of geometrical parameters Ae, Aa with Aa -Ab.
The thickness e was measured with a stylus profilo- meter.For one printing, the value of e was near 15 microns but two printing operations were generally preferred, which is recommended by the ink manu- facturers.In this case, e was near 30 microns with Ae -2 microns on a series of 20 capacitors.We also measured Aa -50 microns.In these conditions for 0,63 mm z area capacitors (a 0,9 ram; b 0,7 mm) we estimated: Ac 20% c in fact dispersion on measured capacitances gave generally Ac <15% c A smaller error can be predicted when the area of the capacitor increases.For instance, for a 0. Capacitance vs. frequency for 950C firing temperatures for several samples.E.S. K1000.Loss factor vs. frequency for 950C firing temperature for various substrate temperatures.E.S. K1000. 900and 950 firing temperature).For the same component, the capacitance was quite constant for the frequency range considered.For the same firing temperature, the two pastes yield practically the same capacitance.Their mean values are 32 + 2 pF for 850C 36 + 6 pF for 900C 64 + 4 pF for 950C For one dielectric printing (firing temperature 850 C), we obtained C 50 + 3 pF, but in several cases, a short circuit occurred.No significant variation of C was observed when the substrate was heated (from room temperature to O0 C) but the loss-factor decreased (Figures 3 and 4).
Dielectric paste Electro-Science K 1000 was fired at 950C and 975C with 8 minute, 13 minute and 18 minute firing times.At 50 kHz, the mean capaci- tance was C 213 + 25 pF (25 components).In the present conditions, no influence of firing parameters was observed.Examples of variation of C and tan for several samples are shown in Figure 5 and 6.For this high permittivity dielectric, a slight increase of permittivity and a decrease of loss factor were noticed when the substrate was heated from room temperature to 100C (Figures 7 and 8).
Dielectric paste Cermannoy 9100 was previously studied, a Capacitance was smaller than 2 pF/mm 2 and loss factor could reach 0.1.

CONCLUSION
Choice of the dielectric paste depends on the applica- tion.For instance capacitor fabrication needs low loss materials with high permittivity which has to be constant versus frequency and temperature.The cross over demands the lowest permittivity material; loss factor value is not so important in this case as in the previous example.
vs. frequency for 950C firing temperature for various substrate temperatures.