STATE OF THE ART OF THIN FILM COMPONENTS

The main reasons for the development of thin film technology are miniaturisation, reliability and high frequency applications. A comparison with conventional components, and with the demands of system engineers shows that the thin film resistor problem is solved and that there are thin film capacitors for many applications, whereas coils of only small values and low Q can be made. As the distances between thin film circuit elements can be extremely small, low conductor resistance between components is possible. The inherent advantages of thin film technology are good reliability because of fewer connections, the possibility of functional trimming, small TCR and TCC variation in a circuit, and TCR-TCC balancing. Furthermore parasitic effects can be well controlled, distributed elements realized and large scale integration carried out.


INTRODUCTION
In modern electronics the use of thin film circuits is increasing. It is worthwhile therefore to consider briefly the state of the art of thin film components. We will try to answer these questions: Has the research and development of thin film technology been successful? Are we able to fulfill the wishes of system engineers? Let us begin by noting the reasons which led to the development of thin film circuits. 1) After the war we were successful for some twenty years in the miniaturisation of discrete components. In many cases the volume was reduced by a factor of 2 every 5 years. In the sixties it became obvious that it was impossible to continue in this way. It was not so much a problem of physics as a problem of our fingers, and our machinery. We produced thousands or even millions of f2 or pF in an astonishingly small volume, but we needed nearly the same size for components with a few 2 or pF, because a very great deal of the volume was needed for contacts, leads and encapsulation.
2) The reliability of discrete components was originally adequate, but the requirements of our custorners ecame more stringent, and the increasing number of components per device demanded a corresponding improvement in their reliability. In many cases even higher reliability was necessary 'This paper was presented at the Thin Film Technology Symposium held at the University of Stuttgart, Germany, March 18-19, 1974, organized  3) Soldered joints were found to be an important source of failure. The number of connections rose with the number of components per device, and reliability decreased correspondingly. 4) Discrete components change their function with increasing frequencies: capacitors act as coils, coils as capacitors and resistors as coils or capacitors. Leads often have inductances higher than those of capacitors or resistors themselves. 5) In the range of microwaves, discrete components fail entirely because the dimensions of components and wires are greater than the wave length. 6) Conventional circuits suffer from parasitic effects especially in the field of high frequencies. The troublesome task of reducing such effects by the bending and shaping of lead wires is well-known to system engineers.  components. We will consider only typical values, as special units are often considerably better than those commonly used. Figure 1). Thin film techniques make possible resistors of 12 or smaller up to about 1062; higher values require too great an area. Low values up to about 1000 fZ can be realized on an extremely small area of about 1 mm , whereas conventional resistors need 6 mm minimum. Tolerances can be as close as with discrete metal film resistors. The possibility of functional adjustment is a special advantage of thin film circuits, and stability is very good. The  10"

1) Let us begin with resistors (
10 Hz 10 f FIGURE 3 Dissipation factor as a function of frequency.
10.10-6/K. The inductance of thin film resistors depends on their geometry; values as low as nH are possible, a great advantage especially in the field of high frequencies. Noise is extremely low. In summary, it can be said that the resistor problem is solved.
2) Now let us compare thin film capacitors on the base of Ta2Os, silicon monoxide or silicon dioxide with discrete polystyrene and ceramic capacitors.
( Figure 2). The range of thin film capacitors from 10 pF to 60 nF fulf'flls many demands. Higher values need too large an area. Tolerances, stability and insulation resistance are sufficient. The TCC is determined by the properties of the film material and is therefore very consistent, especially for capacitors on the same substrate, an advantage which is not possible with discrete components. Figure 3 shows the dissipation factor as a function of frequency. Thin film capacitors are not as good as high Q ceramic capacitors or polystyrene capacitors with a dissipation factor of about 10 -4, but are general!y better than high K ceramic capacitors. Now let us compare the area required by different capacitors (Figure 4) To summarize: There are good thin film capacitors for many purposes, but an enlargement of the range of capacitance particularly to higher values would be useful, and lower dissipation factors would be welcome.
3) Thin film coils ( Figure 5) reach only to about 1/H, which is a very small value compared with discrete coils up to Henries. The stray field of thin film coils is large, and Q is low. Here thin film components are clearly at a disadvantage. 4) It is obvious (Figure 6) that the resistance of a wire or a copper sheet in a printed circuit is far smaller than the resistance of a thin Film conductor. However, it is not the resistance per mm length but the resistance between two components, which is significant. In thin film circuits, the distance can be made extremdy short-for example 0.2 mm. In this case a gold conductor of 1/am thickness and 0.5 mm width has a resistance of only 20 m2, reducing to 2 m2 with a solder layer of 30/am thickness. However, with a resistor of I2 and a distance to the next component of 6 mm, the gold conductor will give an additional resistance of 720 m2 and an increase in the TCR of 1600.10 -6 [K due to the TCR of gold. Such effects decrease with increasing resistance in the circuit. The inductance of a capacitor between two components becomes extremely small in thin film circuitry because of the dose proximity of components. Figure 7. We have already mentioned that soldered joints are an important reason for failures. Discrete components generally need 4 connections, two outer ones to connect them with other elements, and two interior ones between the component itself and the wires. The reduction in the number of joints depends upon the number of elements and outer connections. Typically, only 1 connection per film component is needed, which makes thin film circuits generally more reliable.

5) Further aspects of integration are shown in
The capacitance of a crossover in printed circuits-one conductor on each side-will amount to 0.04 pF (conductors of 1 mm width and 1 mm distance, permittivity of the insulator e 4).
Astonishingly the capacity of an air gap crossover proposed by Bell Laboratories for thin film circuits is even smaller: 0.006 pF (conductors of 125/am width and 25/am distance, permittivity of the air gap e 1).
Finally let us note some advantages of thin film circuits as a whole. For functional trimming, conventional circuits need potentiometers or trimmers-relatively expensive components. These costs can be saved in thin t'dm circuits by functional adjustment of resistors or capacitors. As thin film components and conductors have a very definite geometry, parasitic effects are exactly reproduceable. Large scale integration is not possible with conventional circuits, but possible with thin film circuits, as very fine and complex patterns can be produced by using thin Film techniques.
Finally, an important feature of thin film technology is the possibility of realizing components with distributed functions, for example with distributed R and C. duction of this paper, we can say: with regard to (1). The miniaturisation of discrete components which slowed down in the sixties, could be continued by using thin film techniques.
With regard to (2) and (3). The reliability of thin film components themselves is at least as good as the reliability of discrete components, and the danger of failures in connections and conductors can be reduced substantially With regard to (4). The critical frequencies, where discrete components change their functions, could be shifted to far higher frequencies by miniaturization of the components and shortening of the conductors.
With regard to (5). The field of microwaves is open for thin f'rims, especially with the strip line technique.
With regard to (6).Because the parasitic effects are exactly reproduceable they can be compensated in most cases.
In this paper we have compared thin film components with conventional components but we did not consider thick film techniques. We are aware of the importance of thick film circuits and that in many practical cases they can be as effective as thin film circuits. A comparison between the two techniques should be the object of a separate paper. On the whole, we think that thin film capabilities are superior, for instance on the grounds of precision, fine pattern generation, low noise and TCR-TCC balancing.
After this very short summary of the state of the art, system engineers are asked to avoid the weak points and use the advantages of thin film techniques.