Thin Film Preparation

Thin film sensors are usually produced by the deposition of platinum onto a suitable substrate. In order to avoid any problems that might arise at the interface between the platinum layer and the substrate, the following criteria have to be taken into account (3):

Since the coefficients of thermal expansion of platinum and polycrystalline alumina are almost equal, being 7 X 10^-6 /K over the temperature range 50 to 300 K (4), alumina was chosen as the substrate for these sensors.

The production technology has been described elsewhere (5). Platinum layers were deposited, at 580 K, by magnetron sputtering of a platinum target onto alumina substrates. The meander shape was then drawn by laser evaporation of the layer, using a pulsed Q-modulated Nd:YAG laser (λ = 1.06 um) with pulse repetition rate of 20 Hz and train energy (for 10 pulses in a train) of 50 to 60 mj. This technique is faster and more cost effective, and avoids the mask writing and chemical treatment that are necessary for lithographic patterning.

However, laser treatment of the layer parameters is not controllable, and so the final characteristics of the layer cannot be predicted.

Temperatnre/Resistance Values of the Layer

An assessment of the purity of the platinum target was made in accordance with ITS-90. It was found that the resistance ratio at the boiling point of water, W(H₂O) b.p.) > 1.3927 (W = R_100°C/R_0°C, where R is the electrical resistance of the platinum layer). However, contrary to this, the sensors that were produced had a resistance ratio W(H₂O b.p.) = 1.385. In addition the resistance ratio at 4.2 K, W = R_4.2K/R_0°C was about 10_-2, which is two orders of magnitude greater than that proposed by ITS-90 for the most accurate thermometers.

An earlier investigation with this type of sensor showed that they required individual calibration in order to achieve an accuracy of ±10 mK (6). The results for the percentage variation of the resistance with temperature for two typical sputtered platinum/alumina sensors, prepared by the same method, serial numbers 023 and 028, are presented in the Table. Their behaviour is characteristic of this type of sensor.

The sensors show significant differences in their values for resistance below 50 K. This can be explained (via Matthiessen’s rule) by any residual relative resistance. However, there is no reason to suppose that there are any impurities present in the layer during the platinum deposition process.

Eleven sensors of the same type were repeatedly subjected to thermal cycling between 60 to 320 K, and measurements of resistance were taken at 48 temperature points during every cycle. A visible shift in the temperature/resistance, T(R), function was observed up to the sixteenth cycle and only after 20 cycles was saturation achieved. These results may be due to both the laser processing of the layer during fabrication of the meander shape and/or to interaction between the layer and the substrate. The effects of the laser processing can be observed by scanning electron microscopy (SEM).

Effects of the Laser Treatment

A SEM micrograph of a part of sensor 028 is shown in Figure 1 (a). The sensor was covered with a glass frit to isolate the layer electrically. However, where the glass frit acted as an electrical isolator the SEM micrographs of the platinum layer were spoilt. Therefore the cover was removed, but small amounts of glass frit which had been deposited in the laser-drawn lines still remained. For this reason, in Figure 1 (a), Figure 1 (b), Figure 1 (c) and Figure 1 (d), lighter (white) lines appear on part of the sensor.

Thus laser treatment of the sensors appears to create some problems. First, the rough edges of the treated layer are clearly visible; these cause a reduction in the cross-sectional area of the platinum film. In Figure 1 (b), which has twice the resolution of Figure 1 (a)., it can be seen that in some places the dimensions of the vertical cross section have become too small. Any mechanical tensions which have arisen as a result of the high temperature laser treatment will be increased in these cross-sectional areas and affect the temperature/resistance ratio.

During laser patterning evaporated material is redeposited onto the nearby film; this is shown by the spots in Figure 1 (c). Some of these drops have dimensions equal to the width of the meander line layer, see Figure 1 (d). It is therefore important to know both the structure and the composition of the drops, but this analysis is made difficult by the narrowness of the layers. If the drop is predominantly composed of alumina the stress will be larger, since alumina vaporises at a higher temperature than platinum. However, as a result of the large temperature difference between each drop and the layer on which it is deposited, strong mechanical tensions will occur in the cross section, causing the platinum lattice to become overstrained. It is suggested therefore that the electrical properties of the layer in this section become the determining ones for the sensor.

Some of these defects could be avoided if techniques used in the preparation of high transition temperature superconducting films (7) were employed. The film could, for example, be covered with photoresist prior to laser treatment and any redeposited material can then be removed by dissolving the resist in ethanol.

To sum up, we have found that the T(R) function in thin film platinum sensors used for temperature measurement is affected in an uncontrollable manner by laser treatment of the layer. Changes in the meander shape are caused by laser processing.

For this reason sensors fabricated by this method display a non-uniform temperature/resistance dependence. It is also suggested that due to the diffusion of aluminium atoms into the platinum lattice, changes in the electrical properties of the layer may also take place.