Access Official Technical PDF

Share Summary:LinkedInEmail

For high-power silicon impact ionization avalanche transit-time (IMPATT) diodes, the quality of an ohmic contact affects how the device performs under demanding operating conditions. Pulsed operation can produce substantial heating, making it important to understand how specific contact resistance changes with temperature. A temperature-based diagnostic method for multilayer silicon structures offers manufacturers a way to assess contacts during diode development and fabrication, including at intermediate process stages.

Why contact resistance matters in IMPATT diode manufacturing

An ohmic contact is intended to conduct without acting as a rectifying junction. In a high-power diode, excessive resistance at the metal–semiconductor interface can contribute to electrical losses and influence device operation as temperature rises. The study sets a target of less than 10-5 Ω·cm² for the specific contact resistance of ohmic contacts to n+-silicon in powerful silicon IMPATT diodes.

Contact behavior depends not only on the metal but also on the transition layer formed where the metallization and silicon interact. In the investigated structure, the metal stack was Pd–Ti–Au, deposited on heated silicon. The silicon itself had an n+-n-n+ doping profile, so the vertical test structure included layers with different electrical properties. This non-uniform doping makes it difficult to determine contact resistance using classical approaches designed for simpler structures.

Measuring specific contact resistance in multilayer structures

The diagnostic approach uses the temperature dependence of the total resistance measured through a vertical structure. Because the measured value includes contributions from both the contacts and the semiconductor layers, interpreting the data requires accounting for the silicon’s volume resistance as it changes with temperature. The proposed method generalizes the Cox–Strack approach for multilayer structures with non-uniform doping.

In practice, this means contact resistance is not inferred from a single room-temperature measurement. Instead, measurements across a broad temperature range are compared with a model of the semiconductor’s temperature-dependent electrical resistance. This provides a route to estimate the contact contribution while recognizing that the layered structure cannot be treated as a uniform semiconductor sample.

The study examined structures from 100 to 360 K. It considered both unetched structures and etched mesa structures, reflecting different vertical geometries that can arise in diode fabrication. The method is intended to support evaluation of ohmic contacts between etching steps as well as during development of process conditions.

What temperature reveals about current flow

Temperature-dependent measurements can help identify the dominant mechanisms by which carriers cross a metal–semiconductor barrier. These mechanisms include field emission, thermal-field emission, and thermionic emission. Their relative importance depends on factors such as doping concentration, temperature, and barrier properties.

For the heavily doped silicon in the tested structures—where the donor concentration was approximately 1020 cm-3—a field-dominated mechanism might be expected across the full temperature range. The analysis, however, found that the measured resistance did not follow a field-emission model across the range, nor could it be explained by a thermal-field model alone.

A mixed mechanism across the temperature range

At lower temperatures, particularly around 100 to 150 K, the temperature dependence was consistent with thermal-field emission. At higher temperatures, above approximately 300 K, the resistance showed a dependence resembling thermionic emission. The study’s interpretation is that thermionic current passes through localized regions whose effective area is much smaller than the full contact area. Thermal-field current, by contrast, contributes across the broader contact area.

Combining these contributions as parallel current paths produced a model that better described the observed temperature dependence. This is an important measurement lesson: a contact may not have one dominant transport mechanism across all operating temperatures. A model based on only one mechanism can therefore miss relevant behavior or misrepresent the contact resistance.

Process implications of the Pd contact layer

The metal stack used a 20 nm palladium layer as the contact-forming layer, with titanium and gold above it. The authors compared the results with earlier work on contacts using a thicker palladium layer, near 30 nm, where current flow through metallic shunts associated with dislocations had been observed. In the 20 nm-layer structures, the measured resistance decreased with increasing temperature, and the analysis did not indicate the same shunt-dominated behavior.

The proposed explanation is that a thinner palladium layer may produce smaller crystallites in the palladium silicide contact-forming layer. Smaller crystallites could reduce mechanical stress concentrations and, in turn, the likelihood of dislocation formation. This is presented as an explanation for the observed difference, not as a universal rule for every contact process. The results underline why metallization thickness and the resulting interface structure should be considered when evaluating electrical performance.

Using temperature testing for quality control

For device manufacturers and process engineers, the practical value of the method is its application to structures with layered doping and temperature-sensitive resistance. A useful evaluation workflow can include:

  • Measure vertical resistance over a relevant temperature range rather than relying only on a single-temperature reading.
  • Model the semiconductor layers so their contribution to total resistance is accounted for.
  • Assess transport mechanisms across the range, allowing for different mechanisms to dominate at different temperatures.
  • Compare process variants, such as etched and unetched structures, to identify differences in contact behavior.

This approach supports electrical characterization during process development and offers a potential means of monitoring contacts at intermediate manufacturing stages. It is particularly relevant when a simple contact-resistance method cannot adequately represent the structure’s non-uniform doping.

Conclusion

Reliable ohmic contacts are a key consideration in high-power silicon IMPATT diode fabrication, especially where operation involves substantial heating. Temperature-dependent resistance testing, combined with a model for the multilayer semiconductor, can provide a more informative estimate of specific contact resistance than methods intended for uniform structures. By distinguishing contact behavior from the resistance of the silicon layers—and by recognizing that current transport can change with temperature—manufacturers can better evaluate contact formation and strengthen electrical process control.

Access Official Technical PDF

About The Author