Access Official Technical PDF

Share Summary:LinkedInEmail

Semiconductor measurement methods provide the evidence engineers need to characterize materials, monitor fabrication processes and evaluate finished devices. A coordinated measurement program can help suppliers and users apply consistent specifications to wafers, thin layers, interconnects and devices. Work described in a 1969 National Bureau of Standards quarterly report illustrates a still-relevant principle: measurement results are useful only when the test method, its limitations and the sources of variation are understood.

Why semiconductor measurement methods matter

Semiconductor characteristics are influenced at several stages, from crystal and wafer properties to processing conditions and device construction. Measurements therefore serve different purposes: some characterize the starting material, others control fabrication, and still others assess device behavior. Comparing results across organizations also requires repeatable procedures and a shared understanding of what each test actually measures.

The report grouped its work into these three areas. Its investigations included resistivity and carrier lifetime, electrical nonuniformity, metallization adhesion, die attachment, wire bonds, thermal properties and microwave devices. The broader operational lesson is to treat measurement as part of process control—not merely as final inspection—and to select methods appropriate to the material or structure under test.

Material characterization: resistivity and carrier lifetime

Resistivity depends on method and specimen

Four-probe measurement was being evaluated for silicon wafers across a wide resistivity range, as well as for epitaxial layers. Researchers examined the effects of measurement current, probe force and surface finish because these variables can influence the result. In tests on lapped wafers, most samples showed less than a one-percent change across the currents used, although a slightly higher reading at the lowest current was generally observed. That finding did not remove the need to establish suitable conditions for other materials and surfaces.

Alternative techniques addressed different measurement needs. Spreading resistance was used for depth and radial profiling, while capacitance-voltage measurements estimated carrier density as a function of depth. The latter was particularly valuable as a reference for thin layers on conducting substrates, even though it could be more time-consuming and had limits on the layer thickness it could probe. For production or laboratory use, method selection should reflect the structure being measured, and results from different techniques should not be assumed interchangeable without comparison.

Lifetime measurements require careful comparison

Carrier lifetime was studied using photoconductive decay, surface photovoltage and metal-oxide-semiconductor capacitance techniques. Measurements on some silicon and germanium samples did not agree between surface photovoltage and photoconductive decay. The reason was not yet clear, so repeat measurements and changes to surface conditions were planned. This illustrates why a discrepancy between methods should prompt investigation rather than an immediate conclusion that one result is correct.

Diode measurements raised a related issue: lifetime values obtained by voltage decay and reverse recovery differed. The report found no theoretical basis for the observed difference and noted that earlier measurements had used substantially different current levels. Circuits were to be modified so that the techniques could be compared under matched conditions. For engineers validating a measurement procedure, controlling test conditions is essential to separating real device behavior from differences in how tests are run.

Monitoring electrical inhomogeneity

Resistivity gradients across circular wafers can affect material uniformity and device suitability. The report examined a photovoltaic method that relates photovoltage and photoinduced resistance change to radial resistivity variation. Using knife-edge contacts was considered more practical for production monitoring than specially prepared contacts, while also allowing average wafer resistivity to be measured using the van der Pauw technique.

The analysis identified an omitted factor in the earlier equation, and correcting it improved agreement between photovoltaic and probe-based resistivity profiles. Differences remained, particularly near wafer edges, where measurement resolution and contact effects matter. The practical takeaway is that process-control methods need validated calculations as well as repeatable hardware. Contactless or minimally damaging approaches may also be useful where preserving a polished surface is important, but their measurement limits must be established.

Process control: adhesion and wire bonding

Use a meaningful criterion for metallization adhesion

For aluminum-film adhesion, the scratch test was assessed using a threshold adhesion failure concept. Rather than defining failure as complete film removal, this criterion identifies the load at which any film removal becomes detectable. The distinction changes how test results are interpreted and can make comparisons more sensitive to the onset of adhesion failure. Stylus-tip shape was also investigated, including examination with a scanning electron microscope, because tip condition affects the test interaction with the film.

Characterize the ultrasonic bonding system

Wire bond strength is influenced by the bonding system, not only by the nominal settings on the equipment. Measurements of transducer and tool vibration showed that the mechanical response differed among power supplies. The report also examined ways to reduce mechanical quality factor, or Q, including modifying the transducer horn, and assessed swept-frequency operation as a way to limit the effects of resonant-frequency shifts.

These system-level studies supported efforts to produce reproducible bonds before evaluating pull strength and other bond tests. Scanning electron microscopy was used to inspect bond features, including heel cracks, clamp marks and patterns left after a bond was removed. For manufacturing teams, combining strength tests with process characterization and physical inspection can provide a more informative picture than relying on a single pass/fail result.

Device measurements and practical implementation

Device characterization included thermal resistance measurements and investigations of transistor behavior associated with hot-spot formation. The thermal measurement apparatus was being partially automated, while microwave-device measurement work included preparation for mixer-diode noise-parameter testing. These activities underscore the need to connect electrical measurements with thermal behavior where device performance is affected by both.

Across materials, process control and devices, the report’s common theme is measurement discipline. Define the test objective, control relevant variables, understand spatial and structural limits, and compare methods under equivalent conditions. Round-robin testing and standards work can help determine whether a procedure is sufficiently consistent for use between laboratories, suppliers and customers.

Conclusion: build measurement into process control

Semiconductor measurement methods are most effective when they are treated as validated tools for decision-making. Resistivity and lifetime tests need controlled conditions; wafer-uniformity methods need sound analysis and known resolution; and adhesion and wire-bond evaluations benefit from examining both test results and the physical process. For automation engineers, integrators and manufacturing leaders, the practical priority is to select methods suited to each material or device, document the conditions that affect results, and use comparable procedures to support reliable specifications and process control.

Access Official Technical PDF

About The Author