High Voltage Characterization That Holds Up

A power semiconductor can survive a brief bench check and still fail the measurement that matters. During high voltage characterization, the device, probe station, cabling, fixtures, instrument settings, and safety controls all become part of the result. At leakage currents in the nanoamp or picoamp range, a contaminated insulator or poorly routed cable can be more visible than the device under test.

For engineers evaluating MOSFETs, IGBTs, diodes, GaN devices, SiC devices, or high-voltage integrated circuits, the goal is not simply to apply more voltage. The goal is to obtain repeatable IV, CV, breakdown, leakage, and reliability data without creating a measurement artifact, damaging the sample, or exposing personnel to avoidable risk.

What high voltage characterization must establish

High-voltage testing establishes how a device behaves as electric field increases across its active structures and isolation regions. Depending on the device and development stage, this may include drain-source breakdown, gate leakage, reverse leakage, dielectric integrity, capacitance behavior, dynamic switching performance, and parameter shift under electrical stress.

The distinction between functional testing and characterization matters. A production screen may verify that a device exceeds a specified breakdown voltage at a defined current limit. Characterization work asks more difficult questions: Where does leakage begin to increase? Is the increase repeatable after thermal cycling? Does a guard ring reduce surface leakage? Is the apparent breakdown location on the die, in the package, or in the test fixture?

Those questions require more control than a high-voltage supply and a handheld probe. A usable system must manage applied voltage, current compliance, low-current resolution, shielding, grounding, mechanical spacing, and sample access as one environment.

Build the measurement path before selecting voltage range

The rated voltage of a source-measure unit or power supply is an obvious starting point, but it is not the complete specification. Select the measurement path from the expected device behavior outward.

A 1,200 V SiC MOSFET, for example, may need a voltage source with margin beyond its expected blocking voltage, a current measurement range that resolves off-state leakage, and a compliance setting that protects the device when avalanche begins. A high-voltage diode may require a different balance of voltage range, current capability, and sweep speed. For dielectric or gate-oxide studies, current resolution and guarded connections can matter more than high-current delivery.

The instrument should also match the test method. A slow DC sweep is useful for leakage and breakdown curves, but it can create self-heating near the onset of conduction. Pulsed measurements can reduce thermal effects, although pulse width, duty cycle, cable inductance, and triggering become additional variables. There is no universal preference. The appropriate method depends on whether the engineering question concerns static blocking behavior, electrothermal limits, or transient response.

Source, measure, and protect at the same time

At high voltage, compliance is a protection setting, not an afterthought. Set voltage, current limit, ramp rate, and abort conditions before contacting the device. A current limit that is too high can turn a characterization run into destructive failure analysis. One that is too low may obscure the actual breakdown mechanism or prevent the curve from reaching the region of interest.

Use controlled ramps rather than abrupt steps where possible. Ramping limits charging transients in device capacitance and cabling, gives operators time to observe unexpected behavior, and makes it easier to identify the onset of leakage. For automated testing, software should record the actual source value, measured current, compliance state, timing, and any abort event. A curve without context is difficult to defend later.

The probe station is part of the high voltage characterization system

A high-voltage probe station must provide more than positional access to pads. It needs adequate creepage and clearance around energized points, insulated probe arms where required, stable sample mounting, and an enclosure strategy appropriate to the voltage and lab workflow.

On-wafer testing adds practical challenges because the device is exposed. A probe tip may approach a high-potential drain pad while adjacent structures remain near ground. The chuck, substrate, probe holders, and nearby microscope components need a defined electrical relationship. A poorly considered arrangement can introduce parasitic leakage paths or place accessible hardware at an unintended potential.

For this reason, system configuration should account for the full current return path. Determine which terminal is driven, which terminals are grounded or floated, where shielding is connected, and how the chuck is biased. For back-side bias applications, a conductive chuck and appropriate substrate contact may be required. For front-side-only structures or packaged parts, a custom substrate mount, socket fixture, or guarded test board may produce more reliable results than forcing a wafer-probing setup to do work it was not designed for.

Micron Probing typically approaches these requirements at the system level, pairing probe stations, high-voltage instrumentation, accessories, enclosures, and custom mounting when the application calls for them. That approach reduces the common procurement problem of individually compatible components that do not form a safe or low-leakage test environment together.

Guarding, shielding, and cable selection

When measuring low leakage at high applied voltage, insulation quality is often the limiting factor. Humidity, fingerprints, flux residue, probe-card contamination, and worn cable dielectric can all create leakage that looks like device behavior.

Guarding helps redirect surface leakage away from the sensitive measurement node. It is particularly valuable for gate leakage, dielectric testing, low-current reverse-bias measurements, and capacitance work. The correct guard implementation depends on the instrument and fixture topology, so it should be treated as a designed connection rather than a generic grounding practice.

Shielding serves a different purpose. It reduces noise pickup and can stabilize low-level current measurements, but a shield connection must be compatible with the voltage potential and safety scheme. Use cables rated for the applied voltage, maintain clean connectors, and keep high-impedance paths as short as practical. Triax connections can be beneficial for low-current guarded measurements, while high-voltage coaxial or specialized cable assemblies may be better suited to higher-voltage paths. The right choice depends on voltage, current resolution, connector rating, and instrument architecture.

Control environmental variables that change the curve

A high-voltage curve is not automatically a material property. Temperature, illumination, humidity, and prior electrical stress can alter the result substantially.

Temperature-dependent characterization is common for wide-bandgap power devices because leakage, threshold voltage, on-resistance, and breakdown behavior may shift across the intended operating range. A thermal chuck or temperature-controlled probe station must be evaluated for both temperature accuracy and high-voltage compatibility. Condensation, thermal expansion, and changing contact force can complicate low-temperature or cryogenic work, while elevated-temperature work can accelerate surface contamination and probe wear.

Light-sensitive devices and exposed junctions may require dark testing. Ambient light can generate photocurrent and distort leakage data, especially in small-area devices and photonic structures. A light-tight enclosure is often less expensive than repeated troubleshooting of unexplained current offsets.

Humidity deserves equal attention. If leakage changes between morning and afternoon, the device may not be the variable. Clean insulators, controlled ambient conditions, and a documented pre-test conditioning procedure can improve correlation across operators and labs.

Safety design should not interrupt engineering work

High voltage requires physical safeguards that allow engineers to work efficiently without normalizing risk. A suitable test area generally uses an interlocked enclosure or guard, visible energized-status indication, emergency shutoff capability, properly rated connectors, and a defined discharge procedure. Stored charge in cables, chucks, capacitors, and device structures can remain hazardous after the source is turned off.

The safety strategy should also support the actual workflow. If probe adjustments require frequent enclosure access, the system should de-energize and discharge predictably before access is allowed. If automated sweeps run unattended, the system needs clear fault handling and a safe state after an instrument or software interruption. Avoid improvised high-voltage setups that depend on a single operator remembering every connection sequence.

Documentation matters here as much as hardware. Record maximum voltage, expected stored energy, compliance limits, grounding points, cable ratings, enclosure behavior, and lockout procedures. This is useful for EHS review, but it also prevents a proven setup from becoming an undocumented one-off experiment.

Common reasons high-voltage data fails review

Questionable results often have recognizable causes. The measured leakage may be fixture leakage rather than device leakage. Breakdown may occur in air near a probe tip rather than in the semiconductor. A sudden current rise may be caused by compliance or range switching. Thermal drift may be mistaken for hysteresis. A device may look unstable simply because contact resistance changes during a long sweep.

A disciplined validation sequence helps separate these effects. Measure the open fixture and a known reference structure. Verify cable insulation and instrument zero behavior. Confirm that the result is stable when voltage is held at key points. Repeat sweeps in both directions where hysteresis is relevant, and compare devices across locations on the wafer. If the result changes after cleaning, shielding, repositioning, or changing humidity, investigate the test environment before assigning a physical failure mechanism to the device.

For advanced work, combine electrical data with optical inspection or failure analysis. Visible surface damage, edge termination defects, probe marks, and localized heating can explain electrical behavior that a curve alone cannot. The most efficient characterization workflows use these tools together rather than treating them as separate departments.

High voltage characterization earns confidence when the system can explain its own limits. Start with the device physics, define the measurement uncertainty and safety boundary, then configure the source, station, fixture, enclosure, and software around those requirements. That discipline gives design teams data they can use to make decisions, not just curves that look complete.