Microwave photoconductivity decay measures how quickly a wafer's excess free-carrier population, injected by a short laser pulse, relaxes back toward equilibrium, but instead of touching the wafer with a probe or a contact, it senses that relaxation through the way the wafer reflects a microwave signal. A silicon wafer with more free carriers present reflects microwave power differently than the same wafer at equilibrium, so watching the reflected microwave power fall after a laser pulse traces the same recombination physics a contact measurement would reveal, without the risk of introducing a contact-induced defect or leaving a probe mark on production material. Because the technique never touches the wafer, it has become the workhorse contactless lifetime method across silicon wafer manufacturing, solar-cell processing, and fab-floor contamination screening alike.

A short laser pulse injects excess carriers into the wafer, and the resulting change in bulk conductivity is sensed as a shift in the microwave power reflected from a waveguide antenna positioned above the surface. A typical µW-PCD system pumps the wafer with a pulsed diode laser near 904 nm to 1064 nm, chosen for a penetration depth that samples a useful fraction of wafer thickness, while a microwave source operating near a 10 GHz X-band frequency continuously illuminates the same spot through a waveguide or antenna. As the injected excess carriers recombine, the wafer's conductivity falls back toward its equilibrium value, and the reflected microwave power tracks that conductivity change point for point, giving a transient signal that can be digitized and fit without ever placing a physical probe on the wafer surface. Spot size on the wafer is typically held between 1 mm and 3 mm, small enough to resolve localized defects while still averaging over enough area to produce a stable, repeatable reading.

Extracting a lifetime value from the recorded decay transient requires fitting the reflected-power-versus-time trace to an exponential model over a window chosen to avoid both pump-pulse artifacts and long-time noise floor effects. A single-exponential fit through the decay tail typically yields a lifetime figure with the fit window starting roughly 1 µs after the pump pulse ends, skipping the fast initial transient that reflects instrumentation response rather than genuine recombination physics. A representative silicon wafer site might show a decay lifetime near 40 µs, with the reflected power falling to below 5% of its initial value by the time the fit window closes. When the log of reflected power departs from a straight line partway through the decay, that curvature itself is diagnostic, since it commonly signals trap-assisted recombination through a defect level rather than simple band-to-band recombination, and a fit that ignores the curvature will report a lifetime number that depends on an essentially arbitrary choice of fit window.

Scanning the pump-and-probe spot across the wafer converts a single decay measurement into a full lifetime map that reveals spatial non-uniformity invisible to any single-point reading. A production mapping recipe typically steps the measurement spot at a pitch of 2 mm to 5 mm across a 300 mm wafer, building a map with enough points to resolve a contamination ring or a process-tool signature while still completing the scan in a practical amount of time. Low-lifetime clusters on the resulting map frequently correlate with specific wafer positions handled by a particular chuck pin or edge-contact fixture, letting a process engineer trace a lifetime anomaly back to a specific piece of hardware rather than treating it as a generic yield loss. Because the measurement never contacts the wafer, the same site can be rescanned repeatedly through subsequent process steps to track how a lifetime signature evolves, something a destructive or contact-based method cannot offer.

Microwave photoconductivity decay is exceptionally sensitive to trace iron contamination, and the iron-boron pairing effect gives it a built-in method for confirming an iron signature rather than merely inferring one from a low lifetime alone. Interstitial iron pairs with substitutional boron in as-grown p-type silicon, and that pairing measurably raises the recombination lifetime relative to the same wafer with iron in its dissociated state; illuminating the wafer briefly to break the iron-boron pairs and then remeasuring lifetime produces a characteristic before-and-after shift that a simple contamination source cannot mimic. A lifetime that recovers by 50% or more after an iron-boron dissociation split is strong confirmatory evidence of iron contamination rather than another metal species or a generation-lifetime artifact. Because that diagnostic split can be run in a couple of quick rescans rather than a full destructive analysis, it lets a fab confirm an iron excursion on the same shift it was first flagged. A typical dissociation illumination step runs for tens of s at moderate intensity, after which the wafer is left in the dark for a comparable interval before the confirmation rescan, and a shift of less than 5% on that rescan is usually treated as within normal measurement noise rather than a genuine iron signature.

Surface recombination velocity and bulk recombination lifetime both shape the measured µW-PCD decay, and separating the two requires either a passivated reference sample or a model that accounts for surface effects explicitly. A poorly passivated wafer surface can hold surface recombination velocity above 1000 cm per second, enough to dominate a measured lifetime and mask a genuinely clean bulk behind an artificially fast decay, while a well-passivated surface with recombination velocity below 10 cm per second lets the bulk lifetime dominate the signal instead. Passivation films used for dedicated lifetime test wafers are typically 50 nm to 100 nm thick silicon nitride or thermal oxide layers, thick enough to suppress surface states without materially perturbing the microwave reflection measurement itself. Comparing a µW-PCD scan taken before and after a passivation anneal step is a standard way to confirm whether a low lifetime reading reflects genuine bulk contamination or simply an unpassivated surface. A forming-gas anneal near 400 °C for a fixed interval is a common passivation-quality lever, and a wafer that gains more than 20% in measured lifetime after that anneal step is generally judged to have been surface-limited rather than bulk-limited before the anneal.

Microwave photoconductivity decay, quasi-steady-state photoconductance, and photoluminescence lifetime mapping each occupy a different sweet spot, and a mature metrology program uses more than one to cross-check results. Quasi-steady-state photoconductance typically outperforms µW-PCD for lifetimes below 5 µs because its steady-state excitation avoids some transient-fitting ambiguity, but it requires a coil-coupled contact geometry that µW-PCD avoids entirely. Photoluminescence lifetime mapping resolves spatial detail down to an excitation spot near 5 µm, finer than the 1 mm to 3 mm spot typical of µW-PCD, but it requires optical access and a more elaborate detection chain. Correlation studies comparing Semilab µW-PCD scans against PL-mapped lifetime on the same wafer set typically show agreement within 15% to 20% once both are calibrated against a common reference standard, giving engineers confidence to treat the two methods as complementary rather than redundant checks on the same physical quantity. A four-point probe sheet-resistance map and a Hall effect mobility measurement are frequently pulled from the same lot to separate a doping-concentration signature from a genuine recombination-lifetime effect, since a low µW-PCD reading can sometimes trace back to a resistivity outlier rather than a contamination event. AFM surface imaging is occasionally added on a flagged site to rule out a topographic artifact, such as a scratch or particle, before a lifetime anomaly is escalated as a bulk contamination excursion.

Laser pulse injects excess carriers into wafer → Microwave antenna senses reflected power shift → Digitize reflected-power decay transient → Fit exponential model over selected time window → Extract lifetime τ at each measurement site → Raster scan to build wafer lifetime map → Cross-check against QSSPC and PL mapping references → Flag low-lifetime sites for SIMS and DLTS contamination follow-up

Viewed through a contactless-lifetime metrology engineering lens, microwave photoconductivity decay earns its permanent place on the wafer-quality bench because it turns an entirely non-invasive microwave reflection measurement into a quantitative, repeatable proxy for the same recombination physics that governs solar-cell efficiency, transistor leakage, and downstream yield, letting the same wafer be rescanned through an entire process flow without ever risking the contact-induced damage a probe-based lifetime method would introduce.