PL, EL And AOI: Complementary Perovskite Module Defect Checks

A perovskite defect inspection system needs several forms of contrast because a module defect can affect appearance, recombination, electrical injection, or more than one of these at once. PL maps the response to optical excitation. EL maps emission under electrical injection. AOI examines reflected or scattered light from physical features. Combining them helps a production team locate abnormalities and choose the next verification step. The useful output is a registered defect record tied to the process stage and electrical impact, rather than three unrelated images or a single brightness threshold applied to every module.

Match the inspection method to the defect question

PL, EL, and AOI interrogate different aspects of a module. Their agreement strengthens a diagnosis; their disagreement often reveals which follow-up measurement is needed. None is a universal substitute for the others.

Common inspection targets include coating nonuniformity, local dark regions, particles, pinholes, cracks, patterning anomalies, and electrically inactive areas. Some features are visible before contacts are completed; others become interpretable only after a functioning device exists. Decide which process intervention is possible at each stage before specifying the cameras and scan sequence.

A practical perovskite defect inspection system should preserve the distinction between a detected feature, a suspected mechanism, and a confirmed failure. For example, “low PL region” is a measurement label. “High nonradiative recombination” is an interpretation that needs controlled optical conditions. “Reject” is a production decision that should depend on a validated acceptance rule.

Use PL to screen optical and recombination nonuniformity

PL imaging is useful before electrical contacting and after device completion because excitation does not require injected current. Under controlled conditions, it can reveal spatial differences associated with recombination and absorber quality.

However, PL intensity is not a defect concentration map. A region can become darker because it absorbs less excitation, emits into a different optical path, extracts charges more efficiently, or suffers stronger nonradiative losses. Thickness variation, texture, interference, and spectral shifts can all change the collected intensity without producing the same electrical consequence.

For process comparisons, hold excitation wavelength, irradiance, collection geometry, exposure, and electrical boundary condition constant. Open-circuit and extracting conditions can produce different images of the same device. Record whether the sample was electrically floating, contacted, biased, or loaded rather than treating all PL maps as equivalent.

Useful follow-up measurements include PL spectra, excitation-dependent PL, TRPL, and calibrated luminescence analysis. Spectral measurements can reveal whether a fixed camera band misses shifted emission. A lifetime map adds timing information, but neither intensity nor lifetime alone uniquely identifies every failure mechanism.

At the coating stage, combine PL with an optical appearance image. A dark region that also contains an obvious coverage anomaly should be investigated differently from a dark region with no visible surface feature. This comparison narrows the diagnostic path while retaining uncertainty about microscopic origin.

Use EL to expose problems under electrical injection

EL requires a device configuration that can accept controlled electrical injection. It is particularly useful for checking whether different regions respond consistently once contacts and interconnections are present.

Weak EL can reflect poor injection, contact or interconnection problems, local recombination losses, or nonuniform voltage distribution. A region that remains bright in PL but becomes dark in EL is therefore a candidate for electrical-path investigation. It is not automatic proof of one specific contact failure.

Measure and record injection current, voltage, polarity, dwell time, and sample temperature. Comparing modules at identical terminal voltage may not mean comparing them at identical injected current or local operating conditions. A carefully selected current series can reveal whether a contrast pattern changes as injection increases.

PL and EL can be complementary even when both images look similar. If both show the same low-emission area, absorber or interface losses become plausible, but optical shading and collection differences still need to be excluded. If a feature appears only after contacting, align it with the process history before deciding whether the cause lies in the deposited contact or an earlier latent defect.

A perovskite defect inspection system should distinguish camera acquisition time from the time required to contact the module, establish bias, stabilize, acquire, and release it. Short exposure specifications do not by themselves establish that the complete station fits the production takt.

Use AOI for physical features and pattern inspection

AOI uses image contrast from illumination and collection geometry to identify physical or appearance deviations. Its role includes locating features that luminescence measurements may not characterize clearly, such as surface particles or pattern irregularities.

Bright-field and dark-field configurations emphasize different contrasts. Select illumination angles, polarization where useful, magnification, and focus for the actual substrate, coatings, and encapsulation. Reflections and texture can create false features if an inspection recipe developed for one stack is transferred without validation.

Pattern inspection should use the intended geometry as a reference. Scribe continuity, line placement, edge coverage, and dimensional deviations need suitable optical resolution and validated image processing. A nominal pixel pitch is not the smallest defect that can be detected reliably: contrast, blur, motion, sampling, and classification rules also matter.

Keep AOI results separate from PL and EL until the spatial registration is verified. A surface feature may obscure emission, coincide with an electrical issue, or be harmless at the observed size. Correlation helps prioritize verification; it should not silently turn an appearance label into an electrical failure label.

Build a registered workflow from coating to finished module

A useful workflow captures the earliest actionable evidence and follows the same location through subsequent operations. Registration and measurement repeatability are prerequisites for interpreting defect evolution.

1. After absorber formation: use AOI and PL to record coverage, appearance, and emission nonuniformity before later processing hides the original state.

2. After relevant patterning and contacting: repeat optical checks and introduce EL when the electrical structure supports it.

3. After interconnection or packaging: inspect for newly introduced changes using the appropriate optical access and electrical conditions.

4. For flagged locations: compare registered PL, EL, and AOI regions, then select spectral, electrical, or microscopic verification.

5. For process feedback: connect confirmed defects to lot, position, tool, recipe, and process stage before changing production settings.

Use fiducials or stable geometric features and account for image distortion, different magnifications, and the side of observation. If one modality observes through glass and another views the opposite surface, orientation mistakes can create convincing but false correlations. Store the coordinate transform and its residual error.

Classify findings with confidence levels and retain uncertain cases. Training labels should come from verified examples rather than being inferred from brightness alone. Measure false calls and missed defects using representative modules, including acceptable texture and edge regions. A classifier that works only on clean demonstration images is not a production acceptance method.

Specify an inspection station around process decisions

Choose inline inspection when the required measurements can be completed reliably within the available handling and cycle time. Use offline investigation when diagnostic flexibility and additional measurements matter more than immediate line disposition.

Define the minimum relevant defect size, acceptable contrast variation, module dimensions, optical access, handling constraints, and required data connections. Validate the complete cycle, including transport, contacting, exposure, analysis, and reporting. Recheck recipes after material or stack changes because a different emission spectrum or reflectance can alter contrast without changing the defect population.

Time Tech Spectra’s integrated inline inspection system explicitly combines PL, EL, and AOI, with line-scan and area-scan configurations. Its listed capabilities include bright-/dark-field imaging, defect annotation and classification, and tracking across process stages. Those are distinct measurement channels within an integrated perovskite defect inspection system, rather than AOI functions being inferred from PL or EL alone.

The listed typical spatial sampling is approximately 70 μm/pixel, with sample dimensions and configurations customizable. Treat this as a configuration parameter, not a guaranteed minimum detectable crack size. For a production evaluation, use representative good and defective modules and agree on measured detection performance, repeatability, and total station cycle time.

Conclusion

PL, EL, and AOI provide complementary evidence about a perovskite module. PL reveals emission nonuniformity under optical excitation, EL tests response under electrical injection, and AOI locates physical and appearance features. A perovskite defect inspection system becomes most useful when registered maps, controlled conditions, and validated acceptance rules connect those signals to process action. Time Tech Spectra, a spectroscopy company, provides an integrated inline platform covering the three modalities. Qualify the complete workflow on representative modules so that defect maps support consistent decisions rather than unverified explanations based on image brightness.

FAQ

Can a perovskite defect inspection system replace electrical testing?

Imaging complements electrical testing. Validate any image-based acceptance rule against the electrical properties and reliability outcomes relevant to the product.

Can PL inspect a film before electrodes are deposited?

Yes, if the film can be optically excited and its emission collected. Interpretation still depends on substrate, absorption, and measurement conditions.

Should all dark pixels be rejected?

No. Mask known edges and geometry, correct instrumental nonuniformity, and use validated defect rules. A dark pixel alone is neither a unique mechanism nor a universal rejection criterion.

How should an inspection recipe be transferred to a new stack?

Revalidate excitation, spectral collection, appearance contrast, bias conditions, registration, and classification thresholds using representative samples of the new stack.