How to Learn Lock-In Thermography (LIT): From Periodic Heating and Infrared Phase Maps to Subsurface Defects, Solar Cells and Electronic Failure Analysis

## Wait, What? A Phase Image Can Reveal a Buried Defect Even When the Surface Temperature Barely Looks Different Ordinary thermography asks which pixels are hotter. Lock-in thermography asks which pixels oscillate at the heating frequency, and how delayed they are. Periodically heat the sample. Record an infrared movie. At every pixel, extract the response at exactly the driving frequency. Two new images appear: – amplitude; – phase. > **LIT is not merely an IR camera with repeated heating. It is a frequency-domain thermal-wave experiment whose depth sensitivity and defect contrast are controlled by modulation frequency, diffusivity, emissivity and heat-source geometry.** ## The One-Sentence Answer **Learn LIT by tracing periodic heating → thermal diffusion → infrared temperature oscillation → lock-in amplitude and phase → defect contrast, then add emissivity, nonuniform heating, lateral diffusion, modulation frequency and inverse thermal modelling before converting a phase lag into defect depth or a hotspot into one failure mechanism.** # Beginner Layer — Periodic Heating ## Stage 1: Apply a Repeating Thermal Excitation Possible heat sources include lamps, lasers, electrical dissipation, eddy currents and ultrasound. ## Stage 2: The Surface Temperature Oscillates The response follows the driving frequency with an amplitude and a phase delay. ## Stage 3: An Infrared Camera Records the Time History Each pixel becomes its own thermal time series. # Lock-In Demodulation ## Stage 4: Multiply the Signal by Reference Sine/Cosine Functions Pixelwise lock-in processing extracts the response at the excitation frequency. ## Stage 5: Build an Amplitude Image Amplitude reports how strongly the pixel participates in the periodic thermal response. ## Stage 6: Build a Phase Image Phase reports how delayed the thermal oscillation is relative to excitation. ## Stage 7: Phase Often Suppresses Slowly Varying Background Better Than Raw Temperature This is one reason buried defects can emerge from weak signals. # Thermal Diffusion Length ## Stage 8: Periodic Heating Produces a Characteristic Thermal Depth For a homogeneous material: **μ = √(α/πf)** where α is thermal diffusivity and f modulation frequency. ## Stage 9: Low Frequency Probes Deeper But requires longer measurement time. ## Stage 10: High Frequency Probes Shallower It can also improve lateral localization. > **frequency is a depth-control knob.** # Subsurface Defect Layer ## Stage 11: A Void or Delamination Changes Heat Flow The defect reflects, blocks or redistributes the thermal wave. ## Stage 12: Surface Amplitude and Phase Change Above the Defect The defect can appear even though it is physically buried. ## Stage 13: Defect Contrast Depends on Depth and Size Small deep defects become increasingly difficult. # Phase Versus Amplitude ## Stage 14: Amplitude Can Be Strongly Affected by Nonuniform Heating ## Stage 15: Phase Can Be More Robust to Multiplicative Intensity Variation ## Stage 16: Phase Is Not Immune to Every Artifact Emissivity, nonlinear heating and complex boundary conditions can still affect it. # Emissivity Layer ## Stage 17: IR Cameras Measure Radiance, Not Temperature Directly Surface emissivity and reflected ambient radiation matter. ## Stage 18: Lock-In Processing Reduces Some Static Emissivity Effects But changes in dynamic emissivity or surface coatings can remain. ## Stage 19: A Black Paint Layer Can Improve Optical Heating and IR Uniformity It can also change the thermal boundary and may be unacceptable for heritage or device samples. # Nonuniform Heating and Lateral Diffusion ## Stage 20: Illumination Is Rarely Perfectly Uniform Lamp angle, laser profile and surface reflectivity create spatial variation. ## Stage 21: Heat Spreads Sideways A small defect can appear larger than it is. ## Stage 22: Spatial Resolution Worsens at Low Frequency Deeper probing comes with more lateral diffusion. ## Stage 23: Defect Size Must Be Reconstructed With a Forward Model # Frequency Sweeps ## Stage 24: One Frequency Gives One Thermal Scale ## Stage 25: Multi-Frequency LIT Builds a Depth-Sensitive Dataset A defect that emerges at low frequency but disappears at high frequency is likely deeper. ## Stage 26: Frequency-to-Depth Conversion Is Not One Universal Formula Finite thickness, anisotropy and convection modify the simple semi-infinite model. # Composites and Aerospace NDT ## Stage 27: Delaminations in CFRP Are Classic LIT Targets A delamination changes through-thickness heat flow strongly. ## Stage 28: Fibre Orientation Makes Thermal Diffusion Anisotropic In-plane heat spreads faster along conductive fibre directions. ## Stage 29: 2026 Work Continues Quantitative Delamination Sizing Recent work explicitly tackles geometric sizing of wedge-like hidden delaminations. # 2026 Model-Based Frontier ## Stage 30: Finite-Element Thermal-Wave Models Are Becoming More Central Current 2026 work uses unified structural/thermal formulations to model LIT sensitivity to subsurface defects. ## Stage 31: Quantitative LIT Is Moving Beyond “See the Defect” The stronger question is: > what size and depth are consistent with the amplitude–phase response across frequency? # Electronics Failure Analysis ## Stage 32: The Device Itself Can Be the Periodic Heat Source Pulse electrical bias. Weak leakage or short-circuit regions heat periodically. ## Stage 33: Lock-In Detection Can Localize Heat Sources Far Below Raw Camera Noise The method is widely used for semiconductor and power-module failure analysis. ## Stage 34: A Thermal Hotspot Is Not the Electrical Root Cause by Itself The hotspot localizes where power is dissipated. FIB, SEM and electrical analysis identify the mechanism. # 2026 Electronic Packaging Frontier ## Stage 35: Periodic Eddy-Current Heating Can Be Combined With Lock-In Analysis Current work on electronic packaging uses multi-pulse eddy-current thermography, analyzed at the fundamental frequency, to expose subsurface defects in multilayer packages. ## Stage 36: Internal Heat Sources Change the Forward Model Optical surface heating and eddy-current volumetric heating do not have identical depth response. # Photovoltaics Layer ## Stage 37: Illuminated Lock-In Thermography Can Map Solar-Cell Losses Periodically illuminate or electrically bias a cell. ## Stage 38: Local Heating Can Reveal Shunts and Recombination Losses ## Stage 39: Thermal Power Is Not Directly Electrical Efficiency A quantitative solar-cell model is required. # Pulsed Thermography Versus LIT ## Stage 40: Pulsed Thermography Uses a Broadband Thermal Excitation A short pulse contains many temporal frequencies. ## Stage 41: LIT Uses a Narrow Frequency Band It trades measurement time for phase-sensitive noise rejection. ## Stage 42: Neither Method Is Universally Better Professional science does not turn one technique into a winner in every regime. # Vibrothermography / Ultrasound LIT ## Stage 43: Ultrasound Can Heat Defects Through Friction or Crack-Face Interaction ## Stage 44: The Resulting Periodic Heating Can Be Lock-In Detected The heat-source physics differs from optical heating. # Thermal-Diffusivity Layer ## Stage 45: Phase Propagation Can Constrain Thermal Diffusivity ## Stage 46: Anisotropic Materials Require Tensor-Aware Interpretation # Machine-Learning Layer ## Stage 47: ML Can Segment Weak Defect Signatures ## Stage 48: It Can Also Learn Surface Texture, Paint or Illumination Artifacts ## Stage 49: Multi-Frequency Physics Constraints Improve Robustness A credible learned defect should follow expected frequency/depth behavior. # Professional Layer ## Stage 50: Separate Five Objects 1. true defect/heat source; 2. periodic thermal excitation; 3. thermal diffusion field; 4. IR amplitude/phase image; 5. inferred defect geometry or failure mechanism. ## Stage 51: Professional LIT Is a Heat-Source–Diffusion–Inverse-Depth Problem > **Which subsurface defect, leakage site or thermal property remains identifiable after emissivity, nonuniform heating, anisotropic diffusivity, lateral spreading, boundary conditions and alternative defect geometries are all allowed to explain the same amplitude–phase dataset?** # Evidence: What Makes an LIT Claim Strong? Stronger evidence combines known modulation frequency, calibrated frame timing, amplitude + phase, multiple frequencies, blank/reference regions, emissivity control, known artificial defects, finite-element modelling, ultrasound/X-ray/CT comparison and device electrical or structural failure analysis. # Misconceptions Worth Hunting – LIT is ordinary thermal imaging with more frames. – Phase is unaffected by emissivity under every condition. – Lower modulation frequency only means slower acquisition. – Thermal diffusion length gives exact defect depth directly. – A defect’s image width equals its physical width. – Every hotspot is the electrical root cause. – One modulation frequency is enough for quantitative depth. – LIT is always better than pulsed thermography. – Surface heating and eddy-current heating have identical transfer functions. – Deep learning can determine defect depth without thermal diffusivity. # Transfer Check A defect appears strongly at 0.1 Hz but disappears at 10 Hz. Is a deeper feature plausible? **Yes. Lower frequency has a longer thermal diffusion length.** A phase feature stays fixed while illumination amplitude varies spatially. Does that strengthen the defect interpretation? **Yes, though emissivity and boundary conditions still need checks.** A hotspot pinpoints one IGBT cell in a power module. Is the microscopic gate defect identified? **No. The hotspot localizes dissipation; root-cause microscopy is still needed.** A CFRP defect appears much broader when modulation frequency is lowered. Did the delamination grow during the scan? **No. Lateral thermal diffusion increased.** # How We Know the Learning Has Held A learner should be able to explain periodic thermal excitation, compute thermal diffusion-length scaling, distinguish amplitude and phase images, explain frequency-dependent depth sensitivity, identify emissivity and nonuniform-heating effects, explain lateral diffusion, interpret composite delaminations, explain electronic failure LIT, explain photovoltaic LIT, compare LIT with pulsed, eddy-current and vibrothermography, explain multi-frequency inverse modelling and identify ML limits. # Model Limits LIT measures periodic surface radiance produced by a thermal diffusion field. It does not directly give defect depth, defect size, electrical failure mechanism or thermal diffusivity without a model and calibration. Professional LIT keeps **heat-source type + modulation frequency + camera calibration + emissivity + thermal diffusivity + anisotropy + boundary conditions + amplitude/phase + frequency series + orthogonal NDT evidence** visible together. # Teaching Guide Teach in this order: **periodic heating → IR movie → lock-in demodulation → amplitude → phase → thermal diffusion length → defect contrast → frequency sweep → emissivity/heating artifacts → composites → electronics → photovoltaics → alternative thermography modes → inverse modelling → validation.** # Connect This to the eduKate Learning Estate – Infrared Radiation and Thermal Imaging — IR fundamentals. – Scanning Thermal Microscopy — nanoscale probe thermal mapping. – Ultrasound and Acoustic Imaging — acoustic NDT. – Semiconductor Devices — electrical failure mechanisms. – Solar Cells — photovoltaic loss mechanisms. # Research Foundations and Further Learning – Lock-in thermography for semiconductor and power-module failure analysis. – Breitenstein and collaborators on LIT of solar cells and electronic components. – CFRP delamination detection with LIT. – Thermal-wave diffusion relation \(μ=\sqrt{α/π f}\). – 2026 model-based subsurface-defect LIT work. – *Sizing Wedge Delaminations Using Lock-in Thermography*, 2026. – 2026 electronic-packaging work combining periodic eddy-current heating with lock-in thermographic analysis. # The Quiet Ending The beginner asks, “Which surface pixels oscillated with the heater?” The developing NDT scientist asks, “How delayed was the thermal wave at each pixel?” The advanced learner asks, “What defect depth and geometry could reproduce the frequency-dependent phase?” And the professional asks: > **Which hidden structure survives after excitation, thermal diffusion, emissivity and the inverse heat-transfer problem are all treated as one experiment?**