Among the thermal infrared (IR) microscopies, Lock-in Thermography (LIT) is a decisive technical improvement [6-9] and is commercially available for failure analysis by different vendors, e.g. in iPHEMOS systems. This LIT technique allows the detection of local heat sources at the surface of a few μW corresponding to a local temperature modulation of a few μK [10] so that this new generation of IR-based fault localization techniques provides an improved sensitivity. [1, 3] It outperforms other thermal imaging methods, such as liquid crystal imaging, fluorescent microthermal imaging, Raman IR-thermography, and steady-state IR thermography by 2 to 3 orders of magnitude. Therefore, the LIT becomes a powerful nondestructive fault localization technique drastically. For instance, lock-in thermography (LIT) can be used for microscopic failure analysis of integrated circuits (ICs). Results obtained by LIT was able to show point defects with emissions in the range of mid-wave infra-red (MWIR), which Photo Emission Microscopy (PEM), with the wavelengths of near infrared (NIR) to short- wave infra-red (SWIR), cannot detect. [5] Therefore, ELITE (Enhanced Lock-In Thermal Emission) system becomes a powerful tool for fault isolation of functional failures because the resistive heat emissions beyond the wavelength of 2000 nm can be detected (see comparison between different techniques with different wavelengths at page4915).

As shown in Figure 4914a, the principle of LIT technique is that the device under test (DUT) is powered up using periodically modulated signal and then its surface temperature response is analyzed using an IR thermo camera. [1, 2] In this technique, both lock-in technique and thermography measurement are employed. That is, during LIT data collections, the power dissipated in the object under investigation is periodically amplitude-modulated and the resulting surface temperature modulation is captured by a thermally sensitive camera running with a certain frame rate (f_fr); the generated IR images are digitally processed according to the lock-in principle. Therefore, the sum of thermal images captured while the device is powered off can be subtracted from the sum of thermal images captured while the device is powered. Figure 4914b shows the setup of LIT.

In LIT, the generated IR images are digitally processed according to the lock-in principle. If each pixel of the IRT image would be connected to a 2-phase lock-in amplifier, the amplitude image of the in-phase signal is S^0°(x,y) and that of the out-of-phase (or quadrature) signal S^90°(x,y). In LIT measurements, the -90° signal is used instead of the +90° one [8]. From these two amplitude signals, the image of the phase-independent amplitude A(x,y) and the phase image ϕ(x,y) of the measured temperature can be given by,
          -------------------------- [4914a]
          -------------------------- [4914b]
The contrast of the amplitude image of a heat source is proportional to its dissipated power, while the phase signal is inherently emissivity-corrected since the phase image relies on the quotient of the 0° and the -90° image, e.g. for isolated heat sources, it is independent of the power of the heat source and emissivity. [8] The values in the phase images in Figure 4914c measure the time delay of the surface temperature modulation referred to the power modulation and are actually independent of the magnitude of the heat source. An alternative to the phase image in Equation 4914b, S^0°(x,y)/S^90°(x,y) is often used for imaging due to its better spatial resolution.

Figure 4914a. Thermal lock-in measurement (also called Lock-in thermography (LIT)). [4]

Figure 4914b. Setup of LIT. [5]

As shown in Figure 4914c, the obtained LIT results are presented in two formats:
         i) Amplitude (total increase in temperature during repeated power cycling),
         ii) Phase (time delay between powering up and subsequent heating).

Figure 4914c. LIT results: amplitude and phase. [5]

In LIT operation, for obtaining the highest possible detection sensitivity, the IR camera runs at its highest possible frame rate (f_fr) and the
lock-in frequency (f_lock-in) is adjusted by choosing an appropriate number of frames per period. [10] A limitation of the conventional LIT correlation is that at least 4 frames per lock-in period have to be evaluated. [10] Therefore, for a typical frame rate of 100 Hz, the maximum lock-in frequency is 25 Hz. If several local heat sources locate close to each other, it may then be necessary to further increase lock-in frequency to further reduce thermal blurring. However, this reduction of blurring reduces the sensitivity, since the magnitude of the LIT signal always deceases with increasing lock-in frequency. [8] On the other hand, as the lock-in frequencies increase, there is a blurring effect with a halo surrounding this small spot.

Figure 4914d shows a comparison between amplitude image, phase image, 0°/-90° image and power distribution obtained from a LIT measurement and its data extraction, obtained from a Hall sensor circuit at its normal operation.

	◆ Amplitude image ◆ Strongly affected by the emissivity contrast caused by the metallization pattern.  ◆ The use of a lock-in technique significantly suppressed the steady-state IR image (topography image)        → The bright regions outside of the heat source positions, modulated by the local emissivity contrast, are actually caused by the lateral heat conduction-induced halo of the temperature modulation around the heat sources. ◆ It is difficult to judge which of the bright regions are real heat sources and which are regions of high emissivity.
	◆ Phase image ◆ The emissivity contrast is removed. ◆ This image shows a stronger halo around the heat sources and displays heat sources of different intensity in a comparable brightness ("dynamic compression" feature).
	◆ 0°/-90° image from the region indicated in (b) ◆ The emissivity contrast is removed. ◆ This image shows a lower blurring and shows heat sources of different power intensity with different brightness, in comparing with (b).
	◆ Power distribution, numerically deconvoluted from (c) ◆ It reveals many more details than the original images.

In summary, the main characteristics of LIT are:
         i) it allows the detection of local heat sources at the surface of a few μW corresponding to a local temperature modulation of a few μK. [10]
         ii) Due to its dynamic nature, lateral heat diffusion (blurring) is considerably reduced compared to steady-state techniques, depending on the chosen lock-in frequency. [10]
         iii) it allows a relatively coarse (3-5 μm) but very sensitive localization of any leakage current or local heat source in an IC with a very high success rate without any preparation expense. [10]
         iv) it is applicable for backside inspection and for detecting sub-surface heat sources.
         v) its spatial resolution can be improved down to 1 μm by applying a solid immersion lens.
         vi) some defects can be localized which are not visible in laser-based techniques such as TIVA, OBIRCH, and light emission microscopy.
         vii) lateral heat spreading is more problematic for spatially extended heat sources [8].
         viii) one main issue is the ability of conventional microscope objective to spatially resolve objects, in the mid-IR range (3 to 5 µm wavelength), due to physical limitation from diffraction.
         ix) The phase image (approx. -45°) can be used to distinguish whether heat is generated in the on-state or in the off-state (in positive or negative response) of the some devices, eg. triggers in an intact step motor controller. [10]

Table 4914. Comparisons of accuracies of fault localization with LIT technique and other techniques.

[1] Breitenstein, O & P Rakotoniaina, J & Altmann, Frank & Schulz, J & Linse, G. (2019). Fault Localization and Functional Testing of ICs by Lock-in Thermography.
[2] X.P.V. Maldague, Theory and Practice of Infrared Technology for Nondestructive Testing, Wiley, New York (2001).
[3] O. Breitenstein, J.P. Rakotoniaina, F. Altmann, T.Riediger, and M. Gradhand, “New Developments in IR Lock-in Thermography,” Proc. 30th International Symposium for Testing & Failure Analysis, (2004).
[4] Overview of function of iPHEMOS series.
[5] Paul Hubert P. Llamera and Camille Joyce G. Garcia-Awitan, Thermal Failure Analysis of Functional Failures by IR Lock-in Thermal Emission, ISTFA™ 2019: Conference Proceedings from the 45th, (2019).
[6] Kuo, P. K., Ahmed, T., Jin, H. and Thomas, R. L., "Phase-Locked Image Acquisition in Thermography", SPIE 1004, (1988), pp. 41-45.
[7] Maldague, X .P. V., Theory and Practice of Infrared Technology for Nondestructive Testing, Wiley (New York 2001).
[8] Breitenstein, O., and Langenkamp, M., Lock-in Thermography — Basics and Use for Functional Diagnostics of Electronic Components, Springer (Berlin, Heidelberg 2003).
[9] Breitenstein, O., Rakotoniaina, J. P., Altmann, F , Schulz, J., and Linse, G., "Fault Localization and Functional Testing of ICs by Lock-in Thermography", Proc 28th Int'l Symposium for Testing and Failure Analysis (ISTFA), 2002.
[10] Paiboon Tangyunyong and Christian Schmidt, Thermal Defect Detection Techniques, Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.