Lock-in thermography enables solar cell development

Lock-in techniques greatly increase the sensitivity and image resolution of thermography used in PV cell defect detection.

Markus Tarin, MoviTHERM, Irvine, CA USA;
Ross Overstreet, FLIR Systems, Torrance, CA USA

Lock-in techniques greatly increase the sensitivity and image resolution of thermography used in PV cell defect detection.

Currently, PV cells suffer from various manufacturing problems that limit their conversion efficiency. Additionally, conversion efficiency varies according to the technology employed, with silicon PV cells achieving conversion efficiencies between 15% and 25%, while typical metallic thin film cells have efficiencies in the 5% to 20% range (depending on materials used).

Much of the industry’s R&D efforts are aimed at reducing production defects. Too many defects in the semiconducting material structure go undetected before PV cells are put into solar panel assemblies. Identifying these defects requires efficient, cost-effective test and measurement methods for characterizing a cell’s performance and its electronic structure.

Sources of defects

A PV cell is typically modeled as an ideal diode in parallel with a photocurrent source, plus parasitic resistances, such as shunt resistance (RSH) and series resistance (RS) Fig. 1).

Figure 1. Circuit model of a PV cell.

The conversion efficiency of silicon PV cells is limited by free carrier recombination, due to bulk material defects. This is especially true in multicrystalline silicon (mc-Si) wafers, which have significant concentrations of crystallographic non-uniformities, such as dislocations, grain boundaries, and impurities.


In thin metallic film PV development, lateral non-uniformities in current flow across a cell are troublesome [1]. Since larger solar panels are constructed by connecting individual PV cells, a few bad cells can affect the performance of the entire assembly. Often, a high RSH results from:

  • Improper handling during processing;
  • Diamond saw scribing at cell boundaries;
  • Over-firing during cell metallization;
  • Poor edge isolation processes [2,3];
  • Random shunts inherent in most production processes.

The dominant sources of RS are contact resistance, bus bar resistance, screen-printed “fingers,” and lateral conductions in the emitter. The relative importance of each source depends on bias level and current flow.

To determine the sources and magnitudes of defects, the parameters most commonly measured include resistivity (to screen wafer material) and characteristics of production cells, such as I-V and C-V curves, charge carrier characteristics/current density, free charge recombination lifetime, bulk material lifetime, and effective lifetime.

Test techniques vary greatly in terms of complexity, equipment cost, and the time required for a typical set of measurements. The three broad areas of test technology are spectroscopy, electrical (contact) measurements, and infrared (IR) imaging. Frequently, multiple techniques are used [4].

Electrical C-V, I-V, and resistivity profiling in early production stages require wafer probing and thickness measurements. The latter require additional optical or capacitive gauging techniques. Some of these may also require time-consuming sample preparation.

Conventional IR imaging methods

Testing via infrared imaging has been used for more than a decade, and is growing in importance because it is relatively fast and with moderate equipment cost. The IR cameras are basically video devices, but each video frame is accessible as a still thermographic image, whose digital content is also available—including actual temperature data. Standard thermographic imaging of a PV cell quickly reveals major shunt defects during the application of reverse bias Fig. 2), or by just observing the temperature of the cell under typical operation.


Figure 3. Non-contact LIT test system (SolarCheck by MoviTHERM, uses modulated light as the PV cell stimulus. (Optional VOC measurements do require electrical probing of the cell). The system uses a FLIR IR camera with an uncooled microbolometer detector.

The sensitivity and thermal resolution of standard thermography, however, is limited by an IR camera’s inherent detector sensitivity or noise equivalent temperature difference (NETD). The NETD for cameras with cooled indium antimonide (InSb) detectors is ~20mK; it’s ~80mK for an uncooled microbolometer detector. Only severely shunted areas are visible. The darker orange regions in Fig. 2 result from weaker shunt defects. Locating the origins of weaker shunts is extremely difficult, if not impossible, due to the thermal diffusion (spreading of thermal energy over time), as well as the weak thermal radiation of the defect itself.

Figure 2. IR image of 60x60mm silicon solar cell showing shunt defects (orange areas) under steady state reverse bias conditions. (Triangular shapes on the left and right are alligator clips that apply bias voltage; the circular blue area is a reflection of the IR camera lens.)

Visible light cameras are largely ineffective at revealing even major defects, but provide useful reference points alongside an IR image. Today, IR cameras are available that combine both thermographic and visible light imaging, making fast steady-state testing of solar cells very convenient.

A variation in conventional IR imaging of defects is to move a heat lamp and camera attached to a fixture across the surface of a PV cell or a solar panel. This methodology can improve the crack detection rate and reduce inspection time. The drawback of using a “slow” heat or excitation source is that the resulting thermal diffusivity will be significant, negatively affecting the spatial resolution and definition of the crack.

Refinements to conventional IR imaging

To minimize the thermal diffusion that occurs with slow stimuli, pulsed or sinusoidal stimulation can be used. This can take the form of applied electrical signals or light.

Electroluminescence (EL) and photoluminescence (PL) techniques. EL and PL are techniques used to generate spatially resolved images of solar cells that reveal localized shunts, series resistance, and areas of charge carrier recombination [2,5]. EL applies a forward voltage and current to cause localized irradiance due to carrier recombination. PL uses light irradiation for the same purpose. In both cases, the stimulus can be applied as a pulse.

In EL testing, current flow causes the PV cell to emit light in the near infrared (NIR) range of the spectrum. The resulting thermographic image provides a visual representation of a PV cell’s uniformity with respect to its ability to convert photons into electrons. Care must be exercised to avoid applying a destructive amount of current to the cell.

Since EL and PL techniques only work in the NIR region, both types require a camera with a cooled NIR detector. (Uncooled microbolometer detectors are longwave IR instruments and, therefore, not suitable.)

Lock-in thermography (LIT). Commercially available lock-in thermography systems are overcoming the limitations of conventional thermal imaging. Typically, they use a xenon or halogen flash lamp, or a modulated laser as the excitation source. In LIT measurements, the test system synchronizes the excitation source to the camera’s data acquisition Fig. 3), which collects a sequence of hundreds of images.

Advantages of LIT

By stimulating a PV cell with pulsed light, heat, or electrical signals, a lock-in amplifier tuned to the stimulus’ excitation frequency allows the system to detect subtle thermal responses beyond the noise floor limitations of an IR camera. The increased sensitivity brings the system’s detection threshold down below the noise floor by a factor of 100 to 1,000. In addition, this type of system has the distinct advantage of eliminating problems due to reflections from other heat sources, such as human body radiation, overhead lights, etc.

Figure 4 is an image of the same PV cell used for Fig. 2, but collected with the LIT system shown in Fig. 3. Note that the image resolution in Fig. 4 is much better (i.e., not diffused and blurry), with localized shunt defects more sharply defined by the orange areas. The sharper image provides other information, such as non-uniform heating of the cell, as revealed by lighter and darker blue areas. In addition, the reflection of the camera lens and outlines of the alligator clips no longer obscure large portions of the image, as they do in Fig. 2.

This LIT technique allows mapping of forward current density distribution, and can also reveal series resistance and sites where there is heightened carrier recombination [6-8]. It requires significantly less energy input to a solar cell compared to conventional thermography.

LIT test variables

In using LIT for shunt detection, the stimulus’ modulation frequency is important because it affects thermal diffusion and image resolution. In conventional electrical measurements using lock-in amplifiers, the tendency is to lower the stimulus frequency to only a few Hz to get below the frequency of most noise sources. This has to be modified somewhat in LIT. Typically, the order of magnitude for a stimulation frequency is ~100Hz. If the stimulation frequency is an order of magnitude lower (~10Hz), thermal diffusion becomes so great that defects tend to disappear from the LIT image.

With proper selection of the stimulation frequency, thermal image resolution is limited primarily by the pixel resolution of the camera’s focal plane array detector and its optics. For the camera used in Fig. 3,
the detector pixel size is 25µm. Microscope optics are available that are capable of 6µm/pixel spatial resolution.

Figure 4. Image of the same solar cell in Figure 2, now showing shunt defects more clearly (orange areas) when mapped with an LIT technique. (Image from a SolarCheck system.)

Images and cell parameters are calculated by the system software running on a PC. With appropriate software, the processed signal from the IR camera’s detector can be used to make quantitative measurements of I-V characteristics associated with a localized shunt, calculate the reduction in cell efficiency due to shunts, and map saturation current density and ideality factor over the entire cell.

Quantifying charge carrier behavior

With LIT systems and software, charge carrier behavior in PV wafers and cells can be characterized. Charge density imaging (CDI) is particularly advantageous as it can rapidly map saturation current density and other parameters over an entire PV cell [9,10].

Flash CDI is based on free-carrier absorption of photo-generated excess carriers, and thus allows the imaging of charge carrier lifetime properties. Carrier generation is controlled by adjusting the laser intensity to approximately a 1-sun level. With lock-in processing of the signal, much shorter lifetimes can be measured. Test time can be on the order of seconds, and an entire CDI wafer map can be created in a minute or so. (Actual CDI test times depend on the length of the effective lifetimes being measured.) Image resolution is better than that obtained with many other techniques, some of which are an order of magnitude slower.

Typical PV cell parameters of interest include short circuit current (ISC), open circuit voltage (VOC), fill factor (FF), ideality factor (η), series resistance (RS) at VOC, shunt resistance (RSH) at 0V, and reverse voltage breakdown. FF and η are usually derived from I-V measurements.

To a great extent, cell efficiency is largely a function of ISC, VOC, and FF. VOC and FF can be characterized by the shape of the forward biased dark I-V curve. Frequently, distribution of forward current density over the entire cell is inhomogeneous. If current density through a given region is higher than the cell average, this is indicative of a shunt. Therefore, characterizing shunts under forward-bias conditions is important in the efforts to increase efficiency.

Parasitic RS is important as it contributes to FF losses and ideality factor (η). RS is typically determined from a set of illuminated and dark I-V measurements. While these are straightforward, collecting a complete set of I-V measurements can be time consuming.

Parameter extraction is done by assuming carrier lifetime is homogeneous over a particular sample width, and applying a sinusoidal correlation procedure. The intensity of each pixel, IA, can be determined by:

IA=k Δm W,

where k is the sinusoidal correlation factor, m=excess minority carrier concentration, and W is the sample width.

Basing the camera’s frame rate on the lock-in frequency allows for carrier density measurements to be taken under a steady-state condition during each half period of the stimulus signal. The steady-state measurement of Δm values can be used to calculate effective carrier lifetime according to:

τeff = ΔmW/G = IA/–kG,

where G is the local generation rate for the sample area.

Because a PV cell can be modeled as an ideal diode in parallel with a photocurrent source, LIT can be used to thermally measure local I-V characteristics that reveal non-ideal behavior (i.e., parasitic series and shunt resistance).

Non-ideal diode properties of the PV cell are also expressed in the ideality factor,

η = δV/δln(I).

By accounting for RS and RSH, the relationship between η and PV cell voltage can be established, which leads to a better understanding of charge carrier transport mechanisms.

LIT can be combined with illuminated and dark I-V measurements to derive other PV cell parameters. For example, FF is lower than ideal due to RS and RSH, and can be expressed as:

FF = Impp Vmpp/ISC VOC

where mpp is the maximum power point.


A major advantage of LIT compared to many other test methods is the short time required to complete a set of measurements without elaborate sample preparation. Once an LIT system is configured, significant amounts of data can be acquired in seconds, compared to minutes or hours with other methodologies. This makes LIT a good candidate for process-related testing, as well as for use in the R&D lab to detect cracks, shunts, parasitic series resistance, and localized charge carrier characteristics.


[1] D. Shvydka et al., “Lock-in Thermography and Non-uniformity Modeling of Thin-Film CdTe Solar Cells,” Appl. Phys. Lett., Vol. 84, No. 5, (Feb. 2004).

[2] M. Tajima et al., “Characterization of Multicrystalline Silicon by Photoluminescence Spectroscopy, Mapping and Imaging,” Institute of Space and Astronautical Science/Japan Aerospace Exploration Agency (2006); tajima@isas.jaxa.jp

[3] A. Hauser et al., “Comparison of Different Techniques for Edge Isolation,” 17th EU-PVSEC –Munich (2001; alexander.hauser@

[4] C. Ballif et al., “Efficient Characterization Techniques For Industrial Solar Cells and Solar Cell Materials,” Proc. 12th Workshop on Crystalline Silicon Solar Cell Materials and Processes, NREL, Breckenridge, CO, USA (2002), pp. 136-146.

[5] M. Kasemann et al. “Comparison of Luminescence Imaging and Illuminated Lock-in Thermography on Silicon Solar Cells,” Appl. Phys. Lett. 89, 224102 (2006).

[6] J. Rakotoniaina et al., “Quantitative Analysis of the Influence of Shunts in Solar Cells by Means of Lock-in Thermography,” 3rd World Conference on Photovoltaic Energy Conversion (2003), Osaka, Japan, ref. 4P-A8-62.

[7] O. Breitenstein et al., “Quantitative Evaluation of Shunts in Solar Cells by Lock-In Thermography,” Progress in Photovoltaics Research and Applications (2003), 11:515–526.

[8] I. Konovalov et al., “Activation Energy of Local Currents in Solar Cells Measured by Thermal Methods,” Progress in Photovoltaics Research and Applications (1998), 6:151-161.

[9] S. Riepe et al., “Carrier Density and Lifetime Imaging of Silicon Wafers by Infrared Lock-in Thermography,” 17th EU-PVSEC, Munich (2001), paper VC1.51.

[10] J. Schmidt, et al., “Advances in Contactless Silicon Defect and Impurity Diagnostics Based on Lifetime Spectroscopy and Infrared Imaging,” Advances in OptoElectronics, Vol. 2007, Article ID 92842, Hindawi Publishing Corp.

Markus Tarin received a BS in electrical engineering and an MS in computer science from the Higher Technical College in Hanover, Germany, and is president/CEO of MoviTHERM, a division of Epsilon Technologies International, LLC (ETI), 15540 Rockfield Blvd., Suite C-110, Irvine, CA 92618 USA; 949-699-6600; m.tarin@movitherm.com

Ross Overstreet received his bachelors and masters degrees in mechanical engineering from Auburn U., and is the science camera sales engineer covering the western US for FLIR, 21143 Hawthorne Blvd, #445, Torrance, CA 90503; 1-800-905-9557; Ross.Overstreet@FLIR.com; www.goinfrared.com/PVtest

More Photovoltaics World Issue Articles


Previous articleIII-V thin films for PV applications
Next articleHydropower Receives $531 million in Clean Renewable Energy Bonds

No posts to display