Int. J. Nanoparticles, Vol. 6, Nos. 2/3, 2013 81

Copyright © 2013 Inderscience Enterprises Ltd.

Illuminated versus dark lock-in thermography investigations of solar cells

Otwin Breitenstein Max Planck Institute of Microstructure Physics, Weinberg 2, D-06120 Halle, Germany E-mail: breiten@mpi-halle.mpg.de

Abstract: In this paper, after describing the basics of lock-in thermography investigations of solar cells, the application of this imaging technique for the characterisation of silicon wafer-based solar cells is reviewed. In particular, the differences between various variants of this technique working with and without light illumination are discussed. It is found that, for imaging and a detailed local investigation of leakage currents, both under forward and reverse bias, the application of dark lock-in thermography is most useful. On the other hand, illuminated lock-in thermography provides some unique possibilities, like the investigation of shunts in non-metallised solar cells or without contacting the cells, local efficiency imaging, and the imaging of the local avalanche multiplication factor under reverse bias. A special variant of illuminated lock-in thermography enables the imaging of the minority carrier lifetime distribution and of trapping centres in bare silicon wafers.

Keywords: solar cells; silicon; leakage current; imaging technique; lock-in thermography; LIT; carrier density imaging; CDI; infrared lifetime mapping; ILM; dark lock-in thermography; DLIT; illuminated lock-in thermography; ILIT.

Reference to this paper should be made as follows: Breitenstein, O. (2013) ‘Illuminated versus dark lock-in thermography investigations of solar cells’, Int. J. Nanoparticles, Vol. 6, Nos. 2/3, pp.81–92.

Biographical notes: Otwin Breitenstein received his PhD in Physics from the University of Leipzig, Leipzig, Germany, in 1980. Since 1992, he has been with the Max Planck Institute of Microstructure Physics in Halle, Germany, where he investigated defects in semiconductors. Since 1999, he is using lock-in thermography for detecting internal shunts in silicon solar cells. In 2001, he has introduced this technique on a microscopic scale for isolating faults in ICs. He is giving lectures on photovoltaics at Halle University, Halle, Germany, and is author of a book on lock-in thermography.

This paper is a revised and expanded version of a paper entitled ‘Illuminated versus dark lock-in thermography investigations of solar cells’ presented at Proceedings of the 11th International Workshop on Beam Injection Assessment of Microstructures in Semiconductors, Annaba, Algeria, 25–28 June 2012.

82 O. Breitenstein

1 Introduction

The worldwide production of solar cells has risen from about 400 MW/year in 2001 to more than 37 GW/year in 2011, which corresponds to an average yearly increase of about 60% (Hering, 2012). Thus, the photovoltaics (PV) industry represents a big business and is important for securing an alternative energy supply in the future. Besides looking for new solar cell concepts, an important goal of PV research is the improvement of the conversion efficiency of existing and proven solar cell technologies. The largest part of all solar cells produced today is based on mono- or multicrystalline silicon wafers (Hering, 2012). The electronic properties particularly of multicrystalline solar cells are not homogeneous due to an inhomogeneous distribution of grain boundaries and dislocations in the material. Therefore an important contribution for improving the efficiency of such cells is a detailed local characterisation of the cells. Besides camera-based luminescence imaging techniques, lock-in thermography (LIT) is a very successful technique for imaging and locally analysing the electronic properties of solar cells (Breitenstein et al., 2011a, 2011b).

In this contribution, first the physical and technical basics of the LIT technique are reviewed. Also a special LIT-method is described, which enables a fast imaging of the minority carrier lifetime in bare solar wafers. Then typical results of different variants of the LIT technique are introduced, some of them working in the dark [dark lock-in thermography (DLIT)] and some working under pulsed or continuous irradiation of light [illuminated lock-in thermography (ILIT)]. It will be shown that both variants have their individual advantages, limitations, and special application fields.

2 Physical and technical basics

A solar cell is a large-area semiconductor diode (p-n junction), where the upper contact is usually formed as a grid structure for enabling illumination of the semiconductor material. In the dark, this diode shows a typical rectifying current-voltage (I-V) characteristic, hence under forward bias an exponentially increasing forward current flows, and under reverse bias only some weak leakage current flows. Under illumination, due to the absorption of photons in the semiconductor material, a photocurrent is induced. This photocurrent is a reverse current, its magnitude is independent of the cell bias, and it tends to bias the cell into forward direction. Note that a negative current flowing under positive bias means that the device is producing electric energy. The net current of the cell is the sum of the (bias-independent) photocurrent and the (strongly bias-dependent) forward current, which flows under illumination just as in the dark. The so-called open circuit voltage appears at the device if no current is extracted. Then the whole photocurrent is balanced by the dark current. The larger the amount of the current for a certain voltage, the lower is the open circuit voltage to be achieved, and the lower is the efficiency of the cell. Generally, the forward current represents a limitation for the efficiency, since any forward current flowing in operation within the cell represents a loss in current to be fed out of the cell. If the electronic properties of a solar cell were homogeneous, only the cell-inherent physics and technology would determine its efficiency. However, in reality no industrial solar cell is really homogeneous. It is well-known since many years that, especially in multicrystalline silicon solar cells, the dark forward current flows very inhomogeneous (Breitenstein et al., 2011a). It is rather a rule

Illuminated versus dark lock-in thermography investigations of solar cells 83

than an exception that some sites of locally increased forward current (so-called shunts) are dominating the whole forward current and thus limiting the efficiency of the cell. Therefore, for optimising the efficiency of solar cells, it is important to image and investigate such leakage current sites and generally to perform a local analysis of efficiency-relevant parameters of solar cells.

If a bias V is applied to a solar cell and the flowing current is not too high, it can be assumed that the local bias is everywhere in the cell the same and equal to the applied bias V. Then, if in a certain position a certain current density J flows in the dark, it leads to a locally dissipated power density of p = J*V. This leads to a local increase of the cell temperature in this position. Therefore thermography, which is the imaging of the surface temperature by an appropriate infrared (IR) thermo-camera, has often been used for imaging leakage currents in solar cells. However, the detection sensitivity even of today's best thermo-cameras of about 20 mK is too low for detecting weak leakage currents under forward bias. Because of this sensitivity limitation, in the past, thermographic investigations on solar cells could be performed only under several Volts reverse bias (Simo and Martinuzzi, 1990). Another limitation of conventional (steady-state) thermography is caused by the relatively high thermal conductivity of silicon. If there is a local heat source in the device, due to this thermal conductivity, the heat spreads not only into the depth but also laterally, which leads to the typical 'blurred' appearance of steady-state thermograms. This inevitable thermal blurring limits the realistic spatial resolution for silicon devices to several mm.

Figure 1 Comparison of standard (steady-state) thermography and LIT of a multicrystalline solar cell (a) scaled from 22–70°C (b) 10 Hz, scaled from 0 to 10 mK T-modulation amplitude (see online version for colours)

2 cm

(a) (b)

These limitations are overcome by using LIT instead of conventional (steady-state) thermography. LIT simply means that the heat sources are pulsed square-shaped between zero and a certain bias V at a certain lock-in frequency flock-in (typically 10 Hz), which leads to a periodic temperature modulation at the surface at this frequency. The thermo-camera is running at its specific frame rate (typically 100 Hz) and all images are fed to a computer. By appropriate evaluation and time-averaging of all images over a certain acquisition time (typically some minutes), images of the local surface temperature modulation are obtained (Breitenstein et al., 2010). Due to the averaging nature of this technique, the detection limit drops with the inverse of the square root of the acquisition

84 O. Breitenstein

time and reaches values below 100 µK temperature modulation amplitude after 1/2 hr acquisition time. Moreover, due to the dynamic nature of this technique, the lateral heat diffusion is strongly suppressed, leading to a considerably improved lateral resolution of LIT, compared to steady-state thermography. Note that LIT was introduced already by Kuo et al. (1988) and was used later on extensively in nondestructive testing (Maldague, 2001). Only since 2000, LIT was also used for imaging of shunts under forward bias in solar cells (Breitenstein et al., 2000). Figure 1 shows a comparison of steady-state thermography and LIT for imaging pre-breakdown sites at a reverse bias of –14 V in a typical multicrystalline solar cell, which demonstrates the advantage of LIT in spatial resolution. It is clearly visible that, due to the strongly reduced thermal blurring, the local information obtained by LIT is dramatically higher than for standard thermography.

2.1 Physical origins of heat sources in solar cells

There are various physical mechanisms by which heat may be generated in differently biased and/ or illuminated solar cells. The simplest one is Joule heat, which is generated if current flows in a conductor showing a significant ohmic resistance. This may be due to horizontal current flow in a grid line or in the emitter or due to vertical current flow in an ohmic shunt. Physically more interesting are the heat sources due to current flow across the p-n junction. These heat sources are reviewed in detail by Breitenstein and Rakotoniaina (2005). In the dark and under forward bias, it depends on the magnitude of the bias, which heat dissipation mechanism dominates. For low forward bias (up to approx. 0.5 V) the so-called recombination current or 'second diode' current dominates, which is due to recombination of carriers in the depletion region of the p-n junction. Above this voltage the so-called diffusion current or 'first diode current' dominates, which is due to recombination of minority carriers in the bulk and at the upper and lower wafer surface. Hence this current is due to the injection of minority carriers into the p- and n-region, where they loose their energy by recombination. Thus, by selecting either a lower or a higher forward bias, either the recombination current or the diffusion current may be predominantly imaged. Under reverse bias up to –20 V, actually no measurable reverse current should be observed, since a typical silicon solar cell with a base doping concentration of 1016 cm–3 should break down only at about –60 V (Sze and Gibbons, 1966). In reality, however, a significant reverse current with a typical super-linear breakdown characteristic appears for most multicrystalline silicon cells already beyond –5 V. It has been shown that several different mechanisms may be responsible for this pre-breakdown behaviour (Breitenstein et al., 2011b). Breakdown sites do not affect the efficiency but may lead to 'hot spots', which even may destroy solar modules. A cell in a solar module may easily get reverse-biased e.g. by shading this single cell under illumination of the whole module. Therefore breakdown in solar cells is an important reliability issue and has to be investigated, e.g. by applying LIT.

If a solar cell is illuminated, in addition to the heat dissipation mechanisms described above, two more heat sources appear, which are both acting essentially homogeneous across the area. The one is the thermalisation of hot carriers immediately after the light absorption event. Note that the photon energy is usually higher that the gap energy. This energy difference is converted into heat immediately after absorption by thermalisation

Illuminated versus dark lock-in thermography investigations of solar cells 85

of the hot carrier. The second additional heat source in illuminated solar cells is the energy loss of photo-induced minority carriers on their way across the p-n junction. By passing the p-n junction, each carrier (electron or hole) dissipates an energy given by the sum of the equilibrium barrier height plus the magnitude of the reverse bias (Breitenstein and Rakotoniaina, 2005). If there is any avalanche-type carrier multiplication, the dissipated heat gets larger by the amount of the avalanche multiplication factor (MF). To summarise, in all LIT experiments of solar cells made under illumination, an additional homogeneous heating appears.

2.2 LIT-based lifetime imaging

The well-known standard method for imaging the minority carrier lifetime in semiconductor wafers, which is the most important material parameter of semiconductors, is Micro-Photoconductivity Decay (µ-PCD). This is a sequentially measuring method, hence in each position a focused and pulsed laser beam generates minority carriers, and their recombination is observed by detecting the microwave reflectivity in this position. By fitting the time-dependence of the conductivity to an exponential function, the local lifetime is calculated. For obtaining a high-resolution lifetime image with a good signal-to-noise ratio, an acquisition time of several hours may be necessary. Another limitation of µ-PCD is that, due to sensitivity limitations, it cannot work with a low illumination intensity. Even an intensity of one sun, which is typical for solar cells, is hard to realise; typical µ-PCD intensities are between 10 and 100 suns.

A thermo-camera-based technique called ‘infrared lifetime mapping’ (ILM) was proposed by Bail et al. (2000). In this technique a wafer is homogeneously illuminated by above-bandgap light (e.g., 850 nm) and the thermo-camera detects the local modulation of the IR emissivity. Note that, in the wavelength range of thermo-cameras (3–5 or 8–12 µm, respectively) the silicon material is nominally transparent, but it shows free-carrier absorption. If the free carrier density is changed by homogeneous light irradiation, any inhomogeneity of the lifetime leads to an inhomogeneous IR emissivity, which may be detected by the thermo-camera either in absorption or in emission mode. According to Kirchhoff's law, the absorption coefficient equals the emission coefficient. Later on this technique was further developed, made more sensitive and practicable by implementing the lock-in technique, and named ‘carrier density imaging’ (CDI) (Riepe et al., 2001). The advantage of the CDI/ILM technique, compared to µ-PCD, is a considerably reduced image acquisition time and a much higher sensitivity. This enables intensity-dependent lifetime imaging down to an illumination intensity well below one sun, where even carrier trapping effects may be investigated (Schubert et al., 2006). The limitation of CDI/ILM compared to µ-PCD is that its results are strongly influenced by the surface roughness. Hence, for enabling a quantitative scaling, the average of the lifetime or the lifetime in a certain position has to be measured by an independent method. This limitation also holds for photoluminescence (PL)-based lifetime imaging, which today has widely replaced CDI/ILM, since it needs only a silicon-based camera and not a more expensive thermo-camera. Therefore CDI/ILM results will not be introduced in this contribution, but at least the existence of this physically very interesting method should be mentioned here.

86 O. Breitenstein

3 Typical results

3.1 Dark lock-in thermography (DLIT)

The solar cells used for these investigations are typical industrial cells made from mono- or multicrystalline material having efficiencies between 15 and 17 %. Their size (125 × 125 or 156 × 156 mm2) is visible from the scaling bars. Figure 2 shows typical DLIT results (in arbitrary units) of a monocrystalline silicon solar cell. A typical measurement sequence implies one measurement at a forward bias of +500 mV [Figure 2(a)], which is close to the operation point of such a cell, one measurement at a weak reverse bias of –500 mV [Figure 2(b)] for checking the conductivity type of the shunts (ohmic or diode-like), and one measurement at a larger reverse bias [Figure 2(c)] for detecting pre-breakdown sites. Only if a shunt shows the same signal height (brightness) under + and –500 mV, it is an ohmic shunt. The non-ohmic shunts are usually recombination-induced and are caused by local accumulations of recombination centres (Breitenstein et al., 2004). As mentioned above, pre-breakdown sites are high-field phenomena and may be caused by completely different physical mechanisms (Breitenstein et al., 2011b). However, as Figure 2 shows, the ohmic shunts usually also lead to hot spots under reverse bias.

Figure 2 DLIT images of a monocrystalline solar cell taken (a) at +0.5 V, (b) at –0.5 V, (c) at –12 V (see online version for colours)

2 cm

linear (ohmic) shunts

(a) (b)

(c)

Illuminated versus dark lock-in thermography investigations of solar cells 87

In Figure 3, DLIT results obtained at different forward biases are collected. At +0.5 V [Figure 3(a)] the current flows only in some local sites, whereas at +0.6 V [Figure 3(b)] it flows more homogeneously. This is due to the fact that the depletion region recombination current, dominating at +0.5 V, is due to local extended defects crossing the p-n junction, whereas the diffusion current, dominating at +0.6 V, is the classical diode current. Its magnitude is governed by the local minority carrier lifetime. If the spatial resolution of the method is increased by detecting at a higher lock-in frequency and displaying the in-phase (0°) image instead of the amplitude one [Figure 3(c)], it becomes obvious that the diffusion current flows predominantly in the position of some lines. These lines are recombination-active grain boundaries in this multicrystalline material, which may be imaged in this way by DLIT.

Figure 3 DLIT images (in a.u.) of a multicrystalline cell taken at (a) +0.5 V, (b) +0.6 V (both amplitude images taken at 10 Hz) (c) in-phase (0°) image taken at +0.6 V and 40 Hz (see online version for colours)

2 cm

(a) (b)

(c)

In Figure, 4 results of an investigation of pre-breakdown sites in another multicrystalline cell are presented. Figure 4(a) is a reverse-bias thermogram taken at –14 V, similar to that in Figure 1(b). If two of such thermograms are taken at different temperatures, their evaluation allows one to calculate the local temperature coefficient of the leakage

88 O. Breitenstein

currents in units of ‘%(current change)/K’ (Breitenstein et al., 2008), which is shown in Figure 4(b). In the same way, if two thermograms are taken at two slightly different biases, the slope of the current change in units of ‘%/V’ is obtained, which is shown for this cell in Figure 4(c). It is visible that in regions of negative temperature coefficient the slope of the characteristic is particularly high, which points to an avalanche breakdown mechanism. Thus, by performing such investigations, a detailed physical analysis of pre-breakdown sites may be performed, which enables an identification of the breakdown mechanisms (Breitenstein et al., 2011b).

Figure 4 (a) DLIT image taken at –14 V (a.u.), (b) temperature coefficient of the reverse current, scaled from –3 to +3 %/K, (c) slope of the reverse current, scaled from 0 to 200 %/V (see online version for colours)

2 cm

(a) (b)

(c)

It was mentioned above that the DLIT signal is proportional to the locally dissipated power density. By taking and evaluating DLIT images at various voltages, local dark I-V characteristics of each single image pixel may be obtained. Since, as described above, the illuminated characteristic is strongly influenced by the dark characteristic, such DLIT investigations enable a detailed local efficiency analysis of solar cells. For lack of space these very recent results cannot be introduced here, details of this DLIT-based local efficiency analysis of solar cells are published by Breitenstein (2011, 2012).

Illuminated versus dark lock-in thermography investigations of solar cells 89

3.2 Illuminated Lock-in Thermography (ILIT)

Whenever light is irradiated in a LIT experiment, we speak from ILIT (Breitentein et al., 2010). Note that here always the whole solar cell is illuminated homogeneously, which is in contrast to other beam injection methods, where usually a focused light beam is applied. In ILIT experiments, the spatial resolution is obtained by the thermo-camera detecting the local thermal response in the device. Generally, shunt investigations can be performed also if the cell is exposed to light pulses instead of bias pulses in the dark. However, due to the inevitable homogeneous heating described in Section 2.1, the ILIT signal is not zero even in positions where no dark current is flowing. Hence, for measuring such currents, DLIT has to be preferred. On the other hand, shunts in a solar cell in an early technological state, where it does not have any metallisation yet, can only be investigated by ILIT with pulsed illumination, without any contacting the cell. This variant is called ‘Voc-ILIT’. Voc-ILIT is also able to image shunts in a production line without contacting the cell, one example measured within only one second is shown in Figure 5(a) (Gupta et al., 2007). For compensating the inevitable homogeneous heating and improving the spatial resolution, here, instead of the amplitude image, the in-phase (0°) image is displayed.

Another interesting ILIT application is the imaging of the efficiency according to Ramspeck et al. (2008). If a cell has no serious series resistance problems, it can be assumed that, under illumination under short circuit, the complete irradiation energy is converted into heat. On the other hand, if this cell is illuminated under its maxim power point (mpp) condition, some of the irradiation energy is converted into electric energy and fed out. This leads to a lower amount of heating, whereby the regions producing most electric energy dissipate the lowest heat and vice versa. By subtracting the power distribution under mpp-condition from that under short circuit condition, an image of the locally generated electric energy is obtained, which may be converted into a local efficiency image. Figure 5(b) (kindly provided by K. Bothe, ISFH Hamelin) shows such an image of the same cell as used for Figure 4. Note that this is a monochromatic efficiency, measured at a wavelength of 850 nm, which is significantly larger than the efficiency measured at natural solar irradiation. The dark spots in Figure 5(b) are local shunts, similar to that visible in Figure 2(a), and the extended regions of slightly reduced efficiency are poor crystal quality regions.

Finally, an interesting ILIT technique working under reverse bias will be introduced. This ‘MF-ILIT’ technique is used to image the avalanche MF as a function of the reverse bias. Note that at least three different pre-breakdown mechanisms may be active in multicrystalline solar cells, only one of them being the well-known avalanche breakdown (Breitenstein et al., 2011b). Only this mechanism leads to the multiplication of photo-induced carriers. It is well-known that avalanche multiplication shows a sharp onset voltage, below which there is no multiplication and above which the MF steeply increases. The MF-ILIT technique works in such a way that one ILIT image with pulsed reverse bias and steady-state illumination is taken at a low reverse bias, where no avalanche multiplication is expected, and another image is taken in the same way at the reverse bias where the MF has to me imaged. Then, by appropriate evaluation of these two images, an image of the avalanche MF is obtained (Breitenstein et al., 2008). MF = 1 means no avalanche multiplication. An MF-ILIT image of the cell used also for Figures 4 and 5(b), taken at –15 V, is shown in Figure 5(c). It is visible that, indeed, the regions showing a negative TC and a high slope also show a large avalanche MF. The

90 O. Breitenstein

comparison with Figure 5(b) shows that this breakdown behaviour has no significant influence on the local efficiency.

Figure 5 (a) Voc-ILIT image (a.u.) of a cell containing shunts, measured within one second, (b) monochromatic efficiency image (–10 to +44%), (c) avalanche MF image (0 to 3) (see online version for colours)

2 cm

(a) (b)

(c)

4 Conclusions

The imaging technique of lock-in thermography implies a whole group of techniques, some working in the dark (DLIT) and some under illumination (ILIT), which can be used to image various electronic properties of solar cells. By using DLIT, all kinds of shunts (sites of locally increased dark current) can be imaged, both under forward and reverse bias, and their electronic properties may be investigated. Thus, it may be checked whether a shunt shows a linear or a non-linear characteristic, recombination currents may be distinguished from diffusion currents, and the temperature coefficient and the relative slope of a local characteristic can be imaged at a certain bias. DLIT even may be used for a detailed local efficiency analysis of solar cells. ILIT performed under open circuit condition (Voc-ILIT) also allows to image shunts, even without electrically contacting the cell, but this works only under forward bias. These Voc-ILIT images always contain a certain homogeneous background signal, which may disturb their quantitative evaluation.

Illuminated versus dark lock-in thermography investigations of solar cells 91

If only local shunts have to be detected, this disturbance may be avoided by displaying the in-phase (0°) signal. There are two special ILIT techniques, where this homogeneous signal is properly compensated or even used for imaging, which is (monochromatic) efficiency imaging and avalanche MF imaging. ILIT can also be used for a special kind of minority carrier lifetime imaging (CDI/ILM). It may be concluded that DLIT and ILIT techniques can be used for different imaging tasks and supplement each other.

It should be noted that LIT is a relatively slow technique needing acquisition times between some minutes up tens of minutes for forward-bias investigations. Therefore, LIT is less appropriate for in-line performance testing of solar cells, e.g., for accompanying or replacing the usual I-V characteristics measurements. Instead, if in such I-V measurements some cells with poor efficiency parameters are detected, LIT techniques may be used for finding out the root causes responsible for such poor parameters. Only if these root causes are known, corresponding actions may be taken for avoiding them and thus improving the efficiencies of the produced cells. LIT investigations under reverse bias, on the other hand, can be made very quickly in less than a second. This technique may be and is already used for detecting strong ohmic shunts in solar cell production lines.

Acknowledgements

Many thanks to K. Bothe (ISFH Hameln) for providing the monochromatic efficiency image used for Figure 5(b). This work was financially supported by the German Federal Ministry for the Environment, Nature Conservation and Nuclear Safety and by industry partners within the research cluster ‘SolarWinS’ (contract No. 0325270C). The content is the responsibility of the authors.

References Bail, M., Kentsch, J., Brendel, R. and Schulz, M. (2000) ‘Lifetime mapping of Si wafers by an

infrared camera’, in Proc. 28th IEEE Photovoltaics Specialists Conference, Anchorage 2000, USA, pp.99–103.

Breitenstein, O., Langenkamp, M., Lang, O. and Schirrmacher, A. (2000) ‘Shunts due to laser scribing of solar cells evaluated by highly sensitive locl-in thermography’, Solar Energy Materials & Solar Cells, Vol. 65, Nos. 1–4, pp.55–62.

Breitenstein, O., Rakotoniaina, J.P., Al Rifai, M.H. and Werner M. (2004) ‘Shunt types in crystalline silicon solar cells’, Progress Photovolt: Res. Appl., Vol. 12, No. 7, pp.529–538.

Breitenstein, O. and Rakotoniaina, J.P. (2005) ‘Electrothermal simulation of a defect in a solar cell’, Journal of Applied Physics, Vol. 97, No. 7, 074905.

Breitenstein, O., Bauer, J., Wagner, J-M. and Lotnyk, A. (2008) ‘Imaging physical parameters of pre-breakdown sites by lock-in thermography techniques’, Prog. Photovolt: Res. Appl., Vol. 16, No. 8, pp.679–685.

Breitenstein, O., Warta, W. and Langenkamp, M. (2010) Lock-in Thermography - Basics and Use for Evaluating Electronic Devices and Materials (Second Edition), Springer, Heidelberg/New York.

Breitenstein, O., Bauer, J., Bothe, K., Hinken, D., Müller, J., Kwapil, W., Schubert, M.C. and Warta, W. (2011a) ‘Can luminescence imaging replace lock-in thermography on solar cells?’, IEEE Journal of Photovoltaics, Vol. 1, No. 2, pp.159–167.

92 O. Breitenstein

Breitenstein, O., Bauer, J., Bothe, K., Kwapil, W., Lausch, F., Rau, U., Schmidt, J., Schneemann, M., Schubert, M.C., Wagner, J-M. and Warta, W. (2011b) ‘Understanding junction breakdown in multicrystalline solar cells’, Journal of Applied Physics, Vol. 109, No. 7, 071101.

Breitenstein, O. (2011) ‘Nondestructive local analysis of current-voltage characteristics of solar cells by lock-in thermography’, Solar Energy Materials & Solar Cells, Vol. 95, No. 10, pp.2933–2936.

Breitenstein, O. (2012) ‘Local efficiency analysis of solar cells based on lock-in thermography’, Solar Energy Materials & Solar Cells, Vol. 107, pp.381–389.

Gupta, R., Breitenstein, O., Zettner, J. and Karg, D. (2007) ‘In-line shunt detection in solar cells by fast lock-in infrared thermography’, in Proc. 22nd Photovolt. Solar Energy Conference’, Milan, Italy, pp.1975–1978.

Hering, G. (2012) ‘Zellproduktion 2011 (in german)’, Photon 4/2012, pp.42–63. Kuo, P.K., Ahmed, T., Huijia, J. and Thomas R.L. (1988) ‘Phase-locked image acquisition in

thermography’, SPIE, Vol. 1004, p.41. Maldague, X.P.V. (2001) Theory and Practice of Infrared Technology for Nondestructive Testing,

Wiley, New York. Ramspeck, K., Bothe, K., Schmidt, J. and Brendel, R. (2008) ‘Correlation between spatially

resolved solar cell efficiency and carrier lifetime of multicrystalline silicon’, J. Mat. Sci: Mater. Electron., Vol. 19, No. 19, pp.S4–S8.

Riepe, S., Isenberg, J., Ballif, C., Glunz, S.W. and Warta, W. (2001) ‘Carrier density and lifetime imaging of silicon wafers by infrared lock-in thermography’, in Proc. 17th Eur. Photovoltaic Solar Energy Conference, Munich 2001, Germany, pp.1597–1599.

Schubert, M.C., Riepe, S., Bermejo, S. and Warta, W. (2006) ‘Determination of spatially resolved trapping parameters in silicon with injection dependent carrier density imaging’, Journal of Applied Physics, Vol. 99, No. 11, p.114908.

Simo, A. and Martinuzzi, S. (1990) ‘Hot spots and heavily dislocated regions in multicrystalline silicon cells’, in Proc. 21rd IEEE Photovoltaics Specialists Conference, Kissimee, USA, pp.800–805.

Sze, S.M. and Gibbons, G. (1966) ‘Avalanche breakdown voltages of abrupt and linearly graded p-n junctions in Ge, Si, GaAs, and GaP’, Applied Physics Letters, Vol. 8, No. 5, pp.111–113.