Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 92 (2016) 10 15

6th International Conference on Silicon Photovoltaics, SiliconPV 2016

Local solar cell efficiency analysis performed by injection-

dependent PL imaging (ELBA) and voltage-dependent lock-
in thermography (Local I-V)
Otwin Breitensteina, Felix Frhaufa, Jan Bauera, Florian Schindlerb, Bernhard Michlb
a
Max Planck Institute of Microstructure Physics, Weinberg 2, Halle 06120, Germany
b
Fraunhofer Institute for Solar Energy Systems (ISE), Heidenhofstrae 2, Freiburg 79110, Germany

Abstract

In this contribution two methods for performing local efficiency analysis of solar cells are compared with each other by applying
them to a solar cell and a neighboring wafer. The first method called "ELBA" is based on injection-dependent photoluminescence
(PL) imaging of a passivated wafer. The second method called "Local I-V" is based on voltage-dependent dark lock-in
thermography (DLIT) on a solar cell. The results of both methods with respect to the influence of the bulk on the local solar cell
parameters are comparable with each other. However, since only "Local I-V" is investigating a finished solar cell, it may image
also local ohmic shunts, inhomogeneous front- and backside and depletion region recombination, and Rs effects, whereas
ELBA is suited for assessing bulk-related efficiency losses in detail, which are not concealed by the aforementioned cell-
related loss mechanisms in this method.

2016
2016TheTheAuthors. Published
Authors. Publishedby Elsevier Ltd. Ltd.
by Elsevier This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG.
Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG.
Keywords: Local efficiency analysis; PL imaging; lock-in thermography; ELBA; Local I-V

1. Introduction

In multicrystalline silicon solar cells the carrier lifetime is very inhomogeneous. Here low lifetime regions limit
the performance of the whole cell by (1) limiting the short circuit current and (2) increasing the dark current and thus
reducing the fill factor (FF) and open circuit voltage (Voc). Several methods have been developed to study and
evaluate this effect. Two of them are Efficiency Limiting Bulk recombination Analysis (ELBA [1]) and voltage-
dependent dark lock-in thermography (Local I-V [2]). ELBA is an injection-dependent PL image evaluation method,
which is performed on passivated wafers that have experienced the same thermal processes as solar cells. Local I-V,

1876-6102 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG.
doi:10.1016/j.egypro.2016.07.003
 Otwin Breitenstein et al. / Energy Procedia 92 (2016) 10 15 11

on the other hand, is performed on complete solar cells and thus also implies the influence of the emitter and the
metallization. In this work both methods are compared quantitatively on an industrial solar cell and a parallel wafer
exposed to the same thermal history. The differences can be well explained by regarding the different regions of the
device where both methods gain information for the analysis.

2. Experimental

The Local I-V measurements have been made on a multicrystalline silicon solar cell of PERC technology (laser-
fired contacts on the rear, 190 m thick 1 :cm p-type bulk, acidic texture, 90 :/sqr P-doped emitter), the cell data
are shown in Table I. The wafer used for the ELBA analysis was a parallel wafer, which was exposed to the same
diffusion and annealing history as the cell. Then the emitter was etched away and the wafer was passivated by a
stack of aluminum oxide and silicon nitride on both sides. The PL measurements used for the ELBA analysis were
performed by using a 1000 nm short-pass filter. The used LIT system was an InfraTec PV-LIT system, which
enables local IR emissivity correction. The EL measurements for measuring the local diode voltage in the dark were
performed by using a 1000 nm long-pass filter.

2.1. ELBA

Efficiency limiting bulk recombination analysis (ELBA) [1] allows for a prediction of solar cell efficiency
limited by injection-dependent bulk recombination for a given cell concept. The local bulk recombination is
assessed by injection-dependent lifetime imaging [2] of a surface-passivated lifetime sample that has been subjected
to the same thermal process steps as the solar cell under investigation. This ensures that the bulk recombination in
the lifetime sample corresponds to that in the finished solar cell. Photon blurring in the PL images was reduced here
by the use of a 1000-nm short-pass filter. We have checked that deconvolution of the photon scattering by a
measurement of the point spread function after [4] does not lead to significant changes in the mean lifetime of the
different regions anymore. In combination with local PC1D cell simulations accounting for the characteristics of the
chosen cell concept, high-resolution images of Voc, Jsc, FF, and the efficiency limited by bulk recombination are
calculated. Global values for the whole wafer are calculated considering balancing mechanisms between the areas
on the wafer in the model of independent diodes. The PC1D cell model was adjusted to a monocrystalline solar cell
of the same cell batch. It has to be noted that the ELBA results do not rely in any way on measurements of the final
multicrystalline cell, but only on measurements of the bulk lifetime on a wafer. Therefore, the results are real
independent predictions of the bulk limited efficiency. Furthermore, the wafer based ELBA simulation allows a very
detailed investigation of the materials efficiency limits, which would be not possible by a cell based method: An
arbitrary cell concept can be applied, an arbitrary cell thickness can be investigated, even the impact of base doping
can be estimated, and by a combination with measurements of defect concentrations, the efficiency limit due to
specific defects in the wafer can be assessed.

2.2. Local I-V

"Local I-V" is a dark lock-in thermography (DLIT) image evaluation method, which exploits the proportionality
of the DLIT signal and the locally dissipated power density [3]. Four DLIT images are evaluated to fit each pixel to
a 2-diode model, leading to images of J01, J02, n2, and Gp = 1/Rp. The local series resistance Rs is regarded by
entering an electroluminescence- (EL)-based local diode voltage image taken at the highest forward bias (here 610
mV). The EL images, taken for wavelengths above 1000 nm here, were corrected for light scattering in the detector
according to [4], and the local diode voltage was artificially blurred (thermal blurring) for having the same spatial
resolution as the DLIT-based J images. Using the DLIT-based local J01 data, for a given global value of Jsc, local Jsc
expectation data are calculated according to [5]. Once this and the local 2-diode parameters are known, the method
calculates efficiency-related images like the local Voc, the local FF, and the local efficiency K [6], which may be
compared with the corresponding ELBA data. All these images are calculated, like the ELBA images, assuming
electrically isolated pixels, but considering lateral balancing currents in the model of independent diodes. Therefore
these local efficiency data are expectation values of the corresponding efficiency parameters, which would hold for
12 Otwin Breitenstein et al. / Energy Procedia 92 (2016) 10 15

a homogeneous solar cell having the properties of the pixel in position (x,y). The 'Local I-V 2' software performing
these evaluations is available [7].

3. Results

Figure 1 shows lifetime images resulting from PL images of the investigated wafer for various illumination
intensities and resulting ELBA-based results of the expected cell, all scaling ranges are indicated. We see significant
and locally different changes in the lifetime over the whole range of illumination intensities from 0.004 to 1.54 suns.
Fig. 2 (a) shows results of a PC1D analysis of this cell type, which is used to predict local Voc and Jsc data from the
lifetime images. Fig. 2 (b) shows some examples of carrier-dependent lifetimes from three typical positions marked
by the corresponding symbols in Fig. 1 (c) and from the whole wafer. The predicted local cell data in Fig. 1 (g) and
(h) were calculated by assuming a homogeneous series resistance of 1.05 :cm2.

(a) W (0.004 suns) (b) W (0.1 suns) (c) W (1 sun) (a) W (1.54 suns)
0 to 50 s 0 to 50 s 0 to 250 s 0 to 250 s

(e) ELBA Voc (f) ELBA Jsc (g) ELBA FF (h) ELBA K
560 to 660 mV 32 to 38 mA/cm2 60 to 83 % 13 to 20 %
max

2 cm
min

Fig. 1. (a) - (d): Lifetime images of the wafer for various illumination intensities. The symbols in (c) mark the positions of the carrier-dependent
lifetime curves in Fig. 2 (b). (e): ELBA-based Voc (expectation value image, see text), (f): Jsc, (g): FF, (h): efficiency expectation image, the latter
calculated under the assumption of a homogeneous Rs of 1.05 :cm2. The color bar in (g) holds for all images.

(a) (b)

Fig. 2. (a) PC1D results for the investigated cell type, (b) carrier-dependent lifetime data from three positions marked by the corresponding
symbols in Fig. 1 (c).
 Otwin Breitenstein et al. / Energy Procedia 92 (2016) 10 15 13

Figure 3 shows the most relevant images coming from the DLIT- and EL-based Local I-V analysis. Images (e) to
(h) may be directly compared with the ELBA-based images in Figure 1. Figure 4 shows local I-V characteristics of
some positions in the cell obtained from evaluating the DLIT images, which have led to Figs. 3 (a) to (c). These
characteristics are from the position of a J01-type shunt (a), from a 'good' region (b), from a J02-type shunt at the left
edge (c) and from an ohmic shunt (d). The different current contributions (J01, J02, ohmic) are shown separately, the
symbols mark the measured points. The positions of these local characteristics are indicated in Figs. 3 (a) to (c).

(a) J01 (b) Jrec(610 mV) (c) Jshunt(-1 V) (d) Vd(610 mV)
0 to 4 pA/cm2 0 to 50 mA/cm2 0 to 20 mA/cm2 580 to 610 mV

ohmic shunt max

good region J02 shunt

2 cm
J01 shunt min

(g) FF (h) Local I-V K

(e) Local I-V Voc (f) Local I-V Jsc
60 to 83 % 13 to 20 %
560 to 660 mV 32 to 38 mA/cm2

Fig. 3. (a): DLIT-based dark saturation current density, (b): depletion region recombination current density at 610 mV, (c): ohmic shunt current
density at -1 V (note the different scaling range), (d): Local I-V based local diode voltage at 610 mV, (e): Voc expectation image, (f): Jsc image
calculated after [4], (g): FF image (with Rs based on the Jdark image (a) and the Vd image (d)), (h): efficiency expectation image. The color bar in
(c) holds for all images.

4. Discussion

The comparison between Figs. 1 and 3 shows good correlations but also distinct differences, which all may be
well interpreted. The spatial resolution of the ELBA results is generally much better than that of the Local I-V
results, since only the latter are thermally blurred. Note that these PL results are obtained on a wafer, hence they are
not disturbed by horizontal balancing currents. Both evaluations clearly indicate the typical crystal defect regions in
the material and regions of lower lifetimes at the upper and the left edge of the cell, which are due to the fact that the
wafers stem from a corner brick. Hence the contamination with impurities (e.g. iron) of the material at these edges is
stronger limiting the lifetime there. The Local I-V based Voc image predicts clearly lower values in the crystal defect
regions than the ELBA-based one. A possible explanation is that the crystal defects not only influence the bulk
lifetime but also the emitter and/or backside recombination. This would become visible in Local I-V, but not in
ELBA. It was detected in [8] (a similar PERC cell process was investigated there) that the emitter recombination can
be significantly higher in dislocated areas. Using this increased recombination (2000 cm/s for effective surface
recombination of emitter and front surface) in the ELBA simulation leads to a 30mV lower Voc potential in a
dislocated area. In [9] it was observed that, in recombination-active defect regions, also the front metallization
contribution of J01 increases. Due to its limited spatial resolution, DLIT cannot distinguish between regions below
and between gridlines. Thus, these effects could explain the main differences between ELBA and Local I-V. The Jsc
images of both methods look quite similar, except that the influence of the top and left edge regions is stronger
predicted in ELBA than in the J01-based analysis after [5]. This may be due to the fact that the parameters in [5] are
fitted to regions of recombination-active grain boundaries, whereas in the edge regions of the cell presented here we
have a homogeneously low lifetime. Hence, in this region other parameters may apply. The two FF images differ
14 Otwin Breitenstein et al. / Energy Procedia 92 (2016) 10 15

most strongly. The ELBA analysis cannot take into account the Jrec and Jshunt images in Fig. 1, which both strongly
influence the FF image of the cell. Moreover, since in the ELBA analysis a homogeneous Rs was assumed, the
position-dependence of Rs is not considered. Thus, as expected, the ELBA-based efficiency expectation image
correctly considers the influence of the bulk lifetime, but not that caused by locally inhomogeneous ohmic shunts,
emitter, backside, and depletion region recombination, and Rs effects. Table I shows the measured global cell
parameters and that predicted by ELBA and Local I-V.

80 80
(a) J01 shunt (b) good region
70 70
60 60
J [mA/cm2]

J [mA/cm2]
Jsum[mA/cm] Jsum[mA/cm]
50 Jrec[mA/cm] 50 Jrec[mA/cm]
40 Jdiff[mA/cm] 40 Jdiff[mA/cm]
<Current>[mA/cm] <Current>[mA/cm]
30 30
20 20
10 10
0 0
0.50 0.52 0.54 0.56 0.58 0.60 0.50 0.52 0.54 0.56 0.58 0.60
Voltage[V] Voltage[V]
20
80
(c) J02 shunt (d) ohmic shunt
70
60 Jsum[mA/cm]
J [mA/cm2]
J [mA/cm2]

Jsum[mA/cm] Jrec[mA/cm]
50 Jrec[mA/cm] Jdiff[mA/cm]
40 Jdiff[mA/cm] 10 Jsum[mA/cm]
30 <Current>[mA/cm] <Current>[mA/cm]
20
10
0 0
0.50 0.52 0.54 0.56 0.58 0.60 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Voltage[V] Voltage[V]

Fig. 4. Local dark I-V characteristics: (a) J01 shunt (position shown in Fig. 3 a), (b) good region (position shown in Fig. 3a), (c) J02 shunt
(position shown in Fig. 3 b), (d) ohmic shunt (position shown in Fig. 3 c). The symbols mark the measured points.

Table I. Measured and simulated global cell parameters.

efficiency parameter flasher cell data ELBA Local I-V

Voc [mV] 631 630 632
2
Jsc [mA/cm ] 36.6 36.3 36.6 (fitted)
FF [%] 76.6 77.0 76.6
efficiency [%] 17.7 17.6 17.7

Acknowledgements

O. Breitenstein, F. Frhauf, and J. Bauer acknowledge the cooperation with InfraTec (Dresden) [10] and the
financial support by the German Federal Ministry for Commerce and Energy and by industry partners within the
research cluster "SolarLIFE" (contract no. 0325763 D). The content is in the responsibility of the authors.

References

[1] B. Michl, M. Rdiger, J.A. Giesecke, M. Hermle, W. Warta, and M.C. Schubert, "Efficiency limiting bulk recombination in multicrystalline
silicon solar cells", Solar Energy Materials & Solar Cells 98, 441-447 (2012).
 Otwin Breitenstein et al. / Energy Procedia 92 (2016) 10 15 15

[2] J.A. Giesecke, M.C. Schubert, B. Michl, F. Schindler, W. Warta, Minority carrier lifetime imaging of silicon wafers calibrated by quasi-
steady-state photoluminescence, Solar Energy Materials & Solar Cells 95, 1011-1018 (2011).
[3] O. Breitenstein, "Nondestructive local analysis of current-voltage characteristics of solar cells by lock-in thermography", Solar Energy
Materials & Solar Cells 95, 2933-2936 (2011).
[4] D. Walter, A. Fell, E. Franklin, D. Macdonald, B. Mitchell, and T. Trupke, "The impact of silicon CCD photon spread on quantitative
analyses of luminescence images", IEEE Journal of Photovoltaics 4 (1), 368-373 (2014).
[5] O. Breitenstein, F. Fertig, and J. Bauer, "An empirical method for imaging the short circuit current density in silicon solar cells based on dark
lock-in thermography", Solar Energy Materials & Solar Cells 143, 406-410 (2015).
[6] O. Breitenstein, "Local efficiency analysis of solar cells based on lock-in thermography", Solar Energy Materials & Solar Cells 107, 381-389
(2012).
[7] www.maxplanckinnovation.com/en/ (Febr. 2016).
[8] B. Michl, M. Padilla, I. Geisemeyer, S. T. Haag, F. Schindler, M. C. Schubert, and W. Warta, "Imaging techniques for quantitative silicon
material and solar cell analysis", IEEE Journal of Photovoltaics 4 (6), 1502-1510 (2014).
[9] S. Riland, T.M. Pletzer, H. Windgassen, and O. Breitenstein, "Local thermographic efficiency analysis of multicrystalline and cast-mono
silicon solar cells", IEEE Journal of Photovoltaics 3 (4), 1192-1199 (2013).
[10] www.infratec-infrared.com (Febr. 2016)