Relation between the local soft breakdown and the global re- re-verse I-V characteristics

The global breakdown behavior is determined by the concatenation of the behavior of all local breakdown sites. Hence in the following, the local breakdown from the very first stages of pre-breakdown up to high reverse currents (~ 10 A) is studied with the help of spatially resolved measurements taken over the full solar cell area. The aim is to be able to explain the large variety in the global reverse characteristics as described in section 6.3.1.

In Figure 6.32 (b), the global reverse I-V characteristics of various solar cells selected from ingots Stand. 2-5 and from the UMG-Si ingot (description see Table 6.1) are plot-ted⁵⁵. All solar cells differ in their global breakdown voltage, that is their point of maxi-mum curvature of the reverse IV-characteristics, depending mainly on the base resistivity of the wafers. While the solar cells taken from ingots Stand. 4 and 5 do not show any early reverse current increase in the pre-breakdown regime (see Figure 6.5), a significant current flows through the solar cells from ingot Stand. 2 and 3 and the UMG-Si ingot be-fore the global breakdown voltage is reached.

In the graph Figure 6.32 (a), for comparison, the fraction of the breakdown light emitting surface is plotted versus the reverse voltage; this fraction corresponds to the percentage of the solar cell area which takes part in the breakdown process. These curves, too, can be divided into two regimes: In the beginning, at low reverse bias, the number of break-down sites increases rather gently. This is followed by a sudden multiplication of the spots, as soon as a critical reverse voltage is reached. The critical reverse voltage of abrupt increase in BD sites of three examples is marked by the arrows. Note, that in the scale necessary to plot all curves the onset points are not as obvious as from the data values. By comparing graph (a) with the global I-V curve (b), one sees that the rapid BD site multiplication coincides with the global breakdown voltage. The correlation is obvious for the solar cells made from UMG-Si and ingots Stand. 2 and 3; it is not so good for the investigated solar cells from ingots Stand. 4 and 5.

There are two facts to be learned from this comparison: Firstly, in the beginning the total reverse current is determined by the number of breakdown sites, each carrying only a small current. Only later, the exponential behavior of the soft breakdown and hard breakdown sites (if there are any) makes its contribution to the global reverse character-stics⁵⁶.

55 Measurement by E. Schäffer.

56 Recent microscopic investigations of the bias-dependent BD light intensity of single breakdown sites by P.

Gundel (Fraunhofer ISE) and by M. Schneemann (FZ Jülich) [179] suggest that the light intensity grows linearly with increasing reverse bias – contrary to the results measured via silicon CCD cameras in usual EL measure-ments, see Figure 6.7. The macroscopically soft behavior is explained by the close vicinity of many similar breakdown sites, each of which starts to emit light at a different reverse bias, then adding to the macroscopi-cally observed light intensity. Note that the EL silicon CCD camera setup offers a maximal resolution in the order of 100 µm; therefore, one camera pixel may well cover dozens of individual soft breakdown sites.

If the light intensity of one BD site is in fact proportional to the local reverse current flowing through that site, then the above result translates into a linearly increasing reverse current with increasing reverse bias. This would be for example the case if the current is limited by the series resistance due to the transport of carriers toward the breakdown channel. A similar prediction was made in reference [149].

Secondly, one can see that the pre-breakdown regime in the global reverse characteris-tics is related to the gentle increase of the number of breakdown sites, before a rapid multiplication sets in.

The reason for the moderate amplification of the number of breakdown sites in the pre-breakdown regime is best visualized in the maps of the local pre-breakdown voltage. As an example, the solar cell from the UMG-Si ingot (25% ingot height) is chosen; its reverse bias behavior is plotted in Figure 6.32 (red stars).

Figure 6.32: Plots of the fraction of the surface area emitting breakdown light (a) and of the global reverse current (b) versus the reverse voltage. Both graphs show measure-ments of the same selection of different solar cells taken from ingots “Dop mc 2-5” and from the ingot “UMG mc 1” (see Table 6.1), which illustrate the variety of the breakdown behavior. For the explanation of the arrows, please refer to the text.

Figure 6.33: (a) Breakdown voltage map of the UMG-Si solar cell taken from 25% ingot height (compare with Figure 6.28) from ingot UMG mc 1. The measurement of the frac-tion of the light emitting surface area and of the global reverse I-V characteristics of this cell are plotted in Figure 6.32 (“UMG mc 1”, red curves). (b) The global reverse I-V characteristics is plotted of the entire cell (black continuous curve) and, after laser cut-ting along the white dashed lines in (a), of the recombination active area A at the left wafer edge (red dashed curve) and of the cell center B (blue dashed and dotted curve).

For comparison, reverse bias dependent DLIT measurements were performed before the cell was cut. The current density was analyzed for the entire solar cell area (black trian-gles), the recombination active area A (red squares) and the cell center part B (blue circles).

In Figure 6.33 (a), the breakdown voltage map reveals several groups of breakdown sites, nevertheless all belonging to the soft BD type: At the left and right wafer edges, the local breakdown voltage is decreased as a result of the contamination due to the cru-cible (see also Figure 6.28). Also in the center of the solar cell, there are a few small re-gions with a lower breakdown voltage due to highly recombination active crystal defects similar to the highlighted features in Figure 6.22. All these regions break down between -5 to -8 V (yellow to dark orange in image (a)), which makes them responsible for the slow increase in the pre-breakdown regime.

The second set of breakdown sites appears at around -8 to -8.5V. In image (a), this volt-age corresponds to the colors dark red and violet. These sites are mainly found in the solar cell center away from the wafer edges. By comparing the breakdown sites with the forward EL image, Figure 6.28 (a) and (b), one can see that light is emitted at recombi-nation active defects which are less pronounced. They consist of straight dark lines asso-ciated with grain boundaries. In total, there are only very few recombination active de-fects left which do not break down in the voltage-current range up to -10 A.

Therefore, the relation between the local soft breakdown behavior and the global reverse I-V characteristics is summarized thus: Regions of high contamination and of high re-combination activity (which may be due to a certain configuration of crystal defects, es-pecially dislocation clusters, appearing as dark “nests”) result in “pre”-breakdown and induce a “soft” global breakdown behavior.

This statement in verified in Figure 6.33 (b), in which two different bias dependent measurements are plotted. At first, the solar cell was divided into three parts, see image (a): Part A contains mostly the highly contaminated left wafer edge while part B consists of the solar cell center, in which only a few spots break down at a low reverse voltage (part C is disregarded for reasons of lucidity). DLIT measurements were performed and calibrated to current densities which are shown for the entire cell and for part A and B (symbols). Then, the reverse I-V curve of the entire cell was measured before the solar cell was cut along the dashed white lines shown in image (a). Afterwards, the I-V charac-teristics of parts A and B were reassessed (lines)⁵⁷. Taking into account the different con-tacting methods of the sample, having different contact resistances, the DLIT and the I-V measurements are in reasonable agreement.

Through the wafer edge (part A), reverse current starts to flow already at a very low bias around -3 to -4 V. The fact that breakdown light is not detected before ca. -5 V is due to the way the detection threshold is set (see section 6.2.1). The I-V characteristics are very soft resembling IFE diodes. On the other hand, the less recombination active cell (part B) center shows a very hard breakdown behavior resembling avalanche breakdown, although breakdown still occurs at recombination active defects.

In fact, the bias-dependent reverse current of the whole, uncut cell matches exactly the sum of the reverse current flowing through the three separate parts A-C.

The analysis shown here yields direct proof that highly recombination active defects sof-ten the I-V characteristics and are responsible for the pre-breakdown in mc-Si solar cells.

6.5.8 Theoretical considerations on the basic mechanism of soft