For detecting local R_S defects that stem from unintended inhomogeneous processing with PL imaging, we explore the influence of well-defined R_S defects on the luminescence imaging signature. Finger interruptions are introduced artificially by laser cuttings, and the interruption is created by cutting the finger twice with a gap width of 1 mm by the laser beam. Figure 2(a) shows the EL image of a 20 mm × 20 mm monocrystalline passivated emitter and rear contact (PERC) solar cell with laser cutting fingers, which show no defects initially. The applied forward bias voltage is 0.676 V. producing a carrier injection level equivalent to 1 sun. The positions of finger interruptions are marked by black arrows. The region of interest (ROI) 1 is affected by a single interrupted finger while ROI2 is affected by two consecutively interrupted fingers. They have a same distance of 5 mm to the busbar. The affected regions appear to be darker than the unaffected regions in the EL image, which means that using the laser cutting can effectively introduce an finger interruption that restrains the charge carriers from transporting through the cutting-off grids. We note that the laser cutting will bring about inevitable recombination defects, and they do not affect the local R_S distribution,^[13] which can also be seen from Fig. 2(b). However, those defects reduce the luminescence intensity as shown in Fig. 2(a) and thus will cause locally disturbance to influence the R_S defects on PL signal. Therefore, while calculating ΔPL bellow, those defective regions should not be taken into consideration.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) Images of a monocrystalline solar cell with interrupted fingers induced by laser cutting. (a) EL image with an applied forward bias of 0.676 V. (b) PL image at an applied voltage of 0.476 V under 1-sun illumination. (c) RS,i image calculated by the C-DCR method.

The R_S,i image according to Eq. (2) is shown in Fig. 2(b). The region with cyan dash rectangles has been rescaled to allow for a better contrast in the R_S,i image. As expected, interrupted fingers only affect the regions of higher R_S,i away from the respective busbar. In the EL image, the luminescence signal is significantly elevated in ROI2 and slightly elevated in ROI1. This indicates severe and slight additional series resistance losses in their respective region. This can be quantitatively confirmed by the R_S,i image, where the peak values of affected regions reach 6.5 Ω·cm² for ROI2 and 1.5 Ω·cm² for ROI1, respectively. This means that a slight locally increased R_S defect (ROI1) and a significantly increased R_S defect (ROI2) are produced by laser cutting. At interrupted fingers, the detected local R_S defects can be explained as follows: the lateral current of the affected region has to bypass the metal interruptions by flowing through the emitter, this non-ideal electrical path leading to a high resistive loss in the affected region.

While a solar cell absorbs the photons from the excitation light, electron-hole pairs are generated in the bulk. These excess carriers will split the quasi-Fermi levels and induce a local diode voltage. When the applied voltage of the solar cell is lower than the local diode voltage, the voltage difference provides the driving force to drive lateral current flows and to extract the excess carriers through the external contact. A finger interruption will induce a region with locally increased series resistance as discussed above, and the extraction of the excess carriers will be impeded in those regions with high R_S,i. The PL signal is proportional to the product of charge carrier densities. Therefore, the regions influenced by interrupted fingers will show higher PL intensities than those in the regions without any R_S defects. Figure 2(b) shows the PL images under current extraction of the same sample shown in Fig. 2. The sample is illuminated at 1 sun (V_OC = 0.650 V) and applied a forward voltage of 0.476 V. It can be clearly seen that ROI1 and ROI2 influenced by finger interruptions show higher PL intensities than those in the surrounding defect-free regions, which is due to higher R_S,i values in those regions as shown in Fig. 2(b). This is consistent with our previous analysis. The contrast between PL intensities of ROI1 and ROI2 can be calculated according to Eq. (1). The values of ΔPL in the influenced regions are 0.33 and 1.83, respectively. This is because a higher R_S has a greater ability to impede the extraction of the excess carriers. In conclusion, the region with elevated PL signal allows us to identify finger interruptions in a solar cell using PL imaging.

As discussed above, detection of finger interruptions using PL images is greatly dependent on excess carrier extraction. Variable forward applied voltage leads to variable driving force under a constant illumination intensity. It then induces different extraction levels of excess carriers and, therefore, different values of ΔPL. To investigate the influence of the external bias control on PL image, the applied voltage of the sample is changed to ensure that the cell works at different working points (from SC to OC). The PL images are obtained at different working points, and the values of Δ PL of ROI1 and ROI2 are calculated as well, respectively. To have a better comparison and analysis, the curve of ΔPL versus the applied voltage and the extracted current (J_extra) and the I–V curve are shown in Fig. 3(a). Some representatively selected PL images at different working points are shown in Fig. 3(b).

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (color online) (a) Influence of external bias control on ΔPL. (b) PL images of the selected working points from SC to OC.

The finger interruption resulting in higher R_S,i exhibits higher ΔPL at all current extractions, which is in agreement with the results for the case of mentioned above. It can be clearly seen that ΔPL is greatly dependent on external bias control, and can be divided into three stages according to the applied voltage. In the first stage (from 0 V to 0.30 V for ROI1), ΔPL keeps a constant value approaching to 0 with V_appl increasing. This is because the extraction of carriers from the defective region and non-defective region are saturated, which can be verified by the fact that the extracted current approximately keep a constant value of 141 mA. In this stage, the local R_S defects of the interrupted fingers cannot be detected. The PL image at short circuit can only detect defects with extremely high R_S,i^[14] that are large enough to impede the extraction of carries and change the local short circuit current extraction, like a completely missed finger or four consecutively interrupted fingers. In the second stage, with the further increase of V_appl, ΔPL increases to a peak value (ΔPL_max) of 0.33 at an applied voltage of 0.476 V for ROI1. This can be understood as the extraction of carries from the non-defective region being still saturated or nearly saturated, whilst in the defective region it decreases rapidly. In this stage, the sensitivity of the PL imaging method to local R_S defect increases to a maximum value. In the third stage, the extracted current rapidly decreases with V_appl increasing, which means that the lateral current decreases rapidly. This leads to reducing the R_S effect. under the open circuit condition, ΔPL approaches to 0 as no lateral current exists. The V_appl corresponding to ΔPL_max of ROI2 has a slight decrease (about 0.04 V) with respect to ROI1. This is the result of higher R_S,i in ROI2. The variation of Δ PL can be visually seen from the PL images (i)–(ix), and the finger interruptions cannot be identified by the PL images under open circuit condition or short circuit condition.

The results discussed above show that in the context of defect detection, an optimal applied voltage range exists for which ΔPL is at a maximum. However, the most suitable range of V_appl relates to various cell aspects, such as the metal finger and the emitter sheet resistance. So it is different for different type solar cells. It is also influenced by the illumination intensity and the external contact. Despite all these influence factors, we note that none of working points at a sufficient current extraction (approaching to SC) and low current extraction (approaching to OC) are suitable for interruption detection because of their low sensitivity to R_S defects.

The illumination intensity is another important experimental variable that changes the working condition of solar cells. It is known that the PL signal depends strongly on the illumination intensity because of the variable carrier injection level. Therefore, we assume that the illumination intensity also has a great influence on ΔPL. To investigate the influence of illumination intensity, PL images of the sample are obtained at illumination intensities varying from 0 to 4.25 suns by changing the laser power, and then the values of ΔPL_max at different illumination intensities are acquired through the method shown in Fig. 3. The variations of ΔPL_max values with illumination intensity of ROI1 and ROI2 are shown in Fig. 4. For better contrast, different vertical coordinates are used. The region with higher series resistance exhibits higher ΔPL_max at all illumination intensities, which is in agreement with the results shown above. With the increase of illumination intensity, ΔPL_max reaches a maximum value at an intermediate intensity and decreases toward higher intensities. The ΔPL_max approaches to 0 at low intensity extreme because of no carrier injection. At extremely low illumination intensities, the lateral currents are small, which results in negligible R_S-effect. As the intensity increases, ΔPL_max increases as well. This is because the extraction of carriers becomes dependent on R_S and the defective regions (PL_R) have elevated R_S,i relative to the non-defective regions (PL_n). For illumination intensity higher than the intensity where the ΔPL_max reaches a maximum, ΔPL_max begins to decrease. This can be understood as that high injection level makes insufficient the extraction of carriers in the non-defective regions. That ΔPL_max approaches to 0 at extremely high illumination intensities (not shown here) can be predicted, since the extracted carriers in both of the non-defective regions and the defective regions are negligible relative to the originally injected carries. The results show that increasing illumination intensity appropriately is an effective way to increase the sensitivity to local R_S defect. However, the corresponding illumination intensity of the maximal ΔPL_max is larger in the region of low R_S,i (about 2.8 suns for ROI1) than in the region of high R_S,i (about 1.9 suns for ROI2). This means that high illumination intensity will reduce the sensitivity to high-R_S defect to some extent. On the other hand, it is difficult to obtain high illumination intensity for industrial size silicon solar cells. Illumination intensity between 1 sun and 2 suns is recommended practically because it is both sufficient to detect the regions with slightly increased R_S,i as shown above and easy to realize the large area solar cells.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) Influences of illumination intensity on ΔPLmax, where different vertical coordinates of ROI1 (left) and ROI2 (right) are used for better contrast.

The results and discussion above show that the detection of interrupted fingers, and any other defects with locally increased R_S, are feasible in a PL image. The regions influenced by finger interruptions show that the PL signal is higher than their surrounding. Identification of finger interruptions is also available with EL images. On contrary to the increase in luminescence intensity in the PL image, interrupted fingers result in lower luminescence intensities in EL images. It is a challenge to use EL images to identify finger interruptions in solar cells with variations in bulk carrier lifetime, particularly in the case of multicrystalline cells. This is because the region with recombination defects (low carrier lifetime) and the region with R_S defects both result in lower luminescence intensity.^[4]

However, it is quite different in the PL imaging method that the regions affected by these two kinds of defects show inverse illumination intensities. Therefore, PL imaging seems to have greater ability to discriminate regions of low carrier lifetime and regions of high R_S,i. This can be confirmed by Fig. 5. Figure 5 shows luminescence images of two samples under different operating conditions. Both samples contain regions of local R_S defects and recombination defects. Sample 1 is an Al-BSF monocrystalline solar cell containing a region influenced by an interrupted finger (ROI3) and a neighboring region of low-carrier lifetime (ROI4). The rectangular region of low carrier lifetime centered over a finger shown in ROI4 is made by scratching the cell with laser cutting. In this region, the low lifetime is dominated by the surface recombination. Sample 2 is an Al-BSF multicrystalline solar cell with obvious recombination defects (dislocation clusters or grain boundaries). The ROI5 is affected by both kinds of defects. The illumination intensity of PL images is 1.5 suns, and the applied voltages of EL images are set to be equivalent to the same carrier injection level of 1.5 suns.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) Images of PL at open circuit, EL, PL under current extraction for two samples with both finger interruption defects and recombination defects (see text for details).

The recombination defects can be clearly seen from the PL images under open circuit condition without the disturbance of R_S defects. It needs to be noted that the PL signal in ROI3 is slightly elevated because the balancing current^[15] exists between ROI3 and ROI4. In sample 1, both ROI3 and ROI4 show lower luminescence intensity in the EL image. On the contrary, ROI3 shows higher luminescence intensity while ROI4 show lower luminescence intensity in the PL image under current extraction. The inversion of luminescence intensity allows the interrupted finger to be more accurately identified. In ROI5, the finger interruption and the recombination defects can also be clearly identified. It is noted that defects in ROI4 seem to have a lower influence on the PL image under current extraction than on the EL image and the PL image at open circuit. This is because the carriers near the pn junction are extracted, which will reduce the influence of surface recombination velocity on luminescence intensity.^[16] In conclusion, the results accord with our previous analysis that PL imaging is more suitable than EL imaging for the identification of finger interruptions in solar cells with variational variable carrier lifetime.