SRIXE measurements were performed by Natalie Palina, Agnieszka Banas, and Krzysztof Banasin atmosphere at room temperature with a phase-contrast image- tomography beam-line at the Singapore Synchrotron Light Source (SSLS) operating with an electron energy of 0.7GeV and with a bending magnet of 4.5T. The critical excitation x-ray energy was 1.47keV. This flux passed through a 500àm thick Be window and 50cm of air before interacting with specific regions of the sample labeled R1, R2, and R3 as shown in Figure 4.3.1. The x-ray beam spans 2 to 12keV with a peak at 3.8keV and maximum brilliance of ≈ 1010 photons/s/mradH/0.1%, and was directed with a 45◦ incidence angle onto the multicrystalline silicon wafer solar cell. This source had a silicon penetration depth of 8.5±1.5cm.
The beam line aperture was adjusted to provide a 1mm2 square spot onto regions of the cell marked in Fig. 4.3.1. The x-ray emission was acquired over 900s on a thermo- electrically cooled Si-PIN photodiode (AMPTEK, XR-100CR Si:Li) with a resolution of 250eV at 5.9keV. The sample to detector distance was 0.5±0.2cm to minimize scattering and increase sensitivity. Recorded SRIXE spectra were analyzed using a software package (Cross-roads Scientific, XRS-FP). The analysis included calibration, smoothing, baseline correction, normalization, extraction of characteristic intensity lines, and peak integration. For reference, a float-zone silicon wafer was studied using this same experimental apparatus.
4.4.1 Experimental results from SRIXE analysis
Figure 4.4.1 shows the resulting x-ray fluorescence spectra from regions R1, R2, and R3, as well as the float-zone reference sample. In both regionsR2 andR3, similar x-ray emission line intensities were observed, including those indicating Zn (K-lines), and Pb (L-lines). These are metals commonly present in the paste used for forming a front- grid contact on a silicon wafer solar cell, and most likely are detected due to the large
2
Figure 4.4.1: (a) Calibrated SRIXE spectra showing both spectra from regions of sub- bandgap (red), and reverse-bias (blue) luminescence emission, as well as defect-free re- gions of the sample (green) after background subtraction. For reference, the spectrum from a float-zone (FZ) silicon sample is shown (black). (b) The inset shows selected concentrations of metals computed using quantitative SRIXE, as labeled.
the fully processed solar cell.
A somewhat different content of Ca (not shown), Co, Fe, and other 3d-transition metals were found for bothR2 andR3 regions. These earth metals could be introduced into the multicrystalline silicon wafer during the production process, for example, during solidification of the multicrystalline silicon ingot, or could be present in the material used to form the ingot. The main difference in the SRIXE spectra is observed to be a presence of strong Co x-ray emission line in the spectrum taken in the region labeled R3. This observation was both repeatable if spectra were taken at the same point, and consistent if spectra were accumulated at different points inside different R2 and R3 areas. Various areas of the silicon wafer solar cell which were studied using the method are marked as yellow dots in the image of Figure 4.3.1. The results indicate that reverse- bias luminescence is related to metal impurities, whereas sub-bandgap luminescence (in the absence of impurity precipitation, as areas of overlap observed in Figure 4.3.1) is related to crystallographic dislocations. These results are similar to results found in the literature by Breitenstein [117, Figures 6 and 7] as well as Kwapil [455], and
others [254, 463–467]. Thus, metallic precipitates or point defects can lead to reverse- bias luminescence. This undoubtedly is related to the proximity of which these defects are to the pn junction.
Since dislocations in silicon wafers tend to penetrate the entire thickness of the wafer, and are not observed to always yield reverse-bias luminescence, the presence of precipitation of metals in dislocations extending into the junction, or the presence of a point metal defect near the junction are the likely candidates for reverse-bias emission.
As well, surface states and contamination due to processing at the top surface, such as laser doping or isolation, and fast firing of ohmic contacts on the upper surface of the solar cell are also likely candidates for this form of defect luminescence.
4.4.2 Summary and discussion on defect luminescence and x-ray fluores- cence studies on multicrystalline silicon wafer solar cells
Multicrystalline silicon wafer solar cells were observed to yield luminescence associated with defects. Defect topographies of both reverse-bias electroluminescence and sub- bandgap electroluminescence were distinct from each other, while their combination correlates strongly with regions of defects found using electroluminescence imaging. The concentrations of elemental constituents found using SRIXE showed regions emitting exclusively reverse-bias electroluminescence or sub-bandgap electroluminescence may be distinguished by the relative presence or absence of metallic impurities, respectively.
The analysis indicates that generally, regions of reverse-bias electroluminescence corre- spond with the inclusion of metals, particularly Co and Fe in this case, while regions of sub-bandgap electroluminescence had chemical constituents closer to concentrations measured on defect free silicon (float-zone silicon). Reverse-bias electroluminescence may thus indicate the presence of metals, presumably near the pn junction of the pho- tovoltaic device, while sub-bandgap electroluminescence may indicate a precipitate-free extended defect in the absence of reverse-bias electroluminescence, at low reverse bias.
This suggests precipitate-free dislocations do not create strong shunts. Presumably, dislocations in silicon do not allow strong reverse currents when their localized D-line
energy levels [125, 468] are introduced in the pn junction. Perhaps the broad spread of energies from metallic constituents (e.g. Co valence configuration 3d74s2) is the most detrimental in allowing reverse current flow across the pn junction of a multicrystalline silicon wafer solar cell, and thus causes shunting of multicrystalline silicon photovoltaic devices, as a precipitate, substitution, or interstitial impurity atoms in the substrate or device.
5 Luminescence spectroscopy for characterization of sili- con wafer solar cells
In this Chapter, the application of spatially-resolved luminescence spectroscopy is stud- ied for characterization of silicon wafer solar cells. An imaging spectrometer instrument is explored for two kinds of characterization. First, textured solar cells are measured so that the pathlength enhancement of the device can be evaluated over the entire device.
The proposal is that the optical pathlength of the luminescence in the solar cell will increase with the texturing, and that this effect may be detected due to the increased reabsorption of luminescence [469]. Secondly, the electrical properties of the solar cell, specifically the minority carrier diffusion length in the absorber [470], are inferred from the measurements of the luminescence spectrum. More information may be found in the literature [3, 4, 469].