Determining the optical properties of solar cells using low cost scanners

www.nature.com/scientificreports OPEN Determining the optical properties of solar cells using low cost scanners Mattias Klaus Juhl1*, Binesh Puthen Veettil1, Giuseppe Scardera2 & David Neil Roger Payne1 This paper investigates the use of consumer flatbed scanners for the use of monitoring solar cell precursors. Two types of scanners are investigated a contact image scanner and scanners with more conventional optical setups. The contact image sensor is found to be more suitable as it does not require additional flat field calibration. The scanners’ ability to monitor variation in sample texture was investigated by monitoring the reflection of multi-crystalline and mono-crystalline textured wafers. For a baseline, a comparison was made to a high-end tool used in industry. Both good qualitative agreement and statistical correlation were achieved between the scanner and industry tool for the isotropic multi-crystalline wafers. The consumer electronics market has resulted in technologically complex equipment designed for specific everyday uses being available off the shelf at a very low-cost (<$100 USD). If such items can be leveraged for unintended niche applications, they can substantially lower equipment costs, lead times, and open up access for more people, e.g. researchers on limited budgets. This paper investigates the conventional office-use flatbed scanners for optical characterisation of solar cells. Flatbed office scanners are widely available, have typical costs in the hundreds of dollars, provide high spatial resolution (40 µm pixel resolution on low cost scanners), and include a built-in white light source enabling measurements across three different colours channels. Office Scanners have found similar unintended uses for monitoring radiation1–4, a range of biomedical imaging applications5 and for monitoring components’ physical d imensions6. The same types of scanners looked at in this paper have also previously been analysed for use in holographic m icroscopy7. High efficiency, low cost solar cells’ performance depends on their electronic and optical properties. As solar cells are large area devices (greater than 220 cm 2 wafers), it can be challenging to maintain uniform optical properties. To monitor these properties, specifically the reflection, solar cells are often measured with stand alone point by point mapping tools, such as tools from S emilab8 or the LOANA from pv-tools as used in this paper. Pointwise spectral mapping of solar cells can take more than 20 min per sample for millimetre resolution. With production speeds of over one sample per second per line, this is much too slow to monitor every sample in manufacturing. Such tools also cost several hundred thousand dollars, and are out of reach for many research groups. The use of flatbed scanners may address both of these disadvantages. In this paper, we demonstrate the optical characterisation of solar cells using a consumer-grade flatbed scanner. We first compare the two common types of flatbed scanner technology widely available, showing significantly different results. Based on the analysis of these results a preferred type of scanner is selected and is then experimentally investigated as a means to monitor the variation of texturing across a silicon solar cell wafer substrate. We compare the results from the scanner, obtained in less than 10 seconds to that of a high quality laboratory based tool that takes 20 min. Results First, measurements on a textured silicon wafer with a contact image sensor (CIS) and non-CIS based flatbed scanner are presented. Significant differences between them are observed, with artifacts present in the non-CIS scanner data. Following on from this, the CIS scanner is used as it does not suffer from this effect. The CIS scanner is then used to examine a variety of textured mono-crystalline and multi-crystalline wafers. Scans are measured and compared to reflection maps obtained on a high end laboratory tool. Comparison of scanner types. To compare the CIS and non-CIS scanners, scans were taken of the same wafers with both scanners. The first wafer imaged was a p-type random pyramid textured mono-crystalline 1 Macquarie University, Sydney, NSW 2113, Australia. 2University of New South Wales, Kensington, NSW 2052, Australia. *email: Scientific Reports | (2022) 12:17697 | https://doi.org/10.1038/s41598-022-21229-w 1 Vol.:(0123456789) www.nature.com/scientificreports/ Figure 1.  Comparison of the signal from the blue channel averaged across the wafer in the scanners direction of movement. Each dot represents the average signal measured by a different pixel for (a) a CIS and (b) non-CIS scanner. Figure 2.  Repeat measurement of the same sample demonstrating the increase in the measured signal at regions away from the center of the scanner head for a CCD based Epson V800 scanner. Note the Epson V800 has a larger imaging axis compared to the Canon Lide 300, as was used to produce Fig. 6. wafer. There was a clear difference between the scans across the axis of imaging. To highlight this, line scans representing an average over the wafer along the axis of motion are shown in Fig. 1. The axis of motion is described in the “Methods” section and labelled in Fig. 6. The CIS sensor measured a relatively constant signal across the sample while the non-CIS scanner measured a large bow and a small wavy feature across the sample. These features remained on this imaging axis if the sample was rotated, suggesting that the bow and ripple are likely the results of differences between the scanners’ imaging optics. To further investigate this effect, additional measurements were performed on a textured p-type monocrystalline wafer coated with SiNx, a material that is used as an anti reflection coating for solar cells. Three measurements were taken of this wafer with the non-CIS scanner (Epson V800). After each measurement, the wafer was translated to the right across the scanner’s imaging axis. The same line scans as in Fig. 1 were extracted from the green channel and are shown in Fig. 2a. A difference in the measured bow occurs between the three measurement locations. The bow is more easily observed when re-plotting the data as a function of distance from the edge of the wafer, as is shown in Fig. 2b. For parts of the wafer close to the centre of the scanner the counts are lower, and these counts increase as the wafer moves to the edge of the scanner. This is most clearly seen in the measured intensity of the left side of the wafer decreasing as the wafer is moved away from the left edge of the scanner. Equivalent effects are observed on the right side of the wafer when it is located further away from the right side of the scanner. Also shown in Fig. 2b is data averaged along the imaging axis rather than the axis of motion and is labelled as Epson-vertical, and the same wafer imaged with the CIS based Canon LiDe 300 is labelled as CIS-middle. The Epson-vertical data was (...truncated)